PopularFX
Home Help Search Login Register
Welcome,Guest. Please login or register.
2019-06-19, 05:52:48
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 ... 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 [19] 20 21 22 23
Author Topic: Graham Gunderson Energy conference High COP demonstration  (Read 151344 times)
Group: Guest
Dear TK,

Here are the support files for the circuit shown in post #446.

Spokane1

Thanks!

As to the edit of your graphics: I use a program called mtPaint, and the Select, Copy, Paste, Rotate, Straight Line and Text tools. You did all the hard work, I just re-arranged what was already there, mostly!
   
Hero Member
*****

Posts: 845
Spokane1,

I've been giving more thought to your interpretation of Graham's output circuit and I'm curious how you arrived at your most recent design? Did you conclude this from his video, posted notes, or did Graham discuss this with you?

The reason I ask is that it suddenly occurred to me that if say secondary #1 is allowed to collapse (causing the current in secondary #1 to reverse) while at the same time secondary #2 is clamped, this would create bucking mmfs in the secondary which would cancel in the primary. IOW, the primary would reflect very little of this secondary activity.  If this is the case, then IMO both secondaries would drive the load in parallel for most of the output cycle until the narrow "off" pulse for the synchronous fets.  Also IMO, the leakage inductance between the two secondaries will be less than the leakage inductance between the primary and the secondaries so the pulse duration will be even shorter.

I have done considerable research into bucking coils with additional drive and output coils and it is amazing what can be done with various physical layouts, K factors, etc.  So, this type of output configuration now makes sense to me!

pm   
   
Jr. Member
**

Posts: 79
Spokane1,

I've been giving more thought to your interpretation of Graham's output circuit and I'm curious how you arrived at your most recent design? Did you conclude this from his video, posted notes, or did Graham discuss this with you?

The reason I ask is that it suddenly occurred to me that if say secondary #1 is allowed to collapse (causing the current in secondary #1 to reverse) while at the same time secondary #2 is clamped, this would create bucking mmfs in the secondary which would cancel in the primary. IOW, the primary would reflect very little of this secondary activity.  If this is the case, then IMO both secondaries would drive the load in parallel for most of the output cycle until the narrow "off" pulse for the synchronous fets.  Also IMO, the leakage inductance between the two secondaries will be less than the leakage inductance between the primary and the secondaries so the pulse duration will be even shorter.

I have done considerable research into bucking coils with additional drive and output coils and it is amazing what can be done with various physical layouts, K factors, etc.  So, this type of output configuration now makes sense to me!

pm

Dear partzman,

Thank you for the interest.  For weeks I didn't have a clue as to how Graham's circuit worked. I started with determining the circuit schematics from to photos. I paid close attention to the backend synchronous diode connections. Once the overall system schematic was determine I started to layout LT Spice schematics and see if I could get them to produce some of the waveforms displayed.  Early on I noticed that according to the simulations there was no overall current flow in the large backend storage capacitor. It took me a few days to figure this out.  Then came the question "If there is no effective AC current flowing through that capacitor then just how does the device develop a 10.5 V charge to run the automotive lamp?".  After playing with some simple sketches I came across one possible method that matched the observed components.

The logic output from the controller board turned out to be exceedingly simple. There was only two independent timing signals going to the synchronous diode. To make my theory of operation work I needed only one signal common to the gates of both MOSFET's. The other signal had to be for something else, so I delegated it to a master clock for the power supply.

I finally got the overall simulations to produce the tank voltage wave forms, so that part of the circuit finally came into focus for me. At least now I feel comfortable in ordering and assembling parts.

The secondary operation is probably going to prove to be more challenging.  My simulations don't produce the waveforms (no where close) to the ones that Graham showed in the presentation.
So really, I can't say if my proposed theory of secondary operation is correct or not.

Your concepts of flux reversals and collapsing fields is probably not far from the actual situation. Something like that has got to be going on in order to cause those huge disturbances in secondary current.

