Here's a circuit of the next controller, i have implemented Poynt's TVS protection and also the pulses width limiting circuit.
The TVS are 500Watt rated.
Let's hope the circuit functionality isn't affected too much Peter and your effect is still there ;)
Is the TVS on the gate a bi-directional type?
.99
Hi Darren
Yes the gate is Bidirectional and the drain are unidirectional, sorry couldn't find a symbol for them in my cad package so really i guess i should of created one ::)
I think the TVS should be good, not yet sure how the pulse width limiter will work but i can always cut them out if they are causing me problems.
Bit's are on order right now, and i will etch the board over the next couple of days.
Are these TVS's selenium or silicon? One was faster than the other.
My phone sucks or I would look it up.
QuoteFast Response time: Typicaly less than 1ps from 0 volts to VBR for unidirectional and 5nS for bi directional
Peter,
Just in case, I want to mention that the leads for the TVS diodes should be clipped as short as possible, and soldered directly to the MOSFET leads if possible. You want to eliminate as much stray inductance as is practical.
.99
OK Darren
Cheers just etched the board, not so easy with a big heat sink, i guess i could tag them on underneath directly across the terminals, i have used really thick tracks though and am using the same bigger FET's i used for Ds board instead of irf840 this time around.
I will go with tagging on bottom then thx.
Just my experience folks but the TVS should have minimum conductor length between the TVS and the source of the spike. You want more inductance after the TVS than before it.
Always place the protection nearest the source of the damaging energy.
Not quiet ready for testing but almost there.
Controller 5
Looks good!
Thanks Ds
Initial tests show it works well, i have 2 scope shoots taken across each FET gate/source
During this initial test i have nothing connected to the drain's
gate5_1 is taken @ 1Mhz
gate5_2 is taken @ 100Hz
The only worry is over 1Mhz the driver chip starts getting hot and the gate drive is about 8 Volts instead of 18 Volts.
Peter,
You could use liquid cooling. Its easier than most people think.
Of course this would change the length of the leads.
8)
Thanks for the advice Ken 'a bucket full of oil' ;D
These new fet's do not seem that good, i have just powered them up and i cannot even see a switch off at my maximum setting of 255nS and on the scope it looks like when i set for a 32nS pulse the pulse is over 400nS DOH so it looks like it's back to irf840A's
IXFH12N100Q
http://www.farnell.com/datasheets/42399.pdf
Here's 2 pictures 1 with 0 phase delay and the other with 255nS delay.
OK swapped to IRF840A's now and that's much better
3 Picture's
1 Across Gate/Source no load to Drain
2 Across load coil with 0 delay phase
3 Across load coil with 255 delay phase - Notice the pulse ;)
That's good news Peter...you're back!
Hopefully the rig is a little more robust as well. Have you pushed it yet?
.99
Hi Darren
Just pushed it up to 70volts and i am not getting my usual crashing spots, not sure what to make of it yet.
First thing i will say is that i need a delay near 250nS to get the pulse to appear so it would seem that the gate drive has slowed down something.
OK i have run with 90Volts for the coil supply and am not getting any noise that i have been getting, the coil sings at the fundamental frequency my Sig Gen is set to with no White Noise as i get with re-triggering, which is because of the lack of Noise being injected.
2 things maybe happening ,either the Gate drive has slowed the fet switching down too much or the EM bursts are a result of very fast pulses way above the TVS rating and can only occur without the TVS.
I guess first i need to remove the Gate drive cap and resistor and see if i can get the Noise bursts back.
The other problem i have is the IRF fets have metal tab's and these dam heat-sinks require a screw so i have no way of insulation without getting some screw isolators, i already have the flat heat washer.
Sometimes with ruggedness built in, performance can suffer.
You may want to try increasing the input cap value, or bypassing it to see how detrimental (if at all) it is to the gate drive. Keep the PW low just in case. You may also want to remove the 20 Ohm resistor temporarily as well if possible. Perhaps with some luck you can eliminate both, and just leave the TVS diodes in place.
.99
Yep no problem Darren
I am going to swap over the heat-sinks from a previous board that have clips so i do not have the Drain connected to the heat-sinks.
First i will try without gate cap and resistor and see what happens.
OK swapped my heat-sinks over and so now my FET's are isolated from the heat-sink metal.
Next up was to test the board without the gate drive resistor and capacitor
See scope shot across Gate/Source with no load below.
Next i started winding the volts to the coil supply up, 70volts no problem, 80 no problem, 90 was running well until the pulse showed up and then my psu shut down due to over current, the Non delayed fet had gone short, the TVS diodes read ok.
There was no sign of the pic crashes and re-triggering i had before to cause the FET to turn on more than it should, so what do you make of that, these TVS diodes should react in a pico seconds or so :o
Interestingly i cant see much difference between the gate drive with or without the protection circuit in place, a bit higher in volts without, but interesting that it didn't blow with the protection.
It is now looking like the TVS that is stopping the crashing of my pic by stopping something, off course to prove this i need to take the TVS off and see if the pic crashing comes back.
Now i have taken the TVS off the Drain and run some tests, i am still unable to get the EM noise that crashes my pic, i went up to 80 volts, i dare not go any higher because there is only one likely hood when the noise starts.
The only thing i have not tried is taking the TVS diode off the gate.
I dont know why some controllers are better than others for causing this noise, the difference with this one is i did not use any SM components, my gate track is 3 times longer than usual because of the larger heat sinks i used and the places for the cap and resistor in the gate control circuit.
It maybe time to switch back to controller 4 and try the TVS diodes on that controller.
It seemed that the pulse across the coil was shorter without the TVS diodes but i didn't grab a shot to prove it.
I am not sure why the TVS diodes didn't protect the fet in the previous test above, someone mentioned not to run over half the fet's rated voltage to be safe
Insert a current limiting resistor between the PSU and the coil. Scope across that and give it the juice. If the voltage drop across the resistor doesn't show an actual over-current condition things might become interesting.
When the FET blew, was it possibly because of an extended time in an over-current condition?
If so, the TVS diodes can't (and aren't meant to) protect against that.
.99
Or against the intrinsic diode being slower than the FET junction ;)
With the TVS diode there, that shouldn't be a problem, unless I'm missing something.
.99
I'm thinking about the diode's reverse recovery part of the cycle. On very short pulse widths there may be a shoot-through, or similar, problem.
Darren
What would you call an extended period of time stuck in the ON position, it was working fine and then BOOM ;D
What i have always said is that the FET's fail cold.
I use an analogue Meter to monitor the current through the coil, and it always reads zero while running normally, the digital psu reads 1 or 2 mA under 10kHz depending on the fundamental frequency being driven at, if i go higher the more on time the higher the current.
My analogue meter has 3 taps 50mA, 500mA and 5 amp ranges, like a fool in the last test i did, i pulled the plug from the ammeter 5 amp range and plugged it into the 500mA range when it was running the result was KABOOM, it knocked the non delayed fet out again DOH Note i did not have the TVS across the Drain/Source when it blew for this test.
I thought this was quiet interesting, when i was a tv engineer, there are common failure causes, things we see all the time, with CRT based tv's the common ones are as follows
1) Dry Joint's
2) Overvolting Psu's due to leaking cap's
3) arcing mains plug pins due to double adapter use or extension lead use.
3 is interesting, here in the UK we use 3 pin mains plugs of quiet large size, double adapters and extension leads are quiet poorly made, they rely on each pin of the plug to push into a u shaped brass socket and when old or over used the pins of the plug get black arc spots on, this more often than not destroys the high voltage stage transistor eventually, any noise on the mains despite filtering causes big problems once the voltage gets stepped up to 32kv, i never did fully understand why because there is filtering for the mains then the psu has filtering and regulation and more filtering and then the high voltage circuits have protection as well, yet the poor old line transistor would always go short.
It became a matter of course to check for adapters and extension leads if the line transistor was short and the LOPTX was ringing ok and dry joints were not visible or could not be found.
Anyway going back to my last FET failure above, when i disconnected the Ammeter what possible conditions could i have had on reconnection to cause failure.
Well if the fet was on for 32nS 1000 times a second then current would have been drawn and i guess an arc could have formed on the plug just as i pushed it into the ammeter so why would this cause the FET to blow.?
Also my psu is only a 1 amp supply, if i was to turn the fet on forever i am still only driving it with 1 amp max which this fet should be able to handle, so apart from getting hot it should survive, although no doubt my psu has a reservoir cap which i guess for a short period of time could deliver much more current than 1 amp.
guys also the interesting thing was the FET blew as soon as the big sharp pulse appeared, not sure what it is but there is something magical about this pulse, it's either way higher voltage than my equipment shows and probably much sharper than we see also.
Also D's idea about resistor in series with the coil is good. we used this technique to fault find line stages in tv's, we use a 240v AC bulb in series with the primary coil, this loads the psu and allows the line stage to power up for fault finding
WW if the reverse integral diode was burning, surely i would not be reading shorts across my gate as well as drain/source although i guess it depends if the diode is a separate part of the substrate.
The intrinsic diode is not separate.
The problem with it is it may not conduct as soon as the fet turns off. On short pulses the diode may not turn off before the fet turns on.
Plugging the meter in is the same as connecting to the computer? KABOOM-wise?
On your 1A PSU.... on very short pulses the current output maximum can be in the kA.
If the big pulse is what I've said it was all all along, it is perpendicular to the first and delayed pulses. Replace the coil core with a short piece of wire and scope across that. Disregard the fact the short piece of wire appears to be a 'short circuit'. If your scope shows a high voltage pulse then replace the short piece of wire with a closed wire loop of a single turn. The fets should stop blowing.
Also something i find interesting is that with the short pulse into the coil see picture below, the neo magnet wants to pull into the alignment shown in the picture if placed 90 degrees from it's pictured position, now i can understand this if i had a dc voltage across the coil, and i do know i am only pulsing it with dc, but such a short pulse pulls the magnet round slightly(Quiet a weak pull) as if there is no collapse of the field or reversal of field to null the pull on the magnet, also interesting is i can hear the fundamental frequency it's being driven at, it starts off quiet and gradually increase in level as the phase increases, which again makes sense as the phase progresses we end up with 2 times the on period as the pulses separate, but the appearance of the sharp pulse doesn't really seem to alter the pitch that much, which to me indicates the driven energy causing the noise does not seem to increase dramatically when the pulse is present.
Increasing the 'effective on period' without increasing the energy required to do it would be OU.
Now that's an interesting thought there WW, with a fundamental frequency at 5khz i can hardly measure any current so it would be hard to tell if the current goes down with the pulse, but maybe worth me trying a few things to see whats going on with the current, maybe try using my uA dig meter.
Here's a video of phase delay, showing scope pulse and current.
I am using Controller 5 with TVS on drain/source but with no cap and resistor on gate.
The idea behind the test is to show current consumption during the pulse event, ok i know i shouldnt use a DVM for showing pulsed current but it's all relative ;D
Scope shows pulse
DVM shows dc current uA scale
Pic shows phase delay nS
Now this one is interesting.
This time i have plugged a mike into the laptop and adjusted the level so i dont get feedback and so the camera can record it.
The fundamental frequency is near 5kHz audible range.
Now listen to this ;D Look at the way the noise increases with the pulse present, now using my ear with the magnet near to the coil i can hear the 5kHz gradually increase but nothing like the microphone is picking up especially the way it increases with the pulse, my ear does not hear this sharp increase only a gradual increase from 0 delay to 255 delay, so therefore the mike is picking up the fundamental frequency 5kHz by some other means other than sound waves.
OK The noise only seems present when i have the scope connected as well as the mike, so there must be a sort of feedback involved between the scope connected to the usb port and the strange sound picked up by the mike.
Quote from: Peterae on 2010.06.02, 10:30:07
Also something i find interesting is that with the short pulse into the coil see picture below, the neo magnet wants to pull into the alignment shown in the picture if placed 90 degrees from it's pictured position, now i can understand this if i had a dc voltage across the coil, and i do know i am only pulsing it with dc, but such a short pulse pulls the magnet round slightly(Quiet a weak pull) as if there is no collapse of the field or reversal of field to null the pull on the magnet, also interesting is i can hear the fundamental frequency it's being driven at, it starts off quiet and gradually increase in level as the phase increases, which again makes sense as the phase progresses we end up with 2 times the on period as the pulses separate, but the appearance of the sharp pulse doesn't really seem to alter the pitch that much, which to me indicates the driven energy causing the noise does not seem to increase dramatically when the pulse is present.
You can see it better with a compass. On the most basic level of explanations, there is a force perpendicular to the wire.
Also, if you use 10kv and break the wire so that you can place the compass between the ends, the compass will align to be parallel to the wire rather than perpindicular to it.
Now, does the compass or magnet align itself because it is magnetic or because it is conductive?
I have now started increasing the fundamental drive frequency and as before with controller 4 above 350-400kHz i do start getting the crashing at 50 volts, now this is good because now i maybe able to start the crackling if all goes to plan.
As before over 350-400kHz the first pulse sharpens dramatically to the size of the large pulse as the frequency increases.
I believe the sharpening of the pulses due to frequency drive increase is related to the crashing problem which leads to the explosions occurring.It's all down to getting sharp pulses and then moving them in nS delay steps to phase them.
Quote from: Peterae on 2010.06.02, 10:46:50
Now that's an interesting thought there WW, with a fundamental frequency at 5khz i can hardly measure any current so it would be hard to tell if the current goes down with the pulse, but maybe worth me trying a few things to see whats going on with the current, maybe try using my uA dig meter.
Keep in the back of your mind that "total current" includes
all forms, not just "conduction current". So, when you see OU based on measurement of "conduction current", the mystery goes away when you add in the other forms of current. Moving space is the same as moving a charge and is therefore a current and can appear as both a conduction current and a displacement current.
Quote from: Peterae on 2010.06.02, 11:51:35
Now listen to this ;D Look at the way the noise increases with the pulse present, now using my ear with the magnet near to the coil i can hear the 5kHz gradually increase but nothing like the microphone is picking up especially the way it increases with the pulse, my ear does not hear this sharp increase only a gradual increase from 0 delay to 255 delay, so therefore the mike is picking up the fundamental frequency 5kHz by some other means other than sound waves.
'
is it a capacitive mic?
It's an electret microphone, but as stated after that post the noise seems to be related to the scope being powered up in the laptop usb port, i am currently not sure why this is, it maybe because with the scope powered down but scope probes still connected puts some sort of different load across the coil, i know when i power up the scope i hear relays click and also when different ranges are selected also.
OR
it maybe just some sort of weird feedback.
Quote from: Peterae on 2010.06.02, 13:58:35
It's an electret microphone, but as stated after that post the noise seems to be related to the scope being powered up in the laptop usb port, i am currently not sure why this is, it maybe because with the scope powered down but scope probes still connected puts some sort of different load across the coil, i know when i power up the scope i hear relays click and also when different ranges are selected also.
OR
it maybe just some sort of weird feedback.
what if you change the switch on the scope probe to 10x?
Quote from: Grumpy on 2010.06.02, 13:42:52
You can see it better with a compass. On the most basic level of explanations, there is a force perpendicular to the wire.
Also, if you use 10kv and break the wire so that you can place the compass between the ends, the compass will align to be parallel to the wire rather than perpindicular to it.
Now, does the compass or magnet align itself because it is magnetic or because it is conductive?
I found this repeatable on my bench when you shared it a couple of years ago. I also found it true that the needle doesn't have to be a compass. You can use a spinning needle (conductor, magnetized or not) to map out charge fields between two points.
Peterae, what is the polarity of that magnet in the photo and which end of the coil is going to positive PSU?
Right hand side of that picture with the coil goes to positive psu.
but i dont know which pole on the magnet is north, and i dont have a compass to hand, maybe another day.
These pictures show how the first pulse sharpens up with different drive frequency's.
First one is a snap @ 100kHz
2nd one is a snap @400kHz
and the interesting thing is my lap top is doing weird things without any phase delay and just the sharp single pulse on above 400kHz, and when i carried on winding up i heard the coil start crackling until i lost my bottle and switched off LOL.
does the noice change when the scope probe switch is changed from 1x to 10x?
Here's a weird effect.
at 650kHz drive frequency the magnet starts banging against the coil, this was with the scope connected and powered by the laptop.
When i disconnected the laptop from the scope the noise level dropped and the magnet stopped banging against the coil but i could still hear the banging noise just much quiet, i then disconnected the scope probe from the scope and powered up and still had the quiet banging noise the same, but when i disconnected the unterminated scope probe from the coil end the noise stopped totally.
Interestingly with the scope connected to the laptop(Running on battery only) as in the video when i touched the heat sinks on the fets it stopped totally and the heat sink is isolated from the fet drain DOH very strange.
Quote from: Peterae on 2010.06.02, 14:42:38
Interestingly with the scope connected to the laptop(Running on battery only) as in the video when i touched the heat sinks on the fets it stopped totally and the heat sink is isolated from the fet drain DOH very strange.
Not at all ;D
Your body soaked up much of any charge collected by the heat sinks. The new discharge path through your body is providing a virtual ground. Your heat sinks are floating?
edit:
Just remembered... most folks tie one side of power bus to a chassis ground. I avoid that, personally, and connect heat sinks, transformer cores, etc. to chassis only.
Am I correct in stating -- The frequency of the 'clicks by magnet application' does not change with a change of magnet proximity to the coil? The only time this frequency changes is when you change the polar orientation between the magnet and coil?
QuoteYour body soaked up much of any charge collected by the heat sinks. The new discharge path through your body is providing a virtual ground. Your heat sinks are floating?
My heatsinks are floating not connected to anything.
I have no grounding that i know of, but all my 0v lines are tied together, but the scope earth lead which is on one end of the coil, is connected to the scope which i guess is also connected to the usb metal or earth which is also connected to my laptop, so my laptop is in effect connected to one side of the coil ::)
QuoteAm I correct in stating -- The frequency of the 'clicks by magnet application' does not change with a change of magnet proximity to the coil? The only time this frequency changes is when you change the polar orientation between the magnet and coil?
No the magnet has no effect on the clicks frequency what so ever it just makes them louder to my ear.
It's worth noting that in previous tests when i first did the Explosions in the wire experiment, i used a bench scope, i didn't have my usb scope at that time, and yet the microphone on the laptop still picked up the weird noises from a foot away, it have opened a thread to discuss this, in many of the videos i have made with all sorts of configs , 2 scopes, totally different controllers and psu's powering them there is always this weird sweeping noise appears every now and then as i do a phase sweep, in some cases the laptop is not connected in anyway yet the mike still picks up the weird noises from 1 foot away, yet my ear hears nothing.
http://www.overunityresearch.com/index.php?topic=279.0
In the dancing magnet test above an un connected scope lead was needed to create the noises, thats a scope lead connected across the coil but with the bnc end not connected to the scope, when i removed the clips from the coil the faint clicking/vibrating stopped so i am guessing the scope lead was acting as an unterminated delay line.
touch the floating heatsink and you will feel a tingling or prickly sensation.
The heatsink is a collector and will retransmit to anything of different potential - including you - like a continuous spectrum reciever.
You should also be able to pull a tiny arc off the heatsink with an insulated metal object such as a screwdriver.
Quote from: Peterae on 2010.06.02, 17:50:48
My heatsinks are floating not connected to anything.
I have no grounding that i know of, but all my 0v lines are tied together, but the scope earth lead which is on one end of the coil, is connected to the scope which i guess is also connected to the usb metal or earth which is also connected to my laptop, so my laptop is in effect connected to one side of the coil ::)
Interesting. I would wonder if making an electrical connection between the sinks and USB shield would do the same as when you touched the sink.
Quote
No the magnet has no effect on the clicks frequency what so ever it just makes them louder to my ear.
While watching you move the magnet I detected a noticeable change in audio frequency when you twisted the magnet around the horizontal plane.
I would expect the frequency to raise when you twist the magnet right or left but only if the magnet poles were parallel with the coil poles.
This relates to how I make an air core coil resonate at lower frequencies.
i cant say i noticed a frequency change but the loudness varied enourmously, maybe i wasnt looking for it
Listen to your video off u tube. Maybe it is another sound not heard at the bench.
I had similar not heard but seen on a spectrum analyser with magnetic pickup.
Peter,
I think your "exposions" are "ion acoustic waves" in the audible range.
If so, they will strongly scatter radio waves of certain wavelengths and produce Langmuir waves (longitudinal electric waves).
http://rrsl.ee.washington.edu/ionosphere.htm
(right click and save to see this one)
http://mr-fusion.hellblazer.com/pdfs/ion-accoustic-waves-and-langmuir-waves.pdf
Basically, this is the "force" and you have to combine it with a conductor (to carry a current) and a magnetic field to convert it - aothough it should impart charge to conductors. (Same as the thing everyone calls RE).
Looks like i can now produce the crackles using 2 white noise channels, i can adjust the occurrence of them also ;D
Here's the video of the spectrum, what do you make of that 11mHz peak that pops up now and then?
Got Crackles. What's next?
Not sure yet G, this is the first time i have produced them using a bifilar coil, normally needs a monofilar.
I am not yet fully sure that these crackles are the same, the occurrence of these seem to be coil voltage dependent, the more volts the more you get.
I guess i need to try capturing their energy next to get a handle on their power, the coil only seems to draw 4mA @ 40volts when i have them occurring in the below video.
Just tried driving as a monofilar with both fets driving 1 coil, the crackles were still there, probably more pronounced, but as i tried sweeping the phase the pic started crashing and the delayed fet died.
Quote from: Peterae on 2010.06.08, 14:20:53
Not sure yet G, this is the first time i have produced them using a bifilar coil, normally needs a monofilar.
I am not yet fully sure that these crackles are the same, the occurrence of these seem to be coil voltage dependent, the more volts the more you get.
I guess i need to try capturing their energy next to get a handle on their power, the coil only seems to draw 4mA @ 40volts when i have them occurring in the below video.
capture with a cap as it is a force like "pressure" - try a ceramic plate cap - think of it like using it to induce current rather than fill something with magical energy
You have "A" and you need "B" and "C" to convert it
Try trifilar and use the middle conductor as the output.
I wish i had some small compasses to work out which direction the tug is in, if i hold the magnet close to the face of the coil i can feel the BOOM tugging at it.
A look on ebay for them i think.
I can try a ceramic cap indeed, good idea.
I dont need a trfilar now i can get it using just 1 coil and try the other as the collector although i have a feeling the collector needs to be 90 degrees from the wire direction.
I am still trying to get my head around whats going on here, the idea behind the whitenoise was to create the crackle, but infact it creates the noise that makes the pic counter unstable which in turn creates the crackle.
The whitenoise has helped as i now only need 40volts to make them appear but i do need the pic counter connected, it's the ribbon cable that connects the pic to the digital delay, it must be picking up noise and injecting it into the parallel port of the delay chip, because if i drive the 2 digital mono's with dip switches and no pic i dont get the crackles, and with the pic and ribbon cable if i touch the ribbon cable the crackles appear much more pronounced and occur much more often.
It's interesting that with 40volts on my coil and when i tried a sweep the fet dies straight away, this is with the TVS diodes in place but without the pulse width limiter circuit of Poynts in place.
At least now i am in a position where i have a setup and can easily produce the effect i wanted to produce and study.
The event that occurs when the fet dies is still unknown, but i feel it's got energy ;D
Quote from: WaveWatcher on 2010.06.08, 16:46:30
Try trifilar and use the middle conductor as the output.
That may work, but not sure how it will affect the interaction of the bifilar pair if it is directly between them.
Here is the deal: everything has to have the correct orientation to work together
1. the force is radial (this is the kick or explosion force) and must be perpendicular to a static magnetic field AND to the collector
2. rotating the force in a circle has a cummulative effect as other parameters come into play
(deleted)
EDIT:
Correction: with the single core wire for static dc field the field is circular so the collector is in the core position - like the early units were presumed to be when SM didn't know about the static DC field. Hence when you loop it on itself to supply the DC it increases the static field and you more DC = more field = boom
the outer collector is with a solenoid shape inner static DC coil
with the bifilar or other toroidal coil supplying the force field (kick) the force spirals along the coil windings as it propagates along the coil.
1. make a long bifilar coil with a core wire
2. bias the core wire with a little DC to get things started
3. pulse the bifilar with the split supply and delay to get the kicks
4. DC on the collector/bias coil will rise and burn up if tuning is too good - load is across this biased coil
and there is your TPU, but my recommendation is to separate the fields and sources to get better control and not feed the beast it's tail
Something just occurred to me.
The false triggering of my digital mono, makes the phase shift random, now when i do a sweep the fet blows, so what is happening with a random phase shift is that this spot that causes the fet to die is not held longenough when random and that explains the explosion noise.
So randomly the correct phase is hit an explosion occurs but before the level builds to blow the fet, it's already moved on to the next random phase delay setting NOW THIS EXPLAINS EVERYTHING.After all whats the difference between randomly selecting different phase delays or doing a sweep, they both cover all possibility's so should be the same.the only difference is the held time on each nS phase delay .
It doesn't matter if i tune to the exact setting as long as i don't hold that position, i have in a way controlled the energy released by controlling the time period.
The kick is slow to form, thats why the TPU winds up, it maybe i am not yet seeing much free energy because i am terminating the kick to soon, this also explains why i have seen different loudness from the explosion, it's because each controller is different and picks up this retriggering differently and therefore the randomness is at different speeds, the slower the randomness the louder the bang, GOD this all makes sense now.
