Rosemary Ainslie Quantum Magazine Circuit COP > 17 Claims

[MarkE]

Now let's take a look at my traces, using the Plain Q17 circuit and the MarkE Gate Booster.

The Link DSO can only display 2 live traces, but the data can be dumped into a spreadsheet and the same Math trace can be calculated, just as Ainslie's LeCroy does internally. I plotted the first thousand data points (out of 32k total) on the Math trace, it only captured two complete cycles but I think it's clear that I am seeing here the very same "negative mean power product" that Ainslie produces from her measurement setup, and for the same reasons. The "mean power product" here is about -2.15 Watts for the entire 32k data set, which is equivalent to 20 full scopescreens worth of data.

Yet my load gets warm nevertheless, as well it should.... since it has around 13.5 - 14 ohms total resistance and a supply of 33-36 Volts, and the mosfet is "ON" for around 1/3 of the time. Yet this clearly present power isn't visible in the computed Math "instantaneous power" trace. Just as in Ainslie's measurement system.

Those oscillations really look like Miller capacitance to me.

[MarkE]

And finally let's go to the Steve Weir "Shifting Paradigms" board with its precision 0.25 ohm, properly connected CVR and its on-board Vbatt filtering. The same input settings as before give us these scopetraces and spreadsheet-computed Math (Instantaneous Power) traces.  Finally we see something that makes sense! A realistic amount of power is shown during the "ON" times and there is no "negative mean power product", the mean power computed across all 32k samples is about 31.3 Watts.
The peak power is right around 80 Watts, just as it should be with a supply of about 33.3 Volts and a total circuit resistance of about 14 Ohms.

I would be interested, and I am sure Steve would too, to see how things look probing the current sense at the compensated test point on the test board.  I expect that the negative spikes will be further attenuated.  Even that nice precision current shunt resistor that is on the board still has enough inductance to cause peaking on fast edges.  The compensated test point exhibits flat bandwidth, presenting an even more accurate waveform.  You should use a 10X probe with it to limit probe capacitance effects.

The other thing that you should do and can do with these traces is first subtract the current sense voltage from the battery voltage.  That board uses the Q1 source as the common.  So the battery voltage is reading the battery plus the current sense.   Given the 8 bit resolution of the oscilloscope, it won't make a lot of difference, but it should flatten out the power during the off intervals.

[TinselKoala]

Here's the comparison between the two probe positions, the "uncompensated" one designated for the Scope, and the "FComp" one which is supposed to be for a DMM, I think. I slid the VBatt probe over to the TP3 to make the reading. See the "normal" and the "FComp" probe point images below. I didn't invert the traces for this shot, so that the numbers in boxes would be easy to interpret.

The FComp position does reduce the negative ringing even more.

It will take me a while to do the trace subtraction shots you recommend. I know that the board works that way, I've just been too lazy to set up the subtraction on the scope.

All these recent shots are using the MarkE Gate Booster, with +15 V supply from the HP721a.

[MarkE]

Here's the comparison between the two probe positions, the "uncompensated" one designated for the Scope, and the "FComp" one which is supposed to be for a DMM, I think. I slid the VBatt probe over to the TP3 to make the reading. See the "normal" and the "FComp" probe point images below. I didn't invert the traces for this shot, so that the numbers in boxes would be easy to interpret.

The FComp position does reduce the negative ringing even more.

It will take me a while to do the trace subtraction shots you recommend. I know that the board works that way, I've just been too lazy to set up the subtraction on the scope.

All these recent shots are using the MarkE Gate Booster, with +15 V supply from the HP721a.

I think that Steve's original intent for the FComp was for DMM filtering.  I think he changed his mind.  My recollection was that he and Josh spent an evening or two tweaking and testing boards they had built up.  I saw some really clean capture waveforms.  I may have one around here someplace.

[orbut 3000]

It looks like Ms. Ainslie can't suffer polite an reasonable people today.
Could it be influenced by the lunar phase?