PopularFX
Home Help Search Login Register
Welcome,Guest. Please login or register.
2021-10-23, 16:25:30
News: Forum TIP:
The SHOUT BOX deletes messages after 3 hours. It is NOT meant to have lengthy conversations in. Use the Chat feature instead.

Pages: 1 2 3 4 [5]
Author Topic: Self running coil?  (Read 75601 times)
Group: Moderator
Hero Member
*****

Posts: 901
Hi MileHigh,

thank you again for all these details you have posted.

I'll see if I can add a shunt across the pulse coil so we can see what is going on.

Thanks for sharing all this amazing amount of information.

Luc
   
Group: Moderator
Hero Member
*****

Posts: 901
Here are new scope shots of the coil without magnet at 3.9KHz in self pulse mode with 3vdc source and drawing 14uA

The 1st scope shot below is with adding a 100 Ohm 5% carbon resistor as shunt in series between feed and mosfet source. Green probe is across drain and source. Yellow is across the 100 Ohm resistor.

The next one is with 100 Ohm 5% carbon resistor in series between source and pulse coil. Green probe is across drain and source. Yellow is across the 100 Ohm resistor with probe ground at source side of resistor.

The next shot is a close up of the above

The last one is even closer of the bump of the above

http://i944.photobucket.com/albums/ad290/gotoluc/IRF640at379KHz3vdcat14uA100Ohmimput.png
Self running coil?


http://i944.photobucket.com/albums/ad290/gotoluc/IRF640at379KHz3vdcat14uA100Ohmpulse.png
Self running coil?


http://i944.photobucket.com/albums/ad290/gotoluc/IRF640at379KHz3vdcat14uA100Ohmpu-1.png
Self running coil?


http://i944.photobucket.com/albums/ad290/gotoluc/IRF640at379KHz3vdcat14uA100Ohmpu-2.png
Self running coil?

   
Group: Tech Wizard
Hero Member
*****

Posts: 928
...
For your second comment I just have a few things to say without getting too deeply into all of the design issues.  For the gate input let's say that you can define an "OFF" region, a "linear" region, and an "ON" region.  This is a function of the source-drain current and source voltage.

So for example if the source-drain voltage is 10 volts and the current is 100 milliamperes, then "OFF" might be 0-2 volts, "linear" might be 2-6 volts, and "ON" might be 6-10 volts.  Therefore if you excite the gate input with a sine wave, it all depends on the voltage sweep of the sine wave to determine how the MOSFET is going to react to this stimulus.

Just as a caveat, I am being pretty general and probably oversimplifying things here.  Certainly if I was testing a similar circuit I would be referencing the datasheet for the MOSFET and investigating all of these aspects.   

Hi MH,

Thanks for your detailed answer and again mostly I agree with it.  What I meant on linear operation is that if you input a sinusoidal voltage at the gate, you receive it at the drain with maximum a few percent distortion. 
And if you consider the 100mA (or any real drain current) in your example you can set the 100mA current with DC bias only at the gate, and no sine wave input yet. Then you can input a certain amplitude of sinewave drive to the gate which will change the drain source voltage from the 5-6V mid range down to 0-2V and up to the 8-10V.  And the AMPLITUDE of the input sine wave is important: depends on the gm transconductance of the FET and the (load) impedance at the drain. If you think in advance the drain current times the load drain impedance already CONSTITUTES the 5-6V middle range needed for linear operation, (in the dynamic case the AC impedance at the drain what counts).  If this condition is not met, distortion in the output signal is bound to occur, except for very smal input amplitudes that do not try to swing the drain voltage lower than 0 or higher than 10V in the example case.

Thanks,  Gyula

   
Group: Tech Wizard
Hero Member
*****

Posts: 928
Hi Gyula,

I was away for the weekend and now back.

Thanks for taking the time to make a schematic of my test setup and posting it with your explanation as to possibly why the 3.9KHz circuit uses more current then the 5KHz circuit with magnet. That is a good point!... however if we multiply the  78us X 3.9 = 304  and if we take the 56us X 5 = 280  so we have a possible difference of 25Us

I don't know if this small a difference would account for the total current differences?

