Testing the TK Tar Baby

[TinselKoala]

Thank you Stefan.

By the very nature of things, though, we are still trying to resolve several issues pertaining to the Ainslie device and claims. Since Ainslie refuses to cooperate by answering simple questions with simple answers, we feel free to speculate, test, experiment, and speculate some more.

1. We would like to know the "approved" final circuit, minus timer or FG. Until otherwise informed, since RA has endorsed the crossover circuit with the Q2x4 stack on the left side of the diagram, with its Source pins connected to the Positive marked lead of the FG, that is the version Tar Baby will continue to use, with the addition of my timer circuit in place of the FG. If it later turns out that RA is claiming some other diagram for her experiment, I'll be slightly upset but not at all surprised.

2. We would like to have a real answer to PicoWatt's question about the possibility of the RA circuit having a bad transistor in some of the scope shots presented in the two papers. I think everyone who has real experience in these matters must agree that there is something wrong with those shots as PW has pointed out, and Rosemary's explanations are not satisfactory.

3. We would like to know the real reason that one battery was removed from the 60 volt stack, leaving 48 volts, for the "high heat" positive going drive pulse portion of the Ainslie demo video. Since the schematic of Paper 1 is supposed to be correct, a positive-going drive pulse would turn on the single Q1 mosfet completely, and with 60 volts and a total circuit resistance of under 14 Ohms, this single mosfet would be near its absolute maximum current, and on the small heatsink the Ainslie crew provided, would likely exceed its maximum heat dissipation value, hence would fail quickly. It is my hypothesis that they removed batteries (they claimed to have used 72 volts in some experiments) to avoid blowing this mosfet, and perhaps don't even understand the reasons. But whether or not they do, they have never given any explanation for this glaring fact. A 72 volt supply and a positive-going drive pulse would certainly have blown this single, poorly-mounted mosfet fairly quickly.

4. The issue of "phase shifts" or phase differences has also not been adequately explained by Ainslie. She has complained about phase relations in Tar Baby, yet her own scope shots show the same phase relationships that she finds objectionable in Tar Baby. Clearly.... this is inconsistent, nonsensical, and requires an explanation from Ainslie. Since power calculations done the way RA has done them depend critically on the phase relationship between the signals used, this issue is obviously critical to any replication or non-replication duplication efforts.

5. Ainslie still seems to deny the possibility of current flow through, or current sourced by, the FG (or even the 555 timer) in the NERD (and Tar Baby) circuit. Even in the face of experimental demonstrations, circuit diagrams of the FG she herself used, and documents from major test equipment manufacturers, she continues to maintain her mistaken position. Since MY function generator, and everybody else's FG, does indeed act as a current source or a 50-ohm impedance to external current flow.... it will be difficult to make any "replication" or duplication non-replication of her device that she will agree is valid. This issue needs to be resolved, by Ainslie understanding, finally, how these things actually do work.

Other issues with reference to Ainslie, the NERD device, and the RATs claims may also arise and become significant in the context of duplication. These will obviously be discussed here, since there is no other thread available to do so.

[MileHigh]

TK:

I think that the issues that you raised in your previous posting are all relevant.  Don't be surprised if Rosie does not answer them directly but instead tries to address these issues in some future presentation of new data.  The big question being does she have some people to help her or is she biting off more than she can chew?

About the thermal stuff:

But the Weak Claim will require some crude calorimetry, at least. (At least.... if you don't believe in Clarke-Hess, the demon god of power measuring, it will.) So... I can do that. By heating up the container of oil and letting it cool, I can quantify the heat-leak rate of the insulated container, since I know that the specific heat of mineral oil is 1.67 Joules/degree/gram,and that there are 250 mL of oil in there,  and a Watt is a Joule per second, so if I heat the oil to 100 degrees and let it cool back to ambient at 20 degrees, and this takes an hour exactly, or 3600 seconds.... that would be 80 degrees x 250 mL oil x 0.83 gm/mL x 1.67 Joules/degree/gm = 28000 Joules, and per 3600 seconds... or 28000 J/3600 s = about 8 Watts average dissipation over the hour of cooling. So.... then I can use that figure in conjunction with a heat _rise_ caused by the Tar Baby and also a DC control.
First, I can use the CH2330 to measure the input power to the circuit, and see how hot the load heats per unit time. Then I can use a regulated DC power supply to give the load the same average power, but DC, and without the intervening circuit's dissipative elements. Comparing the two graphs generated from the time-temperature data will likely produce a set of curves like these that I got the last time I did this same sort of thing, a couple of years ago. This will tell me the "efficiency" of the device as a heater. And knowing the container's loss rate and the other parameters, I should actually be able to get close to the total energies involved as well.

Yikes, you are being too crude in your estimates here in my opinion.  For starters, recall that your insulated container will decrease in temperature over time in essentially the same way that a capacitor discharges.  In that sense it will "never" cool down to an ambient temperature of 20 C or more realistically it will take a long time, like 10 hours or more.  It all depends on how much precision you have for reading the temperature.

So I have a few suggestions.  Heat up the container and monitor how long it takes for the delta temperature between the heated oil and the original ambient temperature oil to drop by 63%.  So you will get your time constant for your thermal system.  Supposing for the sake of argument that it's one hour.

Once you have that number and you know the thermal capacity of your container then in theory you know very precisely what the heat power dissipation rate is at any given temperature.  So you can drop the "8 watts" number that you averaged out, I can't see any use for that information.

For the thermal capacity of your container, I think it would be worth it to factor in the glass beaker also.  I don't know if you know what type of glass it is.  If you can weigh the container and you know the specific heat of the glass then you can calculate the thermal capacity of the glass.  On the other hand, it it's a real beaker made for a chemistry lab, I assume the manufacturer would give you that information.