This also explains another weird thing thats happened sometimes, on my pic controller i have the ability to turn the sweep onto manual increment, this means that each time i press the button i go up 1nS step, well it starts at 0 phase delay, but everytime i tried it i reach about 18nS delay and them suddenly the display reads 30-40 as if i had pressed the button really fast to step past those delay settings, but in truth whats been happening is the port line which is pulled high is getting pulses of energy which steps it past as soon as i get near the delay that causes the em field to expand, this shows that the field is alternating or sending bursts of energy out at high frequency.
tune slightly off the sweet spot - deliberately
measurable movement of the medium is slow to form, the kick is there and gone in a few ns
it takes time to wind up because you are only accelerating it during the rise of the pulse
it's just a force at this point, not usable current - it isn't a battery - you apply the force to move the medium in combination with a static magnetic field to convert the moving medium into current in a conductor/insulator
come on bro!, coil it up and don;t get hung up on the retriggering and noise
OK let's cut the BS and get to the fun stuff:
form a bifilar coil into a ring with a core wire inside - plenty of distance between the core and bifilar and keep this distance the same so use some sort of even spacer so the field is not varying all over the place - make it large enough to be easy to work with - say 4 to 6 inches diameter
place a compass inside the ring and pulse the bifilar - adjust the phase shift to get rotation of the compass CCW - do not tune for clicks, noise, or other stuff like that - just a little rotation don't try to burn up the compass
put some DC on the core coil - say 24v and and a few ma via a battery and see if the battery voltage goes up- pull off extra power for lights and stuff - pull too much and the battery gets pulled down, field goes down and power does down (but it is still spinning like a top - see?)
Now, tune closer to "oh crap" and the battery get's more juice for the load...do not loop the damn thing back because you will feed this bias field which feeds the converted current which feeds the field - see? If you try to regulate the power to the bias coil - keep in mind that cold current is there and you are not familiar with this.
Quote from: Grumpy on 2010.06.08, 18:07:23
That may work, but not sure how it will affect the interaction of the bifilar pair if it is directly between them.
Conductors 1, 2 and 3. Current on 1 and 3 going the same direction. The signal on 3 is a delayed copy of 1. Resistive load and scope probe on 2.
The idea is 3 is a delayed copy of 1. This creates angular velocity between them (very much like circular polarization of a radio signal coming from the same signal sent out dipole perpendicular to one another.)
Any time you send a rotating signal round and along the length of a conductor, with a electric bias, or a length of dielectric, holding a static charge, with the correct pitch, a current will be generated (not classically induced) through that path. This is how I believe large currents can be generated in a short piece of wire being used as a coil core.
Basically, you are building the reciprocal of a current carrying wire.
A little primer on truly 'active' antenna systems (denied by most, even the ones incorporating it):
We all look at the wave on a scope and consider the energy between the zero line and the trace (DC pulses, other types not crossing zero, just to keep it simple).
What about the energy being displaced by your scope signal? Look at it in the inverse, like a photo-negative.
With the right pulse width and delay between pulses(of the same signal) you are creating a displacement frequency(don't know what else to call it).
The short story: That 11mHz burst was probably the energy of a distant transmitter. One you would not hear with a conventional receiver. Your delay isn't very steady, at these resolutions. The slightest variation in delay will cause a frequency drift across a radio band.
A 'DELAY' of 20ns should get you around 50mHz. 90ns, around 11mHz. 160ns, around 7mHz. You are increasing the Q of the coil and shifting that massive increase in Q up and down between HF and lower VHF.
The secret of truly active antennae is the injected, high current signal is NOT a copy of the received signal but the spatio-temporal inverse of it.
Your coil isn't likely generating these frequencies. You are sucking them in.
I was dead serious when I said you invented an energy sucking antenna. ;)
OOPS!
I forgot.... I'm already known as nutty as a fruit-cake :D
Quote from: WaveWatcher on 2010.06.08, 23:50:24
Conductors 1, 2 and 3. Current on 1 and 3 going the same direction. The signal on 3 is a delayed copy of 1. Resistive load and scope probe on 2.
I've always thought of the interaction of the delayed and base signal which produces a spike as a form of pulse compression.
The current on 1 and 3 can be almost negligible and the force will still manifest as it is an electric force produced by an electric field that is time-dependent and spatially varying.
Do as I have suggested and let's get on with this.
Sorry Grumps.
My INet went down when I was finished with that post. When the connection came back up, I sent it without reading the several new posts.
'Pulse compression' is a good way to put what most describe.
However, 'THE' electric field isn't time-dependent and spatially varying unless it is perturbed. The two pulses can never meet because they are Hertzian and traveling down the wire in separate helical paths.
This may be the difference between a so-called 'static' field (electric charge - static magnetic) and what we call 'usable' energy. There is no helical motion in the 'static', even if it is moving. So, we are stuck with changes in density to give us induction.
What should happen between these two pulses is stretching the electric field between them. Among other things this produces a magnetic field around the coil core similar the magnetic field around a single current carrying conductor.
At that point, if you wish more amps just use more strands in the core.
In other words, I'm agreeing with you .....because it works under other circumstances, should be even better with the works of Peterae.
Quote from: WaveWatcher on 2010.06.09, 01:48:53
Sorry Grumps.
My INet went down when I was finished with that post. When the connection came back up, I sent it without reading the several new posts.
'Pulse compression' is a good way to put what most describe.
However, 'THE' electric field isn't time-dependent and spatially varying unless it is perturbed. The two pulses can never meet because they are Hertzian and traveling down the wire in separate helical paths.
This may be the difference between a so-called 'static' field (electric charge - static magnetic) and what we call 'usable' energy. There is no helical motion in the 'static', even if it is moving. So, we are stuck with changes in density to give us induction.
What should happen between these two pulses is stretching the electric field between them. Among other things this produces a magnetic field around the coil core similar the magnetic field around a single current carrying conductor.
At that point, if you wish more amps just use more strands in the core.
In other words, I'm agreeing with you .....because it works under other circumstances, should be even better with the works of Peterae.
I hope spherics or SM is ready this.the changes in density require manipulation - they are not directly converted - and they only exist during the perturbation
Conversion requires a magnetic field and the choice of force fields is your own - then it is just a matter of application. Create an artificial gravitational field and you won't need pulses any more as you will have a continuous force - assuming it is strong enough to induce enough current.
See?, the universe is a big damn place with a lot going on and there is more than one way to OZ.
force rotated perpendicular to a magnetic field and both perpendicular to the collector - that't it!
Frankenstein is about to live, if Peter will raise the lightning rod.
Good work!!
yes load it or you may miss the good part - it need a path and a load
you are looking for mainly current at this point
Let's not expect a complete solution from only Peterae.
When I said 'it works under other circumstances', I was referring to generating DC in a conductive core. The best I've had is a couple of volts, never enough energy to close the loop.
I agree the next stage is to try and capture some energy and will start what G say's
OK so let's build a 6 inch coil.
Now G you are saying a central conductor with 24volts dc running through it at a few mA, how long is this conductor and how thin, that seems like a lot of voltage with little current to push through i single conductor, so we are talking many turns of this conductor?
You then say wind the bifilar around this equally distanced away from the conductor, i need to think how to do this it's not so easy to keep the distance equal within say 1mm, i need some sort of foam frame around the central conductor to do this.
The central conductor needs to 2 things it needs to be biased but also act as a load, so could i maybe place a bulb in series so this lights with the 24vdc applied and watch it's brightness as the kicks are started up.
You mention a compass, i have just ordered 5 of these.
Diagrams and sketches are welcome :) any ideas on construction of the outer bifilar are also welcome.
Hi,
Is your bifilar wound up as a pair of wires side by side (twisted) or have you constructed the
coils with 2 identical singles but physically spaced on the former to create an angular displacement? (the start for each coil is some degrees apart on the former)
Im assuming its a torus former. The origins of the 'original device' had coils spaced and this came to mind as a possible component for your success.
The interest here will go on an exponential increase if just one second of excess energy is created.
Be safe.
Steve.
Quote from: Peterae on 2010.06.09, 10:07:32
I agree the next stage is to try and capture some energy and will start what G say's
OK so let's build a 6 inch coil.
Now G you are saying a central conductor with 24volts dc running through it at a few mA, how long is this conductor and how thin, that seems like a lot of voltage with little current to push through i single conductor, so we are talking many turns of this conductor?
You then say wind the bifilar around this equally distanced away from the conductor, i need to think how to do this it's not so easy to keep the distance equal within say 1mm, i need some sort of foam frame around the central conductor to do this.
The central conductor needs to 2 things it needs to be biased but also act as a load, so could i maybe place a bulb in series so this lights with the 24vdc applied and watch it's brightness as the kicks are started up.
You mention a compass, i have just ordered 5 of these.
Diagrams and sketches are welcome :) any ideas on construction of the outer bifilar are also welcome.
One stout loop for the core. You must maintain a circular magnetic field around the core with this configuration. The static magnetic field must be perpendicular to the radial force from the kick and also perpendicular to the direction of propagation of the radial force (wether just propagating or rotating) and perpendicular to the conductor being induced. Lay all this on a paper and you will see that you have a couple of different ways you can go depending on how you place the static field and collector.
For a quick test, you can stick two wires into the coil - right into the side of it so the ends are inside the coil - one toward the top and one toward the bottom (stacked but separated), add a static mag field (with a magnet) - place it under the coil - and see what you can measure with a high impedance meter or probe. Don't expect anything spectacular, just an indication.
a moving divergence (charge), bound or not (dielectric or conductor), produces a magnetic field - hence the compass - this is a current as long as it is moving and has to be perpendicular to the collector and the static mag field to cause particle precession in the collector (conduction current)
you can try using LEDS placed around and into the coil space as well or small bulbs (never tried it - just thought of it)
A single loop for the inner core, then that sounds easier, if i make it first as a long straight wire for the core then use pipe over this and wind my bifilar over this pipe and then bend the whole thing into a circle to form a toroid then would that do, i am just thinking of construction techniques here and that would be the easiest, but is that config what you are saying as i'm having trouble understanding exactly what you are saying.
Yes.
OK G thanks that what i thought, i will work on doing exactly this.
I'll work on a diagram.
that will be great G
I'm working on the coil
OK I have 3 sizes of hose each fits inside the other this makes a nice thick walled hose with all 3 inside each other.
Realistically the inside diameter of the toroid will be about 20-21 cm diameter, the largest hose is 15mm outside diameter, the inside hose is 3mm inside diameter, which i can push an insulated 2.5mm copper wire into or is this too stout.
Quote from: Peterae on 2010.06.09, 17:44:48
OK I have 3 sizes of hose each fits inside the other this makes a nice thick walled hose with all 3 inside each other.
Realistically the inside diameter of the toroid will be about 20-21 cm diameter, the largest hose is 15mm outside diameter, the inside hose is 3mm inside diameter, which i can push an insulated 2.5mm copper wire into or is this too stout.
what kind of hose? you need something like a foam dielectric that is easily penetrated like in some coax cables
you can use plastic (or cardboard) sheet rings fixed to inside and outside of the core coil and then roll palstic strips for the top and bottom to make four "ribs" around the core coil
no problem G i will go to plan B Carboard formers.
Quote from: Peterae on 2010.06.09, 18:47:02
no problem G i will go to plan B Carboard formers.
ouch! I used to make little cardboard/paper space ships when I was in elementary school.
Funny how SM's cork-looking-stuff makes sense now. Still wondering how he came to his conclusions.
You could also cut styrofoam rings and fit the core coil into the center - foam ring could be square-ish with core coil in center surrounded by foam. I would not go with anything much denser than this. Might be easier than flat forms.
this is supposed to be correct
15" 2.7 turns collector attention to the collector outs being overlapped by the controls which are 3 and turn for turn on top of each other , all tinned multistrand wire
Quote from: Lindsay Mannix on 2010.06.10, 00:37:48
this is supposed to be correct
15" 2.7 turns collector attention to the collector outs being overlapped by the controls which are 3 and turn for turn on top of each other , all tinned multistrand wire
That makes sense.
Why is the collector in series rather than three in parallel?
G, they are not terminated in the picture.
I may make another collector section to get 3 but 1 should be enough to see something in the centre wire for the sake of Petare's great work .
That is IF the popping still occours with this config. Im optomistic about it .
Seperate dc supply to the collector just might allow a measurable current increase rather than a high voltage.
The black stuff is plastic lawn edging. Easy to use and form ,but im sure will be usless as far as heat goes .. walk first!
Thanks for the info Lindsay.
Not sure i can visualize exactly how you have wound the wires in the above pictures.
I am not sure that the wire length is so critical for the way i am producing the kick.
Over the last few days i have mentally been refining a much better design to achieve my controller, i am pretty much down to 1 8 pin pic and a couple of fet's , i also believe i can easily control the power level in code, i also think i can do away with the delay chips now as well, but need to spend some time on the dev.
Offcourse it's all down to energy capture to see if it's viable.
So first i need to try capturing the energy and hopefully it's going to be easy to provide feedback to the system for controll purposes.
Anyway onto coil design, i am going to make the first one as simple as possible, i will play today and see how it goes.
First Coil 163mm center wire diameter.
I used the inner wire and insulation from a TV coax lead stripped of the coax copper sheath, then wound a bifilar over the top.
First Kicks ;D
I am getting the crackling nicely now in the new coil, if i scope across the center wire i am getting 50-80 volts fast spikes in a random fashion this is unbiased and unloaded.
If i place a 20 Ohm resistor straight across the center wire the kicks seem to go -/+20volts
I make those kicks @ 80Watt spikes
The below video is of the center wire unloaded, the scope is 10v/div
Looks good Peter.
Is this with white noise drive to the toroid (outside) coil?
.99
I am using 2 white noise generators 1 for each fet stage.
I was going to take a video of it with the 20 Ohm load but it died after 3 or 4 minuets of running, looks like my non delayed fet has gone, i am only running at 33 volts coil supply.
yes all drive circuitry is outside on my controller pcb.
Sorry misread your last question, yes the yellow wire is a bifilar wound down the length of inner coax core.
tune for rotation - which may not occur with the white noise
make a compass with a needle floating on something in water - the water may also spin since it is polar
I think you need higher voltage but try what you have now
The Effect is greater if i short the 2 bifilar's together but it increases the FET's going short >:(
Tomorrow i will retry Poynt's cap and resistor in the gate drive circuit to see if it stops the dying fet's but keeps the effect happening.
I also tried the 120pf mica caps across the drain/source and as soon as i place even 1 on one of the 2 channels the effect stops.
Still waiting my compasses to arrive.
Quote from: Peterae on 2010.06.10, 15:46:01
The Effect is greater if i short the 2 bifilar's together but it increases the FET's going short >:(
Tomorrow i will retry Poynt's cap and resistor in the gate drive circuit to see if it stops the dying fet's but keeps the effect happening.
I also tried the 120pf mica caps across the drain/source and as soon as i place even 1 on one of the 2 channels the effect stops.
Still waiting my compasses to arrive.
While interesting, the explosive sounds are not a requirement and may actually be a loss of energy from the system.
You need a sharp electric impulse, wether you create by an interaction of fields or apply it directly. This creates a force that can cause the medium to move. You then apply this force the medium to rotate it in a circle to create a sort of current.
Build a pulser that works reliably to produce the highest pulse that you can, crackles or not.
Sorry. Can't see pics on this phone.
Is the output mostly one side of zero?
Is your estimated watts out considering the off times?
Since the cap kills it I suspect 'no' is the answer to both.
Welcome to perpendicular induction 8:) or 'squeezing the hose'.
Quote from: Peterae on 2010.06.10, 15:46:01
The Effect is greater if i short the 2 bifilar's together but it increases the FET's going short >:(
try using a coax delay line to turn the pulses off
Short the bifilars together where?
(Might want to think about something other than FET's.)
Quote from: WaveWatcher on 2010.06.10, 17:15:28
Sorry. Can't see pics on this phone.
Is the output mostly one side of zero?
Is your estimated watts out considering the off times?
Since the cap kills it I suspect 'no' is the answer to both.
Welcome to perpendicular induction 8:) or 'squeezing the hose'.
The output seems to swing both positive and negative in the same pulse.
For watts measurement i am using a 20 Ohm Non inductive resistor across the output coil, i am still getting large pulses and therefore when the kick occurs that is what i estimate the wattage of the kick, i need to look at it a bit closer yet, it's probably a bit under what i said as i have not done an exact measurement.
Quote from: Grumpy on 2010.06.10, 18:05:59
try using a coax delay line to turn the pulses off
Short the bifilars together where?
(Might want to think about something other than FET's.)
One end of the bifilars are joined together and connect to the coil supply
Each coil has a fast diode across it, each fet then drives each other end of the coil, if i short them together so the fets, diodes and bifilar are in parallel a get bigger kicks but my fets die faster.
I actually had a fet start working again for 3 or 4 seconds earlier after it went short, when i went back after power down i measured each fet and they read ok, so i powered up again and it work ok but then went off.
Grumpy i would love to have another type of output stage what can i try?
Quote from: Grumpy on 2010.06.10, 17:07:38
While interesting, the explosive sounds are not a requirement and may actually be a loss of energy from the system.
You need a sharp electric impulse, wether you create by an interaction of fields or apply it directly. This creates a force that can cause the medium to move. You then apply this force the medium to rotate it in a circle to create a sort of current.
Build a pulser that works reliably to produce the highest pulse that you can, crackles or not.
I believe the explosion sounds are created by me starting the process of kick generation, but the random pulse collisions terminate the process, it is this fast termination of the kick formation that creates the crackle, if i hang around too long while a kick forms my fets blow, but the longer i do hang around the more energy released.
The use of white noise has increased my chances of stumbling on the conditions to create the kick, but i still don't know the true and correct way to create them and because of this i cannot de tune my system to gain less power, i have to control by termination.
I have seen the kicks before by mixing fundamental, 2nd harmonic and 3rd harmonic in a ferrite core but it was weak crackling inside the core.
A suggestion:
Use a coax delay line for the delay. I would estimate 14 inches for every ns delay required. There are ways to shorten that.
Then use the two fets together to fire both wires of the coil. Just series the delay line between those two fets and the coil wire needing the delay.
Might want to ask Grumpy about coaxial delays used in pulse formers and saturated core experiments.
A couple of inverters will give me the correct delay propagation period.
The problem i need to get over is the dying fets.
To be honest G said i could have a worker by the weekend ::)
And it's been very interesting seeing how much available energy is in this kick, but i still don't know how to produce it correctly, only when i understand this will i get more than random occurrences of it.
Random occurrences are ok if they are huge in size, but i need quantity otherwise the components driving it are being over rated, only when i can produce it at will can i continually fire that cannon, at the moment i am only missfiring.
EDIT:
I have not yet tried a bias voltage to the core wire, i have run out of power units, so need to try an revive an old gel cell tomorrow.
You may find that random is the best that can be done. The solution then would be to make them continue in a circle with the pauses being filled by other kicks.
Can't wait to be home so I can see your scope shots.
I've never heard one being described as both sides of zero at the same time.
I suppose it might look that way on a digital scope. Not sure. I stick with analog for that and other reasons.
;D
I think i am beginning to appreciate the problem with harnessing the power from the kick.
I tried a fast diode to rectify it and it didn't even see it, no wonder SM had to rotate it first to produce dc
I'm not sure you can rectify current from an 'A' field. I am pretty sure the only thing you can do toward conventional current is to bias the rotation and/or direction.
Quote from: Peterae on 2010.06.10, 20:06:20
One end of the bifilars are joined together and connect to the coil supply
Each coil has a fast diode across it, each fet then drives each other end of the coil, if i short them together so the fets, diodes and bifilar are in parallel a get bigger kicks but my fets die faster.
I actually had a fet start working again for 3 or 4 seconds earlier after it went short, when i went back after power down i measured each fet and they read ok, so i powered up again and it work ok but then went off.
Grumpy i would love to have another type of output stage what can i try?
try avlaanche transistors - if you were not so far away I would send everything to you
you need a stack of transistors, I use the 2N5551 and can send you some of those
then you have to find the point where the transistor goes into full avlanche and makes sure it is a low resitance avalanche (the noisy lossy avlanche isn't good) - I do this with an adjustable supply and a small neon bulb and a scope - bulb light when it avalanches, check scope to see if it is clean or noisy - circuit is delay line running wide open and not triggered
Same with diodes but diodes are not cummulative and can't be stacked, but they have high avalanche treshold so using one can be very helpful to reduce the numbers of transistors in the stack.
The delay line approach forces the transitor off when it reflects and then you retrigger it again.
radial force implies all directions but to get a repeatable firing pattern you need repeatable switching so fix that first
get the schematics from my avalanche thread
The energy is NOT in the wire - it is in space (can you say displacement current?) diodes and other SS devices work for conduction current
try coax as a feed line from the FET and ground the shield - WW may have suggested this
http://cmosedu.com/jbaker/papers/RSI621991.pdf
I use the series stack in this paper with 2N5551's and a 100pf cap across each transistor except the triggered one.
keep the current low.
(Maplin has them in UK for .22 each) http://www.maplin.co.uk/Module.aspx?ModuleNo=32952&C=AffilWin87832&T=!!!gid!!!_14453907
Thanks G i will get some and have a play, they are easy to buy over here i can get 150 from China cheaply as well
Also my compasses arrived today, and show no movement whats so ever.
Not sure what to make of this, i just swapped over my high bandwidth whitenoise generators(20Mhz Pic) for low bandwidth(4 Mhz Pic)
The resulting video was recorded with a 20 Ohm load across the scope probe, the scope is on 20Volts/div, instead of spikes i am getting spikes followed by bursts.
Quote from: Peterae on 2010.06.11, 12:13:55
Thanks G i will get some and have a play, they are easy to buy over here i can get 150 from China cheaply as well
Also my compasses arrived today, and show no movement whats so ever.
Not sure what to make of this, i just swapped over my high bandwidth whitenoise generators(20Mhz Pic) for low bandwidth(4 Mhz Pic)
The resulting video was recorded with a 20 Ohm load across the scope probe, the scope is on 20Volts/div, instead of spikes i am getting spikes followed by bursts.
it isn't rotating - you just have bursts of energy in all directions which should still show an indication on a capacitor or battery if they are positive only or biased above zero line (it can't reverse) - look for the high freq oscillation on a cap and battery (separate experiments)
As usual, I can't see the vid yet but...
If the spikes are another view of your big result spike then the bursts are the undulation after the big spike.
I relate that big spike to the high energy TPU arc and the folowing solitons to the loaded DC output.
The aprox. 5khz tone I relate to the rotation frequency.
A good thing to make moving charge change direction is a magnet.
G wouldn't one of these be better it's an avalanche transistor i can get these for 2-3 GBP
http://www.diodes.com/datasheets/FMMT415.pdf
Quote from: Peterae on 2010.06.11, 21:08:01
G wouldn't one of these be better it's an avalanche transistor i can get these for 2-3 GBP
http://www.diodes.com/datasheets/FMMT415.pdf
They will be more consistent between devices and you probably won't have any bad ones. Keep in mind that you will need several at higher voltages. My 2N5551's av at about 240v - 260v and cost about 10 cents each.
Hi,
As avalanche semiconductors have been mentioned there exists avalanche photo diodes. These can operate as high as 1500 volts and operation speaks for itself. Wider bandwidths do give rise to more noise. I have never used them and unsure of their speed or current ratings at present. They may be of use in this project so are worth a mention.
Steve.
G i see you ask about Oscillators on the other thread, whats your requirements, i guess the 555 is the easiest for low frequency's.
I prefer the LTC1799 but you need to make a PCB for this little beast and have steady hands as it's a 5 pin chip the size of a sot23 SMD transistor, but with 1 resistor and 1 cap it has a range between 1kHz to 33MHz, the trick is to batch solder the leggs then wick the excess up if there are any shorts, use a 10 turn pot.
Let me know i maybe able to help you, if you are interested i could knock some pcb's up and mail them, i would'nt mind having some more myself.
Also LTC sent me some as freebies ;D
or option 2 build it yourself
http://cgi.ebay.co.uk/LTC1799-1kHz-33MHz-Oscillator-555-timer-alternative-/360266067165?cmd=ViewItem&pt=UK_BOI_Electrical_Components_Supplies_ET&hash=item53e187ecdd
or direct from them
http://eshop.tirnaelectronics.co.uk/index.php?main_page=product_info&cPath=6&products_id=163&zenid=fbf86089979b368426a0c949a2fdaae1
maybe there's a source of these in the US
Peterae,
I just understood you are using two separate noise generators. This may be cause for the rare occurrence of bursts and would eliminate any chance of rotational effects around the core conductor.
I suggest you feed one signal into both bifilar conductors. Insert a delay before only one of the bifilar coils. The resulting two separate signals must match exactly, with the exception of phase, to create a helical path around the core conductor.
Swapped the delayed and instant coil feeds to see if there is improvement possible.
Apologies if you've already done this.
Quote from: Peterae on 2010.06.12, 08:26:15
G i see you ask about Oscillators on the other thread, whats your requirements, i guess the 555 is the easiest for low frequency's.
I prefer the LTC1799 but you need to make a PCB for this little beast and have steady hands as it's a 5 pin chip the size of a sot23 SMD transistor, but with 1 resistor and 1 cap it has a range between 1kHz to 33MHz, the trick is to batch solder the leggs then wick the excess up if there are any shorts, use a 10 turn pot.
Let me know i maybe able to help you, if you are interested i could knock some pcb's up and mail them, i would'nt mind having some more myself.
Also LTC sent me some as freebies ;D
or option 2 build it yourself
http://cgi.ebay.co.uk/LTC1799-1kHz-33MHz-Oscillator-555-timer-alternative-/360266067165?cmd=ViewItem&pt=UK_BOI_Electrical_Components_Supplies_ET&hash=item53e187ecdd
or direct from them
http://eshop.tirnaelectronics.co.uk/index.php?main_page=product_info&cPath=6&products_id=163&zenid=fbf86089979b368426a0c949a2fdaae1
maybe there's a source of these in the US
Thanks - need about 400kHz
Not sure if i mentioned that i had managed to scope the pulse directly across an Avremenko plug which in turn was connected to one wire of the monfilar that was being driven by both fet's.
and then i have the strange feedback i am getting that causes the re triggering, even now although i have 2 totally separate channels, 2 separate whitenoise generators driving 2 separate fet stages, it still makes a difference if i advance the phase of one of them.
Looks like my next action is a logical one to take, and that is to try and capture the energy using an Avremeko plug.
Also i hadn't realized in the Spherics lab Patent for the audiospatial patent it clearly says SM used grey noise injection to enhance the sound spatially.
Grey noise is where frequency amplitudes are adjusted so they all 'sound' equal to the human ear.
