Graham Gunderson's Energy conference presentation Most impressive and mysterious

[Spokane1]

Well.... hmmm.

Now I'm confused. In previous posts you've said that the harvest pulse is turning the mosfets _off_ for a very brief time (see the diagram attached below) but now your mosfet Gate trace (CH4) looks like it's going HIGH and turning the mosfets _on_ for a very brief time and leaving them off for the remainder of the cycle.

So is the substantial current you are reading here, actually going through the mosfet body diode? What am I missing here?

What is that 50R, 10W load resistor? Is it a wirewound resistor?

Yes, I'll bet that ringing you are seeing is a "real artifact" in the sense that it is really happening but is something you don't want happening and may be caused by stray inductances due to wiring length or maybe the inductance of that load resistor.

Dear TK,

I have to check my wiring again. My intent was to have the MOSFET's on 99% of the time and then apply a short pulse to open them. But as you have noticed my pulse logic appears reversed. I even connected the gate drive logic to the complement output of the last timer to achieve this. I shall get right on this and see what is happening.

The load resistor is 2" long and about 0.250" in diameter. It is dipped in a brown ceramic like substance. It could be a carbon resistor, but it just as well could be something else. Would my new DE 5000 LCR Meter determine how much inductance we have here. Certainly worth a try. I shall also get a photo. Most all the power resistors I have above 5 Watts are the classical wire wound type. I understand that even the aluminum ribbed Gold ones are actually of the wire wound variety.

When my computer gets back from the evacuation exercise (The wife took off with it in her van) I shall do some simulations with added inductance across the back end storage capacitor.

I really doubt that I have enough bias magnets to saturate that huge iron core so I would suspect that this apparatus is still operating in classical mode, which is fine for now. I should probably look into making a purchase of some neo magnets.

If there is enough inductance to form a parallel tank circuit with the back end storage capacitor, then that would imply that those larges currents are actually oscillating around the loop as the Current Transformer reports.  I wonder how this is possible since that HF current (166 KHz) would have to flow through one winding. I would think that there would be a lot of attenuation there. I suppose that one winding could be in some kind of brief oscillation and not really impact the slower steady state of the core's magnetic flux very much.

Thanks for taking a look at this. I'm sure this issue will be one of many as we dig deeper.

Spokane1

[Spokane1]

Dear TK,

That last scope shot was suffering from a misallocated ground reference connection (i.e. I plugged it into the wrong hole). The circuit was working properly but the scope was reading the wrong logic signal. The attached photo is what the last one should have looked like.

I adjusted the timing of the harvest pulse to a different part of the secondary current trace. In the second photo you can see the harvest pulse, but there is no impact to the secondary current. If there were some kind of parasitic resonance then I would think that I should get that ring no matter where the pulse took place, but I'm not sure of that.

The last photo is the load resistor in the circuit.

The next photo is the LCR meter measuring the resistance.

The next photo is the LCR meter reading the inductance of the resistor at 3.7 uH.  I haven't calculated what kind of resonate frequency takes place between a parallel 3.7 uH and a 1500 uF filter capacitor.

Perhaps this information will help determine what is going on.

Spokane

[TinselKoala]

@Spokane1:

OK, thanks for the information. Your resistor is definitely a wirewound one with substantial inductance, but running the R and C values through my favorite resonant frequency calculator gives a frequency of about 2.1 kHz, much too low to explain the ringing you are seeing.
http://www.1728.org/resfreq.htm

Moving the harvest pulse as you have shown to a point near the zero-crossing produces much less ringing, since there isn't much current being interrupted. Plus, you've changed the vertical scale from 50 mV/div  in the ringing shot, to 100 mV/div in the non-ringing shot, so any ringing would show up correspondingly smaller anyhow.

[Spokane1]

Dear TK,

After I replace some chips in my logic controller (that Smoke thing again) I shall explore making other measurements around the secondary and see what else is going on in the neighborhood.

I'm going to have to write up something to explain to people that we are working with two topologies here:

1. The Simplified single switch approach that I have been exploring because of its cost advantage.

2. The full Bridge and two logic signal approach that partzman is developing all those detailed simulations from.

Each approach serves its own purpose, but it is important that people new to this thread realize the difference.

Spokane1

[Spokane1]

Dear TK,

Maybe I will not be swapping out my synchronous diode MOSFETS for the 200 Vdds models you recommended. Take a look at the Drain to Source voltage spike that is developed when the backend FETS turn off for 3 us. That spike is 575 Volts high, no wonder Graham uses the expensive 1.2 kV devices.

Normally I would be considering snubbing networks, MOV's, Zener Diodes, or added capacitance, but at this part of the circuit I fear that any of this could potentially diminish the OU effect we are looking for.

Do you happen to have a part recommendation that can handle these kinds of sharp voltage pulses? I suppose the price goes up with the Vdds rating.

Spokane1