Testing the TK Tar Baby

[TinselKoala]

TK,

Try the original circuit again, with the 3.3V zener, and LED load, but place a 470p cap in parallel with the 1N4007. See if that gets it to oscillate and be more stable. In the sim, the Fo goes down.

What sort of weirdness are you seeing with your simple DC zener test?

The original circuit, with LED load, 3.3 V zener and 470 pf in parallel with the rectifier: nothing. No oscs. Without the zener, the cap in parallel with the 4007 causes the f0 to go down.

The zeners seemed to have a lot of leakage in the reverse-biased direction when still below the Zener voltage using my crude test. But when I put two in series and put them in the circuit, they behave like they should, decreasing the f0 to about 1.875 MHz and providing a little more room for the ringdowns to show in the traces. 

[polln8r]

Any DBT scheduled for the Altoids machine?

[TinselKoala]

Any DBT scheduled for the Altoids machine?

No, I have no plans for such, although I will do some capacitor-powered runs when I take the DSO measurements later this week.
I know the battery runs down already. Did you know Radio Shack gets nearly five dollars for a 9v battery?

The Altoid demonstrator is sort of an exercise in clear thinking. The claim has been made several times by several people that even a negative mean _current_  would indicate battery recharging and overunity performance. This little Altoids box, based on the NERD oscillator circuit, shows that a negative mean current measurement, along with the overall negative mean power product and a decreasing energy integral, can be easily obtained from a circuit that nobody pretends is OU. The measurements are obtained by taking an ordinary feedback oscillator and modelling the wire inductances in the NERD circuit with discrete components in Altoid, and requiring that measurements be made in the same manner that the NERD team performs and analyzes them.

Now... strictly speaking..... one would need to do a comparison of such parameters as LED brightness or resistor temperature (output) and compare those to control circuits using only DC, and see  which is more efficient... maybe the battery simply lasts longer making the same output.... Well, I'll happily pass that testing torch to someone else. .99 designed it, it works in the sim, I built it, it works in hardware, so let someone else test Altoid for overall efficiency.

What I want to do next is some actual heat profiles on Tar Baby.  This will require that I have some input as to just which of the many waveforms and operating modes I am supposed to be testing. The NERDS, of course, are all over the map and since I do not believe anything from them other than the raw data, I'm not sure just what is considered the "optimum" for getting high heat in the load plus optimum battery....er..... non-use.

When I do the heat profiles I intend to do them like this: I will "tune" the device for the desired waveform using a "dummy" identical load. Then when the device is properly tuned I will switch to the "running" load in its oil bath at a good low starting temperature, and I'll let the circuit run, recording a time-lapse video of the times and temps. I'll either run for a fixed time or until the temp reaches equilibrium. Then I'll let the load cool back down and repeat, except I'll use DC power at the same level that I measure/calculate as was supplied to the circuit during the experimental run.
I'd like to do at least three runs at each waveform setting, starting with freshly charged batteries.

All of this will take some time.

But I'd like some opinions as to which waveforms to use. Here are some of the ones I've been considering: Straight oscillations only, negative bias, supplied by a regulated PS (this simulates the FG set to make the negative gate signal as shown in many of the NERD shots); or the FG set for a bipolar pulse with 18-20 percent ON duty cycle as claimed for some of the high heat trials. I'll do them using a battery supply of 60 volts, even though the NERDS have used anything from 48 (for their later high heat trials) to 72 volts nominal.

[PhiChaser]

@TK: Wouldn't the ideal circuit 'tuning' be at the lowest percentage 'on' duty cycle while still getting the circuit to 'ring' between those 'on' cycles?
Is that a sharp sawtooth or a short/long bipolar pulse?

PC

[TinselKoala]

@TK: Wouldn't the ideal circuit 'tuning' be at the lowest percentage 'on' duty cycle while still getting the circuit to 'ring' between those 'on' cycles?
Is that a sharp sawtooth or a short/long bipolar pulse?

PC

I dunno. What do you mean by ideal?

Here's what I think: When the duty cycle is "ON" the mosfet(s) in the Q1 role can pass substantial current and there are no oscillations. This is when the load can heat substantially. And the Q1 heats too, so this is why I say _role_ and not position. That is, I now believe that later, high-heat trials likely used the schematic in Paper 2, not Paper 1, so that the positive gate signal turns on the stack of 4 and the negative signal turns them off and makes oscillations in the now lone Q2 role. In the demo video, you will recall, it is certain that the positive drive went to the lone Q1.... but they also had to pull out one battery and use only 48 volts for the high heat demo, to avoid destroying this lone transistor on its tiny heatsink.

So.... I think that according to their "thesis" the oscillations are what charge or prevent discharging from the battery. Right? But I think that we have shown that these oscillations, at least in Tar Baby where experimental procedures are known, do not produce substantial load heading. All of the NERD trials where high heat is confirmed and credible waveforms are available show substantial "on" duty cycles and substantial DC power being dissipated during the "on" time when no oscillations are happening.

So the oscillations are pretty special indeed, as I have been trying to emphasise before, with all those DC power analyses. The DC portion of the duty cycle is unremarkable.... how could it be  otherwise? In a 2.5 minute period as the NERDs sometimes used, the DC power is on for 20 seconds, about. That is DC, unmagical, unspecial current flowing from a big battery through some ordinary wires and a fully ON mosfet... nothing special.... and heating up a heater element designed to be heated by a current. (The element is probably a 48 volt DC water heater element from an RV installation). 
Now.... the 20 seconds of DC, heavy current drain is over and the oscillations begin.  What must they now accomplish? They have to produce enough NEGATIVE POWER to offset the high DC power and bring the overall average down to the NERD's scoposcopy.  And they have to charge up the battery, somehow, to offset the drain of the 20 seconds of 72 volt DC supplied to a 13 ohm circuit. And they have to do this without any indication (LED, moving coil meter) of reverse current flow other than a scope's computation of undersampled and corrupted waveforms.

But.... notice how the disinformation has tried to lead us away from turning Q1 on at all.... and concentrating on oscillations only, or as you suggest, very very short "ON" times. Of course this mode will fail to produce the spectacular heats in the load.... and so the NERDs will claim that the "replications" have "failed" on that account-- once again demanding that replications show something that they themselves have never shown and in fact admit is not possible.... or they used to so admit.