Testing the TK Tar Baby

[TinselKoala]

Meanwhile, work continues, as I explain and demonstrate more features of the Tar Baby. Now, this is not a replication of the Ainslie NERD RAT device, is it. Even though it uses the same circuit diagram, the same component types and values, and makes the same oscillation and load heating (allowing for its lesser power, of course).... Tar Baby is NOT a NERD RAT device. You see, it's not on a white pegboard, so that disqualifies it totally.

Nevertheless, it performs exactly as that other device does, for some reason, and the various tests and demonstrations that I have shown here could just as well have been performed on that device..... and perhaps THEY SHOULD BE so performed, especially the drawdown test that I have illustrated, or the more complex ones that others have mentioned. But certainly.... if anyone is talking about claiming a prize, a monetary prize, their claims should be tested in a manner that would indeed hold up in a court of law.

My only claim about Tar Baby is that it performs just like the NERD RAT device in all significant respects. If they are allowed to analyze their data in the way that they have shown..... then so am I, and when that's done, Tar Baby is just as "overunity" as their device.  If I am required to perform a battery draw down test... then so are they.

Wait... I've already done that. Where is the NERD RAT test?

And... Stefan.... as we have shown to my satisfaction, even though I have shown a FG charging a battery.... even so, I see no problem any more with allowing her to use the FG in her tests. Let her simply do the same demonstration shown in the second half of her video from last year, where the device is heating a resistor element to nearly 200 degrees C. Let her run the device in that mode continually for 48 hours, then perform a battery draw-down test. Let her use the function generator.
Take away all possible excuses for her not to run the test. Make sure the device stays in the "large heat" mode though, by videoing the scope, with it set in an intelligent and interpretable manner.

It is important, when doing her runs leading up to the battery draw-down test, to make sure that the circuit is actually doing "something", like heating up a load significantly. Using the settings shown in the first part of their video, where essentially no power is being drawn, the big battery pack will last a long time and will need to be run for days before a draw-down test might show some difference.

Meanwhile...
the effect of added inductances is again demonstrated:
http://www.youtube.com/watch?v=4xffhPdoNK0

[TinselKoala]

Why, in the second part of the NERD video, do they remove one battery from the pack, leaving only 4 behind, for a total voltage (nominally) of 48 volts? This has never been explained by the NERDs or anyone else that I can find. (Please correct me if I'm wrong).

Consider the fact that I uncovered by examining the scope trace. They are oscillating the Q2 stack and very little current if any is coming through there. But.... they are turning on the Q1 mosfet by the positive FG offset. Turning it ON, without oscillating.

Let's assume that it turns fully on. I think it is only partially on in the scope shot they've shown, but with just 1 or 2 volts more on the positive offset of the gate signal this mosfet would be fully on.

And the load resistor is quoted to be 11.11 ohms.  So, neglecting other circuit resistances, at 60 volts, the original battery voltage, the mosfet would have to conduct I = V/R = 60/11 = around 5.5 Amps. Call it 5 amps allowing for other resistances in the circuit.
At 48 volts, this figure drops to 4.3 Amps.

Referring to the IRFPG50 data sheet, we find the following:

Under Absolute Maximum Ratings, we find that at a temperature of 25 degrees C the maximum drain current is 6.1 amps, and this falls to 3.9 amps at 100 degrees C.

If they had used the full 60 volts battery in the second part of the demonstration, it is very likely that they would have blown the Q1 mosfet.... and that is why they pulled one battery out AND NEVER EXPLAINED WHY.

A more precise calculation including the other resistances isn't much better. The internal resistance of the IRFPG50 mosfet is 2 ohms when the device is carrying 3.6 Amps, from the data sheet.  The circuit, with the 48 volt battery, the roughly 2 ohm mosfet , the 11.11 ohm load and the 0.25 ohm CVR has a total resistance--neglecting clipleads-- of about 14 Ohms. By Ohm's law I = V/R = 48/14 = 3.4 amps. And at 3.4 amps, a 2 ohm mosfet must dissipate P = I^2 x R = 23 Watts. At 60 volts, the current is 4.3 amps and the mosfet must dissipate over 36 Watts. The mosfet will not survive long, since it is not adequately heatsunk and its current is over the absolute maximum for a hot mosfet. Since they are using a 50 percent duty cycle the average power is half that, but they are still at the edge of the performance envelope of the mosfet, and that is with a good heat sink, which their Q1 does not have.

The fact that they have heavy heatsinks on the Q2 mosfets... when they never carry much current unless not oscillating and turned properly on -- while the Q1 mosfet is not heatsunk except for that tiny bit of aluminum u-channel ... and does carry lots of current when the device is getting a positive offset gate signal--- this is another bizarro feature of the NERD device that makes no sense at all. Or rather...  it makes sense in that it once again demonstrates the incompetency and lack of understanding on the part of the NERD demonstrators.

Tar Baby's Q1 mosfet on its small heatsink, still better cooled than theirs is likely to be, nevertheless runs quite hot and passes even MORE current if it is allowed to get too hot, approaching thermal runaway. The result of putting a fan on it is dramatic... as it cools off the current might drop from 850 mA to 500 mA, more in line with the expected value for the drive setting.

[TinselKoala]

One can even see the progression. In the early still photos of RA's setup in the broom closet, we see the board with the single, Q1 mosfet installed on a tiny heatsink and no mention of other paralleled mosfets, along with the 11.11 ohm "custom" load, a common water heater element.

This mosfet would clearly be overstressed with 60 volts of batteries and the positive offset gate drive that is necessary for "big heat" in the load.

So, somebody said, well, let's parallel a bunch of mosfets together, and don't forget to heat-sink them. So they slapped the other four on there, hooking them up backwards by accident, and then got caught up chasing oscillations.... and STILL wound up blowing the Q1 mosfet when trying for load heating using the 60 volt battery, because they STILL had no clue as to what was actually happening, AND they screwed up the paralleling plan which would indeed have saved their Q1 if only they had done it properly.

[TinselKoala]

OOhh. Scary. Where did this come from, suddenly?

[fuzzytomcat]

OOhh. Scary. Where did this come from, suddenly?

Humm .... kinda fuzzy .... hehehe

The date appears to be  4 APR 2012 ..... a older Tektronix DPO 4034 ..... me thinks