I'm going to be devoting more time to actual circuit measurements and less time to simulations. I've had my junk box mockup circuit running for two evenings at 12V. I don't even know if I can get meaningful operation at 12V. The simulation said I wouldn't harvest enough to run a LED. For now I just want to see if the primary circuit can tick like the working simulation does. I haven't had much luck. The wave forms observed are a mess. I can't even determine the tank resonate frequency for sure, I think it is around 26 kHz. Fortunately nothing is getting hot. My FETS can only handle 6A, and 1,200V PIV.  The transformer is not very close to what Graham used. It is an 5" square laminated steel core that came out of some analytical instrument I salvaged a couple of years ago. It is wound similar to Gram's setup in that it has a large primary and two equal secondary's placed in the same location that Graham has. There is no gap (at least not yet). I don't think that iron cores can go up to 50 kHz, so I'm going to have to increase the tank circuit capacitance and reduce the overall frequency of operation down to something that the core can respond to. Who knows I might have to go all the way down to 60 Hz.

The attached photos are of the mockup transformer and the setup layout from last weekend.

Spokane1
   
Jr. Member
**

Posts: 79
Spokane1,

I have done considerable research into bucking coils with additional drive and output coils and it is amazing what can be done with various physical layouts, K factors, etc.  So, this type of output configuration now makes sense to me!

pm

Dear partzman,

Have you had the opportunity to review the E.V. Gray technology? Now that was a verified powerful OU technology that was based on opposing electromagnets. In fact the forces generated  were so strong that in my opinion I doubt that they were actually magnetic. Graham's system points to a similar fundamental physics.

Spokane1
   
Hero Member
*****

Posts: 845
Dear partzman,

Thank you for the interest.  For weeks I didn't have a clue as to how Graham's circuit worked. I started with determining the circuit schematics from to photos. I paid close attention to the backend synchronous diode connections. Once the overall system schematic was determine I started to layout LT Spice schematics and see if I could get them to produce some of the waveforms displayed.  Early on I noticed that according to the simulations there was no overall current flow in the large backend storage capacitor. It took me a few days to figure this out.  Then came the question "If there is no effective AC current flowing through that capacitor then just how does the device develop a 10.5 V charge to run the automotive lamp?".  After playing with some simple sketches I came across one possible method that matched the observed components.

The logic output from the controller board turned out to be exceedingly simple. There was only two independent timing signals going to the synchronous diode. To make my theory of operation work I needed only one signal common to the gates of both MOSFET's. The other signal had to be for something else, so I delegated it to a master clock for the power supply.

I finally got the overall simulations to produce the tank voltage wave forms, so that part of the circuit finally came into focus for me. At least now I feel comfortable in ordering and assembling parts.

The secondary operation is probably going to prove to be more challenging.  My simulations don't produce the waveforms (no where close) to the ones that Graham showed in the presentation.
So really, I can't say if my proposed theory of secondary operation is correct or not.

Your concepts of flux reversals and collapsing fields is probably not far from the actual situation. Something like that has got to be going on in order to cause those huge disturbances in secondary current.

I'm going to be devoting more time to actual circuit measurements and less time to simulations. I've had my junk box mockup circuit running for two evenings at 12V. I don't even know if I can get meaningful operation at 12V. The simulation said I wouldn't harvest enough to run a LED. For now I just want to see if the primary circuit can tick like the working simulation does. I haven't had much luck. The wave forms observed are a mess. I can't even determine the tank resonate frequency for sure, I think it is around 26 kHz. Fortunately nothing is getting hot. My FETS can only handle 6A, and 1,200V PIV.  The transformer is not very close to what Graham used. It is an 5" square laminated steel core that came out of some analytical instrument I salvaged a couple of years ago. It is wound similar to Gram's setup in that it has a large primary and two equal secondary's placed in the same location that Graham has. There is no gap (at least not yet). I don't think that iron cores can go up to 50 kHz, so I'm going to have to increase the tank circuit capacitance and reduce the overall frequency of operation down to something that the core can respond to. Who knows I might have to go all the way down to 60 Hz.

The attached photos are of the mockup transformer and the setup layout from last weekend.

Spokane1

Thanks for explaining your logic and sharing your bench work.  Sims are OK but it is really the bench results that are important.

Your salvaged transformer does look like a suitable candidate and I agree that the frequency should be able to be lowered with Graham's concept and still work. I'm convinced that a U core is not necessary as I have worked on a configuration using common E cores which are readily available in all sizes and material types. They lend themselves to PM biasing and are easier to wind as well!

