Eldarion and Bruce's build of Bob's Energy Converter

[eldarion]

All,

I have completed parts stuffing for now.  The board is complete enough for me to resume testing with--I did find a major error with regards to the ADCs, so before the system is put into production I will need to run another batch of boards.  I have reenabled two of the ADCs with some hand work, but the other six will probably have to stay disabled.  Other than that the boards are in good shape.

The pulse trains to the IGBT gates are now pretty much perfect, and exactly according to the diagram Bob provided.

I have tried firing the coil, but am still having no success (COP << 0.5 and no DC on the longitudinal windings).  I will double-check everything and try again soon.

Eldarion

[Bob Boyce]

All,

I have completed parts stuffing for now.  The board is complete enough for me to resume testing with--I did find a major error with regards to the ADCs, so before the system is put into production I will need to run another batch of boards.  I have reenabled two of the ADCs with some hand work, but the other six will probably have to stay disabled.  Other than that the boards are in good shape.

The pulse trains to the IGBT gates are now pretty much perfect, and exactly according to the diagram Bob provided.

I have tried firing the coil, but am still having no success (COP << 0.5 and no DC on the longitudinal windings).  I will double-check everything and try again soon.

Eldarion

Yea I hate when that happens. I've had my share of board mistakes. This is one major reason why I want to thoroughly test a board layout prior to releasing it.

So I should consider buying one of those Spartan 3 starter kits with the -1000 gate option? When I was reading some material about the 32 bit microprocessor, it looked like it took 900k gates to emulate the processor. This was real old material, does it still tie up that much resources for the newer code? Also, would I have to get the EDK to be able to load already written code? This shows how little I know about complex FPGA systems. I hate to have to spend more for development tools than the hardware costs.Cost of development tools was one of the primary reasons I chose the Atmel AVR for the hydroxy gas stuff over some of the better choices available from other mfrs. Also, which character LCD would be a good match for that Spartan 3?. I was thinking about ordering one of these 20 X 4 with backlight shown here

http://www.futurlec.com/LED/LCD20X4BL.shtml

I'm wondering if the Spartan 3E-1600 might not be a better choice for this and future experimentation, despite the much higher cost. Being so unfamiliar with the FPGA has me undecided in which direction to consider going. I like learning, but I also hate making a first purchase only to regret not having spent a bit more for a much more usable device. Oh, is the Spartan 3 board small enough to fit within the hole in the center of the core if stood on end?

Bob

[eldarion]

All,

I have completed parts stuffing for now.  The board is complete enough for me to resume testing with--I did find a major error with regards to the ADCs, so before the system is put into production I will need to run another batch of boards.  I have reenabled two of the ADCs with some hand work, but the other six will probably have to stay disabled.  Other than that the boards are in good shape.

The pulse trains to the IGBT gates are now pretty much perfect, and exactly according to the diagram Bob provided.

I have tried firing the coil, but am still having no success (COP << 0.5 and no DC on the longitudinal windings).  I will double-check everything and try again soon.

Eldarion

Yea I hate when that happens. I've had my share of board mistakes. This is one major reason why I want to thoroughly test a board layout prior to releasing it.

So I should consider buying one of those Spartan 3 starter kits with the -1000 gate option? When I was reading some material about the 32 bit microprocessor, it looked like it took 900k gates to emulate the processor. This was real old material, does it still tie up that much resources for the newer code? Also, would I have to get the EDK to be able to load already written code? This shows how little I know about complex FPGA systems. I hate to have to spend more for development tools than the hardware costs.Cost of development tools was one of the primary reasons I chose the Atmel AVR for the hydroxy gas stuff over some of the better choices available from other mfrs. Also, which character LCD would be a good match for that Spartan 3?. I was thinking about ordering one of these 20 X 4 with backlight shown here

http://www.futurlec.com/LED/LCD20X4BL.shtml

I'm wondering if the Spartan 3E-1600 might not be a better choice for this and future experimentation, despite the much higher cost. Being so unfamiliar with the FPGA has me undecided in which direction to consider going. I like learning, but I also hate making a first purchase only to regret not having spent a bit more for a much more usable device. Oh, is the Spartan 3 board small enough to fit within the hole in the center of the core if stood on end?

