Graham Gunderson's Energy conference presentation Most impressive and mysterious

[Spokane1]

Dear All,

Here is one amateur's attempt at measuring the currents that are circulating around the synchronous diode. I'm sure that the well funded folks would be using three hall effect probes (at $1,500 to $3,000 each). In this approach I'm using three surplus current transformers that I got from eBay for $40 each. I figure the five (5) turns of wire through each probe adds some shunt inductance that is very small when compared to the 39 mH of the conversion transformer (mockup) winding.

What this experiment tells me is that I don't have any current circulating around one loop - the yellow trace. From the simulations I should have equal currents in each loop - less the 166 kHz burst, and that is not happening here.

Something to work on this weekend.

Spokane1

[TinselKoala]

Are all three of the Pearsons on the secondary the same model 6164 with 0.05V/A sensitivity? Is the associated mosfet still working? The driver chip still working?

[Spokane1]

Are all three of the Pearsons on the secondary the same model 6164 with 0.05V/A sensitivity? Is the associated mosfet still working? The driver chip still working?

Dear TK,

I believe that all the Pearson current probes are from the same manufacturing lot. Some one in the distant past scored a large lot of them (about a 100 or so) and was selling them at fire sale prices. I made a "Best Offer" and got lucky when I bought three of them.

I checked the Drain to Source voltage of both synchronous diode FET's and both reported the same 575V pulse, so it appears to me that the FET might be working, but I probably need to disconnect them and test them separately. I might have a bad (or improper) transformer connection as well.

This issue will keep me busy this weekend.

Thanks for the comments.

Spokane1

[Spokane1]

Dear All,

Here are two scope shots contrasting opposed secondary's vs. aiding secondary's. The aiding topology results in greater current swings while the opposing approach reduces current swings. In each case the timing of the harvest pulse has to be adjusted since the peak currents take place at different times during the base period. By manually "tuning" the timing for maximum output the aiding connection will yield a 2 Watt output while the opposing output can only generate about 200 milliwatts at its maximum adjustment.

Please Understand that this is using a single front end switch (The simplified approach). If a full four switch H-Bridge were used then the results could be entirely different.

This exercise verifies that the three secondary currents can be observed. The impact of the current transformers on the circuit seem minimal with only 0.8 uH added on each transformer lead.

At this stage of exploration this mockup circuit is not even a very good DC-DC converter with 12 Watts in and only 2 Watts out for a COP of about 0.164.

The synchronous diode MOSFET Drain to Source voltage between the two approaches changes drastically. The third scope photo compares the two responses. The red trace is with the two secondary's opposing and it appears to suppress the voltage drop across the MOSFET's. The pink trace shows the same measurement with the secondary's aiding. Here the voltage drop is in excess of 250 Volts. This means that I will not be able to employ the more efficient 200V Vdds devices at this time. However I was able to swap out the input switch with the new device and got a 20% increase in output probably due to the lower Rds resistance.

First Photo - Aiding secondary's with harvest pulse timing in green at the bottom - blue trace is current through backend storage capacitor.

Second Photo - Opposing secondary's

Third Photo - Vds voltage rise during the MOSFET "off" time

Spokane1

Last Friday's scope trace was missing the yellow information due to a connection to the wrong tap.

[k4zep]

Hi Mark,

Well you have your circuit working perfectly!  2 watts out, additive, 200 mw out, opposing.  I wonder what the output is with no harvest pulse, that is let the addition of the output pulses simply integrate into the output filter cap. This would show if the harvest pulse is adding to the mix!  It looks just like what a normal transformer would do under these conditions.  I'm way behind you in building, testing, etc.  Obviously we all are missing something here.  I would expect under normal conditions (without the implosion effect), we could get at least 70-80% output.  Have you considered looking at the output with only a capacitance that would resonate the output coils and see what the actual waveform is without the heavy integration. Perhaps this would also show if rectification of the ringing with the harvest pulse is occuring.  I'm sure all these questions are already on your mind. 

Thanks again for all the prodigious work you have done.

BTW, my K4ZEP email was hacked this weekend and everyone in all my saved list was sent an email with a funky message.
I fixed the problem (new 16 bit random generated password and 2nd level notification of susp. activity.)  It might slow the @@$@$ holes down that do this, I don't know.

Ben K4ZEP