Which primary circuit are you currently using in your bench testing?

pm
   
Hero Member
*****

Posts: 845
Dear partzman,

Have you had the opportunity to review the E.V. Gray technology? Now that was a verified powerful OU technology that was based on opposing electromagnets. In fact the forces generated  were so strong that in my opinion I doubt that they were actually magnetic. Graham's system points to a similar fundamental physics.

Spokane1

I am aware of the E.V. Gray motor but have not really studied the technology to have any real familiarity with his device.

Yes Graham's secondary is rather challenging at the moment and my intuition is saying there are bucking fields involved during the short "off" time somehow.  This would explain some things including Graham's comment to the effect that there was two different magnetic operations going on in the core assembly.

pm 
   
Hero Member
*****

Posts: 845
Below are three sim variations using different means to create a bucking condition between the secondaries L2 and L4 while using the H bridge as the front end.  The primary current I(L1) is shown reversed with a minus sign for comparison to Graham's scope trace. The secondaries are connected in parallel to the load with L2 being switched identical in all three versions and L4 controlled to produce the bucking conditions during the short "off" period pulse interval.

In the first version "A", M6 that switches L4 has zero volts on the gate and therefore acts as a diode using the substrate for conduction. Although not shown for clarity, this results in L4 conducting current to the load during the time the field in L2 is collapsing thus creating bucking mmfs between L2 and L4.  The result is seen as a reversal of current in L2 with less change seen in L1. L4 does collapse immediately following the collapse of L2.

In the next version "B", M5 and M6 alternately switch L2 and L4 respectively for each cycle. IOW, when one secondary is switched off for a brief period and allowed to collapse, the other secondary remains conducting and vice versa. This action creates the bucking mmf fields with the alternating current reversals in each secondary more pronounced with less affect on the primary.

In version "C", M5 and M6 are switched with the same timing on their gates. The collapse of L4 is delayed by the addition of C3 which effectively spreads the collapse over ~10us. IOW, L2 collapses rapidly while L4 slowly collapses which again creates bucking fields between L2 and L4. The other result of this design is that I(L2) recovers more slowly in an upward ramp which most resembles Graham's wave forms. IMO however, it is the presence of the PM bias in the cores that create Graham's output wave forms.

So, my conclusion is that Graham's device has bucking fields existing between the secondaries during the brief "off" period. Oddly enough, if the gate drive voltage to M6 is the inverse of the gate drive to M5, the results will be identical to version "A".

pm     
   
Group: Elite
Hero Member
******

Posts: 3517
It's turtles all the way down
WOW! you fellows are posting stuff faster than my old brain can absorb, but I'm trying.

Deciphering  each of the V source generator pulse commands and keeping it all in ones head is easy if you wrote it, but more difficult if you are trying to decode and picture them with respect to each other.

What would be a helpful view would be to post the timing diagram for each of the pulse generators separated, not overlapping, so that the timing can be easily pictured. If you post the .asc files we can play with that and get a better understanding of the overall timing relationships for each screenshot.

I noticed you reduced the output storage cap as I also did to get shorter settling time. I guess the disadvantage of doing that is it unloads the secondaries from having to pump a high current, and what effect that has is TBD. I know you need to run the sim for quite a long time to get the true picture of the output voltage after settling. I tried playing with low values and a two pole filter, but that really bogged down the execution time for some reason. I'm new at all this.

Regards, ION

 


---------------------------
"Secrecy, secret societies and secret groups have always been repugnant to a free and open society"......John F Kennedy
   
Hero Member
*****

Posts: 845
WOW! you fellows are posting stuff faster than my old brain can absorb, but I'm trying.

Deciphering  each of the V source generator pulse commands and keeping it all in ones head is easy if you wrote it, but more difficult if you are trying to decode and picture them with respect to each other.

What would be a helpful view would be to post the timing diagram for each of the pulse generators separated, not overlapping, so that the timing can be easily pictured. If you post the .asc files we can play with that and get a better understanding of the overall timing relationships for each screenshot.