Unless the TPU is psychoelectric I can't imagine how Grey noise would have been part.
With his obvious experience in the use of noise, I can imagine him trying all colors :)
G i would personly go for the LTC1799 see if you can buy it on ebay on a pcb or something in the US.
If not i can mail some pcbs but you will need to solder them
or
go for a simple approach and use a 555 timer.
EDIT WW yes i see your point
grey noise
(http://upload.wikimedia.org/wikipedia/commons/thumb/7/75/Gray_noise_spectrum.png/220px-Gray_noise_spectrum.png)
If I must choose a noise color it would be red, pink or white.
The reasons are related to things said by SM (if I remember correctly):
1. One ring had a higher Q than the other
2. Frequencies of 35 & 245k were mentioned
3. 245 was mentioned as if it was the upper end of a bandwidth
4. Lightning and signal at a distance
and many other things relating to one long and one short or low & high.
My deduction is any noise used is probably pink. The lower frequency would be 35.705kHz and highest amplitude. The highest frequency would be near 245kHz with lowest amplitude. The bandwidth of the noise generator would be around 210kHz.
Pink noise is the noise common in nature.
Brownian has possible application, too. Any way taken just requires the correct filter after the white noise generator.
SM may have discovered the effect with the grey noise and determined that it has extra energy, then went on to figure out how to utilize it. If you are ever in close proximity to these "explosive sounds" you will definitely agree that they have a lot of energy.
555 is supposed to go to 500khz...
thats right G 500kHz although i am sure in the past there was a variant that went to 1MHz but i would need to check that
This one goes to 3MHz
http://www.national.com/ds/LM/LMC555.pdf
Motorola MC4024P goes up pretty high.
Got some 555's a few minutes ago at the Shack. (Hey! It was next to the Grocery Store and I needed beer.)
OK - got to wire one up
I've just realized i have an area of time the mono's cannot sweep due to propagation delay, but with random and noise feedback it would be covered maybe, just maybe this is where there is some energy.
What happens at the moment is one mono output feeds the other mono's input, which is OK but there maybe a very short time before the 2nd mono responds to the output of the 1st.
Not sure where to find the info, but Spheric said exactly what the timing was in the bifilar experiment, if anyone has it to hand could they post it.
Mean while i need to re study the delay chip datasheet and then have a browse through Spherics stuff.
EDIT
OK had a study between the phase shift and pwm mode delay chip i could potentially have a delay of up to 40nS dead phase built into the current setup, this actually allows me to extend the sweep period to 255+40 which has been needed in the past, but it does mean i have had a blind spot from 0-40nS.
Now to find Spherics info.
--OK found it very interesting
Trouble is i am not sure i am meant to post it in the open.Still i need to get rid of the propagation delay for sure ;D
Christ even Jitter could be causing the kick to be random.
Priority for me now is to make a 1200v pulser based on Grumpys Avalanche technique.
Just had a Russian guy respond to my whitenoise video on youtube he posted a link to a you tube video of his
http://www.youtube.com/watch?v=HUEw5XLoono
He wants my approval, as i cant hear sound or understand his message in Russian it's a bit hard LOL
translation
QuoteCircular coils are often used as the primary windings in the coils of Tesla. Studies of field using Z-magnet show that such use is not accidental. In the disk coils at low frequencies (less than 50 Hz) field interaction at the edge. At frequencies higher than 50 Hz is shifted to the center. At a frequency of 80 Hz the whole plane circular coil represents a field of interaction in which the rotating Z-magnet.
Thanks Ds
I think he is showing the lower HZ reaction to the magnet - speaker effect - and not the hf effect you are showing
By Using an endless toroidal coil as in my large coil, i can make the crackling appear with just one randomly driven channel, the crackling is loudest in the driven coil but can be heard all around the endless coil as well with the aid of a magnet, previously i have always needed a static pulsed channel and a varying phase channel.
If i use my whitenoise generator i cannot get the crackles but if i drive using a randomly phased delay chip that's driven from the whitenoise generator then they appear, so the bandwidth of the whitenoise gen is not high enough.
I imagine the reason i can get the crackles using one channel and an endless coil is there maybe residual pulses travelling around the endless coil for the random pulses to knock against.
The other interesting thing is that by widening the pulse width more energy goes into the system and the more crackling i get and the louder they are, the same can be said for turning up the power supply voltage.
The thing i don't yet understand is why the crackles appear all the way around the endless coil, i need to make a way of scoping them at 2 different points to see if there is a phase difference when they occur, if not then maybe they are longitudinal in nature.
Here's 2 FFT shots of what happens periodically when a crackle happens.
One shot is across the spare drive coil on the opposite side of the endless toroidal coil to the drive coil.
Another shot scoped across the 4 turn horizontal coil.
By disconnecting or putting a break in the endless coil, the crackling stops in the endless coil itself, but i can still produce crackling in a single coil.
I expected the crackling to stop in the single coil when the endless coil was broken but this is not the case. Not yet sure why this is the case.
Probably a very simple question
Here's a scope shot of noise picked up by my 2 scope probes when the crackling happens, the probes are not connected and are being held near my coil when the crackling happens, i have unclipped the earth leads on the probes and taken the clips of the nib's, so the probes are not physically connected to anything, so what does it mean when i have a large positive peak 'see the end of the trace' there is a couple of very positive pulses, so how can i have a positive pulse when all the others are not so positive, does this mean anything significant.
You can see the probes just resting against the coil each side
Peter,
It could mean you are hitting on the right combination once in a while. :)
.99
Thanks for replying Darren
That would be brilliant news, because that now means i have a way of monitoring what the right way forward is. ;)
Chan 1 is connected to an Avramenko plug
Chan 2 just near the coil
You might be inducing charges into the insulation and surrounding air with the crackling occuring at discharge.
I suggest making a Faraday Cup and attempt to detect charged particles.
http://en.wikipedia.org/wiki/Faraday_cup
You may notice the resemblence to Tesla's Radiant Energy devices.
At such low votlages, though, anything occuring is probably in the wire, such as the space between the insualtion and the conductor.
In my current build I get a static set of pulses from outside. There is no power applied to my setup but the positive and negative spikes appear then a ring down. I stated 'Static' because the scope can lock on to this signature and it doesn't waver. I have a vid I can post but did not at the time because I thought I wasn't credible. The PC, fluorescent lights, android phone, gold rings, incandescent lights all were on so I deemed it noisy and not submittable. The spikes are in the 650mv range. Now when pulsed I saw 7v spikes in league with this 'Noise'. I don't have a clean environment. A Faraday cage is a good idea though the spikes were very thin.
Darren
Would you say this qualifies, again scope probe was laying not connected near a small coil bring phase driven from 2 fet stages.
The bizarre thing about this trace is that i couldn't trigger on it to save my life, i also now have my old scope handy as well, this also had a probe nearby and verified the same waveform.
1 Channel of whitenoise mixed with random phased pulses does it all the time and the coil did have the crackling noises as well.
here's the coil and probes
Peter,
Yes, I think you are getting close. Seems a little more controlled now.
Back to the original tests with that single coil. ;)
.99
GK,
It looks like you might be getting a similar effect there.
.99
So it's been there all the time LOL
(http://www.overunityresearch.com/index.php?action=dlattach;topic=272.0;attach=3851)
This image is nothing more than a phase delayed pulse but after having adjusted the frequency up and scanning the phase to get more positive pulses, there were certainly more phase settings that also did the same and other frequency/pulse setting as well.
Frequency was 1.6407Mhz phase delayed @ 155nS
but in the video the whitenoise causes the crackling but also allows the effect to be more pronounced giving sharper positive pulses.
I am more convinced than ever that there is a sweet spot for more energy release, and the trick all along was finding a way to see it, looks like it was as simple as not connecting the probe C.C
Now i understand why the cpu's go crazy and the fet's blow from time to time, there's a large dc gradient in the space surrounding the coil, you can see in the video that a large positive pulse is normally followed by a length of oscillation, this is where the large pulse has triggered the fet to go into self oscillation/unstable mode, only when the space has returned to normal does the fet stop self oscillation, It remains stable until the condition is right to create an event again.
The effect has nothing to do with coil parameters, infact it is totaly possible for me to produce this effect with a short piece of wire.
@GK
I have seen this many times myself, the trick is to zoom down on the timebase and see if the pulses are still positive, in the video they went from positive or negative pulses into ringing, whats happening here is that your scope is trying to show a pulse thats beyond it's current timebase setting, you get large offset pulses that seem to follow a sinewave pattern, but when you zoom down on the timebase they are no longer positive.
In order to try and work out what's going on in my chaotic setup that causes the crackles, i thought i would place a 47Ohm resistor in series with the coil, then my flyback diode is placed across this LR, i can then connect my scope across the 47R resistor to monitor current flow within the coil, the video was taken with crackling present.
Some strange stuff here, during crackling the current seems to double periodically, could this be the result of random phase shifting, certainly may explain the crackling noise to be suddenly jolted with fast high current pulses, it looks like the crackling could be a result of mechanical slap after all.
The tricky thing here was to monitor using a probe not connected so i had to use the second scope, and this showed a strange long period of dc offset which seems to be resultant from connecting the laptop and scope across the resistor even though there is no physical connection what's so ever.
Yellow trace is across the 47R resistor 100V/div I=380/47 = 8Amps no wonder the fet blows C.C
The other scope red chan is a floating probe placed near the coil.
That other scope was not lying about large dc offset pulses
I disconnected the scope across the resistor and layed the probe near the coil, it can clearly be seen every now and then a definate long period in excess of 10uS a dc offset exists.
I just dont know what to make of it ???
If you have a neon bulb and a variable power supply that can light it, connect the bulb to the power supply and turn the votlage up to a level where the bulb will almost light. Then place the bulb near your coil. Does it light?
The xplosions I had occured where there was nothing to slap mechanically.
Something like this will give an indication of charged particles:
Simple Geiger Detector uses neon Glow Lamp
Peter Lay
March 18, 2002
In these times of increased terrorist threats, the possibility of some type of nuclear incident exists. By using a few simple components, it's possible to build a low-cost Geiger detector. Detecting α radiation requires a special Geiger tube. But for detecting only β and γ radiation, a neon glow lamp can be used.
Operation is as follows (see the figure): The alternating current is rectified by diode D1. A simple Zener-diode stabilization network circuit using D2 supplies a voltage of 100 V dc. The value of R1 depends on the source voltage and must be calculated with the equation, R1 = (V ac −s 100 V)/5 mA. Potentiometer R2 is used to set the detector voltage very close to, but slightly below, the neon lamp's ignition voltage. The lamp must not ignite. If a radioactive particle then ionizes the gas in the lamp, it ignites.
At that point, resistor R3 drops most of the voltage, so the lamp voltage is lower than the holding voltage. No additional current flows through the lamp until the next radioactive particle ignites it. During the small moments of current flow, a short and quiet clock noise comes out of the loudspeaker.
More circuits like this one can be found in my book about experiments with radiation sources (written in German), Experimente mit Strahlenquellen im Haushalt (www.peterlay.de).
Note: As with all circuits that can be connected directly to an ac power source, suitable insulation, grounding, and polarization precautions should be taken to avoid the risk of potentially lethal electric shock.
here is another circuit that uses an 85A2 votlage reference tube (neon). These tubes are a few pounds/dollars in surplus stores/Ebay.
Nothing magic here G, no electrostatic discharges or glow in the dark skin tan ray's.
I'm only using 25Volts right now and the psu is averaging out at 2-5mA, but the magic is in the DC gradient and what ever is creating it, i seem to be expanding a field without collapsing it, just wish i could work out how.
I can produce them by phase delaying 2 pulses (gives small gradients)
I can produce them using white noise combined with random phase firing(Gives much larger random pulses), it's as if i need to ripple space around the coil in just the right way, get near once in a while and the final pulse does the job.
At the moment i am puzzled by the coil's current waveform and why it has 2 different current drive levels, the higher is only there when the crackles are present, although i believe the high current is the cause of the crackles, but then what's the relationship with the expanding field.
Quote from: Peterae on 2011.02.17, 19:57:50
Nothing magic here G, no electrostatic discharges or glow in the dark skin tan ray's.
I'm only using 25Volts right now and the psu is averaging out at 2-5mA, but the magic is in the DC gradient and what ever is creating it, i seem to be expanding a field without collapsing it, just wish i could work out how.
I can produce them by phase delaying 2 pulses (gives small gradients)
I can produce them using white noise combined with random phase firing(Gives much larger random pulses), it's as if i need to ripple space around the coil in just the right way, get near once in a while and the final pulse does the job.
At the moment i am puzzled by the coil's current waveform and why it has 2 different current drive levels, the higher is only there when the crackles are present, although i believe the high current is the cause of the crackles, but then what's the relationship with the expanding field.
I have figured a few things out and it would take a long time to explain why I have suggested the detectors.
A dc gradient is good reason that you should try a detector. Expanding a field without collapsing it is also an indication. Don't guess and make assumptions or you will lose several years following the wrong directions like everyone else.
What on earth is going on with that last video
here's 2 shots, it looks to me that there is some screen persistence, but what is that large long negative line connecting the bursts, and how on earth can i trigger on it if it's not a pulse that's larger than anything else.
these shots were verified using 2 seperate scopes, the probe was just laying next to the driven coil.
care to venture a guess?
I have no idea G, the DC offset can be longer than 40uS sometimes and once it appears does not diminish that much even though my driving electronics should be still pulsing like mad, so once it's there, it's there no matter what until the conditions are met by a pulse or noise sequence to close it down, It's like I've found a way of turning on a dc gradient tap, i don't believe the energy is coming from my system because it would vary in intensity, so in that case where can the energy be coming from, if i could work out how to trigger on it i might be able to see it more clearly.
What ever is happening i believe it is the reason why i have been getting Dc offset pulses all along, just now i have stumbled on a way of it lasting much longer.
EDIT
Having thought on this some more, There must be a capacitor that's charged, maybe the insulation around the wire conductor, so in that case an event happens that charges the dielectric, holds the charge for a while until the conditions are met to uncharge it.
So then the question would be, can the tip of the probe touching the insulation of a wire be enough to form a capacitor of enough capacity to cause the effect seen.To be honest i am not sure if the tip was touching or just laid by, but i can confirm this.
OK just found this frame so there is definitely screen persistence or the camera is not in sync with the laptop, so now i don't really know whats happening until i can find a way of triggering on it, although it's definitely there though.
here's a clearer shot
here's one that shows i am still pulsing when it's present and is unaffected by the pulses.
So far i have demonstrated the ability to create separately positive and negative gradients.
Now if i had 2 systems running at the same time, one with positive and the other with negative, and phase these gradients correctly i think i should be able to create more larger gradients, i think this could possibly produce a self powering ever growing event as long as the conditions are maintained by the initial pulse pattern, offcourse this is only speculation and until i know exactly how to control this i have no chance.
Peterae,
Your displays of DC gradients appear to me as possible digital processing artifacts. Under the influence of very high transients I have found even the best DSO with an 'analog mode' can't be trusted.
A couple of things to try to help identify what is going on....
1. Terminate that open transmission line ( put a high value resistance between the scope probe body shield and tip. Mount a resistor at the tip in a similar way as the signal ground whisker - from the probe tip to the probe shield).
2. Fire up the old CRT scope for comparison. (An option for me :) I still have some toys older than my children :-[ )
My preferences.... When it comes to anything with transients - take the pseudo equipment out of the equation.
Pseudo = anything with a switch mode power supply, not isolated with a transformer or anything that converts a signal to numbers then back again.
If your scope isn't screwing up the internal math due to the transients, you may be dealing with compression of charge area.
Cheers WW
I've wound a 100K res across the tip and earth of the probe
Unfortunately i don't have an old scope.
Here's the results having tried to find the best trigger.
I Thought i would try a radiant collector on it ;D
It looks to me to be a simple reverse case of extracting harmonics from a square wave, you inject harmonics into free space and get a square wave out :)
can you post a still shot of the scope shots?
Sure here's a few random ones
Note there's a 100K termination resistor and those are over 100V pulses
It's also interesting to note the decay rate of the dc offset has increased with the 100K Res fitted ;)
EDIT
Also something rather strange, after there's been a decay of the initial dc offset, it stabilizes and no longer decays at all but is still DC offset.
It "appears" that "effects" can be produced at just about any voltage if the change is very fast. however the distance to energy is inverse squared so it takes a lot more to get the "effects" out into space where they can be manipulated.
The dc offset that stays sounds like trapepd charges in the dielectric.
I have a strip of aluminium foil insulated and sealed both sides using cut polypropylene sheet, i do have a connection at one end i can connect to.
The foil is taped around 1 half of the outer diameter of the coil, i have clipped a microphone to one side to monitor the crackling noises, bearing in mind the crackling is a mixture of white noise and false triggering from em bursts into my drive system, it can be seen that i can increase the false triggering by holding a metal object to the foil connection at one end or by pressing my fingers against the polypropylene, more interesting is that when i place a magnet anywhere around the circumference of the coil, i get notable crackling from the foil and the magnets position.
Edit dam it looks like the sound didn't sync properly
Seems i am starting to see mains supply regularity
Both traces are connected to an Avramenko plug.
Green trace is connected to the insulated aluminium foil that's tapes around half of the outer circumference of the coil.
Yellow trace AV plug is connected to a wire coiled up and placed near the aluminium foil.
Green trace is 20V/Div there is a 47 Ohm resistor across the AV plug
There's a 306 Watt pulse towards the end of the trace, the average size of the pulses equates to about 34 Watts per pulse.
Yellow trace is 2V/Div no resistor across the AV Plug
The period between the larger pulse grouping is 50Hz with one intermediate size of group pulses, so including the intermediate group of pulses i would have 100Hz
It appears to me that the energy from the crackles appears to behave like and electrostatic impulse, the larger the surface area of the plate the more energy it picks up, if the plate is flat the AV plug gives DC very sharp bursts of pulses and if i use a coiled wire to pick it up the AV plug has trouble producing any noticeable DC.
Put a platic cutting borad between the "plate" and source of crackels and see if the plate still recieves anything.
yep good idea G i will try tomorrow O0
G
I tried taping the alu foil to the back of a 6mm sheet of perspex, then placed this against the coil, i still had pick up, i then inserted another 6mm making a total 12mm and it still picks up the pulses, slight reduction in amplitude each time.
On another note i connected a flat pancake coil up in the Tesla configuration and can spin a neo magnet high speed, sometimes it would fly off the coil onto the bench and continue spinning for 5 seconds or so.
place a grounded sheet of metal (foil is good) between the source and collector plate
then try both the gronded sheet and the plastic and measure between the grounded sheet and collector to see if you have a potential difference between them.
These are tests for beta and gamma (without going into details). Conversion of virtual particles to real ones emulates particles, but any slight differences will shed light on how they are converted. Capacitors do it all the time... O0
I must pick up on this coil soon , with the dual avalanche pulser.
(http://www.overunityresearch.com/index.php?action=dlattach;topic=272.0;attach=1650)
I received 2 PCB's from Peterae which he presented here: http://www.overunityresearch.com/index.php?topic=3345.msg57701#msg57701
In that post you will also find the circuit diagram.
I decided to build on one of those PCB's the original circuit which was intended to drive a special coil to produce crackling sounds and 80Watt spikes across a 20 Ohm resistor.
I used this original thread to continue so that all previous info is close by.
Another short thread which contains coil building specifications is this one: http://www.overunityresearch.com/index.php?topic=366.msg5348#msg5348
The PIC's white noise program comes from this website: http://electricdruid.net/white-noise-source/ but need some tweaking.
I build up the PCB so it has a 5V regulator for the PIC12F629 controllers (generating the white noise) and a variable LM317 regulator using a 10K potmeter to set the MOSFET driver to 20V
MOSFETs are put on print connectors so can easely be swapped out in case of damage or if needing different type of MOSFETs (currently 2x IRFP260N's are in).
I will continue building the special coil, but it needs prefferable some thick copper cladded steel inner conductor coax cable (I have RG316, but it is thin, so could perhaps use
multiple turns to build a rigid base to wind the bifilar coil around it).
Anyway, here a short video of the PCB setup and the MOSFET driver output signals on the scope: https://www.youtube.com/watch?v=7Uq8Y7qtfkk
Regards Itsu
Dear Itsu.
I'm not sure if the info following will be of any use.....
DIY mig welder wire is Copper coated steel and comes in 0.6 and 0.8 mm diameter. The coating is very thin.
Just musing whether you could easily substitute the centre conductor?
Kind regards, Graham.
PS . My apologies if this has been previously mentioned.
Hi Graham,
well i did not know that, so thats good, but i think another requirement is that it needs to have an isolation around it, like the inner conductor of a coax cable has.
But thanks anyway, Itsu
Quote from: Itsu on 2016.11.19, 13:14:36
Hi Graham,
well i did not know that, so thats good, but i think another requirement is that it needs to have an isolation around it, like the inner conductor of a coax cable has.
But thanks anyway, Itsu
Hi Itsu.
What length of Coax is needed for the experiment? This might sound daft but you could sleeve the mig wire with a tube, fish keeping air tube for example and by carful dissection of some standard coax by pushing the braid back upon itself slide it onto your new inner cable.
This idea would only work for a few meters though, absolute nightmare for a long length.
Kind regards, Graham.
Graham,
the diameter of the coax loop is around 16cm, so that makes the circumference 50cm, so very doable.
Not sure what quality the isolation has to be.
On the other hand, there is a cheap supply of RG11 coax cable available on Ebay (per foot $1) which also has the AWG 14 inner conductor made of copper clad steel.
Regards Itsu
Quote from: Itsu on 2016.11.19, 10:48:04
Anyway, here a short video of the PCB setup and the MOSFET driver output signals on the scope: https://www.youtube.com/watch?v=7Uq8Y7qtfkk
Can these MOSFET gates really tolerate >20V ?
20V seems to be the limit for the IRFP260N, so indeed, i need to tone it down a bit to 15V or so.
Thanks, Itsu
I normally use 18V
try 2kv
Quote from: Grumpy on 2016.11.21, 00:00:27
try 2kv
I foresee alot of great equipment going up in smoke.
It was just this kind of setup that got me divorced. And that was after I blew up the house and neighborhood.
Be very very careful. I cant believe this config is coming back to light.
What works even better is a partial winding just at the ends of the horizontal, thanks to Wattsup.
Is this a test or conspiracy?
Quote from: Grumpy on 2016.11.21, 00:00:27
try 2kv
Did you know that it is actually possible to apply 2kV to the gate of a MOSFET without damaging it?
The key to this is the fact that it is
not the voltage applied to the gate terminal, that breaks down the gate insulator, but the voltage that the gate-source junction gets charged to.
Since the gate is essentially a capacitor of several nF, it is possible to hit it briefly with 2kV without damage, if you interrupt the process before the gate-source junction reaches 20V.
It is a way to quickly turn on a single MOSFET.
P.S.
With an array of parallel MOSFETs, the transmission line effects might charge up some gates before the others.
Interesting, but it probably takes more logic to control the timing then it takes to design an other way of fast switching.
Anyway, i put up something like the needed coil setup.
Basically it a RG316 (thin) inner conductor with its isolation (Silver-Plated Copper Clad Steel) with some isolation / hose over it to create some distance for the bifilar wound (the
green/red dual bonded magnet wire), coils, see picture.
The hose i used has some spiral like shape in which the dual bonded magnet wire falls nicely, but there is some spacing between the wires which i think decreases the
overall inductance of the bifilar coils to much, as they measure only 9uH each.
The in the diagram used R1 (10 Ohm / 10W) gets hot within minutes at 12V and without this R1, the circuit acts like a short.
Any idea what the inductance needs to be for the bifilar coils?
The drain signals in this situation are like in the screenshot1.
The signal across the inner coax cable (loaded with 10 Ohm to) is shown in screenshot 2.
Itsu
Quote from: Itsu on 2016.11.26, 21:53:32
Interesting, but it probably takes more logic to control the timing then it takes to design an other way of fast switching.
If you do it by measuring the voltage to which the gate got charged to and reacting to that information in real time in order to interrupt the charging process, then it is a lot of work ...but it works with any FET.
However, if you do it the lazy way and just precharge some pF cap to HV and then dump it into the MOSFET's gate at a proper moment, then the charging process terminates when the voltage in the pF capacitor equalizes with the voltage across the nF capacitor, that constitutes the gate.
The downside of the latter method is that the pF capacitor needs to be matched to the capacitance of the gate, so it is not universal and works only with the MOSFETs that it was matched to. Also the pF HV cap needs to be one of these high-current caps for radio transmitter applications.
Quote from: Itsu on 2016.11.26, 21:53:32
Anyway, i put up something like the needed coil setup.
It is a very good looking coil.
I do not know if it is suitable for what Peter was suggesting, but it is good looking nonetheless.
Quote from: verpies on 2016.11.27, 01:08:34
If you do it by measuring the voltage to which the gate got charged to and reacting to that information in real time in order to interrupt the charging process, then it is a lot of work ...but it works with any FET.
However, if you do it the lazy way and just precharge some pF cap to HV and then dump it into the MOSFET's gate at a proper moment, then the charging process terminates when the voltage in the pF capacitor equalizes with the voltage across the nF capacitor, that constitutes the gate.
The downside of the latter method is that the pF capacitor needs to be matched to the capacitance of the gate, so it is not universal and works only with the MOSFETs that it was matched to. Also the pF HV cap needs to be one of these high-current caps for radio transmitter applications.
Right, the "lazy way" seems like a novel idea, and there are many high-current pF caps for radio transmitter applications available.
But how do we "dump it into the MOSFET's gate at a proper moment", by using a MOSFET? :o
Itsu
Quote from: verpies on 2016.11.27, 01:14:04
It is a very good looking coil.
I do not know if it is suitable for what Peter was suggesting, but it is good looking nonetheless.
Thanks, i think ;D
The bifilar coils need more inductance and the radius needs to be smaller too, so i will be looking for a smaller radius hose without the spirals so the windings can be closer together.
Itsu
Itsu
Good work.