Thanks for your explanation and time

Luc

Hi Luc,

What if the small difference is even smaller?  ;)   Because you used 3.9 multiplier instead of the real 3.79kHz.  But even so there is still some difference left, even if we consider I estimated the on-times from the scope shot with certain uncertainty.
 And such comparison maybe not meaningful, maybe does not account for the total current differences.
What I still think accounts for it is the switch on time of the FET wrt a full cycle.

rgds,  Gyula
   
Group: Tech Wizard
Hero Member
*****

Posts: 928
Hi Gyula,

I added some circles and lines to the 3.9KHz scope shot from my above post.

If we look at the 2 red circles I added (tops of the sine wave) we will notice the negative is perfect and the positive has a flat top to it. If we look at the main coil sine wave I added a purple circle were a bump appears at every pulse. Could that be the pulse coil being charged?... if it is, then I added a vertical blue line at the point it starts to fall and it coincides with the beginning of the flat area of the + sine wave. I then added another vertical blue line at the end of the flat area and then a blue line horizontally between the two.

I believe this is the on time of the MOSFET (between blue vertical lines) and the purple circle (main coil bump) to be the charging period of the pulse coil. What do you think?

Thanks for sharing.

Luc

http://i944.photobucket.com/albums/ad290/gotoluc/IRF640at379KHzmaintroid1074mHpul-1.png
Self running coil?


Hi Luc,

Yes I accept what you showed it sounds ok with me, though the positive peak is not so flat at all in case of the 5kHz oscillator.
And I had a look at your earlier scope shots on the other frequencies up to 50kHz with your oscillators and they are not flat either. It is sure that the switching must do the bump in the drain voltage waveform where your circled it in pink and the flat or not so flat positive peaks has no important meaning.

rgds, Gyula
   
Hero Member
*****

Posts: 520
I would simply like to thank all of you for this discussion regardless of who or what is really happening. This discussion in itself has given me some real insight into Mosfets and the way they really tick. Can I just ask one question. When you say "gate source" do you really mean "gate feed" or does the Mosfet source end have something to also do with it. This is confusing to neophytes like me. Maybe when you talk about a Mosfet, do not use the word gate, source or drain for anything other then referring to those specific mosfet ends.

Either way Mosfets are a real pain in the ass and if they are used for OU devices with some serious inductive and capacitive strength, you will have to work out a way for the flyback to return back to the feed battery, without it going backward through the drain/source. On the drain side, you will need a flyback switching to get the flyback into the battery without passing the mosfet otherwise you get what I am really good at. Making toasted mosfets.


---------------------------
   
Group: Tech Wizard
Hero Member
*****

Posts: 928
Hi Wattsup,

Normally I listen to be not sloppy and whenever I say gate source I always mean the electrodes of a MOSFET, not a power supply as a source for instance. In the future I will try to use a dash between the electrodes like this: gate-source to denote the electrodes of a MOSFET between which a driving source (or whatever) is connected.  :)

Yes whenever your MOSFET drain-source side is unloaded and you drive the gate-source input with square wave to switch the device on and off and you have a high value inductance in series with the drain or source electrodes, (and of course you apply DC supply too) then all the flyback pulse gets across the drain-source side, hence the MOSFET must have a max drain-source voltage of higher than the expected maximum flyback pulse.
This often means to use 800-1000V drain-source voltage limit neccessity for the MOSFET, unfortunately MOSFET types manufactured for such high voltages are weaker in other desirable features like very low drain-source ON resistances etc and often more expensive.

Thanks,  Gyula
   
Sr. Member
****

Posts: 275
Hi All,

When you bias a mosfet gate for a drain current, the heating of the device causes more current to flow after a short while.
Try biasing a IRF 840 to illuminate a 21 watt 12V bulb dimly and watch the brightness increase as a result.
This is a problem for operating in anything other than a switch condition. I have biased the device just up to starting conduction for a test oscillator which worked until the fet heated. Then I installed a current limiting cct and the test osc became unstable.
I will try a cct with a driver and then build a mopa and hopefully stabilise the thing.
Only fried a few so far. All died as a result of overvoltage, so doin a bit in the right direction.
FETs, give us hope then add complications, just the way we like them, lol

Cheers,

Steve.
   
