another small breakthrough on our NERD technology.

[TinselKoala]

Yes TK.

That gradual ON and OFF is what I mentioned earlier on about your wave forms getting close, but without that characteristic. You didn't understand what I meant then, but you do now.

With the use of your delayed time base, the evidence is there, you got it.

.99, what's the frequency of the oscs in the sim waveform shot? I can't read the timebase on my blurry copy. I'm consistently getting "approximately exactly" twice the frequency stated by the presenter in the video which relates to Rosemary's claims. This is confirmed by both the "manual" method of estimating freqs from the timebase, and also by direct hookup to the Philips PM6676 counter (an excellent performer, by the way: robust, stable and accurate.)

(Since we are now agreed that the "first" paper's circuit is the correct one to use, and since it is THAT circuit which was used in the demo video, I see no reason why the video isn't a valid topic for discussion and info-extraction.)

[TinselKoala]

Rosemary,

It would seem to be much easier for you to just state, "yes, paper one is correct" than all that "commentary".

What on Earth did I do to deserve that response?

PW

She's beginning to detect a hint of scepticism in your endless insistent demands for an explanation of unexplained features of this open-source, community-development project. So she's starting to turn on you.

I'd like to hear her explanation of why one battery was removed from the stack, for the second part of the video which used the now-known-to-be correct circuit, where high load heat was demonstrated. Why did she only use 48 volts for that part of the test? Wouldn't her ...  things ... have gotten even hotter had she used the full 60 volts ? Or the _actually full_ 72 volts she often reports using? Wouldn't that have been even more impressive than showing a non-immersed water heater element getting up to a mere 190 C?
(You do remember her saying that her MOSFETs don't heat up, don't you?)

Why in the world would one drop the voltage for these high-current tests using the "approved" schematic and circuit construction? Was there some kind of problem?

Rosie Poser said,

But either way - as TK has shown - it gives precisely the same result.

Note the distortion to suit her purpose. What I said and demonstrated was that there is little noticeable difference in the waveforms. There is more or less fuzz on certain portions of the trace and the feedback oscillation frequency is different, but not significantly so. But there is a much more important difference that has practical implications for the circuit's live performance.... and that involves the heat and current handling capacity of the mosfet(s) that is (are) turned on by the positive-going gate drive pulse during the high load heat mode of operation.
This is very different from me saying or showing "precisely the same result".

Why are four mosfets on large heat sinks, yet the lone mosfet is still on its bit of aluminum u-channel? Considering how the circuit actually behaves, the second schematic makes more sense (even though they didn't manage to use it for the video demo). So I am VERY happy indeed that Rosemary has decided that the FIRST schematic is the one to use. Remember, replicators: your Q1 mosfet must be on a small bit of U-channel.... and you must use 72 volts at some point, sustaining a load at high temperatures by using a positive-going gate pulse of 12 volts, just as Rosemary has shown. Oh... wait, sorry, she never showed using 72 volts, only 48.

[poynt99]

.99, what's the frequency of the oscs in the sim waveform shot?

TK,

As I recall, the actual scope shots show a Fo of about 1.3MHz. I get about the same in my simulations.

[poynt99]

TK, just an FYI:

For the video demonstration, Rosemary and her team had moved Q1 to a heatsink of lower Rth compared to the U-channel heatsink. It's still not nearly as large as the heat sinks used for Q2-5. See the attached.

[MileHigh]

There is somewhat of an enigma about this "high heat mode."  Assuming that you are looking at a "correct" data capture where the Q1 MOSFET is functioning then you have the gate drive high and the CSR shows significant current flowing through the circuit.  There are no oscillations so I have to assume that the battery voltage is stable.

So during this part of the cycle you are looking at a 100% conventional setup, a single MOSFET, Q1, switches on and DC current flows through the load resistor and the MOSFET.  Certainly there is a likelihood that the MOSFET could be overstressed and is dissipating more power than you would like it to, but everything is 100% conventional and the DSO should record positive power during this phase.

Then if you go into oscillation mode and the other four MOSFETs start to spasm, isn't the implication that the DSO is recording enough "negative power" to completely compensate for the positive power recorded during the Q1 ON phase?

Again, I think that you are still overlooking my sticky point.  The schematic in the first paper IS NOT the true schematic.  The error is that the schematic in the first paper shows the function generator grounded before the current sensing resistor, but in fact I believe that the function generator is grounded after the current sensing resistor.

What do you have to say about this issue Rosemary?

My theory is that in the original single-MOSFET setup you had the function generator grounded after the current sensing resistor.  Then when you added the four extra MOSFETs and miswired them, the function generator remained grounded after the current sensing resistor.  It did not occur to you that the four extra MOSFETs would conduct current straight through the function generator itself.

Is this what happened Rosemary?  You have to give us the straight goods.

As a reminder to all.  Poynt reverse-engineered the schematic and I confirmed that all appeared to be correct.  I did not literally see where the function generator ground lead was clipped in, but I am still assuming that the ground lead was clipped into the same place were all of the scope probe leads were clipped into, and that was the battery ground node.

So my theory is that once you go into negative oscillation mode, you have the wildly fluctuating battery voltage and an AC signal on the current sensing resistor that is not even the battery current.   The DSO in looking at these two junk signals saw what appeared to be very high negative power.  This false negative power measurement was enough to wipe out the positive power that was measured while Q1 was on.

I will repeat my contention:  The true battery current waveform during oscillation mode is unidirectional pulsing DC.   However, what is seen at the current sensing resistor is symmetrical AC that is AC-coupled through the gates of the Q2-Q5 MOSFET array.  Those are two radically different waveforms.

MileHigh