Try to increase the whitenoise bandwidth by adding a 20MHz crystal and 2 22pf caps across the pic pins 2 & 3, i think a cap could be soldered across the under pads, maybe a xtal as well, do this for both pics, it will increase the bandwidth of the whitenoise greatly.
Is there any crackling coming from the coils, try holding a magnet next to the coils.
I think you need to get some thinner wire for your bifilar and greatly increase the turns if possible, but first do the xtal to see how it performs.
Thanks Peter,
you mean adding a 20Mhz x-tal and 2 x 22pF caps to each Pic pins 2 and 3 like in the attached diagram?
Presently both Pics have their pins 2 and 3 directly connected to ground, so i have to remove that and install the xtals / caps (which i happen to have).
But do the pics not need to be reprogrammed as probably they are set to run on their internal clocks (4Mhz)?
Running now with 40V on the drains via a 220 OHm / 25W resistor, but no crackling noise heard, also not when using a stack (4) of ceramic magnets next to the coils.
Itsu
Quote from: Itsu on 2016.11.27, 10:27:54
The bifilar coils need more inductance and the radius needs to be smaller too, so i will be looking for a smaller radius hose without the spirals so the windings can be closer together.
As you probably know every bifilar coils can be connected in two ways - aiding and opposing (bucking). The former yields more self-inductance and the latter yields less (ideally zero).
Also, decreasing the radius decreases the inductance. Bringing the windings closer together will foremostly increase the coil's self-capacitance.
Increasing the turn count will increase the self-inductance squarely (L∝n
2)
Yes, i did know, but it is funny how some things are just taken for granted without giving it any further thoughts.
Off course i need to take a closer look at this bifilar coil, to see how it ticks (is connected) before throwing in my first comments.
Thanks, Itsu
Hi Itsu
Thats right xtal as per the above diagram, there is an unfortunate error in the pcb, with one of the pics pin was connected to pin1, cut that accidental track underneath on the pic that has this error, then you will notice underneath the board there is a way to solder surface mounted caps to earth across pin 2 & 3 of each pic, they are not connected to ground or should not be, if you look at the track layout.
Then you will need to find a way of soldering a 20MHz xtal onto pins 2 & 3 you could solder directly on to the chip pins so the xtal sits on top of the pic, or you could solder underneath the pcb but make sure the metal case does not touch anything and short it.
Then you will need to re-burn the software with the configuration bits set for HS which is how it was before we got it currently going.
I don't remember being able to get the high current spike with the pic running at 4mhz like you have at the moment.
Unfortunately i cannot guarantee you will find the pulse, this pcb setup has never been checked as working by me, but my intention was to give it a go to try and make it easier, when i got the current spike i kept the current low by limiting the whitenoise pulse width to 100nS by feeding 2 digital monostables, It's the clashing of the rising or falling edges occasionally that seems to create the spike but i could never track it exactly down which is why i use whitenoise because it covers every random pattern once in a while creating the spike.
QuoteHi Itsu
Thats right xtal as per the above diagram, there is an unfortunate error in the pcb, with one of the pics pin was connected to pin1, cut that accidental track underneath on the pic that has this error, then you will notice underneath the board there is a way to solder surface mounted caps to earth across pin 2 & 3 of each pic, they are not connected to ground or should not be, if you look at the track layout.
Peter,
i must of have a different (later?) pcb as none of the 2 pcb's i have, show the pin 2 to pin 1 track you mention and which can be seen in your above pcb layout picture.
Concerning the pins 2 and 3 to ground which i presently have, this is clearly stated in the diagram below, there both Pics have their pins 2 and 3 connected to ground
via 0 Ohms (bridge) SMD resistors (R5, R6, R7 and R8), so i used a solderbridge to put them to ground.
Anyway, i will rearrange the PCB so that it runs with the 20Mhz x-tals and reprogram the pics to use this external timer.
Thanks, itsu
Ah ok thats good pin 2 is not grounded, i am only looking at the photo of the layout and it looks like it is shorted, so thats good.
Pin 2 & 3 show a resistor or a zero ohm link, but these have not been fitted, they should 0800 footprints for surface mount resistors which are the same size as the surface mount capacitor, in your case instead of a resistor you want to fit the 2 crystal caps 22pf.
Ok, 20Mhz X-tals inserted, code changed, so running on 20Mhz now.
Again with 40V on the drains (via a 220 Ohm 25W resistor which still gets hot) no crackling noises heard, not even with the stack of 4 ceramic magnets at the coils.
Screenshot shows the drain to source signal of the both MOSFETs.
I will play with it some more tomorrow.
Itsu
As a new coil i used 6 of those RG316 inner isolated copper clad steel wires in parallel as the main loop (kind of litz wire)
Around it i wound about 250 turns each of the bifilar coil yielding an inductance of 11uH each.
I still need 220 Ohm in the MOSFETs supply line to keep the current to the MOSFETs at a 180mA rms at 41V.
No crackling sound noticed, with or without using 4 stacked ceramic magnets or 2 stacked rectangular neo's.
The 50 Ohm load resistor across the main loop shows random max. 5Vpp which equates to about 0.5W (P=U²/R)
Video here: https://www.youtube.com/watch?v=LREIv7nx92o
Regards Itsu
Peter's coil must by critical.
Before you accomplish the increase of your coil's inductance, you can get rid of that big resistor by high-pass filtering the outputs of your noise generators. With only high frequencies getting through, even the reactance of your low-inductance coil will be enough to limit the average drain current.
Interesting idea verpies to cut the lower frequencies, the other way maybe to feed a dual monostable with each whitenoise signal to limit the pulse width which would be adjustable using a pot, Itsu would need to build a little vero board with a retrigerable mono and the 2 pics on, that then plugs into the 2 8 pin sockets, a bit like i did with my use of the pcb in the other thread, this should do the job.
Why is Itsus signal slanted slightly at an angle? looks strange.
Quote from: verpies on 2016.11.30, 17:53:37
Peter's coil must by critical.
Before you accomplish the increase of your coil's inductance, you can get rid of that big resistor by high-pass filtering the outputs of you noise generators. With only high frequencies getting through, even the reactance of your low-inductance coil will be enough to limit the average drain current.
Ok, a high pass filter in the white noise generator output, that would be the both Pic's, and that would make the white noise pink, right?
Anyway, i followed an online high pass filter calculator and settled for a RC of 1.8K and 10uF, meaning a high pass cutoff of 8.8KHz.
I put it between the Pic's output and the dual MOSFET driver.
Indeed, now the big resistor can be removed and 41V put directly on the drains and whalla, crackling noise emitted from the coils.
I put 2 GDT (Gas Discharge tubes) of 230V across the drain/source of the MOSFETs, but they seem not to light up allthough i measure >400V peaks on the drains.
My bench PS seems to have difficulty supplying current quick enough as the current limiter (10A) even when set to max. kicks in frequently, also
om the 30V circuit supply part :D
First screenshot shows the signal across 1 MOSFET drain/source, the second screenshot is from across the 50 Ohm resistor on the main loop.
Video where you can hear the crackling noise here: https://www.youtube.com/watch?v=LpO4JxQTtSg&feature=youtu.be
Itsu
Thats great news Itsu.
A few things to try, use an electret microphone to listen to the crackling, its very interesting.
Connect both fet drains together and use these connected drains to drive a single coil, start at low voltage and be very slow to turn the voltage up.
The other thing worth trying is to wind a 100K resistor around the tip of a scope probe and connect the other end around the scope probe earth and place it as a sniffer next to the coil and look for positive pulses.
Thanks Peter,
i will do these further experiments tomorrow.
I will start by putting in >600V rated MOSFETs as these 200V IRFP260N's will not survive for long.
In the both drains connected scenario, you mean with "a single coil" one of the bifilar coils, right?.
By the way, these crackling pulses on the drains are very small, <100ns, see screenshot.
Regards Itsu
QuoteIn the both drains connected scenario, you mean with "a single coil" one of the bifilar coils, right?.
Thats right O0
QuoteBy the way, these crackling pulses on the drains are very small, <100ns, see screenshot.
This is probably because when the phase delay between the two fet signals is just right you hit on the sweet spot and the result is a large pulse, you have probably seen my phase delay experiments where 1 fet is delayed 0-255nS and swept in 1nS steps at a certain delay a large pulse appears.
I dont ever remember seeing such high voltages on my drians but that could have been the poor quality scope i had at the time and would certainly explain why my fets would blow.
I was using irf840 fets.
Some fets just dont work as they are too slow to turn on and off.
Quote from: Itsu on 2016.11.30, 21:17:38
Ok, a high pass filter in the white noise generator output, that would be the both Pic's, and that would make the white noise pink, right?
Violet (https://en.wikipedia.org/wiki/Colors_of_noise#Violet_noise)
Quote from: Itsu on 2016.11.30, 21:17:38
Indeed, now the big resistor can be removed and 41V put directly on the drains and whalla, crackling noise emitted from the coils.
Interesting. I have to analyze whether the superposition of two noise waveforms can cause these amplitudes. My intuition says "no" but I haven't factored in the reflections from the ends of your "bifilar transmision line" yet.
Quote from: Itsu on 2016.11.30, 21:17:38
I put 2 GDT (Gas Discharge tubes) of 230V across the drain/source of the MOSFETs, but they seem not to light up allthough i measure >400V peaks on the drains.
Turn off all the lights including scopes, tape up all pilot lights and put some camera on long exposure. I bet you will see these GDT light up.
Quote from: Itsu on 2016.11.30, 21:17:38
My bench PS seems to have difficulty supplying current quick enough as the current limiter (10A) even when set to max. kicks in frequently, also
om the 30V circuit supply part :D
We can't have that current limiter acting up like this.
Perhaps a large capacitor bank bypassing the output of the power supply will help. If not, add a huge choke between the caps and the power supply.
What is the " 30V circuit supply part" ? (I haven't been following closely).
Quote from: Itsu on 2016.11.30, 21:17:38
First screenshot shows the signal across 1 MOSFET drain/source, the second screenshot is from across the 50 Ohm resistor on the main loop.
I saw them. Interesting.
What I am missing is the baseline at a different scale , i.e. V and I of normal noise pulses visible when the cracking is not occurring.
This is for comparison between the usual and unusual.
I won't repeat Peter's suggestions only add some more:
1) Null experiment: Does the crackling still happen wit a copper wire of the same diameter arranged on an identical coil former the same way? AFAIK you are using iron wire now...
2) Is the crackling sensitive to the static magnetic field of the permanent magnets (orientation, distance, plurality, polarization) ?
3) Is the signal at MOSFET gate disturbed by anything more than the Miller's capacitance (C
D-G) ?
4) Keep the scope probes away from the coil (only their tips should approach it). That might not be possible to notice easily with the randomness of the phenomena.
Quote from: Itsu on 2016.11.30, 22:04:31
I will start by putting in >600V rated MOSFETs as these 200V IRFP260N's will not survive for long.
Stronger MOSFETs and drivers are a good idea. We do not want some avalanche D-S breakdown to be responsible for the crackling.
As usual, keep the Source-Gate loop
area as small as possible.
Quote from: Itsu on 2016.11.30, 22:04:31
By the way, these crackling pulses on the drains are very small, <100ns, see screenshot.
That's good! Does the static magnetic field affect their width ?
The shear between two MOSFETs and the reflections from the ends of a transmission line (your coil) can interfere constructively or destructively....but that much!?
Do show a long flat line before the pulse. It is significant.
Poynt99 duplicated the pulse in SPICE.
What pulse magnitude came out in his sim ?
I changed the MOSFETs for IRFP460's which can handle 500V, the crackling sound seems to be the same.
Some quick answers:
QuoteWe can't have that current limiter acting up like this.
Perhaps a large capacitor bank bypassing the output of the power supply will help. If not, add a huge choke between the caps and the power supply.
What is the " 30V circuit supply part" ? (I haven't been following closely).
I use my dual bench PS (2x 40V @ 10A) which has 2 separate parts, 1 supplying the 30V to the Pic's (5V) and dual MOSFET driver (15V), the other
part supplying the MOSFETs drain voltage (max. = 41.5V)
Returns are connected together.
I stiffened up the 41.5V volt part with 2x 910uF parallel / 350V caps, but the current limiter still kicks in (these caps stopped the separate 30V PS part
current limiter led from coming on :D ).
When i add a huge choke (primary of a MOT @ 250mH), between PS and caps, then the PS voltage starts to rise when the crackling occurres up till 47V,
but still the current limiter kicks in, so i removed the choke.
Using my current probe shows that the drains current peaks, when crackling, goes of my scope scale @ 5A/Div., so must be more then 50A.
Guess the bench PS simply cannot deliver enough and rightfully activate the current limiter.
QuoteI saw them. Interesting.
What I am missing is the baseline at a different scale , i.e. V and I of normal noise pulses visible when the cracking is not occurring.
This is for comparison between the usual and unusual.
When no crackling sound, the drain voltage is the 41V (see screenshot), and no current flows, the MOSFET is not active.
Perhaps i have the high pass filter set too high.
The NEO's seems to have no real influence on the shape or length of the pulses, they wiggle in rhythm with the crackling when set at 4cm from the coil.
More testing / results lateron.
Itsu
Itsu
take some thin enamelled copper wire and wind an air bifilar, wire length is not important then drive that and see that your 50Amp pulse is still there maybe.
It should also still crackle, my yellow coil was only used to find the anomalous 80 Watt pulse, but even a length of wire can crackle with both fets connected together.
PS if you have a dual chan sig gen then you could take the pics out and drive each fet driver with each sig gen and set pulse width to 100nS and then vary the delay between the pulses, some where you will find the anomalous pulse if you can adjust in nS
Quote from: verpies on 2016.12.01, 03:16:54
What pulse magnitude came out in his sim ?
You'd have to ask Poynt99. I haven't been able to find the posts he made.
Quote from: Itsu on 2016.12.01, 16:41:31
The NEO's seems to have no real influence on the shape or length of the pulses, they wiggle in rhythm with the crackling when set at 4cm from the coil.
Itsu
Try putting a coil around a NEO, matching the magnetic orientation and observe on a scope.
Quote from: Peterae on 2016.11.30, 21:40:02
Thats great news Itsu.
A few things to try, use an electret microphone to listen to the crackling, its very interesting.
Connect both fet drains together and use these connected drains to drive a single coil, start at low voltage and be very slow to turn the voltage up.
The other thing worth trying is to wind a 100K resistor around the tip of a scope probe and connect the other end around the scope probe earth and place it as a sniffer next to the coil and look for positive pulses.
Peter,
i seem to have misplaced my electret microphone / amplifier, so this test has to wait, but should it not sound the same as the sound picked up by the video camera mic?
When connecting both drains together on a single coil, i do not see much difference with the MOSFETs connected to their own coils, see screenshot 1
When connecting a 120K resistor across my scope probe, a get the signals as can be seen in screenshot 2.
All positions around the coil seem to pick up a similar signal.
These signals are with 24V from a series set of battery instead of the 41V from the PS. So we have no current limiting now and a 10A analog current meter on the batteries sometimes goes all the way to 10A!
Itsu
Quote from: Grumpy on 2016.12.01, 19:31:36
Try putting a coil around a NEO, matching the magnetic orientation and observe on a scope.
When putting a 4 Ohm speaker coil around 2 stacked rectangular neo's i get this signal from this speaker coil, see screenshot.
Itsu
Look at those positive pulses in all your scope shots, i would be interested to see what verpies makes of those, especially the 120K terminated scope probe.
QuoteI won't repeat Peter's suggestions only add some more:
1) Null experiment: Does the crackling still happen wit a copper wire of the same diameter arranged on an identical coil former the same way? AFAIK you are using iron wire now...
2) Is the crackling sensitive to the static magnetic field of the permanent magnets (orientation, distance, plurality, polarization) ?
3) Is the signal at MOSFET gate disturbed by anything more than the Miller's capacitance (CD-G) ?
4) Keep the scope probes away from the coil (only their tips should approach it). That might not be possible to notice easily with the randomness of the phenomena.
1) i used my earlier bifilar coil on the hose, but now without any center loop wire.
Even with 24V from 2 series batteries instead of the 41V from the PS i see (not hear) some activity.
Only when attaching stacked neo magnets to this bifilar air coil i can hear some crackling again, but coming from the magnets it seems.
So the inner loop wire (copper or steel) is not needed to get the effect.
2) i can not detect any effect from the magnets on the crackling or shape of the pulses, but as the cracking is very random, its not easy to tell.
3) I can not see any abnormalities on the gate signal during the crackling, except these oscillations just before activating the MOSFET, see screenshot
4) i tried to keep the probes away from the coils as much as possible.
One thing i noticed is that the 2 series batteries seem to gain voltage from 24.8V when starting to 25.1V after some minutes of testing.
I will upload a video which shows some of the above tests lateron.
Itsu
How much delay are you using between each channel? Peter found a sweet spot around 220ns with his setup.
Quote from: Peterae on 2016.12.01, 18:07:01
If you have a dual chan sig gen then you could take the pics out and drive each fet driver with each sig gen and set pulse width to 100nS and then vary the delay between the pulses, some where you will find the anomalous pulse if you can adjust in nS
He does have a dual chan sig gen that is capable of ns resolution, but you'd need to be more specific how to program it.
So far we have 100ns pulse width, but what about the other timings and most importantly is VAR1 equal to VAR2 ?
(http://www.overunityresearch.com/index.php?action=dlattach;topic=272.0;attach=23359)
Quote from: Itsu on 2016.12.01, 20:44:02
I seem to have misplaced my electret microphone / amplifier, so this test has to wait, but should it not sound the same as the sound picked up by the video camera mic?
Not necessarily because electret microphones are also good E-field sensors...besides acoustics.
Quote from: Itsu on 2016.12.01, 20:44:02
When connecting a 120K resistor across my scope probe, a get the signals as can be seen in screenshot 2.
The amplitude seems too much for one loop of such small area. Is there much difference with a shorted loop? How about 20 turns of thin wire wrapped around a small short ferrite rod/bead ?
Quote from: Itsu on 2016.12.01, 20:44:02
All positions around the coil seem to pick up a similar signal.
Which suggests E-field pickup
Quote from: Itsu on 2016.12.01, 20:44:02
These signals are with 24V from a series set of battery instead of the 41V from the PS. So we have no current limiting now and a 10A analog current meter on the batteries sometimes goes all the way to 10A!
Could the MOSFETs be avalanching or getting stuck on? High dv/dt from the drain of one MOSFET is capable of turning-on the other. Investigate which way that current is flowing.
Quote from: Itsu on 2016.12.01, 21:20:52
3) I can not see any abnormalities on the gate signal during the crackling, except these oscillations just before activating the MOSFET, see screenshot
Why are there oscillations before the first two rising edges but not before the third? Weird...
Sorry for the randomness of my answers, it must be contagious :-), or for any missed ones.
Here a video from some of the above tests: https://www.youtube.com/watch?v=-SaiuJTtC8c
More tomorrow.......
Itsu
Quote from: Grumpy on 2016.12.01, 22:24:42
How much delay are you using between each channel? Peter found a sweet spot around 220ns with his setup.
I am not using any deliberate delays, both pic's (channels) are running on their own 20Mhz X-tal, no sync between them.
I will look into a delay setup like also mentioned by Peter.
Itsu
Quote from: verpies on 2016.12.02, 00:02:38
He does have a dual chan sig gen that is capable of ns resolution, but you'd need to be more specific how to program it.
So far we have 100ns pulse width, but what about the other timings and most importantly is VAR1 equal to VAR2 ?
(http://www.overunityresearch.com/index.php?action=dlattach;topic=272.0;attach=23359)
Var1 & Var2 can be equal as long as Var is adjustable, he can adjust this, the width is the same but the narrower the better.
I would pick up ghostly noises on the electret microphone even feet away, sort of howling noises, there are videos of this noise but cannot remember which thread they are in now, if i come across them then i will post a link.
before i found this crackling noise using this method i found it before using 3 555 timers running at 1st ,2nd & 3rd harmonics with a trifilar wound around the ferrite, this time the ferrite seemed to crackle.
What do you make of the positive pulses picked up on the 120k resistor.??
I used a dual spark gap into an air-core Brooks-wound coil, with 10kv DC pulses (arbitrary pulse rate, free-running) and instead of crackling, I got very loud sounds like a rifle crack. There was no flash, just the loud crack sound.
The lack of sync between channels may be why you are getting any crackling at all. Similar to Peter's use of white noise.
Thanks Grumpy, sounds impressive.
I at the moment still use a similar setup as Peter with 2 white noise generators.
Itsu
Some further tests, with a starting voltage of 25.19V (2x 12V batteries in series) on the drains.
First the electret mic. test, which only reveals similar crackling sounds as emitting from the coil / magnets, so no howling.
I tried with both coils (so with and without a center loop wire) and with 24V from the batteries as well as with the 41V from the PS (last one not on video).
Then i tried to measure / show the current through 1 of the MOSFETs.
It shows that the current flows the normal way from drain to source.
Current controller was set to 5A/Div. and the voltage on the MOSFETs was reduced to 12.5V only (single battery), see screenshot 1.
Screenshot 2 is both voltage (yellow) over drain / source and current through (green).
Video here: https://www.youtube.com/watch?v=yyEAJqV4u7g
Next i will use some different pickup probes to show the different signals.
Itsu
I used 3 probes with different sensors to sample the crackling noise.
41V on the drains.
Yellow is a probe with a 120K resistor
Blue is a probe with a shorted loop
purple is a probe with a ferrite bead with 22 turns of thin copper wire
Itsu
Just to clarify Itsu in your last post are you AC or DC coupling on your scope setting.
and maybe a zoom in on the timebase to look at the purple chan oscillations.
Peter,
all are DC coupled.
See screenshot for a zoomed in picture of the purple trace (ferrite bead)
Itsu
Quote from: Itsu on 2016.12.02, 21:23:53
Peter,
all are DC coupled.
See screenshot for a zoomed in picture of the purple trace (ferrite bead)
Itsu
Thanks Itsu O0
The series batteries measured 25.16V today, so we did loose some charge (0.3V) compared with yesterday.
Measuring the outputs of the white noise generators (Pics) after the high pass filter going into the dual MOSFET driver (blue and purple), together with the MOSFETs drain signals (yellow and green).
First screenshot is without crackling, second screenshot is with crackling.
Itsu
Same as above, but now measuring the gates (blue and purple) together again with the drains (yellow and green).
first screenshot without crackling, second screenshot with crackling.
To be noticed is that without crackling, all lines are flat (drains on drain voltage), so also the gates.
To me this means that the crackling selection (when the crackling occurres) is being done by the MOSFET driver, not the MOSFETs itself, right?
itsu
I thought that the PICs are generating noise continuously - even when the coil is not crackling, so why is there no sign of that noise on the no-crackling scopeshots ?
QuoteI thought that the PICs are generating noise continuously - even when the coil is not crackling
correct, and as can be seen in upper scopeshot of post #232, we DO have noise coming out of the Pic's continuously, see blue and purple traces.
But this noise coming out of the Pic's (after the high pass filter) and fed into the ixdd604 dual MOSFET driver does NOT come out continuously, from this driver,
see blue and purple traces of upper scopeshot of post #233.
So therfor my question in post #233:
QuoteTo me this means that the crackling selection (when the crackling occurres) is being done by the MOSFET driver, not the MOSFETs itself, right?
Itsu
Something strange going on then, there should be whitenoise all the time going into the fet driver and into the fet gate but the crackling is random but constant.
When you look at the upper scopeshot of post #232 to the blue and purple traces (white noise output after the high pass filter) you can see
that the signal levels are mostly below the zero lines.
So could it be that the high pass filter is causing this so that the IXDD604 dual MOSFET driver is getting a to low (normal noise) signal and not amplify it?
Only when a certain combination of noise occurres it gets above this threshold (see lower scopeshot of post #232) and then amplifies the signal to the gates.
Itsu
Here i zoomed in on the signals before the high pass filter and after (high pass filter sits between the white noise Pic's and the IXDD604 dual MOSFET driver).
We can clearly see that the HP filter is leveling the signal so only half of its amplitude is above ground, which probably is too less for the IXDD604.
Blue is BEFORE HP filter, purple AFTER HP filter.
Itsu
These are the same signals (with higher vertical setting) but now when crackling occurres.
This must be some kind of feedback as there is no way these Pic's can generate these kind of signals.
How far are the Pic's from the switching devices?
See this physical layout of the used PCB, red are the Pic's, green the both MOSFETs, so about 2cm apart.
itsu
Concerning the quality of the white noise generated by the Pic's, see the screenshot where blue is the signal coming from one of the Pic's, compared to purple which is
the white noise signal generated by my FG (5Vpp, 2V offset).
The purple looks to me really like a white noise signal, while the blue signal more looks like a FM modulated signal.
Itsu
Scope the power rails across the pic when crackling happens, this looks like power rail noise to me, if so you need a high quality 10nF & 100nF in parallel across the pic rail.
Hi Peter,
yes, i guess you are right, but after adding wima 10nF caps parallel to the existing 100nF caps across the Pic's supply, i still have heavy spikes on that 5V rail, see screenshot.
There is not much room to add additional chokes / caps there, so i might need to rebuild the setup to get it quiet.
Itsu
In the below video i removed the MOSFETs drain voltage and just run the Pic's and MOSFET driver.
I monitor with the yellow and green probes the continuous white noise outputs from the Pic's before the high pass filters and with
the blue and purple probes the signals on the gates of the MOSFETs (without any drain voltage on).
We still see the crackling (sort of) occure, but because the MOSFETs are down they cannot push it through to the coils.
To me this means that the decision when the crackling happens is being made by the MOSFET driver when it receives a certain combination of signals from the Pic's.
Video here: https://www.youtube.com/watch?v=KuFZBqQnHjk
Screenshot shows a possible trigger combination.
Itsu
Is your driver a dual driver on one chip?
correct, i followed Peters original design which uses an IXDD604 dual MOSFET driver chip.
I removed the both Pic's and used my dual channel FG to feed white noise (5Vpp, 1.5V offset) to the both MOSFET driver inputs via a 50 Ohm series resistor.
