Thanks Les, you are right. The plates are now moved farther apart, the capacitance became much less (0.xx pf). You do trigger my thinking on stray capacitance along the wires to plates that could eventually increase the load. Anyway, I would rather start with something, then refine it step by step...

Thanks again for your reminder.

 

Could you draw the block diagram for your all system since you are asking here for help, if you can't discribe more clearly then our members can't give you the right solution to suit your needs, and label on the block to show the details.

 

That is going to be a pretty hard design....You biggest issue will be the driver circuit. You will need to drive a transformer that is going to a high reactance (a cable). That is my take from it.

They make transformers that work with that high frequency, but you will probably need to gain them in series to get a step up that high, which will be a headache in itself at currents higher than an open circuit, which in it own will create another issue.

 

Thanks Gibson. Indeed a great challenge. What you said "gain them in series", do you mean to use several stages of amplifiers? My current thought is as Scott suggested to amplify to 24v or higher with mosfet, then a transformer to step up to 3kv. Dose this work?
To Scott, I have not yet a block diagram. I will make it. I did a trial yesterday per your suggestion. I can amplify sine wave to around 60vpp, then supplied to a rewound (40:2200) transformer from an old computer SMPS. Unfortunately the voltage dropped a lot above 100kHz. It is not straightforward. I will need more study or buy a commercial one.(not yet find it). Let me know if you have further suggestion.

 

The frequency response of core of transformer better could reach up to 500Khz.

I'm not sure what's the results that you want, but I'm just thinking if the purpose as what I thought then the power should be can be adjust, maybe from 2KV to 5KV, when the voltage rasing until the spark occurring and the current will drawing from uA to 0.5A or more, because the resistance between two plates will be going to very low.

If do so then the wire of transformer should be bigger and can be flows through more current.

 

Thanks Scott. I resound a ferrite core with primary wire ~1mm, which should be able to drive 2.5A per an article I read. Secondary wire is ~0.1mm, which can be for 25mA current flow. I made the turn ratio 55. I limit the primary current to 400mA only. Current for the secondary won't exceed 8mA in principle, the wire thickness should be sufficient. However, with the thin wire and long length, the wire loss could be big . I may try to reduce the turns and do a trial again. (I calculated minimum turns for primary is 16 in my case, assuming Bmax is 0.15T. I wound it for 40 turns previously, still much room to reduce it)