I noticed you reduced the output storage cap as I also did to get shorter settling time. I guess the disadvantage of doing that is it unloads the secondaries from having to pump a high current, and what effect that has is TBD. I know you need to run the sim for quite a long time to get the true picture of the output voltage after settling. I tried playing with low values and a two pole filter, but that really bogged down the execution time for some reason. I'm new at all this.

Regards, ION

ION,

I've attached plots showing the gate timing for "B" and "C". The timing for "A" is the same as "B" except M6Gate is always off. I have an appointment to go to so I'll have to post the asc's later as I didn't create separate files for each variation. :(

Yes I reduced the value of the output cap C2 to speed up the simulation and there is little ripple with the 50 ohm load.

To perhaps speed up your simulation, go to Simulation>Control Panel>Spice and try changing some of the defaults. For example, I'm using the 'modified trap' for integration and the 'alternate' solver as I had a little problem with convergence on one of the sims. Too small a time step will slow you down as well. I'm using 50ns minimum in the posted sims.

Hope this helps!

pm

Edit-
   
Group: Guest
partzman, looking very good there I think.

A couple of comments:

In Spokane1's latest diagrams of the Synchronous Rectifier portion, he has the "dot" end of one secondary connected to the Drain of its mosfet, and the "non-dot" end of the other secondary connected to the Drain of its mosfet.
You have both "dot" ends connected to the drains. 
I get confused about winding sense CW vs CCW, current direction and the "bucking" arrangement, since whether or not fields are bucking depends on the current strength and direction and the winding sense as well as the time it takes to produce the fields and reverse them in the first place.
So I don't know exactly what effects are produced where. But at a first pass, it would seem that Spokane1's arrangement is more like a center tap in a coil that is all wound in one direction, whereas your arrangement is more like a center tapped coil that reverses winding direction at the center tap. In other words, an "EMJunkie" type arrangement. Am I thinking about this secondary winding arrangement correctly? And what is the functional difference in performance between your version and Spokane1's?

And  ... grrr..... you are not exactly consistent with your trace color assignments in your three plots! I spent some time looking at the first one and could see that it agrees pretty closely with the Gunderson shot, with one significant difference. But then when I started looking at the other two I became confused again because some of the trace colors had changed or you plotted different voltage points or something. Maybe I just need more coffee ... or new glasses.....    :o
   
Group: Guest
.asc files please?
   
Group: Elite
Hero Member
******

Posts: 3517
It's turtles all the way down
partzman:

Thank you so much for including the timing diagrams on the plots. It makes it much easier for us all (well at least me) to understand the sequence of events, and timing changes from version to version as well as  little delays in execution  etc.

Keep up the excellent work, it is very much appreciated as well as a learning experience.

Regards, ION


---------------------------
"Secrecy, secret societies and secret groups have always been repugnant to a free and open society"......John F Kennedy
   
Hero Member
*****

Posts: 845
partzman, looking very good there I think.

A couple of comments:

In Spokane1's latest diagrams of the Synchronous Rectifier portion, he has the "dot" end of one secondary connected to the Drain of its mosfet, and the "non-dot" end of the other secondary connected to the Drain of its mosfet.
You have both "dot" ends connected to the drains. 
I get confused about winding sense CW vs CCW, current direction and the "bucking" arrangement, since whether or not fields are bucking depends on the current strength and direction and the winding sense as well as the time it takes to produce the fields and reverse them in the first place.
So I don't know exactly what effects are produced where. But at a first pass, it would seem that Spokane1's arrangement is more like a center tap in a coil that is all wound in one direction, whereas your arrangement is more like a center tapped coil that reverses winding direction at the center tap. In other words, an "EMJunkie" type arrangement. Am I thinking about this secondary winding arrangement correctly? And what is the functional difference in performance between your version and Spokane1's?

OK, Spokane's most recent posted output circuit has opposing mmfs/currents in the secondaries thru the load (which cancel) during most of the output cycle period and then does an energy capture during the brief field collapse period. In contrast, I'm taking the output of each secondary and applying them in parallel during most of the output period and then during the brief collapse period I force the secondaries to buck each other thru the various means disclosed and I'm sure there are other ways of doing this.

I'm somewhat familiar with EMJ's work but tend to think more towards the energy shuttling in these circuits myself.

I know Graham was limited by the number of current probes he had on hand but if he had included the current thru the other secondary on his scope shot, it would have answered many questions and may have given us the operation of his secondary circuits.
 