There seem to be some (thermal?) runaway condition happening as (with no drain voltage on the MOSFETs), gradually the current into the PCB (driver only) increases untill its that hot i can smell it.
Same when also putting on the drain voltage on the MOSFETs, gradually the current through the MOSFETs increase untill i stop it.
I think the circuit is missing some kind of MOSFET driver output / MOSFET gate clamping (10K to ground and / or zeners).
I will add that tomorrow to see if it stabilize it.
Itsu
Spherics said that output snubber circuits were required for the mosfet drive he proposed.
attached are the docs that reference his snubber
Thanks Grump, looking into it.
Itsu
Quote from: Itsu on 2016.12.06, 21:46:54
I removed the both Pic's and used my dual channel FG to feed white noise (5Vpp, 1.5V offset) to the both MOSFET driver inputs via a 50 Ohm series resistor.
There seem to be some (thermal?) runaway condition happening as (with no drain voltage on the MOSFETs), gradually the current into the PCB (driver only) increases untill its that hot i can smell it.
Same when also putting on the drain voltage on the MOSFETs, gradually the current through the MOSFETs increase untill i stop it.
I think the circuit is missing some kind of MOSFET driver output / MOSFET gate clamping (10K to ground and / or zeners).
I will add that tomorrow to see if it stabilize it.
Itsu
I added a 1K resistor across gate/source of each MOSFET, but with direct white noise input into the both MOSFET driver inputs at a certain point i get again the runaway.
So i added again the high pass filters between the FG and the MOSFET driver inputs.
Now it seems to be stable, as we have violet noise on each gate and some serious noise on one of the MOSFETs drain, see screenshot.
But this does not generate any crackling noises anymore in the coils., even when running on 41V on the drains and with or without the 2 stackted neo's attached to the coils.
Yellow is gate 1 signal, blue is gate 2 signal and purple is MOSFET 2 drain signal.
So i think that the crackling only occurres when using this specific white noise (FM like) generated by the Pic's.
Itsu
Quote from: Itsu on 2016.12.07, 21:01:00
So i added again the high pass filters between the FG and the MOSFET driver inputs.
So now you re brave enough to connect your FG to this circuit ?
...or are you going to order these digital isolators from Silicon Labs or finish the EL2009 buffers ?
Quote from: Itsu on 2016.12.07, 21:01:00
Now it seems to be stable, as we have violet noise on each gate and some serious noise on one of the MOSFETs drain, see screenshot.
Both drains?
Quote from: Itsu on 2016.12.07, 21:01:00
But this does not generate any crackling noises anymore in the coils., even when running on 41V on the drains and with or without the 2 stackted neo's attached to the coils.
Are the two noise channels from your FG phase-synchronized ? I know it is weird to talk about phase synch between two noise signals ...but it is possible in a digital FG.
Quote from: Itsu on 2016.12.07, 21:01:00
So i think that the crackling only occurres when using this specific white noise (FM like) generated by the Pic's.
Maybe and if so that would be highly unusual.
Let's investigate the phase of the two noise signals appearing at the two outputs of your FG, first.
Peter demonstrated repeatable crackling with a delay of around 220ns between his two signals.
To do anything with whatever is happening, you need predictable repeatability.
Well, Itsu has a 2-ch Function Generator that can output any phase offset of any two signals down to nanoseconds with a turn of a dial.
I just do not know if he is brave enough to use it without galvanic isolators (...such as the ones from SiliconLabs).
Quote from: verpies on 2016.12.08, 01:39:30
So now you re brave enough to connect your FG to this circuit ?
...or are you going to order these digital isolators from Silicon Labs or finish the EL2009 buffers ?
I made sure first without any drain voltage on the MOSFETs that there were no abnormall signals on the gates like there are when using the Pics.
But you are right, that is certainly not full proof, so i will first hook up those EL2009's before doing any further tests.
Concerning those digital isolators from Silicon Labs, do you recommend 2 single or 1 dual-channel unidirectional one(s)?
QuoteBoth drains?
yes, both drains.
QuoteAre the two noise channels from your FG phase-synchronized ? I know it is weird to talk about phase synch between two noise signals ...but it is possible in a digital FG.
Hmmm, don't think so as there seems to be no phase sync setting on the FG screen when on noise, just amplitude and offset, but i will take another look, perhaps the CH1=CH2 button?
Itsu
Dear All.
Off topic, I know but...
Does anyone remember the website that sold all those " Exotic " wires ? UK based?
Kind regards, Graham.
Dont connect your FG, it's not worth it, i will as soon as i get a chance send you my uproc signal gen i built especially for this, you will need to add fet stages to it but i cannot see why the pcb fet drivers could not be hooked up to it.
Problem i have is that i only see day light at the weekend, but not this weekend as i am working, so i guess i can only send the weekend after this coming one O0
Grum wires.co.uk sell all sorts of wires, not sure what you mean by exotic.?
Hi Peterae.
Many thanks. That's the one. O0
Anything other than Iron was classed as exotic in my workshop!! Old " in house " joke !!
Cheers Graham.
Thanks Peter,
with the EL2009CT buffer/amplifiers in place the FG will be protected enough.
I just need to make a neat RF tight enclosure (or 2) which will be the main problem together with a clean + and - 16V power supply.
Itsu
Quote from: Itsu on 2016.12.08, 10:11:47
Concerning those digital isolators from Silicon Labs, do you recommend 2 single or 1 dual-channel unidirectional one(s)?
2 single because they offer less crosstalk and more versatile isolation.
Just pick the fast ones. S.L. has some slow ones, too.
Quote from: Itsu on 2016.12.08, 19:53:37
I just need to make a neat RF tight enclosure (or 2) which will be the main problem together with a clean + and - 16V power supply.
Just put some heavy chokes in series with the power supply lines.
The good old solderable
feedthrough capacitors (http://www.maruwa-g.com/e/products/electronic-parts/000355.html) are a good way to get the power lines into metal enclosures ...with RF bypassing!
(http://www.overunityresearch.com/index.php?action=dlattach;topic=272.0;attach=23471)
Quote2 single because they offer less crosstalk and more versatile isolation.
Right, sounds logical.
QuoteJust put some heavy chokes in series with the power supply lines.
with heavy you mean mH's?
Would a simple circuit like the one below (with the heavy chokes and feedthrough capacitors added) be enough?
Thanks, itsu
Quote from: Itsu on 2016.12.08, 21:45:06
with heavy you mean mH's?
Whatever you can fit in the enclosure....and beads, beads, beads on the power lines inside the enclosure BEFORE the bypass cap.
Quote from: Itsu on 2016.12.08, 21:45:06
Would a simple circuit like the one below (with the heavy chokes and feedthrough capacitors added) be enough?
Yes, but the output should
not have that 50Ω resistor in series (just a 0Ω direct connection) because the output impedance of this amplifier is around 1Ω and you want that because such low impedance will be very useful for driving high current loads (coils etc...) in the future.
Also, the 50Ω input resistor might be wrong because this amplifier and enclosure already have some intrinsic input impedances and that parallel input resistor should be selected to make up a total 50Ω impedance in combination with them, in order to match it to the output impedance of your FG and any coax cable.
The way to select this input resistor it is to put a T-splitter at the FG's output and scope short-pulse reflections there, while adjusting this input resistor until these reflections are minimal ...or use a VSWR meter with a CW sine signal.
P.S.
To minimize distortions, the EL2009 might need a resistor to ground on the output pin in order to force it to always source some small idle current. I think that is mentioned in the datasheet
Thanks verpies,
i read about an optional R2 in the datasheet, which seems connected to the minus rail, but only needed when driving multiple outputs.
Itsu
The EL2008 datasheet, which is also applicable to the EL2009, states:
QuoteDRIVING "INDUCTIVE LOADS":
To prevent damage to the EL2008 when the output kicks beyond the supplies it is recommended that catch diodes be placed from each supply to the output.
So a pair of fast Schottky diodes from the output to the supply rails is a good idea, too. Reverse-biased, of course, and rated over 2A. (40V is enough).
It is worth to keep in mind the warnings described in the "CAPACITIVE LOADS" section of this datasheet, when driving purely capacitive loads (such as driving piezos or MOSFET gates directly without a gate driver chip).
The solution to this is simple: a single resistor or an RC snubber, but you have to remember about adding them externally when driving such peculiar loads.
The "OP AMP BOOSTER" section is not applicable to your simple FG buffer, because you do not have any op-amps in it.
The only advantage of adding an op-amp to the EL2009 is that you can get more output voltage than your FG can output.
Watch
this video (https://www.youtube.com/watch?v=_o4ScgRZtNI) if you ever want to play with op-amps.
Hi Itsu
I forgot i had a half working day to day so managed to box up and ship the controller.
The circuit is here
http://www.overunityresearch.com/index.php?topic=31.0
and even a video on it's capabilities and how to use it
http://www.youtube.com/watch?v=tSQFbPGvcgE
verpies,
I did notice the "pair of fast Schottky diodes from the output to the supply rails" when looking into the data sheets of the ELxxxx series yesterday.
I will add them to the circuit.
Thanks for the video on the op amps, it is pleasantly simple explained.
Peter,
thanks for that, it seems a very nice piece of equipment.
I will take a further look into that thread as it seems to have some additional info.
Itsu
Updated EL2009 FG output protection circuit:
Itsu
Looks well but I'd add 10μF SMD tantalum caps in parallel with C1 and C2. You can get away without them if the power supply wires are very short and thick. Tight dead-bug construction is preferable for RF.
I wonder how much R1 will turn out to be.
P.S.
The ± power supply for that thing is much larger than the entire device, isn't it? ;)
I am assuming you made a dedicated ±16V linear power supply for it.
Yes, these tantalum caps (no SMD) where also planned to be installed if needed.
The ± PS is indeed much larger, its a 8A ± 24V circuit board with a bulky toroid transformer.
Not sure yet if i will dedicate it to the EL2009's (adjusted to ± 16V) or that i use the ± 24V to feed into 2 separate LM337/LM317 regulators adjusted to ± 16V.
Presently that ± PS is feeding a whole range of fixed + and - regulators so i have a lot of fixed + and - voltages available for quick test purposes.
I should have (but you never know around X-mass) all the parts the coming week.
Itsu
Quote from: Peterae on 2016.12.09, 17:17:23
Hi Itsu
I forgot i had a half working day to day so managed to box up and ship the controller.
The circuit is here
http://www.overunityresearch.com/index.php?topic=31.0
and even a video on it's capabilities and how to use it
http://www.youtube.com/watch?v=tSQFbPGvcgE
Peter,
i received your digital controller box yesterday, thanks.
Its a very nice piece of electronics / programming.
I put it on 24V and i understand the lower SMA center pins carry the pulse, while their upper counterpart SMA center pins carry the return signal, right?
I increased the frequency to see more then 1 pulse on the scope and display the both signals, see screenshot.
But as the phase setting on the box was still at 0.00nS (see picture) i would expect to see the both signals in phase, but they are not.
It that to be expected or do i need to sync those signals somehow.
Itsu
Hi Itsu
Thanks & glad it arrived OK.
That is right, 1 sma is gnd and the other is the signal, just my way of isolating the pulse from the case ground, i did have fet modules that screwed onto those sma's.
You are indeed correct about an offset delay, i'm trying to remember the full story but it's basically inbuilt into the delay chips, i was going to alter the software in an update to negate this but never got to v1.1 O0
OOPS i forgot to say i have the knob and back in a box in my loft but because i had very little time to wrap and send it, it went without them, but should not really be a problem.
Peter,
thanks for the info, the blue signal is leading the yellow by 11nS, so i will call the blue (right hand side of the box) the primary.
No problem with the knob and back, i will use a piece of double sided PCB to close it up and a knob will be available somewhere here.
I was puzzled by the "storing" line on the display, i know it means "to store" in English, but in plain Dutch (my primary language) it means "failure".
Toke me some time to make the switch :o
Itsu
Quotei know it means "to store" in English, but in plain Dutch (my primary language) it means "failure".
LOL
it would work fine without the bottom on as the electronics is enclosed and facing the top of the case by using a double sided pcb.
Quote from: Peterae on 2016.12.16, 16:32:37
You are indeed correct about an offset delay, i'm trying to remember the full story but it's basically inbuilt into the delay chips, i was going to alter the software in an update to negate this but never got to v1.1 O0
That's right. These delay chips require a calibration to be done in software.
If you have the sources and a matching compiler, you can just hardcode the equalizing delay by adding a small integer in the IO routine. This way you will create a firmware that will be suited only to your box with minimal effort.
Quote from: Itsu on 2016.12.10, 15:39:34
Updated EL2009 FG output protection circuit:
Itsu
Ok, i completed this 2009 protection circuit for my FG.
I added 10uF tantalum caps at the + and - 16V lines to ground.
I was able to adjust (when powered off) the input impedance to give almost zero reflection by setting the parallel 100 Ohms pot to 9 Ohm (both sides the same).
I think this is kind of low, when powering on, i see many signals appear, which looks like oscillations to me.
Does there need to be a physical grounded barrier between the input and output?
Anyway, the buffer seems not to work as i do not see any output signal when applying ±16V and a 10Vpp sine/square wave signal.
According to the EL2009 data sheet, this should be the correct layout, see picture.
Removing the schottky diodes shows some signal, but less then the input, so i wonder of the above mentioned pin layout is the correct one.
The EL2009's get fairly warm to the touch after a while.
Itsu
Quote from: Itsu on 2016.12.17, 20:06:23
I was able to adjust (when powered off) the input impedance to give almost zero reflection by setting the parallel 100 Ohms pot to 9 Ohm (both sides the same).
I think this is kind of low,
It seems wrong
Quote from: Itsu on 2016.12.17, 20:06:23
Does there need to be a physical grounded barrier between the input and output?
No.
Quote from: Itsu on 2016.12.17, 20:06:23
Anyway, the buffer seems not to work as i do not see any output signal when applying ±16V and a 10Vpp sine/square wave signal.
Could you have accidentally swapped the ±16V supply voltages ?
Quote from: Itsu on 2016.12.17, 20:06:23
According to the EL2009 data sheet, this should be the correct layout, see picture.
If you have doubts, then:
With the EL2009 completely out of the circuit, measure the supply pins like you would measure diode's voltage drop with a multimeter. The larger drop indicates the correct polarity and the smaller drop indicates the wrong polarity. (mine has 1.95V and 0.76V, respectively).
Also, measure the voltage drop from the +V supply pin to the input pin, as well as from the +V supply pin to the output pin, when reverse biased (that is: with the negative meter lead connected to the +V supply pin in both cases).
The larger voltage drop indicates the output pin ( mine has 0.930V vs. 0.735V ) . For the cause of this difference see the red path and the blue path on the attached schematic - the extra built-in blue resistors are responsible for the increased voltage drop.
Also the negative multimeter lead indicates the input pin, when the voltage drop measured between the input pin and the output pin, is larger (mine has 1.45V vs. 1.36V ).
Quote from: Itsu on 2016.12.17, 20:06:23
Removing the schottky diodes shows some signal, but less then the input, so i wonder of the above mentioned pin layout is the correct one.
The EL2009's get fairly warm to the touch after a while.
Without a load?
Does it draw around 25mA without a load and with the input grounded ?
P.S.
Thus buffer has an under unity voltage gain so you can expect the output voltage amplitude to be 90% of the input. ...but the current gain is huge.
QuoteCould you have accidentally swapped the ±16V supply voltages ?
Hmmm, i checked it hundred times now, seems ok, unless the layout of the pins as shown above is wrong.
QuoteIf you have doubts, then: .......
Ok, great, i will remove the both EL2009's and make some measurements.
QuoteWithout a load?
Does it draw around 25mA without a load and with the input grounded ?
It draws about 800mA for the 2, so 400mA / EL2009, input was set at 9 Ohm with the pots, no loads
QuoteP.S.
Thus buffer has an under unity voltage gain so you can expect the output voltage amplitude to be 90% of the input. ...but the current gain is huge.
Ok, it seems one side has that when at sinus wave, when switching to square its all oscillations i see.
The other side seems to be dead, i found the 2 schottky diodes shorted causing the ± PS to blow its 1A input fuse.
These are 40V 2A diodes!! (SB240-E3/E4)
Anyway, work to do, thanks for the tips.
Itsu
Meanwhile playing with Peters magic box.
I have hooked up the both outputs of this box to the little white noise PCB without the two Pics, so driving the IXDD604 MOSFET driver.
The 2 MOSFETs are running on 24V from a battery stack and are connected to the special coils.
When setting the frequency of the box to around 200KHz and a pulse width of 36.5ns and a delay between the pulses of about 220ns (including the default 11ns)
i get again the crackling.
Well its more a single crack which makes the double neo magnets jump and the scope shows a HV glitch on both drain signals.
The 24V battery current meter jumps to the 10A end of scale.
It does influence Peters box as it jumps between settings, like it goes into INC mode, or jumps through the selection lines.
Guess i have to close the back side.
Screenshot 1 shows the both MOSFET gate signals (delta 220ns).
Other 2 screenshots are from the MOSFET drains when crackling occures.
Itsu
Hi Itsu
I found this signal generator quiet resistant to noise from the crackling, is the power supply separated from the signal generator and the fets with just the 0v line connected between both or are you using your dual channel psu?
the other thing during my tests i scoped across the driven coils not the fet source drain but you have to be careful about earths, my scope was battery driven from a laptop so totally isolated from the earth of the psu's, you would need to do it this way to see the big pulse appear across the coils.
Are you driving 2 bifilar coils or are both fets driving 1 coil, you will get cracks using a 2 coil bifilar but it gets a lot more interesting when driving 1 coil with both fets but beware that fets can start popping.
Peter,
i have 3 power sources running this setup, your box (24V) and the white noise PCB (30V) from my dual PS (which has isolated outputs, also from ground) with the return lines connected.
Then 24V from 2 series batteries to the drains of the MOSFETs, again with the return connected to the other return lines.
I can scope across the driven coils, i use only my HV probe there, so no problem with earths that way.
I use both MOSFETs to each drive 1 coil, but i will try to drive 1 coil with both MOSFETs.
The IRFP460's can handle 500V and up till now i did not see a such high voltage.
Thanks, itsu
QuoteIf you have doubts, then:
With the EL2009 completely out of the circuit, measure the supply pins like you would measure diode's voltage drop with a multimeter. The larger drop indicates the correct polarity and the smaller drop indicates the wrong polarity. (mine has 1.95V and 0.76V, respectively).
Also, measure the voltage drop from the +V supply pin to the input pin, as well as from the +V supply pin to the output pin, when reverse biased (that is: with the negative meter lead connected to the +V supply pin in both cases).
The larger voltage drop indicates the output pin ( mine has 0.930V vs. 0.735V ) . For the cause of this difference see the red path and the blue path on the attached schematic - the extra built-in blue resistors are responsible for the increased voltage drop.
Also the negative multimeter lead indicates the input pin, when the voltage drop measured between the input pin and the output pin, is larger (mine has 1.45V vs. 1.36V ).
Measurements done, i think the EL2009's are OK and also the layout pins as was marked earlier above.
I have drawn in the measured values see picture below.
Using my Fluke 179 DMM in the Diode measurement setting, red dot = red lead, black dot = black lead.
I have the back of the EL2009 (connected to pin 3 NC) screwed to the case which is at zero level, guess thats OK.
Itsu
By the way, here i envision how to use the protection box with the FG picture 1
And the inside of the now gutted box, so severall components need to be added, see picture 2.
Itsu
Quote from: Itsu on 2016.12.18, 21:11:27
Measurements done, i think the EL2009's are OK and also the layout pins as was marked earlier above.
I have drawn in the measured values see picture below.
Nice hand drawn depiction of measurements.
They seem fine except the in<->out pin measurements, because you have a 445mV difference there, while I have only a 9mV difference ( that's 49 times less! ).
Quote from: Itsu on 2016.12.18, 21:11:27
I have the back of the EL2009 (connected to pin 3 NC) screwed to the case which is at zero level, guess thats OK.
There can be manufacturing differences in the Tab & pin3 connections. Make certain with your multimeter that there is indeed no conduction from the tab to any of the other pins.
Quote from: Itsu on 2016.12.18, 21:51:42
By the way, here i envision how to use the protection box with the FG picture 1
Looks well. I made 2 separate boxes (tubes really).
Quote from: Itsu on 2016.12.18, 21:51:42
And the inside of the now gutted box, so severall components need to be added, see picture 2.
There seem to be construction errors. The beads (or any chokes) should be before any caps (including feedthrough caps) and caps should be the last ones, like this: PS-->Bead-->Cap-->...-->Bead-->Cap-->EL2009
If beads (or chokes) are the last ones before the EL2009 then they will ring and the resulting HV from resonant rise will damage the Schottky diodes and possibly other stuff. Also, to do their job well, ferrite beads need to be loose (not mechanically constrained nor squeezed).
P.S.
Please make higher resolution photos at different angles. 30º variance is enough.
Quote
Nice hand drawn depiction of measurements.
They seem fine except the in<->out pin measurements, because you have a 445mV difference there, while I have only a 9mV difference ( that's 49 times less! ).
thanks, so does that mean that there is something wrong? Seems not as both EL2009's have the same.
Anyway, i have some more coming, so i could measure those too.
QuoteThere can be manufacturing differences in the Tab & pin3 connections. Make certain with your multimeter that there is indeed no conduction from the tab to any of the other pins.
yes, i did that, and no connection to the other pins was seen.
But i feel its kind of strange that only the center pins of the FG is fed into (and out of) the EL2009's, their grounds never make it to the EL2009's.
QuoteThere seem to be construction errors. The beads (or any chokes) should be before any caps (including feedthrough caps) and caps should be the last ones, like this: PS-->Bead-->Cap-->...-->Bead-->Cap-->EL2009
If beads (or chokes) are the last ones before the EL2009 then they will ring and the resulting HV from resonant rise will damage the Schottky diodes and possibly other stuff. Also, to do their job well, ferrite beads need to be loose (not mechanically constrained nor squeezed).
As the above inside box picture tries to show, the mentioned "bypass caps 100nF on power rails" are pointing to a connection between the beads and the EL2009 powerleads to ground, so your "PS-->Bead-->Cap-->...-->Bead-->Cap-->EL2009" was implemented.
I had chokes (1mH) right at the beginning after entering the box, but their 2.5 Ohm DC resistance caused a 2V or so voltage drop (due to the high current draw) so i removed them.
So it was planned like this for each ± supply: "PS-->5nF feedthrough cap-->1mH choke-->10uF tantalum cap-->1nF feedthrough cap-->bead-->100nF bypass cap-->EL2009"
By the way, the beads taken of from some PCB's are having some kind of substance inside their holes which kind of sticks to the wire, i guess to prevent any noise.
I will take some highres pictures lateron.
Itsu
I put the EL2009's back in again, but still without the schottky diodes.
Current pulled on both + and - supply is 120mA (±16V, input grounded, no load)
Seems 1 side is working allthough only 2.5Vpp out when 5Vpp in, but i still need to adjust the input impedance (set at 50 Ohm now).
Sine wave and triangular are steady, square wave signal blinks on / off, like its being protected against shorts.
The other side shows no output........perhaps also need the input inmpedance adjusted.
Some pictures shows the inside.
I have no chokes installed, so PS-->4nF feedthrough-->10uF tantalum-->1nF feedthroughs-->bead-->100nF cap-->EL2009
Itsu
Quote from: Itsu on 2016.12.19, 11:16:55
thanks, so does that mean that there is something wrong? Seems not as both EL2009's have the same.
Anyway, i have some more coming, so i could measure those too.
yes, i did that, and no connection to the other pins was seen.
But i feel its kind of strange that only the center pins of the FG is fed into (and out of) the EL2009's, their grounds never make it to the EL2009's.
As the above inside box picture tries to show, the mentioned "bypass caps 100nF on power rails" are pointing to a connection between the beads and the EL2009 powerleads to ground, so your "PS-->Bead-->Cap-->...-->Bead-->Cap-->EL2009" was implemented.
I had chokes (1mH) right at the beginning after entering the box, but their 2.5 Ohm DC resistance caused a 2V or so voltage drop (due to the high current draw) so i removed them.
So it was planned like this for each ± supply: "PS-->5nF feedthrough cap-->1mH choke-->10uF tantalum cap-->1nF feedthrough cap-->bead-->100nF bypass cap-->EL2009"
By the way, the beads taken of from some PCB's are having some kind of substance inside their holes which kind of sticks to the wire, i guess to prevent any noise.
I will take some highres pictures lateron.
Itsu
Good day Itsu
Just wanted to say I really like the F.G. isolation box you made. Great job!
It's something that every experimenter should have on his bench. Maybe when you get the chance you could post a BOM for the complete build.
Thanks in advance
Happy Holidays
take care, peace
lost_bro
Quote from: Itsu on 2016.12.19, 22:16:29
Current pulled on both + and - supply is 120mA (±16V, input grounded, no load)
Take a look at the "I
S: Supply Current" in the datasheet.
Quote from: Itsu on 2016.12.19, 22:16:29
I have no chokes installed, so PS-->4nF feedthrough-->10uF tantalum-->1nF feedthroughs-->bead-->100nF cap-->EL2009
It seems fine. A bead acts like a small choke.
Quote from: Itsu on 2016.12.19, 11:16:55
But i feel its kind of strange that only the center pins of the FG is fed into (and out of) the EL2009's, their grounds never make it to the EL2009's.
You have good instincts.
The EL2009 treats ½ of the supply voltage between +V and -V as the virtual ground level at 0V.
So to apply the virtual 0V to the input pin you should form a ½ voltage divider out of two equal resistors connected between the +V pin and -V pin.
If the ±16V supply voltages are well balanced then their common terminal also forms the 0V.
Quote from: Itsu on 2016.12.19, 22:16:29
The other side shows no output........perhaps also need the input impedance adjusted.
That asymmetry is worrisome. If both channels are built identically then they should work the same way. Something must be broken.