To be continued....

MileHigh

[MileHigh]

TK:

Here is where I think you could make good use of your thermally insulated container:

Suppose that your time constant for the container is two hours.  So then you can say that if you make tests that involve heating up the container for say 10 minutes or less, that will give you a fairly accurate reading of the heat energy produced by the load resistor.  I also am going to assume that part of your setup is something like an insulated stir stick that pops through a tiny hole in the top of the container that allows you to agitate the oil to even out the temperature.

So, for example, say the oil is at 20 C when you start your test.  You run a test where power is supplied to the load resistor for exactly 10 minutes.  Then you lightly agitate the oil for fifteen seconds, then you wait for fifteen seconds, and then you lightly agitate the oil for fifteen seconds and then you take your temperature reading.

Since the thermal time constant is two hours and you were providing heat for only 10 minutes, you know that the heat dissipation (a.k.a.; lost heat) was relatively minor during the 10 minutes.  So you can get a quite accurate reading for the heat production during the 10 minutes.  The reason I am suggesting two light agitations with 15 seconds in between is to let the glass of the beaker and the oil reach the same temperature.

With respect to the testing itself, we already know there is a kind or conundrum.  The conundrum being that we know ahead of time that massive amounts of power are "lost" in the Q2 MOSFET array in negative oscillation mode.  So he NERD circuit will never be as efficient as a heater as compared to a straight DC connection between your variable power supply and the load resistor.  So with that in mind, the whole thing might be an exercise in futility.

Perhaps, you might want to look at the "COP infinity" claim in a bit more detail before you look into the thermal stuff?  The suggestion would be to set up a nice robust oscillation in negative oscillation mode with the full battery voltage, nice clean waveforms and the whole nine yards.  Such that it at least looks like you have the waveform shapes and the "phasings" (what to what???) that look very RAT-like.

With that robust by-the-book "COP infinity" setup, then try doing an LEDs of DOOM test, and then do the big capacitor test.  Both of those tests will indicate that the battery is discharging, and you can crunch the numbers on the big capacitor test and get a decently accurate measurement of the power consumption in negative offset oscillation mode.

I don't know if Rosemary is going to do much with the positive-pulse high heat mode in her tests.  I just don't see any point, it's just a pure standard implementation of using a MOSFET to switch on and off a resistive load.  It's a no-brainer exercise in under unity if you don't count the heat dissipated in the MOSFET itself.  If you do count the heat dissipated in the MOSFET itself, then it's a no-brainer exercise in unity (or near-unity).

MileHigh

[Farmhand]

What I don't understand with the initial evaluation of these setups that use function generators is,
Why isn't the power consumed by the function generator considered as input ? Without the signal the circuit does not operate.

If they (FG's) are part of the circuit all power consumed by them is input. If the Function generator is plugged into the wall a
simple watt meter should measure the consumed power. The entire circuit would need to then produce that much power again
to break even. If a function generator is used in a battery powered system it should be powered from the batteries, even if an inverter is needed.
The alternative would be to build a low power circuit to provide the signal, which would be powered from the batteries.

If a vital part of the circuit is plugged into the wall (the function generator) then the power it consumes is part of the input.

It really is that simple in my opinion.

Cheers

[TinselKoala]

@MH:
Your calorimetry is a bit more sophisticated than mine... I was just going to approximate the dissipation rate by assuming linearity from hot to cold, but of course you are more correct and precise by using a single time constant and assuming little or no leakage.
KISS is my motto on this project.... the "choir" already knows the tune, but we are trying to convince some others that simply taping a thermocouple to a resistor hanging out in the ambience isn't quite good enough calorimetry, even for high-school science fair standards.

As far as operating in the "correct" robust oscillation mode and so forth.... except for the capacitor test part, I've already done that, twice, and tested the run batteries with the Dim Bulb Test. The only part lacking is some kind of agreement that I'm making the right waveforms and that Tar Baby is a duplication of the NERD RATs circuit and operating parameters. And we know... without the white pegboard and clipleads.... that kind of agreement will never be forthcoming from the "one who matters" in this matter.

(An interesting point that might be made with the calorimetry would be to examine the differences in efficiency between the IRF830a and the IRFPG50 mosfets. The 830a has a substantially lower Rdss and a faster turn-on time, which means that it should dissipate less heat in the mosfet itself and allow more of the total circuit power to be dissipated in the load.)

I'm still willing..... just take me to the highway and show me a sign, take it to the limit one more time.

@Farmhand:
We are indeed considering the FG as input. That is one of the major issues, in fact.

The problem is that the original NERD RATs device was built by Rosemary Ainslie, who denies that the FG can provide power to the circuit or even allow the circuit's power to pass through it. This is the motivation for Stefan's requirement that the circuit be tested with a 555 timer. Unfortunately he doesn't seem to realize that the simple 555 circuits will also suffer from the same problem: they can act as a power source or alternate unmonitored current pathway for the main circuit.

It is possible to get the circuit to oscillate without the FG's or 555's input, if one can accept just a long feedback "squeal" with no superimposed pulsations. In the main paper of Ainslie, they use a very long duty cycle of some minutes of oscillations followed by a short "off" period, so presumably the steady-state squeal would also be permissible. Once the oscillations are started with a little "tickle" they can stay on until something disturbs them. In fact, the main function of the FG or timer (besides injecting more power) seems to be to turn the oscillations _off_ during the off part of the cycle.

There is a lot of background material that you may have missed, including a lot of discussion about the FG's roles in the circuit. The main discussion of the Ainslie device is in another thread, unfortunately now locked.

Keep thinking, keep suggesting, though. It's all good....