Quote
And  ... grrr..... you are not exactly consistent with your trace color assignments in your three plots! I spent some time looking at the first one and could see that it agrees pretty closely with the Gunderson shot, with one significant difference. But then when I started looking at the other two I became confused again because some of the trace colors had changed or you plotted different voltage points or something. Maybe I just need more coffee ... or new glasses.....    :o

Yes you are so right and I apologize for that! :-[ I was a little hurried while doing those sims and I did realize after the fact that I had mixed the plot colors but didn't have the time to make the corrections. I think multiple plot panes would be better with family grouping of traces in each plot. They could be expanded for clarity before posting so I'll give that a try in the future.

I've attached the A,B, and C sims below and also attached the Control Panel settings as C may give convergence problems.

pm
   
Hero Member
*****

Posts: 845
partzman:

Thank you so much for including the timing diagrams on the plots. It makes it much easier for us all (well at least me) to understand the sequence of events, and timing changes from version to version as well as  little delays in execution  etc.

Keep up the excellent work, it is very much appreciated as well as a learning experience.

Regards, ION

ION,

Thanks! :)

pm
   
Group: Elite
Hero Member
******

Posts: 3517
It's turtles all the way down
ION,

Thanks! :)

pm

Dear pm

Thanks for those .asc files.

I am at a loss to figure out how you got those timing signals to appear nicely scaled on your plot panel. I know how to open a separate plot panel, but after having watched the Waveform Viewer tutorial video, and searched other documents, I can't figure out how you did it. (am I thick or is it too simple?) I will examine your .asc files to see if there is a clue.

Many thanks again  O0

Regards, ION

TK said:
Quote
In other words, an "EMJunkie" type arrangement.

Lets hope this doesn't stick, I would hate to see EMJ get credit for an age old non-inductive winding technique.  ;)
« Last Edit: 2016-08-21, 17:41:32 by ION »


---------------------------
"Secrecy, secret societies and secret groups have always been repugnant to a free and open society"......John F Kennedy
   
Hero Member
*****

Posts: 845
Dear pm

Thanks for those .asc files.

I am at a loss to figure out how you got those timing signals to appear nicely scaled on your plot panel. I know how to open a separate plot panel, but after having watched the Waveform Viewer tutorial video, and searched other documents, I can't figure out how you did it. (am I thick or is it too simple?) I will examine your .asc files to see if there is a clue.

Many thanks again  O0

Regards, ION

ION,

I moved the traces by simply adding an appropriate offset voltage to the original plot trace. For example, in the "B" gate plot, the gate drive to M1 is untouched while the M2 gate drive is offset by adding +35v and then gate drive M5 is further offset by adding +70v, etc. Any plot trace can be manipulated by any of the math functions defined in the Waveform Arithmetic of the Help section. Another example, you can invert a trace by changing it's sign say from + to - or vice versa.

pm
   
Hero Member
*****

Posts: 845
Here is what I believe to be an interesting initial power analysis of the GG1_2secC circuit previously posted with these additions, switch S1 and it's control voltage V3. The purpose of S1 is to turn off at precisely 9.05ms and dump the energy stored in inductor L3 after the circuit has stabilized. This removes any outside source of energy to the circuit and we are left with the energies stored in the leakage inductance between L1 and L2/L4 and capacitor C2. Cap C1 starts the measurement cycle at zero volts so has no stored energy and C3 has a typical -12v across at the beginning and end of the cycle which equals ~.5uJ so is ignored.

The schematic is shown below with the changes and includes the timing for V3.

The first plot shows an overview of where the measurement cycle is located in reference to the switching of S1 plus the cursors indicating the sine wave frequency of 50.9kHz. Using this resonant frequency we calculate the leakage inductance Lleak from L1 to output with L1=1/w^2*C1 = 1.78mH.

The second plot shows the measurement cycle and the plot maths of interest.

First it is seen that the average output voltage across R6 is -12v which results in an output magnitude of 2.88w or 86.4uJ for the 30us measurement period.

Next it is shown that the leakage inductance Lleak starts the cycle at .395ma and ends the cycle at .377ma. IOW, Lleak starts the cycle with 138.8uJ stored energy and ends with 126.5uJ for a net loss of 12.3uJ.