Quote from: Itsu on 2016.12.19, 22:16:29
Seems 1 side is working allthough only 2.5Vpp out when 5Vpp in, but i still need to adjust the input impedance (set at 50 Ohm now).
If you scope the input signal at the input pin of the EL2009, what levels do you get with your various input terminations?
What output levels do you get when you apply a ±5V signal to the input pin of the EL2009, without any input termination (no pot/resistor to ground)?
Please put at least a 1kΩ load on the output of the EL2009 when you do the output measurements.
Quote from: lost_bro on 2016.12.20, 00:58:28
Good day Itsu
Just wanted to say I really like the F.G. isolation box you made. Great job!
It's something that every experimenter should have on his bench. Maybe when you get the chance you could post a BOM for the complete build.
Thanks in advance
Happy Holidays
take care, peace
lost_bro
Thanks Lost_bro, but all credits go to verpies, its his idea and design, again here i am only replicating :)
When it is working to my liking i will put up all the info allthough all what is needed is already available above.
Happy Holidays too O0
itsu
QuoteTake a look at the "IS: Supply Current" in the datasheet.
Parameter Description VIN Load Temp Min Typ Max Test Level Units
IS Supply Current 0 ∞ 25°C 30 45 65 I mA
Well, the 120mA was to both EL2009s, so 60mA each is within those specs.
QuoteYou have good instincts.
The EL2009 treats ½ of the supply voltage between +V and -V as the virtual ground level at 0V.
So to apply the virtual 0V to the input pin you should form a ½ voltage divider out of two equal resistors connected between the +V pin and -V pin.
If the ±16V supply voltages are well balanced then their common forms the 0V, too.
Right, that explains, i did isolate now the back of the EL2009's and the pin 3 from ground, just in case.
QuoteThat asymmetry is worrisome. If both channels are built identically then they should work the same way. Something must be broken.
yes, my idea too, allthough they measure the same (diode measurement).
I will concentrate on the working side and try to adjust the input impedance using that time domain measurement and copy the value to the other side.
QuoteIf you scope the input signal at the input pin of the EL2009, what levels do you get with your various input terminations?
What output levels do you get when you apply a ±5V signal to the input pin of the EL2009, without any input termination (no pot/resistor to ground)?
Please put at least a 1kΩ load on the output of the EL2009 when you do the output measurements.
I will do those measurements lateron today, thanks.
Itsu
I realized that this thread is not the correct one to use for this EL2009 protection box circuit so i opened a new topic here:
http://www.overunityresearch.com/index.php?topic=3388.msg58939;topicseen#msg58939
Thanks, itsu
Meanwhile i still work on Peters crackling system.
I have closed the back of the controller by screwing on a single sided PCB which was extended beyond the box to have room for some external mounted MOSFETs, see picture.
The MOSFETs (Cree C3M0065090 900V) and drivers (IXDN614PI 14A) should be able to withstand great abuse.
I drive the drivers on 20V via 2 LM350 3A regulators from a 24V source.
Still waiting for some SMA connectors which will come this week hopefully.
Regards Itsu
Looking forward to see what sort of results you get O0
I got a nasty flu which knocked me out for a week. That reminds me to DO take that flu cocktail shot they offer every end of year.
Anyway, some first tests with 3x 24v supplies (one for Peters box, one for the MOSFET drivers and LM350's (20V) and one for the MOSFET drains (2x 12V batteries)
shows no crackling, with or without magnets.
Setting of Peters box still at the proven crackling position (width 32ns, phase 220ns), so i have to do some adjustments there and / or increase the drains voltage.
Screenshot is from the both MOSFETs gate signals
Itsu
Quote from: Itsu on 2017.01.11, 21:25:01
I got a nasty flu which knocked me out for a week. That reminds me to DO take that flu cocktail shot they offer every end of year.
I consider the many injections I have received over the years to be detrimental to my health.
One time, a few years into soldiering they dosed me up right before Christmas leave. I got the flu xmas eve went to bed, got up January 2nd to go back to work. 23 hours a day in bed with Fever, completely saturated sheets.. dripping wet. I nearly died, I don't do any shots anymore and have not for 10 years and my health has improved year on year.
We all make choices, only some of us know that, sometimes.
:)
Quote from: evolvingape on 2017.01.11, 21:45:26
I consider the many injections I have received over the years to be detrimental to my health.
I do, too.
As far as flu goes, I read somewhere that the ear canal is a reservoir for the virus during the initial stage of infection. The immune system cannot reach the virus there (unlike the nasal cavity where mucus is weaponized in response to the virus). The virus mutates in the ear canal and launches repeated attacks from there to the nasal cavity and the respiratory track, until a working mutation is found.
At the very onset of the infection, it is enough to disinfect the ear canal with any antiseptic (H
2O
2 is my choice) to throw a monkey wrench into this mechanism. I had only one flu in 14 years and that was when I was on the road and did not disinfected it early enough.
Thanks for your comments guys, i know there is a lot of controversy about taking the flu shots (any shots for that matter) and the same seems to be true
for the H2O2 antiseptic for cleaning your ears (google).
The mean problem with the latter i think is to know when the ear canal is infected and thus when to disinfect.
Itsu
itsu glad your feeling better
If I may ?
I was unaware of Verpies knowledge in this area and would like more info ,it would seem a good plan for the virus and convenient arrangement
to launch attacks from .
can these guys be viewed in a home microscope ?
my grandkids suffer from endless ear infections and head colds ,really horrible for them.
makes me wonder if the cold virus can hide there too ?
perhaps a slide sample would show this virus under a scope ?
would be a big aid in the war on Virus and a big drop in antibiotics for the kids if alternative treatment protocol could be worked out ?
Note
will remove this post shortly
Quote from: Chet K on 2017.01.12, 17:55:07
makes me wonder if the cold virus can hide there too ?
Most certainly
Quote from: Chet K on 2017.01.12, 17:55:07
can these guys be viewed in a home microscope ?
No way!
Take a look at the scale of this virus photo...and the flu & cold virii are even smaller (and immune to antibiotics)
Another curiosity about the human ear canal is that it is immune to ear mites. Did you ever wonder why your cat gets them and you don't ?
This is because the human ear wax contains a natural pesticide (that's what gives it its bitter taste) - the cat doesn't have it.
Unfortunately it is harmless to virii.
Increasing the drains (bifilar coils) voltage from 24V to 42V still does not return the crackling sounds.
A mild whispering is all i can detect together with some shown oscillations on the scope.
I tried all kinds of combinations of frequency, pulse width and delays.
Could it be that the present setup with 900V Cree MOSFETs and 14A drivers is to beefy for the effect to occure (like avalanching or so)?
Itsu
Hi Itsu
I never tried the Cree Sics they do have quiet a high on resistance but not sure if that is enough to stop the crackling, do you see the large pulse appear when you do a sweep on the delay.
First screenshot shows the gate signals (yellow and Blue) of the 2 MOSFETs (220ns delay), and the drain signals (green and purple) of the same MOSFETs with attached bifilar coils.
To be noticed is that allthough the gates are switched 220ns apart, the drains show a timed switching.
I would have expected that we would see also a delayed switching on the drains (220ns apart).
(yellow and green are MOSFET 1, blue and purple MOSFET 2).
Second screenshot shows more delayed gate signals (700ns apart) and again the timed drain signals.
So the delayed gate signals do not translate directly to a delayed drain signal, is it the tramsformer function of this bifilar coil causing this?
Itsu
Quote from: Peterae on 2017.01.13, 20:15:13
Hi Itsu
I never tried the Cree Sics they do have quiet a high on resistance but not sure if that is enough to stop the crackling, do you see the large pulse appear when you do a sweep on the delay.
you think 0.065 Ohm is quite high? I am sweeping (+15), but no pulse is seen up till now.
Itsu
Quote from: Itsu on 2017.01.13, 10:12:59
Could it be that the present setup with 900V Cree MOSFETs and 14A drivers is to beefy for the effect to occur (like avalanching or so)?
Yes it could be.
Also, the drain pulse shape can be different (and its frequency content)
Hmmm, i went back to using the 2x IRFP260N MOSFETs driven my a single dual MOSFET driver IXDD604 with 42V on the drains, but no more crackling is heard here too.
It looks like boxing up Peters box has tamed the crackling as that is the only change i did.
I will continue to do some sweeping, but up till now no crackling pulses / oscillations are noted.
Itsu
Bump.
since i boxed up Peters box as mentioned above, i never was able to create the crackling sounds i had.
I will go over this setup again to see why that is.
For now i have again the earlier setup using Peters box (Freq. 200Khz, Pwidth 64ns (max.), phase 220ns).
I have 18V on the MOSFET driver (dual output IXDD604) and 25V (2x batteries) on the drain of ONE MOSFET.
I use only ONE MOSFET now to see how this ONE MOSFET behaves
The other MOSFET has no drain voltage nor is the MOSFET driver input connected.
I use the max Pwidth from Peters box to get the max gate signal.
The both MOSFETs are loaded with a 100 Ohm resistor (so no (bifilar) coil).
The screenshot below shows in yellow the drain voltage of the working MOSFET and in purple the gate signal of
this MOSFET.
We see that the MOSFET opens quite fast/good, but it closes very very late.
We also see that the gate signal (purple) is nowhere near the 18V that the driver is running on.
This causes the MOSFET to linger in off/on state to long i guess.
The Pwidth of Peters box (64ns) has been shown by the vertical cursors.
I think i have to improve on this simple resistive setup before going into the inductive bifilar coil setup.
Why does the gate signal stay on longer then the driving pulse (108ns versus 64ns)?
When using the required 32ns Pwidth, the gate signal is only about 7V, how to improve this to 18V (which is the driver Vcc)?
How to improve the MOSFET close time (adding a schottky diode across the gate resistor?)?
MOSFETs are irfp460's
I added the schematic from Peters PCB i am using showing the IXDD604 dual driver and MOSFETs setup. (be aware that i do NOT use the PIC white noise chips as input, instead
i use Peters magic box)
Thanks, Itsu
Try measuring what this type of MOSFET does with these kind of pulses, like this:
Hi Verpies,
It is good to see you around again, hope you are doing fine.
As an addition to your suggestion, I would like to add the followings to Itsu:
Hi Itsu,
I assume you checked the output pulse from the IXDD604 when the MOSFET gate is not connected, is this so? Did you find the output pulse ok? as its pulse width should be, coming from the box?
If the drive pulse coming from the IXDD is okay in itself, then perhaps the IRFP460 is to blame?
Otherwise, there is a method to improve MOSFET switch off time by using a pnp transistor and a diode between the gate-source, see this link:
http://www.imajeenyus.com/electronics/20111010_40-400V_supply/index.shtml (http://www.imajeenyus.com/electronics/20111010_40-400V_supply/index.shtml)
It shows a different driver IC which has an open emitter output hence the need for the 470 Ohm pull down resistor. Your IXDD should discharge the gate source capacitance without long delay, it is designed to sink current, so normally such speed up circuitry should not be needed for such IXDD type driver.
Anyway, such speed up circuit should help reduce the speed off time for the MOSFET.
Maybe the unconnected driver IC input ought to be connected to the ground if you left it floating, not to disturb the operation of the driver chip.
Gyula
Hi verpies,
nice to see you, and thanks for the diagram (Hmmm, where did i see that before?).
https://overunity.com/12736/kapanadze-cousin-dally-free-energy/msg342647/#msg342647
I thought that what we see in the first 60ns of the gate (purple) signal was due to the Miller effect.
Guess it will be hard to overcome with Peters small (32ns) pulses.
Let me see what i can do with your diagram, i still should have those source/sink output drivers (ucc27511) somewhere.
Regards itsu
Quote from: Itsu on 2018.10.20, 08:39:40
Let me see what i can do with your diagram, i still should have those source/sink output drivers (ucc27511) somewhere.
They are not mandatory. In that circuit, you can use your single output driver, too.
Gyula,
i did check again the output of Peters box, and they are still the same as tested before in post #270 above, see here:
http://www.overunityresearch.com/index.php?topic=272.msg58828#msg58828
I think they are good for the job.
I did use the IRFP460 MOSFETs before and they did produce the crackling sounds, but i think Peter used IRF840's, so i could try them.
Thanks for the link, i will take a look tonight.
Good idea to tie the unused IXDD604 output to ground.
Itsu
Quote from: verpies on 2018.10.20, 08:48:22
They are not mandatory. In that circuit, you can use your single output driver, too.
Ok, Roger that, even better.
Itsu
Quote from: Itsu on 2018.10.20, 08:49:55
...
Good idea to tie the unused IXDD604 output to ground.
Itsu
Dear Itsu,
I wrote input, not the output. Leave the unused output unconnected.
Gyula
Input was i meant to write O0
Initial test done with verpies his diagram as shown below.
input pulse is 100ns wide
driver is ucc27425 (using non-inverted output only, running on 16V (as is max).
single R1/R2 of 3.3 Ohm
2x IRFP460 with 42V on their drains.
200Khz pulse repetition frequency.
Screenshot shows:
yellow: input pulse from FG
blue: gate Q1 signal
purple: drains Q1/Q2 signal
green: output across R4
Not sure where/what to look for.
I have some IXDD614PI drivers which can withstand up to 35V which i could use.
Itsu
Made some extra screenshots with varying input pulse widths (20, 50, 100 above, 200ns).
Last one is a zoomed out screenshot showing the massive ringing.
Itsu
These MOSFETs have a large gate charge.
Could you make a video showing how these waveforms are changing as you are varying the R4 (make it a pot) ?
Yeah, like 4.2nF (Ciss 4200pF) according to the datasheet.
I will change R4 for a 5K pot and vary it showing the waveforms.
Not sure i will get around it to do it today as i seem to have some other obligations :)
Itsu
I used the same setup as yesterday with a 100ns wide pulse.
38V on the drains, 16V on the driver.
Potmeter (5K) was initially set to 1K, and went down in 4 full turns from 1K to 800 Ohm, 580 Ohm, 360 Ohm, 200 Ohm.
Then up to 1K again and from there in 4 full turns up till about 2K.
No change was seen, so i turned it down even further untill a change was noticed which was when the pot went to 0 Ohm.
Video here: https://www.youtube.com/watch?v=gSlTVd-gXgg
Regards itsu
i realize that the video of the screen was not that clear, so i redid the tests with a lowered yellow amplitude
at the 4 positions of the pot and at 0 Ohm (1K, 800, 580, 360, 200 Ohm and 0 Ohm), see screenshots.
Itsu
At 0Ohm the Ch4 should flatline.
Also, it looks like the impedance of the driver is too high to drive the gate stiffly enough.
There is a lot of ringing. How much inductance is between the driver and the MOSFET ?
...and why is the Ch1 signal so wobbly? It is the input from the FG, isn't it.? What's affecting it?
I can use another driver then the now used ucc27425, but then i need to remodify the used pcb i am using now
or build a new setup.
The below layout is from the PCB used (modified), so there is some 1.5cm long link between the 3.3 Ohm
gate resistor and the Q1 gate.
The input to the ucc27425 driver is indeed from the FG, below 1st screenshot is that signal without power to the driver.
2th screenshot is when i put 16V to it.
Itsu
Something is not right. The power supply rails to the driver appear to be unstable.
I recommend your old ground-plane design in which you soldered to an unetched PCB and the driver was soldered directly to the gate terminal and separately choked and bypassed with a dedicated bank of different caps (each cap has a different self-resonance frequency)
The driver must be able to source and sink many amps, to provide stiff gate drive. After the gate goes up, it should stay up despite the Miller current trying to pull it down. The opposite goes for the case when the driver is pulling the gate down. The gate resistors should be experimentally tailored for the MOSFET to be as low resistance as possible without allowing gate ringing and without overloading the driver.
Hi verpies,
i understand and agree, but....... Peter has seen these explosions / crackling effects using the PCB i am using
which has the layout i have added above, meaning without all these precausions you just mentioned.
(it off course verywell could be that thats the reason he saw them as after boxing up Peters box my crackling sounds stopped).
Anyway, this Miller effect test setup i have now, using the same PCB from Peter, obviously is not ideal so i will be
building one which does have these precausions you pointed out.
When successfull, i will incorperate these into a new PCB driven by Peters box and driving the Bifilar coil setup.
Thanks, Itsu
Quote from: Itsu on 2018.10.22, 14:26:15
(it off course verywell could be that thats the reason he saw them as after boxing up Peters box my crackling sounds stopped).
If it is Meyer related, then perhaps the external magnetic field on your bench changed after the boxing.
Meyer, as in Stan Meyer? Not sure, perhaps Peter knows.
Hi Itsu.
I think Verpies is referring to the Meyer/Mace device.
https://tesla3.com/meyer-mace-michel/
Cheers Graham.
Aha Graham, thanks, i do see some similarities.
Itsu
I decoupled and choked the hell out of my voltages, but still i got the ripples on the FG input signal and
Q1 gate signal after applying voltages to the setup.
I did notice that the ripple started when attaching the scope probe, so i went over using these RF like probe tips
see picture.
Now the signals cleared up a fair bit, see screenshot from REF1 (FG input signal) and REF2 (Q1 gate signal).
I used the white REF signals as its a 4 hands business now to probe 4 channels and i thought i could do it pair
by pair, but i lost the trigger point, so the signals (pair by pair) are not in sync anymore.
Are there any specific 2 signals you want to be shown?
Else i need to come up with a trick to use the 4 probes with RF tips all together.
I now lost the 10 ohm resistor, so need to find a higher wattage one.
Regards Itsu
8)
Aaah, the third hand, nice trick O0
I was able to use my 4 probes with RF tips to capture:
yellow: FG input
blue: Gate Q1
purple: Drains signal
green: Gate Q2 @ 1K load.
Using a IXDD614PI at 20V and 42V on the drains of the IRFP460's.
1st screenshot is at 40ns/Div,
2th screenshot is at 2us/Div.
Itsu
That was a difficult probing setup.
Lets see how it looks when the load of Q2's gate = the driver's impedance that drives Q1's gate.
...but before that, it is unacceptable that Q1's gate signal (blue) does not follow the yellow input signal from the FG (40ns of propagation delay is fine but the lack of ANY inkling of a blue falling edge 80ns later - is not).
There is an inkling of a blue falling edge over 200ns later, but that is not only a huge asymmetry in the propagation delay between rising & falling edges, but >200ms is an unacceptably long delay, by itself!
So what you are saying is that the "inkling of a blue falling edge over 200ns later" really should be there after only 80ns.
So it seems that the MOSFET (gate) is somehow smearing out the signal it receives.
If that's due to the IRFP460 Ciss of 4.2nF, then i could try a FQA11N90C which only has a Ciss of 2.5nF (Typ).
Itsu
As mentioned earlier in another thread (http://www.overunityresearch.com/index.php?topic=3655.msg69985#msg69985),
i received some crappy IXDD614PI driver chips from China which could not withstand the 40V Vcc (magic smoke at 27V).
They now turned out to also be crappy in other area's like the smeared out gate signal shown above.
I received a new batch IXDD614PI's from Mouser and not only they can withstand the 40V Vcc, but the Gate signal looks
very different from the crappy ones, see screenshot 1 below.
In white the FG input signal (Ref1) and Q1 gate signal (Ref2) from the crappy one, and yellow / blue the same signals from the good one.
We now do see an "inkling of a blue falling edge 80ns later".
The 4 signals together with the good driver now show as in screesnhot 2 and 3
yellow: FG input
blue: Gate Q1
purple: Drains signal
green: Gate Q2 @ 1K load.
Using a IXDD614PI at 20V and 42V on the drains of the IRFP460's.
Itsu
Reducing R4 from 1K to 222 Ohm kind of flatlines the green trace (across r4) around zero before and after the Q1 gate signal, see screenshot.
Guess thats "how it looks when the load of Q2's gate = the driver's impedance that drives Q1's gate".
Itsu
Same setup as above, but now with 2x FQA11N90C MOSFETs in (R4 = 1K).
It is amazing how crappy the previous drivers were!
Quote from: Itsu on 2018.10.25, 14:39:27
We now do see an "inkling of a blue falling edge 80ns later".
Indeed we do
Quote from: Itsu on 2018.10.25, 15:21:40
Reducing R4 from 1K to 222 Ohm kind of flatlines the green trace (across r4) around zero before and after the Q1 gate signal, see screenshot.
Guess thats "how it looks when the load of Q2's gate = the driver's impedance that drives Q1's gate".
I am surprised, that the driving impedance is around 220Ohm. I would expect teen or single Ohms for these tens of nanoseconds wide pulses.
If R4 is too large then Q2 latches on until its gate charge leaks out.
It is good that the green trace (across R4) flatlines. This trace should look approximately like the 1st derivative of the drain voltage (purple trace) - it shows the charge injected into the gate by the Miller current....for Q2 and for Q1.
QuoteIt is amazing how crappy the previous drivers were!
yes, i am amazed too, this is the first time i notice something is wrong with parts.
Finding out that they blow up on half their rated Vcc voltages is one thing, but it takes a knowledgeable person to
detect such smeared out propagation as being unacceptable, well done verpies! O0
So that 220 Ohm R4 resistance means that the output inpedance of the driver is also around 220 Ohm?
The data sheets talk about output resistance of ±0.4 Ohm, so that 220 impedance comes ontop of that.
So in the original circuit from Peter, we need to match that 220 Ohm output impedance of the driver to the input
impedance of the gate somehow, right?
Itsu
Quote from: Itsu on 2018.10.26, 19:20:50
So that 220 Ohm R4 resistance means that the output inpedance of the driver is also around 220 Ohm?
The data sheets talk about output resistance of ±0.4 Ohm, so that 220 impedance comes ontop of that.
Yes, 220 ±0.4 would be the total
...but I am kinda lost where that 220 Ohm comes from. I did not read the entire thread but the gate resistors (R3 and R4) in
this schematic (http://www.overunityresearch.com/index.php?topic=272.msg70029#msg70029) seem to be 3 Ohms.
3 ±0.4 would make much more sense at these pulse widths...
Yes, the R3 and R4 are around 3 Ohm.
The 220 Ohm comes from me measuring the green trace flatlining roughly to zero (from 1K).
It very well could be that when even further decreasing (to around 3 Ohm) the green trace will flatten out even more
or shape more according to "like the 1st derivative of the drain voltage (purple trace)".
I can check on that.
How about the capacitive reactance of the 4.2nF Ciss?
Could it not add to the total impedance?
According to: http://www.66pacific.com/calculators/capacitive-reactance-calculator.aspx
a 4.2nF capacitance (Ciss) at 170KHz has a reactance of 223 Ohm.
Can this 170Khz somehow be related to the gate signal pulse (rise / fall times) we have?
Itsu
Quote from: Itsu on 2018.10.27, 08:54:30
a 4.2nF capacitance (Ciss) at 170KHz has a reactance of 223 Ohm.
For pure sine wave - yes, but for square waves the reactance is 19% smaller ...and for rectangular waves it depends on the duty cycle.
The purpose of the Miller test circuit is to see how a Miller current influences the gate voltage during switching, so R4 should be the same as the driving impedance of Q1's gate.
Ok, but can we do something with that knowledge (R4 should be the same as the driving impedance of Q1's gate), or is it just "nice to know"?
Itsu
Meanwhile i have been running Peters box/setup again on all kind of phase, frequency and pulse width combinations/sweeps
for the last week or so, but not a single crackle was detected.
Today i removed the back shield from the box and let it run the whole evening again in all kind of combinations, but
still not a single crackle was noted.
Not sure what to do next
Itsu
are you driving a single coil with the 2 fet channels or a bifilar with the 2 channels. ?
I mostly stuck with IRF840 Fets
Driving a bifilar with the 2 channels.
Yes, i was planning to go to IRF840's, lets see how they behave.
Itsu
I don't believe there was ever any magic to the compressed pulse produced by the bifilar (one channel delayed about 220ns).
It was just a means to produce a high voltage pulse.
Crackling and explosion sounds are exciting, but that's all. The explosion sounds with 10kv are like a rifle shot.
Thanks,
i try to figure out why suddenly after boxing up Peters box the crackling stopped.
I changed to IRF840's now and tossed out the (crappy?) IXDD604 driver for a new Mouser ordered one and things look better now (noisy).
Need some further tests.......Itsu
Nope, still nothing, running for 2 days with IRF840's and new 604 driver.
itsu
If you have the pulse compression, there are other experiments you can do besides looking for explosions and crackling.
Supposedly, you can apply the pulse to a bifilar voice coil (speaker coil) and something interesting happens.
Supposedly, magnets have a pulse rate where the magnet will come apart in tiny pieces. You might want a protective container for this.
Aside from these ideas, you can try to figure out how the TPU operated:
How do you use the pulse to create a rotating field?
Do you need to sequentially apply the pulse to coils or can you connect coils in series and space them apart on a loop collector?
Thanks for those idea's, but i do not seem to have any compression or HV peaks when monitoring the bifilar coil.
Just the normally expected pulse / ringing which now and then add up (in phase) or decrease (180° out of phase) when
sweeping the phase between the both pulses.
Figuring out how the TPU operated is a bridge to far as many much brighter people seem to have tried (or are still trying)
without results.
But i will play around somewhat more as i have some voice coils (not sure they are bifilar).
Itsu
Poynt was able to simulate the compressed pulse in Spice if that helps.
Not everyone shared their results so I would not let that deter your own efforts.
Also, a high voltage pulse will produce the same results.
So i used Peters box with the below settings (see picture), but with 100 Ohm induction free resistors as loads.
Frequency: 1.039 Khz
Pulse width: 32ns (both)
Phase: 205.5ns (offset is about 15ns, so effective phase is about 220ns, see screenshot)
The screenshot shows in:
purple and green, the both pulses from Peters box, so input to driver (phase delta 220ns @ width 33ns)
Yellow and blue, the resp gate signals so output from driver (phase delta still 220ns @ width 58ns!)
White Ref1 and Ref2, the resp drain signals (notice both show 2 Pulses each, where i expected to see only 1 per drain).
Also see the drains pulse width is >100ns.
So if the 32ns is the optimum pulse width, the question is where to measure this?