Group: Tech Wizard
Hero Member
*****

Posts: 928
Hi Luc,

What if the small difference is even smaller?  ;)   Because you used 3.9 multiplier instead of the real 3.79kHz.  But even so there is still some difference left, even if we consider I estimated the on-times from the scope shot with certain uncertainty.
 And such comparison maybe not meaningful, maybe does not account for the total current differences.
What I still think accounts for it is the switch on time of the FET wrt a full cycle.

rgds,  Gyula

Hi Luc,

Been thinking on the current difference explanation...

Considering the inductive reactance of the toroidal coil in series with the drain electrode, some notices:

@ 3.79 kHz you measured 1074mH --->  XL=25.56 kOhm

@ 5 kHz you measured 604mH  --->  XL=18.96 kOhm

So even if you have a higher (inductive) impedance in series with the drain electrode at 3.79 kHz, still you have a higher drain current draw.  This strengthens my thought on the importance of the longer switch-on time at 3.79 kHz.

Comments are welcome from others too.

Thanks, Gyula
   
Group: Moderator
Hero Member
*****

Posts: 901
At everyone,

I think I found something important that is preventing the circuit to show its full potential.

I decided to reproduce as close as I can the sine wave signal the self pulsing coil produces when connected between the gate and source of the mosfet. Using my Signal Generator in sine wave I matched the frequency and adjusted the output to the exact vpp the pulse coil produces. Once I had them matched up as best I can I pulled out the pulse coil and connected only the generator. To my surprise the current draws is exactly the same as when the Pulse coil is connected and self pulsing. I would of though it would use less current since the switching is now powered by the generator! However this is not the case and I also found something even more important.

See the first 2 scope shots below to see first the pulse coil (self pulse mode) shot and next the generator pulsing the circuit.
All tests are using the same 3vdc feed and IRF640 mosfet with same dual coil toroid @ 1075mH (no magnet)
Green scope probe is between mosfet drain and source and Yellow scope probe is between mosfet gate and source.

Why is this important!... because ever since I've been using the pulse coil I lost the ability to control the mosfet switch on timing. If you look at all the scope shots I posted before, when using the pulse coil you will see it has this delay. However if I use the generators I can adjust the on time timing.

So without changing the setup I only turned the micro frequency adjustment on the generator to make the mosfet turn on time sooner and bingo!... no current used and as I keep going it starts to send back current (up to a certain point). See those scope shots below the first two.

It is clear that the pulse coil is delayed too much. We will not get it to use no current or send back current this way. We need the mosfet on pulse trigger to be earlier for this to work.

What do you think?... any ideas how to solve this?

@Gyula, please notice that the main coil bump is still there (without the pulse coil) so this is not what we think it is!... notice it goes away when on time is advanced to starting to come back but in negative as it send most current back. What could this be?

Luc









« Last Edit: 2010-04-06, 17:41:10 by gotoluc »
   
Group: Moderator
Hero Member
*****

Posts: 901
Hi All,

below is an update of the best score mosfet to date.

The 1st Scope shot below is so far the best score mosfet IRFBC20
With 3vdc at 10KHz with main toroid at 337mH and pulse coil at 225mH the circuit operates with just 4.3uA

The 2nd Scope shot below is the next best mosfet IRF640
With 3vdc at 10KHz with main toroid at 165.3mH pulse coil at 96mH the circuit operates with 9.4uA

So it does look like a mosfet of a lower capacitance like the IRFBC20 is better for what I'm trying to do.

Luc

ADDED

I added a red circle to each shot so you can see for yourself that it is in fact the advance timing of the pulse coil that gives the saving as I stated in my post above.



