High voltage nanosecond pulse generator

[researcher]

Dear All,

I would need for one of my research project a special nanosecond range pulse generator. This is very easy so far... But the output voltage for the generator must be in 20-40kV range! I have found some very unusual and expensive solution for this task, but I would need some relatively simple and cheap solution.
The required specification:
Peak pulse output voltage: >20kV
Pulse duration: variable 10ns-1000ns
Pulse repetition rate: variable 100Hz-10kHz

The variable ns pulse generator is basically not a problem, I have already find some good circuit for that. The challenging part, how to generate such a high voltage. The most obvious solution would be to drive a fast MOSFET with a ns pulse generator and use a special high voltage transformer.
But I am absolutely open for any kind of solution, opinion, suggestion.
I am not an electrical engineer, so a detailed schematic is highly appreciated!

Thank you very much in advance!

G.

 

[spec]

Dear All,

I would need for one of my research project a special nanosecond range pulse generator. This is very easy so far... But the output voltage for the generator must be in 20-40kV range! I have found some very unusual and expensive solution for this task, but I would need some relatively simple and cheap solution.
The required specification:
Peak pulse output voltage: >20kV
Pulse duration: variable 10ns-1000ns
Pulse repetition rate: variable 100Hz-10kHz

The variable ns pulse generator is basically not a problem, I have already find some good circuit for that. The challenging part, how to generate such a high voltage. The most obvious solution would be to drive a fast MOSFET with a ns pulse generator and use a special high voltage transformer.
But I am absolutely open for any kind of solution, opinion, suggestion.
I am not an electrical engineer, so a detailed schematic is highly appreciated!

Thank you very much in advance!

G.

Hi Research,

For a couple of years I worked next to a lab were they were generating high voltage fast pulses around 6KV and I am afraid to say it is no mean task.

One vital piece of information missing from your request above is how fast are the edges of the pulse, so the the rate of change of voltage (and related current) can be determined.

A MOSFET and high voltage transformer would simply have too much inductance and capacitance to generate a fast pulse.

If I remember correctly, one approach is to use a massive low ESR capacitor with a very high current capacity and discharge it by a very fast trigger device. The resulting pulse is then fed into a delay line device which reflects backward and forward to multiply the voltage. I think this technique is used in radar systems to drive magnetrons.

Would you be able to build such a system at 20KV- afraid not.

spec

 

[simonbramble]

Try Googling Royer oscillator and Cockcroft Walton multiplier. These are 2 methods of generating high voltage, low current supplies.

 

[researcher]

Hi Spec,

Thanks for the reply.

Actually I have already built that system you mentioned. Its really not a big deal. See the image below.
I used a simple flyback to generate 10kV@10kHz then this voltage is multiplied by a simple 4 stage Cockroft-Walton. The 40kV voltage has been switched by a variable spark gap and a Blumlein line pulse forming network did the rest of the job.
The final purpose of this project is to generate ns pulse dielectric barrier discharge plasma in atmospheric pressure, so I have build this system as a "proof of concept" and it works well. As you can see the size of the system is huge and the sound of a discharging spark gap is like a machine gun. This is a good (but very dangerous) toy to playing with, but can not be applied in a practical system.

What I would need now is some kind of compact and semiconductor based solution of the ns HV pulse generation.

Regarding your question, I am not sure, but lets say max. 10ns as a rise and fall time of a single pulse.

Regards,

G.

 

[spec]

Hi Spec,

Thanks for the reply.

Actually I have already built that system you mentioned. Its really not a big deal. See the image below.
I used a simple flyback to generate 10kV@10kHz then this voltage is multiplied by a simple 4 stage Cockroft-Walton. The 40kV voltage has been switched by a variable spark gap and a Blumlein line pulse forming network did the rest of the job.
The final purpose of this project is to generate ns pulse dielectric barrier discharge plasma in atmospheric pressure, so I have build this system as a "proof of concept" and it works well. As you can see the size of the system is huge and the sound of a discharging spark gap is like a machine gun. This is a good (but very dangerous) toy to playing with, but can not be applied in a practical system.

What I would need now is some kind of compact and semiconductor based solution of the ns HV pulse generation.

Regarding your question, I am not sure, but lets say max. 10ns as a rise and fall time of a single pulse.
View attachment 103981

Regards,

G.

Wow I am impressed- you obviously know more than I thought and also have the construction skills.

Could you post a circuit of your set up and maybe someone with the expertise can assist in miniaturizing it?

spec

 

[ronsimpson]

What I would need now is some kind of compact and semiconductor based solution of the ns HV pulse generation.

Easy to say.

I built something like what you show in high school. Rise time was good but the fall time was no good. I see how to fix the fall time. Certainly not semiconductor or small. (snap crackle pop, reset near by computers)
The picture started to come in then went away. I can not get it now.

 

[Dick Cappels]

One possible replacement for the spark gaps is a very long string of avalanche transistors.

icecube.wisc.edu/~kitamura/NK/Flasher_Board/Useful/research/RSI03066.pdf

 

[researcher]

Thanks for the answers.

Well, the final goal is not to miniaturize the existing Blumlein line system (which is basically impossible) but to built a completely new device without spark gap and this absolutely unreliable cascaded transistor system.

I have found some very good solutions, based on these papers:
https://aip.scitation.org/doi/10.1063/1.3572265
and
https://ieeexplore.ieee.org/document/5471156/
(I am not sure I can upload fulltext papers here due to copyright issues...)

They use simple magnetic core pulse transformer and some "magnetic switch", I don't fully understand.
The problem is that the authors didn't give the full, detailed schematic, so for me its almost impossible to replicate.

 

[spec]

Thanks for the answers.

Well, the final goal is not to miniaturize the existing Blumlein line system (which is basically impossible) but to built a completely new device without spark gap and this absolutely unreliable cascaded transistor system.

I have found some very good solutions, based on these papers:
https://aip.scitation.org/doi/10.1063/1.3572265
and
https://ieeexplore.ieee.org/document/5471156/
(I am not sure I can upload fulltext papers here due to copyright issues...)

They use simple magnetic core pulse transformer and some "magnetic switch", I don't fully understand.
The problem is that the authors didn't give the full, detailed schematic, so for me its almost impossible to replicate.

Hi Research,

The links are talking about edges of 10ns and minimum pulse widths of 70ns. That is a far lesser requirement than what I though you wanted: 1nS edges and 10ns pulse width.

Can you in fact, accept an order less in speed?

spec

 

[ronsimpson]

One possible replacement for the spark gaps is a very long string of avalanche transistors.

Dick,
When I first saw you here; "I know this man." We worked on some project. Last time we talked you gave some names but none were right. There are others here I know from years ago.
Try these name: Jim Webb, Display Labs, Data Ray & CRT monitors, TV monitors, LCD monitors.
Ron Simpson

 

[Dick Cappels]

Oh...you ARE THE Ron Simpson!

 

[ronsimpson]

THE

funny

Some people have the job of eating sashimi, drinking saki, and convincing the customer
¨we can do that¨. I have the job of staying at home and making electrons flow. Most find the job we really want, then we complain about it.

 

[Dick Cappels]

researcher You might be looking for like this:

SLAC-PUB-5432 February 1991 (A) PULSE SHARPENING ... of magnetic pulse compression, ... design a multi-stage magnetic pulse compression line using simple ...