Is it the input pulse like we have now, the output/gate pulse which is now 58ns, or the drain pulse (>100ns)?
And why do i see 2 drain pulses on each drain instead of the expected one per drain?
Itsu
crackling was always a lot more prevalent with a monofilar, one wire being driven by 2 fets connected together, something else I did with the bifilar coil experiments which seemed to increase the crackling was to use moving coil meters at the supply end, the other thing I did was to have all earth wire from everywhere the same length and star connected back to the supplies.
The process for tuning for crackles was to start with the coil voltage low, sweeping the phase from 0-255nS and gradually increase the supply voltage to the fets, once the crackles start fet death is not far away, if i remember correctly the voltage was between 30V and 50V.
Do try to work with the bifilar first as the monofilar is very unstable and fet death is always not far away.
Sometimes the crackles can be there and not heard by the ear but a small capacitive michrophone will pick them up, i was recording my videos using a PC with a cheap microphone plugged into the mic port and placed near the coils.
The other way is with a magnet placed near to the coils, you feel the crackling as violent physical pulling of the magnet.
PS
The last experiment i did and did not record properly which i should have really was to look to see if i could generate the big pulse by passively delaying the pulses.
I used 1 fet channel which was switching on and off for about 50nS, i then divided the current path into 2 wires and added fast diodes in one path to delay the pulses, if i remember correctly i then drove 2 coils (bifilar) with each current path, and after much time on the bench did manage to create the big pulse seen in my videos.
Question
Do you see the big pulse appearing when you sweep, because you should at least be seeing that even if the crackles are not there.
Hi Peter,
thanks for the info, good to know, but the strange thing is that i did have the crackling, sometimes very violent,
therefor i switched to 900V IGBT's to be able to withstand it.
But somewhere along the line i lost the crackling.
When i sweep (the phase from 0-255nS) i see on the scope (across the 50 Ohm bifi resistor) the both pulses (first almost overlaid), then gradually
more and more spaced, but both have heavy ringing on them.
This ringing amplitude i see being peaked and downed during the sweep, not sure if you mean this as the big pulse.
Its in the 20-50V range.
Any idea why i see the double drain pull down signal? Could it the MOSFETs are so close they influence each other.
Itsu
Quote from: Peterae on 2018.11.03, 09:38:14
The other way is with a magnet placed near to the coils, you feel the crackling as violent physical pulling of the magnet.
Try this while scoping a coil around the magnet. Wow! Free energy!
Some may recall the Energea Celeste device the had the drive and collector coils around a magnet.
So here is the discovery video of the pulse you should be seeing, this is using an earlier controller than the one i gave you, but it used the same delay chips but only 1nS resolution and only up to 255nS is possible, this pulse must be seen to stand a chance of getting crackles.
https://www.youtube.com/watch?v=uMmtSpgAij4
and driving a monfilar with 2 fets connected and the phase set to a static delay setting, this also used a moving iron core meter in the coil feed and i was also scoping across this meter to monitor current, you can hear the effect of the crackles in the microphone.
http://www.youtube.com/watch?v=dwtPIennXP4
Re your higher voltage fets, they maybe too slow to get the effect required, but if they are fast enough you should be getting the pulse.
Also here is a sound track of a mic close to a crackling coil
https://www.youtube.com/watch?v=n6eDKuF8n44
Current in a monofilar across moving iron coil
https://www.youtube.com/watch?v=xlWxiqcwIgI
Thanks Peter,
i managed to get the crackling back allthough its rather mild.
I see the PS (41V max) jump in and out of the current limiter when the effect (oscillation?) occure.
Also with the scope probe across my 50 Ohm load resistor i see the pulse (and ringing) and the extra noise then.
Looking at your first video i do not see such a nice pulse, and the second pulse never forms into such a peak like signal
Its more the noisy signal i see in your second video.
These pulses shown there are from across the load resistor also?
Itsu
I don't use load resistors?
The pulse is there in the first video here is a snap shot from the first video and I outlined the pulse in orange
Here is a image of a monofilar setup, I would scope across the coil itself.
http://www.overunityresearch.com/index.php?action=dlattach;topic=241.0;attach=1478
I have been trying to find a diagram of how I connect a bifilar, but no luck so far, but I connect the scope earth leads to the positive side of both coils and then use the 2 channels on the drain side of each fet, its the only way you can monitor both coils at the same time.
Hmmm, no load resistor?
Here: http://www.overunityresearch.com/index.php?topic=366.msg38415#msg38415 you mention:
"shown 80 Watt anomalous Pulses when 1 turn loop was loaded with a 20Ohm non inductive resistance."
I have a 50 Ohm resistor there.
So no load resistors?
Itsu
Ah is that the experiment you are performing, I did not realize, so you have a bifilar wound around copper coated iron core coax cable, and the single turn loop of copper coated iron coax did have a load resistor yes ;)
OK, thanks. I will use a 25 Ohm inductionfree resistor as load in trying to replicate your pulse shown above.
Itsu
I realize that the video's from Peter above and the screenshot are from a different setup (no load resistor),
so its of no use trying to replicate, but the below video shows what happens when i monitor my single turn
loaded with 20 Ohm resistor during a 0-250ns phase sweep.
Freq. 200KHz
Pwidth 32ns
Phase 0-250ns
drains voltage 41V
Video here: https://www.youtube.com/watch?v=f62te4yRLXI
No crackling (or very faint) just some oscillations now and then (@ 70ns phase), no HV peaks, just some increase
in 2th pulse amplitude.
Itsu
Those videos are still what you would get across your bifilar coil, the only difference is that you have a single turn loop with a load resistor with your bifilar wound on the loop, if you scope across your load resistor your are looking for a large current pulse, but first you need to see the pulse across the bifilar coils.
OK, so i need to measure across the bifilar coils first to hunt for "the pulse".
This is tricky as there is 40V across the coils without ground.
Itsu
try to see the pulse on the bifilar coil WITHOUT the iron wire core
Yes that's right, scope both cannels across each coil, the positive of each coil should be connected together and this connects to the negative of the scope, it's the only way, but be carefull you don't have an earth connection from your power supply to the scope probe earth leads, I was using a battery powered scope and I also had a laptop based scope which of course was not even connected to mains to totally isolate from the supply
Grumpy,
Its not possible (without breaking up the bifi coil) to remove the iron wire core.
Peter,
my PS has its ground isolated from grid ground, so no problem there, just need to be carefull.
Used 2 HV probes with their ground leads to the common 40V of the bifi coils and the tips to the other end
(across the diodes so to speak).
Inverted the signals to get a positive pulse, see video here:
https://www.youtube.com/watch?v=JTPAhkgoLKI
No crackling or oscillations noticed somehow, probable due to the loads of the probes.
Freq. was 1 KHz, sweeping phase from 0-250ns, pulse width again 32ns.
Itsu
Hi Itsu
well you are getting the sharp pulse but there seems to be large noise after it, what diode type are you using across your bifilar coils
I use ultra fast diodes UF4007
Me too, also across the drain / source.
Itsu
OK thanks
I don't remember using enamelled copper bifilar wire for the coils, for the moment forget the loop with the load resistor as we need to work out why all that noise is following the sharp pulse, try another bifilar coil that's straight but try using small guage PVC wire side by side to form a bifilar 20 to 40 turns each coil, then scope across the coils, if we get that noise then we know it must be the fet stages, if the noise disappears then we know it's the close coupled enamelled bifilar wire, I would guess it's the wire as we don't want strong coupling between each bifilar coil, infact it is possible to create the pulse with 2 monfilar coils 1 foot apart, I have a video showing this somewhere.
Thanks Peter, i will see what i can do.
Itsu
Quote from: Peterae on 2018.11.03, 09:38:14
...once the crackles start fet death is not far away,
Maybe the crackling is from the transistors starting to avalanche.
Quote from: Itsu on 2018.11.03, 09:58:20
...therefore I switched to 900V IGBT's to be able to withstand it.
...and these transistors do not avalanche as easily.
Quote from: Itsu on 2018.11.03, 09:58:20
When i sweep (the phase from 0-255nS) i see on the scope (across the 50 Ohm bifi resistor) the both pulses (first almost overlaid), then gradually more and more spaced, but both have heavy ringing on them.
If this is Meyer related at all, then this ringing would be detrimental to the purity of resonance inside the metal.
Also, in such case, external magnetic fields should influence the effect.
https://youtu.be/EDyxBWXp6IU
Quote from: Peterae on 2018.11.06, 17:19:09
OK thanks
I don't remember using enamelled copper bifilar wire for the coils, for the moment forget the loop with the load resistor as we need to work out why all that noise is following the sharp pulse, try another bifilar coil that's straight but try using small guage PVC wire side by side to form a bifilar 20 to 40 turns each coil, then scope across the coils, if we get that noise then we know it must be the fet stages, if the noise disappears then we know it's the close coupled enamelled bifilar wire, I would guess it's the wire as we don't want strong coupling between each bifilar coil, infact it is possible to create the pulse with 2 monfilar coils 1 foot apart, I have a video showing this somewhere.
Ok, i made another bifilar coil that's straight but using small gauge wire side by side to form a bifilar
100 turns each coil (wire from an ethernet cable which was untwisted).
Same setup on Peters box as before, but signals look different.
Below first picture is a screenshot from the last video above showing the signals with the magnetwire coils.
The second screenshot is from todays test using the straight non magnetwire coils.
I have put up some lines showing what i think are the 2 pulses (250ns apart) while the other 2 pulses/signals are mere
reflections i think.
Video here: https://www.youtube.com/watch?v=Iqv6VW2Exrg
Itsu
Quote from: verpies on 2018.11.07, 19:30:39
If this is Meyer related at all, then this ringing would be detrimental to the purity of resonance inside the metal.
Also, in such case, external magnetic fields should influence the effect.
https://youtu.be/EDyxBWXp6IU
I had a double barmagnet (neo's) next to the bifilar magnetwire coil which has a slight attraction to the inner
copper cladded steel 1 turn coil inside the bifilar coils.
But thats the only magnetic field in the area around this setup.
I had that magnet there during all my earlier tests which did show / produced crackling and oscillations.
It now seem to have no effect at all, or very faint.
Itsu
Higher magnetic flux density (B) is not always better.
Uniformity is often more important than density, because the lack o uniformity causes decoherence.
Don't discount the Earth's magnetic field concentrated by steel furniture elements, pipes and rebar.
Also, power supply transformers can be a source of alternating magnetic fields. They leak the most when they are overloaded.
IMO the pulse repetition frequency should be as low as possible if the purity of resonance in the metal is the goal.
Pulsewidths are OK.
If distortions of the pulses are caused by classical transmission line reflections, then the usual termination techniques are the remedy.
I detect some strong 50Hz emf around my bench, so there could be a strong magnetic field too i guess.
Not sure where it comes from, probably from the amount of test equipment and power supplies.
I can test with a lower (1 Khz) prf, i know already the pulses look different (cleaner) then.
I mentioned "reflections" above, concerning the double pulse seen per MOSFET, but i think i really should
say induction, but finding out the characteristic impedance of my bifilar coil / transmission line could
prove usefull.
Thanks, Itsu
Itsu the new wire is it multistranded or single core, use multistranded.
Verpies yes it is possible that the fets were avalanching due to the size and sharpness of the pulse i suppose, but remember when you have good pulses, things around the bench start mailfunctioning / crashing,i suspect the em pulses emitted were triggering the fets gate to cause noise bursts, the crackling is another thing of debate, i know you can get very thin magnet wire to produce large magnetic fields during crackling.
see video here
http://www.youtube.com/watch?v=O8GgiEeH1W4&feature=BFa&list=ULCjmKlUHU_OQ&index=1
Peter, its multistranded.
Nice video showing the effects, i had similar effects earlier, but not lately.
Itsu
These are the signals coming out of Peters box around 220ns phase setting.
These are being fet into the IXDD604PI dual driver.
I think they will strongly contribute to the double pulse seen by each MOSFET.
Need to find a way to filter out or suppress those ringing signals as they false trigger the MOSFETs.
They are not there or much less between 0 and 180ns phase.
Itsu
is that located near the steel core wire or any other ferromagnetic material?
Quote from: Itsu on 2018.11.08, 21:20:50
These are the signals coming out of Peters box around 220ns phase setting.
These are being fet into the IXDD604PI dual driver.
I think they will strongly contribute to the double pulse seen by each MOSFET.
Need to find a way to filter out or suppress those ringing signals as they false trigger the MOSFETs.
They are not there or much less between 0 and 180ns phase.
Itsu
Hi Itsu
That does not look too good, strange the noise is after the second pulse
Try scoping across the fets supply rails at the chip.
Grumpy,
this is at the input to the MOSFET driver chip, without the steel core wire or any other ferromagnetic material or magnet near.
Peter,
Yes the noise/ringing looks delayed somehow, and its also visisble on the drains and across the bifi coil.
I will use a battery combo (24V) to power your box instead of the double PS.
Perhaps its like verpies mentioned some (delayed) reflections from the bifilar coil / transmission line.
Itsu
It looks like some sort of ground noise to me, the 2 output wires from the generator (0V and Signal out) for each channel should be connected across the driver chip and be as short as possible, so you have 4 wires (2 from each channel) going to each IX driver, I separated the driver and fet into a module so each driver and fet were totally separate circuits and each driver used a 7818 regulator to supply 18V to the driver chip with caps ect.
Ok, i am using the PCB you send me which had the 2 white noise generators (Pic's) on it (not in use now) and
which has only 1 dual MOSFET driver and the both MOSFETs close together with common ground.
So i need to build a new setup using your box and new apart paths with a single driver and MOSFET each.
Itsu
Oh I see, well I never did get round to testing that board but it does look to me that 1 chan of the fet driver is affecting the other, maybe someone with better knowledge might be able to work out why the noise.
Peter,
no problem, the funny thing though is that i already have build this seperate driver / MOSFET setup, see here:
http://www.overunityresearch.com/index.php?topic=272.msg59280#msg59280
The outcome was a nice gate signal (see that post screenshot), but also that the crackling or any other effect
(oscillations?) had disappeared.
So i still have that build, so could use it, however will it be of any use in finding (back) the crackling?
I know, just try it.
Itsu
Hi Itsu
I'm not really sure why they are not appearing, what happens if you go back to the irf40 does the crackling reappear.
Peter,
i very much appreciate your comments and suggestions, and understand that you are anxious to help, but i severall
times now get the feeling that you are not quite following what i am doing.
I know that you are very busy and lack the time to work on your own projects let alone to help me, but it is
not very productive for me nor for you when we are not on the same wavelength.
I am already back to the IRF840 MOSFETs, see here: http://www.overunityresearch.com/index.php?topic=272.msg70150#msg70150
But they do not return the crackling, but presently i am on the straight coreless bifilar multistranded wire coils which
up till now never showed any abnormal behaviour, so i do not expect any crackling now.
I have the IRF840's on another pcb which has seperate paths and seperate single drivers (IXDD614PI) and doing some tests
on this coreless coil to see how the pulses look now.
I still see the noise and dual pulses on the drains, so need to further investigate where they come from.
Itsu
I am starting to loose MOSFETs now.
I have the seperate drivers (IXDD614PI's) to seperate MOSFETs going to the coreless bifilar coils.
At 41V on the drains i lost an IRF840 (5A on the PS and one of the 2 was hot).
After installing another IRF840, the other one went off the same way.
Installed IRFP460's, but again one went off quickly. (all drain / source around 1 to 100 Ohm).
Lowered the driver voltage from 20V to 15V, and set the current limiter to minimum.
Now the short still happens, but the MOSFETs stay alive, and i see that it happens above 35V on the drain and
at about 205ns pulse phase.
Peters box has a 15ns offset (shows 0ns phase, but in reality its 15ns), so the shorts occure around 220ns.
I was able to capture the spike across the coil when the short happens, see screenshot.
The yellow trace is as said from across one of the bifilar (coreless) coil.
The 2 pulses on the left half are the "normal" pulses from the both MOSFETs and increase in distance till here at
about 220ns apart. (still not know why i see the 2 pulses when scoping 1 coil/drain).
The PS went into current limiting and the spike occures.
Guess when i have no current limitation, the spike will be much higher destroying the MOSFET.
Could it be avalanching?
Not sure why it happens about 450ns after the last pulse ended.
Itsu
Quote from: Itsu on 2018.11.13, 21:38:23
I am starting to loose MOSFETs now.
When you scope the drain voltage while varying the delay, do you see the reflection from the coil superimpose on the next pulse occurring 220ns later ? ...or after a multiple of the reflection's round-trip time?
No, not really, some 8Mhz ringing but no reflections, see screenshot 1 of the both drains at about 170ns delay.
The 2th screenshot show a capture with the same settings as above (also the drain signals) but with the glitches active.
Video here (glitches visible almost at the end): https://www.youtube.com/watch?v=NPk-g9kbEjg
I added another screenshot 3 with a wider view of the noise, we see the drain pulses on the far left, then the massive noise.
Itsu
I put in my 900V CREE MOSFETs again like i did before here:
http://www.overunityresearch.com/index.php?topic=272.msg59108#msg59108
And again like before i am not able to summon these glitches.
I increased the driver voltage to 20V, but these CREE MOSFETs won't show any form of noise or glitches.
Back to the IRF460's again (driver to 15V) and again i have these glitches.
Guess that the CREE's 900V threshold or any other parameter somehow prevent these glitches from happening.
Itsu
It is possible with a Test Setup and a Variable Voltage Supply to
evaluate the Breakdown/Avalanche Voltage of the MosFet which
is developing the strange "noise." That might nail it down as far
as the source of the strange behavior. I believe it was Verpies a
little ways back who suggested that may be the cause.
Either the Body Diode or the MosFet region could be responsible.
It would be very difficult perhaps to isolate the actual area.
Quote from: Itsu on 2018.11.14, 14:52:59
Video here (glitches visible almost at the end): https://www.youtube.com/watch?v=NPk-g9kbEjg
Is the ringing after the pulse, changing its frequency or is it my imagination ?
Does waving ceramic magnets in the vicinity alter it ?
Quote from: Itsu on 2018.11.14, 20:29:20
I put in my 900V CREE MOSFETs again like i did before here:
http://www.overunityresearch.com/index.php?topic=272.msg59108#msg59108
And again like before i am not able to summon these glitches.
Either these stronger MOSFETs stop breaking down or their different junction capacitances affect the system differently.
P.S.
How is that coil connected to these MOSFETs ?
Are there any chokes in the supply rails ? Where exactly?
muDped,
yes these glitches are very hard to summon or to manipulate as they are able to damage stuff.
verpies,
QuoteIs the ringing after the pulse, changing its frequency or is it my imagination ?
Does waving ceramic magnets in the vicinity alter it ?
I will have to find out later tonight, i think the frequency stays the same.
Will wave a ceramic magnet also tonight.
The below picture shows how the MOSFETs are connected to the coil, no choke is used.
Peters box and the driver chips are fet by a stack of 2 series batteries at 25V, all grounds connected
together at a single point.
Itsu
Quote from: Itsu on 2018.11.15, 09:57:29
no choke is used.
Not even for supplying the MOSFET drivers ?
The MOSFET drivers are powered from 2 batteries (24V), have their own LM350 regulator (15 - 20V) and are as normally
isolated/decoupled with a 1mH choke and tantalium (39uF) and ceramic (0.1uF) caps.
Itsu
QuoteIs the ringing after the pulse, changing its frequency or is it my imagination ?
Does waving ceramic magnets in the vicinity alter it ?
The ringing frequency stays at 7.8Mhz, only the amplitude changes (minimum at 0ns phase (= really 15ns) and
max. at 260ns phase (= really 275ns).
Waving a stack of ceramic magnets along or perpendiculary to the bifilar coils does not show any effect.
Touching the bifilar coil with my hand decreases slightly the amplitude of the ringing signal.
Below screenshot shows both drain and gate signals of the both MOSFETs while sweeping from 0 - 190ns (just before
flipping to 0ns again) thus avoiding the glitches to occure.
Yellow / purple MOSFET 1
Blue / green MOSFET 2
Itsu
Quote from: Itsu on 2018.11.15, 09:57:29
The below picture shows how the MOSFETs are connected to the coil, no choke is used.
What is this coil wound on ?
The coil is wound on some length of 16mm diameter electra piping, see picture:
Itsu
As i was working with my last pair of IRFP460 MOSFETs, it tried another type which i have more of, the SPW47N60C3.
This is a similar one as the IRFP460 (650V / 47A / 0.07 Ohm), but i was not able to get the effect with them.
So to protect my last IRFP460's, i added a common mode choke (CMC) in the 41V supply line and a 50 Ohm resistor.
The CMC more to protect the PS, and the 50 Ohm to create a voltage drop when amps start to flow.
Now i can see the glitches happening, but my PS does not get into current limiting mode and the glitches are much
more controlled (160V only area).
This allowed me to also scope the gate signal of a MOSFET when the glitch happens.
Screenshot shows in yellow the controlled glitch on the drain signal and in purple the gate signal that goes with it.
This looks to me that a glitch is causing (via what? induction / capacitance) a pulse on the gate which triggers the
MOSFET again etc. etc.
The drain spikes (160V) and the gate spikes (16V) are within the specs of the MOSFET, so no harm will be done.
But imagine what happens when the spikes of +700V we have seen earlier can do to the gate signal.
Not sure if any zeners are fast enough to protect the gate.
Itsu
I lowered the driver voltage to 15V and put in 18V bidirectional tvs's (1.5KE18CA) across the gate / source, but when omitting the 50 Ohm resistor in the
41V supply lead, i again have the HV glitches which damaged 1 MOSFET.
So the tvs's are not fast enough, or the damage is caused at the drains.
Itsu
Quote from: Itsu on 2018.11.15, 21:28:17
This looks to me that a glitch is causing (via what? induction / capacitance) a pulse on the gate which triggers the
MOSFET again etc. etc.
Via the Drain-Gate capacitance (a.k.a. Miller's capacitance).
The purpose of that recent Miller testing setup was to test how much that high dv/dt waveform appearing on the drain, influences the gate voltage when that test gate is terminated by THE SAME impedance as the driving impedance of the other MOSFET gate.
Varying the gate driving impedance by several Ohms will exacerbate or minimize the effect of the Drain-Gate capacitance during high dv/dt Drain waveforms. So this proposition can be empirically verified in that manner.
ANOTHER ISSUE:
From your scopeshot it is evident that there is a double, negative going, pulse when there is a large delay between the two channels.
According to the Lenz law the half of the bifilar winding, that is being driven, should induce a pulse of an OPPOSITE polarity in the other half of bifilar winding, that is NOT being driven.
...but this is not what is being observed.
Fortunately the mutual capacitance of these two halves of winding can be used to explain two pulses of the same polarity, ...meaning that in this coil the capacitve coupling is stronger than the inductive coupling.
In fact the entire bifilar coil can be characterized by the ratio of the capacitive coupling to the inductive coupling ...or treated as a distributed capacitance and inductance, which can be analyzed as a transmission line.
I used that recent Miller testing setup again with my IRFP460 MOSFETs i use now, and changed the load resistor R4 to a 100 Ohm pot.
But varying the pot across its complete range (0 - 100 Ohm) did not show any change in the output gate signal, so it is not so obvious by looking at the output gate signal
what the correct input impedance should be.
So if the drain-gate capacitance is causing this feedback from the drain to the gate i should be able to combat that by a zener across the gate/source or as i did a TVS.
So probably my TVS is not fast enough as i did destroy a MOSFET again yesterday.
I calculated the length of my bifilar coils to be 570cm (114 turns on 16mm diameter).
The tdr measurement i did on this bifilar coil (transmission line) shows a round trip time (open ended) of 60ns, so a one way trip time of 30ns.
Speed of light in free space is 30cm/ns, so 900cm (9m) for my 30ns one way trip.
The velocity factor then of this bifilar coil is 570/900 = 0.63.
But the double negative pulse seen on the drain signal is much farther away, like 172ns farther, see screenshot.
There i have put up the white traces as being the signal from one MOSFET drain while the other MOSFET is not triggered (input signal to its driver disconnected).
White trace R1 is MOSFET 1, white trace R2 is MOSFET 2 (identical).
The yellow trace is the MOSFET 1 signal with both MOSFETs activated at phase 0 (really 15ns).
So this shows me that the double pulse (172ns apart) is not caused by the reflection on the bifilar coil.
Itsu
Quote from: Itsu on 2018.11.17, 21:29:49
So if the drain-gate capacitance is causing this feedback from the drain to the gate i should be able to combat that by a zener across the gate/source or as i did a TVS.
Even if the Zener or TVS had no capacitance and were faster than the MOSFET's gate (which they are not), they are several centimeters (and microhenries) further from the Drain than the drain-gate junction.
Even a REMOTE zero Ohm resistor might not be able to keep the gate down if several microhenries separate it from the gate.
Quote from: Itsu on 2018.11.17, 21:29:49
I used that recent Miller testing setup again with my IRFP460 MOSFETs i use now, and changed the load resistor R4 to a 100 Ohm pot.
A 10 Ohm closely placed non-inductive pot would be better.
Quote from: Itsu on 2018.11.17, 21:29:49
So this shows me that the double pulse (172ns apart) is not caused by the reflection on the bifilar coil.
OK, but the C
DG can positively charge the gate of the undriven MOSFET ( thus turning it on *) unless it is grounded by a very low impedance... or better yet, when the gate is brought -15V below source by a very low impedance (by e.g. an AC gate driver)
*
This can be seen by putting a 1 Ohm CSR in the Drain and scoping across it
Quote from: verpies on 2018.11.17, 23:55:49
Even if the Zener or TVS had no capacitance and were faster than the MOSFET's gate (which they are not), they are several centimeters (and microhenries) further from the Drain than the drain-gate junction.
Even a REMOTE zero Ohm resistor might not be able to keep the gate down if several microhenries separate it from the gate.
Ok, so what you are saying is that in this case the zener/TVS is no cure for preventing the gate to develope spikes that will destroy the MOSFET.