We also see confirmation in that little current exists in L3.

Lastly it is seen that the starting and ending voltages across R6 and C2 is -12v. Therefore, no energy is removed or added to C2 during our measurement so C2 can be ignored.

So, over the 30us measurement period we are left with 86.4uJ dissipated in the load R6 and 12.3uJ lost in Lleak which is the only source of energy.

I'm not sure if this means that with longer measurement periods with circuit settling, etc,  the results will show the same apparent gain but we shall soon find out! This all came about after looking closely at the wasteful energy loss in L3 during the resonant part of the cycle.

pm 
   
Group: Elite
Hero Member
******

Posts: 3517
It's turtles all the way down
Dear pm

Thanks for the explanation re: timing waveforms.

I decided to look at C2 complete settling time so changed the sim command "start collecting data" to no entry).

Looks like it has pretty well settled out to -12 volts, with no upward or downward long time constant trend.

I guess I can try a longer sim time also. I'm sure you have a bag full of techniques  for this I've yet to learn.

I like the idea of solid grounding the secondary, makes things easier to read rather than the 1000Meg to ground.

BTW, just eyeballing I get about 3.15 watts input power, 2.88 Watts output power for an efficiency of around 91.4%. Is this right?

Edit: just noticed I was using the prior "GG1_2secC" sim without the power switch, however the settling time looks the same in either case. Not sure of the input power difference yet.

Regards, ION
« Last Edit: 2016-08-22, 14:31:02 by ION »


---------------------------
"Secrecy, secret societies and secret groups have always been repugnant to a free and open society"......John F Kennedy
   
Hero Member
*****

Posts: 845
ION and all,

The leakage current measurements as I have them indicated in post #466 are incorrect. It is not as simple as the peak currents in L1 but rather the combined currents of L1, L2, and L4. So, more analysis is needed before that test in #466 is accurate.

pm
   
Group: Elite
Hero Member
******

Posts: 3517
It's turtles all the way down
pm

Thanks for the heads up on that. We haven't heard from Spokane1 in a while, hope he chimes in.

I guess it will be time to put something together on the bench soon, with the inclusion of the biasing magnets, but I'm really not up for that yet.

I find it interesting that Graham Gunderson has not a comment (not a peep) on the excellent work being done by yourself and Spokane1. This work is all being done in the open, not on a private thread. There are almost 6000 reads of this thread...who are all those people that don't comment? Granted that you need to be signed in to comment and AFAIK he is not a member, but the work is available to be seen by the public without signing in. I wonder if Spokane1 could urge him to become a member here, or maybe the "fine print" in his contract with A&P prevent him from doing so.

Regards, ION



---------------------------
"Secrecy, secret societies and secret groups have always been repugnant to a free and open society"......John F Kennedy
   
Jr. Member
**

Posts: 79
pm

Thanks for the heads up on that. We haven't heard from Spokane1 in a while, hope he chimes in.

I guess it will be time to put something together on the bench soon, with the inclusion of the biasing magnets, but I'm really not up for that yet.

I find it interesting that Graham Gunderson has not a comment (not a peep) on the excellent work being done by yourself and Spokane1. This work is all being done in the open, not on a private thread. There are almost 6000 reads of this thread...who are all those people that don't comment? Granted that you need to be signed in to comment and AFAIK he is not a member, but the work is available to be seen by the public without signing in. I wonder if Spokane1 could urge him to become a member here, or maybe the "fine print" in his contract with A&P prevent him from doing so.

Regards, ION


Dear ION,

Thanks for the concern but I live in the country with no Internet service on the week ends, so I go off line two days out of seven. Besides we just had a wild fire sweep through the area. The fire got within in two houses of my home. All of our cars were loaded and in the process of evacuation when the wind direction changed. My wife went to stay with her daughter across town while I spent the rest of the evening watching water-bomber planes dropping red fire suppressant clay on the flames.  In the morning it looked more homes had survived than I thought. I understand at least three houses were lost, perhaps few more. It took the fire fighting services about 2.5 hours to mobilize. Once they did we had 10 engines and one tanker scattered around the neighborhood. Certainly a stressful order for the evening.