« Last Edit: 2010-04-06, 23:51:10 by gotoluc »
   
Hero Member
*****

Posts: 520
@gotoluc

I would like to know something. If you are comparing the mosfets with a straight pulse generator performance, there will be a major difference. Basically the mosfets you are using are N-channel so they go on the negative side of the circuit, whereas the pulse generator is cutting on/off power from the positive side of the circuit. In the first regular mosfet way, the energy is already in the coils. With the PG, the energy is always being rushed into the coils. There is a major difference there.

If you can find a mosfet like I am using which is the IRF9540, this is a P-Channel mosfet and can go on the positive side. The capacitance is about the same as your IRF640. Maybe this will then give you a more direct comparison to the PG.


---------------------------
   
Group: Moderator
Hero Member
*****

Posts: 901
@gotoluc

I would like to know something. If you are comparing the mosfets with a straight pulse generator performance, there will be a major difference. Basically the mosfets you are using are N-channel so they go on the negative side of the circuit, whereas the pulse generator is cutting on/off power from the positive side of the circuit. In the first regular mosfet way, the energy is already in the coils. With the PG, the energy is always being rushed into the coils. There is a major difference there.

If you can find a mosfet like I am using which is the IRF9540, this is a P-Channel mosfet and can go on the positive side. The capacitance is about the same as your IRF640. Maybe this will then give you a more direct comparison to the PG.

Hi wattsup,

if you are referring to the scope shots above your post, then they are self powered by the pulse coil (no generator)

I have IRF9640 on hand. Tried them before and now once again and they don't self pulse like the IRF640. I don't know why.

Luc
   
Group: Tech Wizard
Hero Member
*****

Posts: 928
Hi Wavewatcher,

I managed to access to an old inductance meter that can measure the Q factor of the coil too. 
(I refer to my earlier post here I quoted below.)

I used a small multilayer, multiturn air core coil, measured 3.6mH at the 37.6kHz measuring frequency.
The Q measured about 47.  (The L meter forms a calibrated oscillator from the coil to be measured,
oscillating frequency always depends on the actual value of the coil with this meter.)

I placed a cylinder Neo magnet to one of the coil ends and noticed the inductance slightly decreased,
maybe a few tens of uH, judged from the mechanical scale. The Q went down to about 38.
I managed to reach about the same reducement in inductance with a bronze screw when I inserted it
about halfway into the air coil (bronze does not attract to magnets) but the Q reduced to about 43.
(The size of the bronze screw was OD=3mm,  length=20mm.)

Next I placed a rectangular ceramic magnet to the same end of the air coil. The inductance increased
maybe a few uH, yes, increased.  The Q stayed at about the same value, 47, it may have changed very little
upwards but on the mechanical scale it could not be read.

I think the Neo magnet reduced the inductance mainly due to its metal body presence just like the
bronze screw did: due to the eddy current induced in the metal by the coil's field.
It seems the eddy current effect is stronger than the very slightly higher than one permeability effect of
the Neo material's small  Fe content.
Ceramic magnets are basically (saturated) ferrite materials, no or only very little eddy currents, so the
Fe content can have a slight increasing effect on inductance.

Notice: I always used the same pole in case of both types of magnets and changing the poles did not
make any difference in the above results.

rgds,  Gyula




Hi WaveWatcher,

Thank you for clarifying my questions.  In the next couple of weeks I will have access to a Q meter and can check how the Q of an air core coil may change when a strong Neo magnet is inserted or brought very near to it.  Will return to this then.

You are right,  the inductance decreases when a copper slug is inserted into it (I erred) and the explanation is that copper is diamagnetic and has a relative permeability of just below 1:  ur= .999994   
For aluminum which is paramagnetic, the ur=1.000022 

I mainly found Alu cores in the oscillators of Philips FM tuners of table, portable and car radio sets (manufactured mainly for the European market in the 70's and 80's).

Thanks,  Gyula
   
Group: Guest
@gyula,

Thanks for filling your promise.
I'll open a discussion about this in my bench area when I return.
Hope to see you there.
   
Pages: 1 2 3 4 [5]
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2021-10-23, 16:25:30