QuoteOK, but the CDG can positively charge the gate of the undriven MOSFET ( thus turning it on *) unless it is grounded by a very low impedance... or better yet, when the gate is brought -15V below source by a very low impedance (by e.g. an AC gate driver)
So for the double pulse this AC coupled gate driver setup could prevent it.
Is it like shown on page 36 -> of this PFD: www.logosfoundation.org/instrum_gwr/balsi/Texas_slua618.pdf
Itsu
Quote from: Itsu on 2018.11.18, 11:53:26
So for the double pulse this AC coupled gate driver setup could prevent it.
Is it like shown on page 36 -> of this PFD: www.logosfoundation.org/instrum_gwr/balsi/Texas_slua618.pdf
No, I had in mind a driver that is supplied by -15V and -15V rails, so the gate is pulled up to +15V when the driver commands the MOSFET to be ON and pulled down to -15V when the driver commands the MOSFET to be OFF.
But before you start looking for a driver with > 30V supply rating and building a split power supply for it, put a 1 Ohm CSR in the undriven MOSFET's Drain and command its driver to turn it permanently OFF to see whether it becomes turned on by the high dv/dt waveform appearing on its Drain (caused by the switching action of the other MOSFET).
If it turns on ( current starts flowing in its Drain ) then disconnect the driver and sequentially connect the Gate to the Source with a 10Ω to 0Ω non-inductive resistors ...and if it turns on even after you get down to 0Ω then connect the gate to -10V to -18V using some kind of battery (e.g. two 9V batteries in series), etc...
Ok, 1 Ohm csr in the undriven MOSFET's Drain and its driver turned off.
Ref3 is the drain signal from the driven MOSFET for comparison.
Ref4 is the csr signal of the undriven MOSFET.
Disconnected the driver and used 10 Ohm, 1 Ohm, 0 ohm (short) and -9V (battery) across the undriven MOSFET gate/source
Purple is the csr signal of the undriven MOSFET at -9V, but its the same as with the 10, 1 and 0 Ohm resistors.
So there seems to be no or very little effect pulling down or negatively bias the gate.
Itsu
Quote from: Itsu on 2018.11.18, 16:00:43
So there seems to be no or very little effect pulling down or negatively bias the gate.
I agree.
You just saved yourself a lot of useless work making an AC gate driver and split power supplies.
Quote from: Itsu on 2018.11.18, 16:00:43
Ok, 1 Ohm csr in the undriven MOSFET's Drain and its driver turned off.
So, if the undriven MOSFET is not turning on due to the Miller effect, the only current that could be flowing through that Drain terminal can be caused by C
DG and C
DS.
Now, due diligence would require the calculation whether C
DG + C
DS can really cause these ~1A peak currents with these drain waveforms...but I feel too lazy to do that now.
"lazy sunday af... err evening" :)
First screenshot below shows both drain signals with the blue (MOSFET 2) one having no drive and its gate/source shorted.
So the blue (MOSFET 2) one is fully opened/activated without any obvious drive.
Second screenshot shows the drain signals with, additional to above, the blue (MOSFET 2) one disconnected from its coil (so no 41V drain voltage).
Obvious there is no pulse now, but we also see the yellow (MOSFET 1) one having only 1 pulse now.
Extending the sweep range to Peters box max. (1200ns) shows that the glitches keep on happening all the way up to this upper limit, so roughly from 170ns to 1200ns phase.
The Spectrum Analyzer shows massive noise up to 370Mhz when the glitches appear, but also on my 70cm Ham transceiver (432Mhz) i hear the noise still very well.
itsu
Quote from: Itsu on 2018.11.18, 20:21:04
First screenshot below shows both drain signals with the blue (MOSFET 2) one having no drive and its gate/source shorted.
So the blue (MOSFET 2) one is fully opened/activated without any obvious drive.
So the undriven MOSFET is conducting now !?
This is very strange...
yes, i find it strange too.
Here a video of this situation where i have removed the MOSFET 2 (blue) driver and shorted its gate/source.
Scoping both MOSFETs drain voltage and the MOSFET 2 drain current.
https://www.youtube.com/watch?v=DhARKF4-bf4
See also the screenshot.
Yellow is MOSFET 1 (driven MOSFET) drain voltage
Blue is MOSFET 2 (undriven / shorted MOSFET) drain voltage
green is MOSFET 2 drain current.
Itsu
Notice, that during the ringdown, current & voltage are in phase, so this means that the current is not flowing through a capacitance, as it would be 90º out of phase with voltage then.
What do you get on the undriven half of the winding when you ground it through a 1kΩ resistor instead through the undriven MOSFET 2 ?.
Hmmm, yes in phase, but opposing, so 180° off.
So i should not have inverted the green channel then, but if so, then the current is flowing out of the drain when the MOSFET is active and into the coil as my current probe "current arrow" points now towards the drain.
Will do the 1kΩ resistor test later this evening.
Itsu
Quote from: Itsu on 2018.11.20, 09:35:06
Hmmm, yes in phase, but opposing, so 180° off.
Getting 180º with passive elements is tantamount to finding the Holy Grail - the negative resistance.
Undriven half of the winding grounded via a 1KΩ resistor (getting hot quick), current probe arrow pointing to ground, see picture.
Green channel NOT inverted (perhaps the yellow channel should).
White trace is the driven MOSFET drain voltage (notice NO double pulse now)
yellow trace is across the undriven half of the winding.
green trace is the current through the 1kΩ resistor.
Itsu
Not sure what went wrong with the above screenshot, but the current trace value seems odd (253.6mApp) as the
vertical setting shows 50mA/div.
So i did a fresh calibration of the scope and redid the test, but now with the yellow probe across the coil reversed.
No more ringing there (ringing appears when disconnecting the ground tip, so..).
See below screenshot.
edit: added the diagram
Itsu
I was thinking a long time and I could not think of a good reason for this MOSFET to become turned on with the gate shorted to source.
Normally, this happens when there is a excessively high dv/dt on the train (rising edge for N-ch MOSFET) or when the D-S junction breaks down due to overvoltage, ..but none of these things seem to be happening here, unless your scope is missing some huge subnanosecond kV pulse.
The voltage waveform across the 1k resistor is almost identical to the voltage waveform on the driven winding, suggesting simple capacitive coupling, the small differences that are visible could be due to delayed transmission-line like reflections.
Since this is strange, it should be investigated further. Does Tinsel look at his thread ?
QuoteI was thinking a long time and I could not think of a good reason for this MOSFET to become turned on with the gate shorted to source.
Normally, this happens when there is a excessively high dv/dt on the train (rising edge for N-ch MOSFET) or when the D-S junction breaks down due to overvoltage, ..but none of these things seem to be happening here, unless your scope is missing some huge subnanosecond kV pulse.
Not sure to which post / situation you point with this remark, as the MOSFET with its gate/source shorted is not in the circuit in my last post situation.
It was in post #414.
I doubt i would not detect any huge subnanosecond kV pulse
QuoteThe voltage waveform across the 1k resistor is almost identical to the voltage waveform on the driven winding, suggesting simple capacitive coupling, the small differences that are visible could be due to delayed transmission-line like reflections.
On this i agree if pointing to my last post situation. The strong coupling of this bifilar coil could do that.
QuoteSince this is strange, it should be investigated further. Does Tinsel look at his thread ?
Not sure if besides you and me anyone else is looking at this thread.
Itsu
Quote from: Itsu on 2018.11.30, 21:29:12
Not sure to which post / situation you point with this remark, as the MOSFET with its gate/source shorted is not in the circuit in my last post situation.
It was in post #414.
Yes, I was referring to post #414.
I still don't understand what could have turned that MOSFET on...or was it simply a matter of the capacitance between Drain and Source.
ok, thats clear.
The IRFP460 output capacitance is 870pF
So for a MOSFET (IRFP460 here) to activate with a shorted gate/source and no driver chip, it could be due to the its output capacitance (870pF here).
There should be a way to simply test this.
Perhaps some other EE's could confirm this or come up with another cause.
For that, start at post #412 for explanation and screenshots.
Thanks for now, Itsu
added the circuit diagram as it was from post #412
Quote from: verpies on 2018.11.17, 15:13:20
In fact the entire bifilar coil can be characterized by the ratio of the capacitive coupling to the inductive coupling ...or treated as a distributed capacitance and inductance, which can be analyzed as a transmission line.
O0
...or treated as a distributed capacitance and inductance, which
can should be analyzed as a transmission line.
Quote from: verpies on 2018.11.30, 16:14:35
I was thinking a long time and I could not think of a good reason for this MOSFET to become turned on with the gate shorted to source.
Is it turning on? Might that opposite pulse on the undriven MOSFET and 1k resistor be what's expected if one considers the bifilar coil to be an open-ended transmission line?
In 2009 when Peter and I worked on these experiments, the following is some background and suggestions based on the path I followed and results I achieved shortly afterward.
As the test setup consisted of two switches and two coils (in a bifilar configuration), and we found the "pulse" only occurred on the delayed coil, it was hypothesized that we may be able to eliminate one coil. That became the second test configuration, and basically where the collaborative experiments ended as far as I remember.
From here I began working to developing the simulated effect with some success. One key factor was that I surmised the connecting wiring between the coil and DC supply could/should be replaced with a transmission line, aka coax cable. This improved and solidified my results a great deal. In discussing the improvement with Wavewatcher/BEP, he suggested I replace even the coil with a length of coax, and this proved to be a fruitful modification. In fact, best performance was achieved when the two halves of TX line were of the same length. At this point it appeared to me to be a new and novel form of Blumlein/step-recovery-diode hybrid Pulser. The diode is very important, and the pulses are NOT produced without it. Some diodes work better than others, so I recommend trying a few types.
I found that the second, delayed MOSFET can also be dispensed with, as one MOSFET can do the job. It turned out that once the coax length, DC supply voltage and drive pulse period are "tuned" just right, a 50% duty input pulse will drive the circuit into a form of resonance where the pulses are 5 to 10 times the amplitude of the DC supply voltage, depending on the resulting output pulse width and gate drive period. In theory this circuit might achieve extremely high voltage/narrow pulses from ordinary supply voltages. One might consider this circuit a "tuned voltage compression device", where it converts a regular 50% pulsed DC input, to a proportional v x t pulsed output; the smaller that "t" is, the higher "v" is, so that the input and output v x t is always equal. At any rate, you can see one scope shot here showing the circuit hitting a "resonant" point where the output voltage is multiplied and time-diminished by about the same factor. The attached pics are using a 20V supply voltage.
- sim01 is the two MOSFET setup, with single red and violet gate pulses, and resulting delayed MOSFET drain pulse.
- sim03 is hitting "resonance" with continuous gate drive.
- 2-TX Line Schematic was my final configuration and starting point timing values.
Thanks Poynt for looking into this, but the 1K resistor setup was done later to see what would happen (nothing weird)
when using a 1K instead of the shorted MOSFET.
So the strangeness we see is at the posts roughly between post #412 and 415.
It looks to me in posts 412 and 414 screenshots there is evidence of the shorted MOSFET to turn on looking
at the drain voltage and drain current signals.
Anyway, nice idea to have the bifilar coil replaced by certain lengths of coax.
The pulses i see without the 50 Ohm protection resistor at the PS are already huge and fast in the bifilar coil
setup, so i can only imagine what they look like with a coax cable and the proper (DSR?)Diode.
Itsu
Quote from: poynt99 on 2018.12.08, 02:45:38
In 2009 when Peter and I worked on these experiments, the following is some background and suggestions based on the path I followed and results I achieved shortly afterward.
As the test setup consisted of two switches and two coils (in a bifilar configuration), and we found the "pulse" only occurred on the delayed coil, it was hypothesized that we may be able to eliminate one coil. That became the second test configuration, and basically where the collaborative experiments ended as far as I remember.
From here I began working to developing the simulated effect with some success. One key factor was that I surmised the connecting wiring between the coil and DC supply could/should be replaced with a transmission line, aka coax cable. This improved and solidified my results a great deal. In discussing the improvement with Wavewatcher/BEP, he suggested I replace even the coil with a length of coax, and this proved to be a fruitful modification. In fact, best performance was achieved when the two halves of TX line were of the same length. At this point it appeared to me to be a new and novel form of Blumlein/step-recovery-diode hybrid Pulser. The diode is very important, and the pulses are NOT produced without it. Some diodes work better than others, so I recommend trying a few types.
I found that the second, delayed MOSFET can also be dispensed with, as one MOSFET can do the job. It turned out that once the coax length, DC supply voltage and drive pulse period are "tuned" just right, a 50% duty input pulse will drive the circuit into a form of resonance where the pulses are 5 to 10 times the amplitude of the DC supply voltage, depending on the resulting output pulse width and gate drive period. In theory this circuit might achieve extremely high voltage/narrow pulses from ordinary supply voltages. One might consider this circuit a "tuned voltage compression device", where it converts a regular 50% pulsed DC input, to a proportional v x t pulsed output; the smaller that "t" is, the higher "v" is, so that the input and output v x t is always equal. At any rate, you can see one scope shot here showing the circuit hitting a "resonant" point where the output voltage is multiplied and time-diminished by about the same factor. The attached pics are using a 20V supply voltage.
- sim01 is the two MOSFET setup, with single red and violet gate pulses, and resulting delayed MOSFET drain pulse.
- sim03 is hitting "resonance" with continuous gate drive.
- 2-TX Line Schematic was my final configuration and starting point timing values.
Hi Poynt
Looking at your circuit, from a layman's point of view, I would say your two pieces of coax are acting as inductors and capacitors all in one. The mosfet switches on and charges the two inductors in series, the mosfet switches off and there is a discharge through the diode to the junction of the two coaxial wires, from this point on the inductive capacitance of the second coax goes into oscillation, being fed by the capacitive charge of the first coax to keep up the voltage.
It is parametric, you changed the inductance and capacitance with the diode connection, the diode supplied "DC" to the Parallel LC of the second coax cable which then went into oscillation. what is interesting is the "level" of oscillation, not the usual drop to zero.
Regards
Mike
I know this is an old thread regarding pulse generation for the TPU but I thought that this was the best place for what I'm going to post. This is also probably ridiculous in light of Mike's current work with his Steap, but I think there may be some value in what I've found. This discovery was by accident and actually was a problem in my RLE research before I realized what was happening.
Anyway, below is a sim which best demonstrates the concept and that is, when two like coils on a common core [ferrous or air] are charged to dissimilar and opposite currents and then mixed, a relatively large voltage transient is produced that is many times the power supply. This is a result of the dissimilar currents attempting to reach equilibrium via the turn-turn and/or coil-coil self capacitance.
In the sim below, L1 serves as a primary and L2 as a secondary. For the first 10us, L1 is charged from V4, a 50v dc supply, and L2 is shorted via S3 and S4. The current build up in L2 is -.9 * I(L1) and in this case the peak currents at 10us are L1 = .521 and L2 = .468 . At the end of 10us, VL1 is grounded via S1, and VL2 along with the dot end of L2 are released. This forces the currents in L1 an L2 to attempt to equalize via their internal 20pf capacitance's, and the result is the 2.1kv pulse generated at L2. Also note that the currents reverse in L1 and L2 at this time.
In this version, the bucking currents in L1 and L2 are discharged into Vs and returned to the supply V4. This reduces the effective energy from the supply to a relatively low level and in this case it is 6.14uJ .
The amplitude and duration of the voltage pulse is dependent on the coupling factor of the coils and their capacitance's plus the current levels.
I will attempt to apply this technique to some of the orthogonal TPU setups I have laying around to see what may result after simulating some of the modeled TPU coils.
Regards,
Pm
This version has L1 and L2 combined in the "aid" mode by reversing L2 rather than the buck mode in the previous sim. The method of operation is slightly different and the peak voltage reached is less.
One thing to note is the peak reactive power in L1 of 233.5 watts. Another is the ratio of the average reactive power of 12.252 watts to the average input power of .216 watts which is 56.7:1 .
It is my opinion that SM's TPU is basically a 'reactive' to 'real' power converter and that is the reason for the high heat generation. It does not matter what creates the electron flow in the device, but the resultant reactive power is many times greater than the real power generated thus creating heat problems.
Pm
Quote from: partzman on 2022.11.01, 21:56:48
I know this is an old thread regarding pulse generation for the TPU but I thought that this was the best place for what I'm going to post. This is also probably ridiculous in light of Mike's current work with his Steap, but I think there may be some value in what I've found. This discovery was by accident and actually was a problem in my RLE research before I realized what was happening.
Anyway, below is a sim which best demonstrates the concept and that is, when two like coils on a common core [ferrous or air] are charged to dissimilar and opposite currents and then mixed, a relatively large voltage transient is produced that is many times the power supply. This is a result of the dissimilar currents attempting to reach equilibrium via the turn-turn and/or coil-coil self capacitance.
In the sim below, L1 serves as a primary and L2 as a secondary. For the first 10us, L1 is charged from V4, a 50v dc supply, and L2 is shorted via S3 and S4. The current build up in L2 is -.9 * I(L1) and in this case the peak currents at 10us are L1 = .521 and L2 = .468 . At the end of 10us, VL1 is grounded via S1, and VL2 along with the dot end of L2 are released. This forces the currents in L1 an L2 to attempt to equalize via their internal 20pf capacitance's, and the result is the 2.1kv pulse generated at L2. Also note that the currents reverse in L1 and L2 at this time.
In this version, the bucking currents in L1 and L2 are discharged into Vs and returned to the supply V4. This reduces the effective energy from the supply to a relatively low level and in this case it is 6.14uJ .
The amplitude and duration of the voltage pulse is dependent on the coupling factor of the coils and their capacitance's plus the current levels.
I will attempt to apply this technique to some of the orthogonal TPU setups I have laying around to see what may result after simulating some of the modeled TPU coils.
Regards,
Pm
Can you upload the LTspice File? Pulsing coils with high voltage nanosecond pulses like AVEC / TPU devices is also my area of interest. I'm looking for practical ways to achive this expecially the delayed HV pulse on bifilar coils.
This is a sim with the buck mode applied to a symmetrical transmission line that was modeled from a bench design. Peak pulse is ~500v with a 143ns width.
Pm
Quote from: Frederik2k1 on 2022.11.02, 15:12:57
Can you upload the LTspice File? Pulsing coils with high voltage nanosecond pulses like AVEC / TPU devices is also my area of interest. I'm looking for practical ways to achive this expecially the delayed HV pulse on bifilar coils.
Sure, see the .asc file below.
Pm
What happens if the coils are not bucking, but boosting (wound same direction).
The bifilar coils used by Peter were wound as a pair in the same direction.
Quote from: Grumpy on 2022.11.02, 15:57:52
What happens if the coils are not bucking, but boosting (wound same direction).
The bifilar coils used by Peter were wound as a pair in the same direction.
My post #430 has the discharge currents in L1 and L2 aiding but the charging of L1 and L2 is still done in a buck mode so I don't think this is the same as Peter's arrangement. I'll have to scan back thru this thread to see Peter's arrangement to be sure.
I will assume [without testing at this point] that the bucking pulse produced in post #429 would not normally induce an orthogonal coil while the aiding pulse produced in post #430 should. IOW, a pair of wires with many turns toroidally wound over a horizontal coil should act as single turns when pulsed as a symmetrical transmission line as in post #432. This is pure speculation on my part at this point and I hope to prove this or not over the next few days.
Pm
Grumpy,
I did a quick but focused look thru this thread and IMO, the pulsing of the bifilar coils by Peter and others that were done in various physical and electrical configurations with delayed pulse timings, had instances where the conditions I describe above were met and large pulses were generated. I could be wrong of course!
What I don't quite understand because I haven't read all the TPU threads is, what was the application of these high level pulses in the TPU?
Pm
Quote from: partzman on 2022.11.03, 14:10:41
Grumpy,
I did a quick but focused look thru this thread and IMO, the pulsing of the bifilar coils by Peter and others that were done in various physical and electrical configurations with delayed pulse timings, had instances where the conditions I describe above were met and large pulses were generated. I could be wrong of course!
What I don't quite understand because I haven't read all the TPU threads is, what was the application of these high level pulses in the TPU?
Pm
You apply the pulses in a circular sequence to produce a rotating field.
This requires the pulses to also interact with either a static magnetic field or have a positive DC offset (bias).
This rotating field will have the inductive properties of a rotating magnetic field and also the inductive properties of a rotating displacement field.
(Imagine that you pattern space as a magnetic field and then rotate the space.)
You then use a capacitive or inductive output coil.
The inductive output coil cannot be the same direction as the input coils, but should be orthogonal, else it reflects back and doesn't work.
As far as know, no one has applied this pulse properly to date, except Centraflow, and his method is complex.
Quote from: Grumpy on 2022.11.03, 14:43:14
You apply the pulses in a circular sequence to produce a rotating field.
This requires the pulses to also interact with either a static magnetic field or have a positive DC offset (bias).
This rotating field will have the inductive properties of a rotating magnetic field and also the inductive properties of a rotating displacement field.
(Imagine that you pattern space as a magnetic field and then rotate the space.)
You then use a capacitive or inductive output coil.
The inductive output coil cannot be the same direction as the input coils, but should be orthogonal, else it reflects back and doesn't work.
As far as know, no one has applied this pulse properly to date, except Centraflow, and his method is complex.
OK, thanks for that explanation. My question then becomes how can we have conventional induction with orthogonal coil arrangements? IIRC, my experiments with an orthogonal TPU topology resulted in the voltages produced in the toroidal primary windings to appear as a single turn to the secondary orthogonal winding. However I wasn't using high voltages.
I am familiar with Centraflow's design and it is indeed complex. I understand he calls it a plasma ion/electron conversion but the voltages are too low for plasma generation in air!?
BTW, I do believe in an active aether teeming with energy.
Pm
Quote from: partzman on 2022.11.03, 15:06:33
OK, thanks for that explanation. My question then becomes how can we have conventional induction with orthogonal coil arrangements? IIRC, my experiments with an orthogonal TPU topology resulted in the voltages produced in the toroidal primary windings to appear as a single turn to the secondary orthogonal winding. However I wasn't using high voltages.
I am familiar with Centraflow's design and it is indeed complex. I understand he calls it a plasma ion/electron conversion but the voltages are too low for plasma generation in air!?
BTW, I do believe in an active aether teeming with energy.
Pm
If proven, (devices are still being built to verify claims) I believe it will be a sort of Wilson Effect where space is rotated rather than a solid dielectric.
This also means that the explanation that SM provided about a rotating magnetic field is not entirely accurate. Spherics called it "magnetic like", but also commented not to use blocks of delectric materials in the rotating area and to leave it as open as possible.
Quote from: Grumpy on 2022.11.03, 15:53:59
If proven, (devices are still being built to verify claims) I believe it will be a sort of Wilson Effect where space is rotated rather than a solid dielectric.
This also means that the explanation that SM provided about a rotating magnetic field is not entirely accurate. Spherics called it "magnetic like", but also commented not to use blocks of delectric materials in the rotating area and to leave it as open as possible.
OK I understand. Will be anxious to hear any results you have with your Spherics build!
Pm
A current is a number of charges that flow per unit time. They are usually produced by a current in a conductor, but they can also be produced by moving a charged conductor or dielectric. The simplest case is a charged ring that rotates: it obviously creates a magnetic field since we have charges that rotate, it's a known fact. The current is usually small since the number of rotating charges is what the device can carry as a capacitor, so for example 10µC if the capacitance of the ring is 1 nF charged at 10 KV, which gives a current of 1 mA if the ring rotates at 100 revolutions per second.
All the effects in the Grumpy's picture above (https://www.overunityresearch.com/index.php?action=dlattach;topic=272.0;attach=46204;image), not necessarily easy to implement experimentally for significant measurable effect, seem to me to be real and to be the simple consequence of these currents linked to the mechanical displacement of charges, and the reciprocal effects.
Quote from: F6FLT on 2022.11.04, 10:26:08
A current is a number of charges that flow per unit time. They are usually produced by a current in a conductor, but they can also be produced by moving a charged conductor or dielectric. The simplest case is a charged ring that rotates: it obviously creates a magnetic field since we have charges that rotate, it's a known fact. The current is usually small since the number of rotating charges is what the device can carry as a capacitor, so for example 10µC if the capacitance of the ring is 1 nF charged at 10 KV, which gives a current of 1 mA if the ring rotates at 100 revolutions per second.
All the effects in the Grumpy's picture above (https://www.overunityresearch.com/index.php?action=dlattach;topic=272.0;attach=46204;image), not necessarily easy to implement experimentally for significant measurable effect, seem to me to be real and to be the simple consequence of these currents linked to the mechanical displacement of charges, and the reciprocal effects.
The need for polarization by a magnetic field or an electric field is interesting.
This is the aiding current version of the buck pulse generator with various bias currents in L2 prior to the cycle starting. I should add that D3 clamps VL2 so that only a unidirectional pulse is generated. If D3 is removed, bipolar pulses will be generated. Also, the reverse voltage in the sim is far greater than D3 would normally handle, but the model for this particular diode does not include reverse breakdown. In an operational circuit, D3 would obviously have HV ratings.
As can be seen from the table, the peak voltage on VL2 increases with increasing L2 bias current. A 400ma bias is shown and what is also interesting is the net energy consumption. We see 17.512uJ drawn from the supply V4 and at the end of the cycle, IL2 is 0.0ma and IL1 is 406.87ma resulting in a gain over the starting bias of (.40687^2-.400^2)*.005/2 = 13.858uJ . This makes for a net energy consumed of only 3.654uJ !
Pm
What kind of switches are in the sim?
Quote from: Grumpy on 2022.11.04, 22:02:11
What kind of switches are in the sim?
They are voltage controlled switches with a threshold of zero volts which means the switching voltage varies from +1 for "on" to -1 for "off". The on resistance is specified at .05 ohm in this case. They have unlimited voltage and current and also do not have any mosfet capacitance's such as Ciss or Coss. With a relatively small Coss at VL2, circuit performance is not degraded.
In the case of the unipolar single pulse generators shown with a mosfet replacing S4, the BVdss rating should be greater than the maximum peak voltage on VL2 and D3 would be the substrate or body diode.
Pm