Testing the TK Tar Baby

[evolvingape]

Facepalm.

Evolve... you do realise what a word salad that post from RA is, I hope.

(But first, I didn't "discover" this RF oscillation, I merely have demonstrated it unambiguously for what it is. It is feedback caused by a little bit of energy ringing back and forth between all those crazy wires and the capacitances of the mosfet and wiring. Circuit designers know this oscillation well and take pains to avoid and eliminate it. In fact if I just put those mosfets up close on TarBaby's board they would probably go away.)

And of course her whole argument about the battery "discharging its whole potential in an instant" in bold and caps is bogusity of the first magnitude.

Come, let us reason together.

The drain trace drops to or near zero in the oscillations because the mosfet is turning on briefly then.
I have demonstrated what happens to the drain trace when a mosfet turns on.

The current is limited by the circuit's resistance.

The Drain voltage drops to zero when the battery is seeing the minimum resistance in the circuit, which is the mosfet's Rdss of 2 ohms plus the CVR of 0.25 ohms plus the 11.11 ohms of the load. Call it 14 ohms, or 15 to be conservative and allow for cliplead crimps and such.
(Actually since it is the Q2 stack that is oscillating the resultant Rdss for the four mosfets in parallel is only 0.5 Ohm. Like that makes a big difference. Well, so instead of 15 ohms use 13.5 ohms in the following calculation.)

Ohm's Law tells us that V=IR, so I = V/R and so a 60 volt battery will push 4 amps of current through a 15 ohm load, dissipating power at the load of P=I^2R, or 240 Watts continuous power if it was on 100 percent of the time. (Here the "load" of course means the entire circuit including the warm mosfets.)

But it's not.

In the single oscillation period, modelling it as a square wave at 1.2 MHz, the pulse is On for less than a half of a microsecond, but call it 500 nanoseconds per pulse.

So the energy in each pulse of the oscillation is actually (less than, because it's not a rectangular pulse) 0.0000005 second x 240 Joules per second, or about 0.00012 Joule, or a bit less than the battery's "whole potential" of over 10 MegaJoules or so. Ten orders of magnitude ! And she thinks her math might sometimes be a "tad out".

And, as we may have determined earlier in the thread, much of this energy is "recycled" into the next period. Only the radiation and Joule heating losses need to be made up by the battery's power to sustain this oscillation.

TK,

Sure I realise.

A battery would never survive total discharge in such a manner and remain serviceable, and neither would the circuit, which should be the first indicator that it is not happening. Your approach of systematically working through each variable and building a data set to understand what is occurring is the correct one. You are allowing the data to build the complete picture and using your knowledge and experience to interpret it, as opposed to starting from a preconceived conclusion and manipulating the data to fit.

I am really pleased that you are taking the time to demonstrate and explain what is actually going on with this circuit. Thanks.

RM

[TinselKoala]

You're welcome.
Somebody told me what my time and effort might be worth as a hired consultant in these matters, but I'm embarrassed to say just how much it was. Me, I'm just taking a busman's holiday, although it has cost me a couple hundred bux in hardware and running around.


More data:

Using a sensitive moving-coil milliammeter, I have found that there is from zero to about 65 mA of current flowing through the Function Generator in TarBaby's circuit, depending on the amplitude setting and the bipolar/unipolar pulsing. If bipolar pulsing is used the current averaged by the meter is just slightly in the "negative" direction wrt the FG's usual polarity. If unipolar pulsing is used the current direction is in whatever sense the pulse is going.
I don't know if the FG is "sourcing" this current or sinking it or just passing it through. I'd guess that it is acting as a current, and hence power, source, since the output voltage of the FG is drawn down by the circuit's loading.

[TinselKoala]

Battery voltage continues to fall, down to 34.0 volts with 100 mA on the drain current DMM. This is with strict unipolar negative pulsing, and the load temp also continues to drop, down to 168 F at 2201.

When the battery gets down to 33 volts, I'll stop the test, unplug everything and let them sit until I wake up from my naptime. Then I'll do the DimBulb test, comparing these 3 batteries with the fully-charged one I set aside yesterday.

Oh, what's the point. If they measure below 12 volts they haven't recharged, even conjecturally. And the fully-charged one measures 13.15 volts right now anyway.

[TinselKoala]

I've looked at the oscillations now with the base FG pulsing set to 100 Hz, 1 kHz, 10 kHz and 100 kHz. At the highest freq there is only time for 10 or 12 so cycles per pulse but the frequency is still that same 2.25-2.4 MHz that it is at 100 Hz pulsation.

[TinselKoala]

Ok, then, the only difference in performance that I can see (besides youknowwhat) is that they cited , if I recall the narration correctly, a frequency of 1.2 MHz for the oscs and I am finding approximately exactly double that. Yes I know that's an oxymoron, but so is yer brudda.

Is there any data, meaning scope traces or dumps, that support their frequency figure for the oscs?

(ETA: In .99's detailed analysis his sim shows an osc frequency of around 1.125 MHz or so, as best as I can tell, which is much more in line with the Ainslie report. He modeled the inductances right down to the wires connecting the batteries. Why is my osc frequency so high? Do my batteries make the difference?)

ETA2: OK, by adding one through six of those little brown inductors in series at the battery positive terminal I brought the frequency down to 1.5 MHz in steps, so I'm satisfied that it's total inductance that is determining the oscillation frequency.

Will my batteries charge at the 1.5 MHz oscillations, I wonder? One more inductor should bring it down to the "Ainslie Range"... do I dare?