But life goes on. This weekends work showed a little promise. I was able to get the junk box mockup circuit to lock into the proper timing sequence to display the discontinuous sine wave. I just have the primary circuit operational. This evening I shall add the synchronous diode circuit and start to get a look at the backend operation. For what its worth this circuit shows:

1. This circuit will function at 12 VDC (using the single switch approach)

2. Operation can take place as low as 3.2 kHz

3. A classical laminated iron core  will support the novel oscillation pattern.

None of this means that we are in the OU window of operation, but it is nice to see some hardware implementation of the circuit under discussion. Using the component parameters I have in the actual circuit the simulation shows that I might be able to harvest up 6.5 Watts at a COP of .863. I'm sure the real world performance will be far less than this.

The advantage of the 12 V approach is that all the excitation energy comes from one source. All the gate losses, skew losses, and magnetic probe  interference issues can be evaluated. If this circuit can perform as a self runner then this is the direction to go.

The simulation also shows that the circuit performance improves if I were to use a smaller charging inductance choke. The junk box device (from a salvaged X-Ray transformer) has a 1 KHz inductance of 11.5 mH. The simulation shows that I need a 2 mH unit there. Grahams circuit apparently used a .0324 mH custom device.

If any one is interested in a schematic for this temporary evaluation circuit let me know. I'm sure there will be several variations employed as the circuit technique improves.

Spokane1
   
Jr. Member
**

Posts: 79
I am aware of the E.V. Gray motor but have not really studied the technology to have any real familiarity with his device.

Yes Graham's secondary is rather challenging at the moment and my intuition is saying there are bucking fields involved during the short "off" time somehow.  This would explain some things including Graham's comment to the effect that there was two different magnetic operations going on in the core assembly.

pm

Dear partzman,

I reversed one of the secondary coils in my simulation to make it opposing (the only change) and the COP improved from 0.265 to 0.815, so I think you are onto something.

Spokane1
   
Hero Member
*****

Posts: 845
OK, this is a test using the GG1 circuitry as in post #466 but with only one output secondary for ease of understanding and measurement. As a reminder, the transformer action is linear and there is no PM bias used.

The first pix is the schematic and plot showing the measurement points for the test. Also the vertical cursors indicate the resonant frequency to be 50.46kHz which with C1 equates to a leakage inductance Lleak of 1.99mH. This is very close to the general equation for primary inductance with the secondary shorted of (1-K^2)*Lp = 1.9mH.

The second pix is a plot of 4 cycles over 120us after the power is removed from the circuit by disconnecting L3 and dumping it's stored energy. The output power during this measurement period is 1.44w or 172.8uJ. Also seen are the starting and ending differential currents of Lleak between L1 and L2 that are 375ma and 325ma respectively. It can be shown by experiment and simulation that the effective Lleak current is 1/2 the differential currents between windings.

So, the effective starting and ending Lleak currents are 188ma and 163ma respectively which equate to starting and ending energies of 35.2uJ and 26.4uJ. This means only 8.8uJ of energy was consumed from Lleak to produce 172.8uJ in the load. Theoretically if this is correct, the energy lost in Lleak should be able to be replenished on a cycle by cycle basis for continuous OU.

Even using the differential currents in the calculations still results in gain of ~5.

One thing to note is that there is actually an energy increase seen in Lleak when the synchronous fet is not effective in the secondary. I've included the asc file and would recommend that one tries removing C2 and the fet circuit so the loads operates on AC during the same test.

pm



 
   
Jr. Member
**

Posts: 79
Dear partzman,

I just want to thank you for the tips on how to get more out of the LT Spice program. Your method for showing the timing pulses will really help explain this to others. Also the ability to change the trace background color is very helpful as well.  I'm going to snatch your MOSFET model and replace my voltage controlled switches in my future simulations - After I finish playing with the hardware.

Spokane1
   
Hero Member
*****

Posts: 845
Dear partzman,

I reversed one of the secondary coils in my simulation to make it opposing (the only change) and the COP improved from 0.265 to 0.815, so I think you are onto something.

Spokane1

Spokane1,

Thanks for trying that change. O0

pm
   
Pages: 1 ... 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 [19] 20 21 22 23
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2019-06-19, 05:52:48