Bob

Yeah, go ahead and get one of the Spartan -1000 starter kits.  That FPGA actually has 1 million logic gates on it!  (Fully loaded with a MicroBlaze soft processor, nine ADC interfaces and 4 three-phase precision pulse generators, as well as the VGA display system and PS2 core, only 70% of the chip is used.)

Yes, you can load a pregenerated EDK file onto your board.  All I would do is send you a bitstream file, and you could load that straight into Impact (the downloader/programmer software tool) and program the board.

Any LCD will work; I just have to write the hardware interface.  The one you found looks good.

Before I put too much more work into the "extraneous" stuff, I would like to see this thing generate some excess power!   I hooked up some power resistors instead of the primary coils and scoped across the IGBT drains.  These things are switching almost perfectly, and the pattern is exact.  Yet I am still getting tiny amounts of power out?!?   Not sure what the problem is.

Oh, and I put the varistors across the primaries like you did on the PWM3E--they get very hot, so I still seem to have a terrible ringing problem.  I still do not understand why your PWM3E seems to switch the primaries with almost no ringing and why I am having such a terrible time with a very similar output drive system.  Any ideas?   I do not have any snubber diodes connected as you do in your PWM3E, maybe I should add them to the circuit?  I don't see where that would help, however, as my ringing waveform usually does not go negative.

I even tried the HV potential, cranking it up to 160VDC, and still nothing!  I am not sure where to proceed from here.

Oh, all tests were performed with a 7W 120VAC light bulb connected across the output terminals, for some tests (without HVDC potential) I just had the bulb connected across the secondary output leads, and for other tests (with the HVC potential) I had it connected through the HF choke and DC blocking capacitor in their appropriate locations in the output circuit.  It lit very dimly in all cases, and as the frequencies were swept, it would go out completely sometimes.

Any ideas?

Thanks,

Eldarion

[MarkSnoswell]

First, good job Mark!  Are you driving those primaries open-ended or closed?  Also, are you using Bob's pulse sequence yet?

I am driving closed loop primaries on my ferrite core. This is still just prelimenary testing -- it has taken a while (and 6 blown drivers!) to get the switch working  - both fast and reliable. I can now push transition times to just below 10ns measured on the coil

I am working on a design for a monostable switch that will deliver a minimum width pulse. The idea is to get the pulse width so short that it fits within the coil -- traveling along as a pulse down a transmission line. After witching I will then close the loop so the pulse can circulate around the coil IN ONE DIRECTION. This requires a very novel mosfet switch be closed across the coil once the pulse is injected -- the switch is a new configuration of oposing mosfets that form on AC switch with almost no voltage drop at all!

I am also plan to move up to some faster drivers:
Intersil drivers http://www.intersil.com/products/pt/parametric_table_503.asp   *** fastest driver you can get and a resonable price - $4.60 each.
EL7518  8ns rise and fal time into 1000pf, 12A peak current, 40MHz continuous opperation.
http://www.intersil.com/cda/deviceinfo/0,0,EL7158,0.html

I am posting parts, data and supplier information in a public forum here http://www.overunity.com/index.php?topic=3544.msg56820#msg56820

If money is no object then the fastest mosfet gate driver soon to be availble has under 2.5ns rise and fall times with 15A current peak. It can create 8ns wide pulses. It is the DEIC515 gate driver from IXYS RF (was Directed Energy) for $21. http://www.ixysrf.com/pdf/driver_ics/deic515.pdf
Currently shipping is the DEIC420 with a 3ns rise time and 20A peak current http://www.ixysrf.com/products/mosfet_driver_ics.html

IXYS have some great application notes on their web site. Here is a link to the PDF for their low side gate driver evaluation board -- it's worth examining for general tips on how to drive mosfets fast http://ixysrf.com/pdf/switch_mode/appnotes/evic420.pdf


mark.

PS. The HV happens when I dont shunt the back EMF back into the supply rail. So the HV is expected - but there are still a few mysteries one of which is... Why I cant see it on the scope-- I am wondering if I am getting a pulse forming/sharpening action as the pulse travels around the core -- the result being that the pulse sharpens up and is simply too fast to see on the scope.  This would be happening due to the limitation of the ferrite to react fast enough so the pulse piles up...  but why can I draw the same sparks from the core?  capacitive coupling from the windings to the core??

[eldarion]

All,

Just to let you know, I have found yet another large instability in my pulse generator, so the above null results are probably incorrect.  The instability is quite the difficult bug to smash!

Eldarion