Bifilar pancake coil overunity experiment

[forest]

You mean that anomalous effect Tinsel ;-) ? 
https://www.highfrequencyelectronics.com/index.php?option=com_content&view=article&id=1956:an-anomalous-coupling-mode-supports-wide-band-rf-pulses-in-a-high-q-narrow-band-resonator&catid=161&Itemid=189

[TinselKoala]

I hope you will at least agree that the turn-off time of the transistor is critical. The faster turnoff the better, right? This is how we maximize the return pulse from a coil, by cutting off the energizing power as rapidly as possible.


It turns out that the 2sc945 is really slow to turn off in the unmodified circuit. Below behold some scope traces. Yes, I've spent another afternoon on this project, while still waiting for AyeAye to reproduce his circuit and scopetraces so we can engage in some interactive testing.


First trace below is the unmodified circuit with the 2sc945 transistor. The blue trace is the Vce, collector voltage with respect to emitter. When this trace is High the transistor is Off. The yellow trace is the signal from the FG to the base of the transistor. You can clearly see the turnoff delay and the rise time of the collector voltage.



Second trace below is the circuit modified with a 1k pulldown resistor between the transistor base and emitter, still using the 2sc945 transistor.


Third trace below is circuit with 1k pulldown but transistor is now MPSH10. This is a _much_ better result in terms of turnoff delay and rise time, and results in a larger coil return pulse when the circuit is tested as originally specified.



(I'm a political man, but I practice what I preach...)
https://www.youtube.com/watch?v=WOPDzD_P9gg

[ayeaye]

Thank you TinselKoala, we will go ahead.

[ayeaye]

It turns out that the 2sc945 is really slow to turn off in the unmodified circuit. Below behold some scope traces.

Yes, i noticed that myself, the long delay of c945 switching off is actually very weird. MPSH10 is certainly better, mostly because the slope is much sharper.

With oscilloscope i measured that my power supply voltage is 11.6 V, and it remains consant all the time.

It is actually not easy to take the analog oscilloscope screen images, as seen on the figure below, this is almost the best one can get. And yes, the brightness of traces is decreased for them to be sharper, yet this is all one can get. With a web cam, one may get some better images with some special cam perhaps.

My 555 timer somehow doesn't work any more, Arduino is the only other thing i got, so i'm now using that instead. And yes, i'm using it the way that it always stays connected to the computer, so i can always change the frequency and duty cycle from the terminal on the computer. It is not good to connect the circuit ground to the computer ground, but with the R2 resistor 1k, it is completely safe. But one should be very careful with it, and otherwise use Arduino with potentiometers and not connected to the usb or such, as they often do.

This screen image is of the voltage on the collector of the transistor, oscilloscope ground there is the circuit ground, the frequency was 100 kHz, duty cycle was 30%, time scale was 1 us and y scale was 2 V.

As you can see, the voltage on the open transistor starts almost from 0.8 V, and then drops to 0.4 V. By the c945 datasheet, the maximum voltage on the saturated transistor is 0.3 V, so maybe the transistor was not fully open.

Based on the above, i think that one can consider that the voltage on the open transistor was 0.6 V, and that it is always constant when the transistor is open. Thus my last scripts that were written assuming that the voltage on the open transistor is constant, can be used, and only the voltage on R3 should be measured with the oscilloscope. Except that i considered there that the voltage on the open transistor is 0.3 V, as it should be on the saturated transistor according to the c945 datasheet, this should be replaced with 0.6 V.

This makes the measurements also more precise, as R2 is relatively large, and a small change of voltage on it does change a lot. But calculating the input part instead from the power supply voltage, the voltage drop 0.6 V on the transistor, and the voltage on R3, is quite precise. Not to talk about measuring two traces instead of one, which even on digital oscilloscopes where waveform data can be taken from the oscilloscope, may cause some error of measurement or calculations.

[TinselKoala]

Photographing an oscilloscope screen does take some practice. Keep trying, you'll get it eventually.


Here's one important thing: If you are photographing for quantitative purposes, you _must_ indicate where the channel baseline is on the screen. That is, you have to switch the channel input coupling to "ground" for a moment to show exactly where the channel's baseline is set, before turning the input coupling back to DC (or rarely AC) to make your measurement photo. Otherwise nobody knows where you are measuring from.


The scopeshot below is just a miscellaneous one from my files, showing stable display of two unrelated frequencies. I think the graticule and trace are clear enough for precise quantitative work, even though this image isn't intended for that (which is why I didn't bother to indicate baselines).