Don Smith device, Zilano final, Korean patent

[kloakez]

Hi, I didn't want to hijack Andy's thread so I started mine. It is based on this post by Andy.

The circuit I plan to build is approximately like this:
   

V1 is a signal generator used to switch the n-channel MOSFET M1, I used IRF840A because I have it available. V2 is a DC source here but in reality I have a VARIAC with a FWBR and a 100uF 500V capacitor for smoothing. R1 is just for convergence of the simulator. D5 makes sure there is no flyback to the MOSFET, D6 is a zener diode with breakdown voltage of 100V to simulate a spark gap (I have 75V, 90V, 470 V, and 1kV). C1 determines the energy that will be released when the spark gap fires, C2 is used to set resonance with the L1. The L1 is a 853 turns of 0.3mm wire on a electrical steel core. L2, and L3 (not on the diagram but otherwise identical to L2) are on the same core with 100 turns of 0.3mm wire.

A circuit like this should help me set the input DC voltage for different spark gaps, and set the switching frequency to virtually any voltage up to about 10kHz, I hope, which should make it possible to do experiments with resonance.

Compared to Andy's circuit I have put the MOSFET on the charging side, rather than the discharging side as ChatGPT advised this should be kinder to the MOSFET as it is not on the discharging path and it is protected by the D5 diode against flyback.

[kloakez]

Using the VARIAC was not a great idea. Sparks, explosions, smoke. Not recommended.

[kloakez]

I found another downside of the so called "high-side" switching, of the MOSFET being on the high voltage side, which necessitates isolation of the signal generator, PWM module and the like.

Surprisingly, learning this didn't kill my signal generator, trying different combinations of spark gaps did, because I attached the signal generator to a circuit with charged capacitors. And the component tester went out the same way.

So my experiments ended before they could properly start. Maybe I will return to this when I can afford new equipment.

[Sandy]

Hi Kloakez. Sorry to hear your equipment died.
If NDR in the switching element is the only source of OU according to the Korean patent, then maybe its better to avoid the spark gap (and LC until NDR is observed in switching alone). In my experiences static SGs are too unstable for reliable tuning anyway. Another experimenter told me that SCRs can exhibit NDR and that this is well known although it was new to me.
You can definitely experiment without the equipment you lost (Hope oscilloscope is OK). If I was starting over I would build high quality decade boxes of capacitance (highest voltage I could afford), resistor or load bank decade box and zener diode decade box.
Another lesson is perhaps that AI is good time saver for calculation and writing code, but is not a voice of experience regarding OU. In my opinion, AI is just practical extension of the internet. It is a resource for learning but is also absolutely full of shit. And the more we jump to google or AI for answer before thinking deeply first, the more stupid and dependant we become.
Hope you are not too discouraged.
Kind regards, Sandy.

[kloakez]

Hi Sandy,

thank you for the message. Oscilloscope is fine, I use 100:1 probes, it should withstand up to 60 kV, but the probe was attached to a wire that got hot, so it melted. The rest is not expensive to replace.

I wonder how an SCR would operate in this circuit. I'll try to simulate it in LTspice, I have some SCR's at home that should work to about 800 V. What I am concerned about is that when I connect a HV supply, the voltage will be rising fast and if I don't release it, it can reach very high voltage and destroy some of the parts in the circuit, whereas a GDT would fire. And the other thing I am concerned about is what will happen in the case of resonance, then the voltage would be moving up and down and then maybe the SCR will not close. I am not very experienced with SCR's but let's try.

Agreed that LLM's are not a voice of experience regarding OU. What I use it mostly for is circuit design and learning. It would be best to learn from someone who knows electronics, but using LLM's is better than nothing.

Currently I am trying to figure out what experiments to do, getting a new component tester, and looking for other parts. The capacitor decade would be great to have.

[Sandy]

Hi Kloakez.
You could use a low power HV supply like from a cheap ion generator or cheap bug killer and a high ohm resistor (10 megaohms +)between your cap and the HV +. If negative resistance in the switch is what you seek then single discharge events will be enough to find that effect, you don't need to pump a resonant circuit, so component overvoltage will happen slowly enough for power to be switched off manually before a problem occurs.
Kind regards, Sandy

[kloakez]

Hi Sandy,

I have a couple of HV power supplies from China, they have various capabilities and some are better for one circuit and another one is better for another circuit. Or so it seems to me. Not sure what could be the cause of overunity. But what I believe is that the dv/dt discharge causes disruptions in the electromagnetic medium and that creates a sudden magnetic field change which induces current in the surrounding conductors. When I was trying to tune my circuit to resonance, I was trying out various spark gaps - 75V, 90V, 470V with everything else being the same. With increasing spark gap voltage the frequency of pulses went down, but the LED's which were connected to a rectifier and a large capacitor for smoothing on the secondary, were shining brighter. I think that could be caused by:

1. higher current induced due to greater magnetism due to the higher dv/dt
   
2. better impedance matching for more efficient energy transfer
   
3. some other effect like the negative resistance
   

or a combination of all three. So I am trying to optimize these parameters. With the LC on the spark gap tuned to resonance on the primary, but on the secondary too, there must be excellent energy transfer and I want to see how that will go.