Devoloping a Variable High Voltage DC pulse Help.

[Qmavam]

I want a 0V to 10kV variable DC pulse. The frequency 1Hz to 100Hz, maybe 1000Hz.
At this time I'm developing the HV DC with a Variac driving a neon sign transformer and a bridge rectifier.
 I'm thinking of setting up a 555 with variable frequency and pulse width and driving a gate to
supply the HV pulse. But that is where I run out of knowledge.
  How do I gate a high DC voltage?
Or is there another way to do this? (low cost is wanted, so far had everything scavenged)

                                 Thanks, Qmavam

[Twospoons]

What's the application?  It makes suggesting solutions much easier if we know what you want to do with this pulse.
Missing in your spec are : duty ratio, output current, pulse shape, pulse energy.

Id start by suggesting you create the pulse at lower voltage (anything up to a few hundred volts) then use that to drive a step-up transformer.

[Qmavam]

What's the application?  It makes suggesting solutions much easier if we know what you want to do with this pulse.
Missing in your spec are : duty ratio, output current, pulse shape, pulse energy.

 Electrophoresis, very low current, 30ua, but have seen 500ua. I am thinking 1Hz to 100Hz with a 10% to 70% duty cycle, easy to do with a 555.

I'd start by suggesting you create the pulse at lower voltage (anything up to a few hundred volts) then use that to drive a step-up transformer.
[/quote]

 If I drive a pulse transformer with lower voltage square wave, It would still need rectification, what would that pulse look like? I was hoping for pretty fast edges under microsecond, but not nanosecond.
                                 Thanks, Qmavam

[Twospoons]

Hm, in that case maybe a cascoded BJT stack would work better for you.
https://commons.wikimedia.org/wiki/File:Cascode-voltage-ladder.png
BJTs over 1000V are not too expensive. eg https://nz.mouser.com/ProductDetail/STMicroelectronics/STN0214?qs=pmcUWs1B1IKOCzUPe9wAgg%3D%3D

The killer in your application might be off-state leakage current. I'd suggest using LTSPICE (or similar) to check before building.

The transformer approach is made difficult by the volt-second requirement, and the high bandwidth (short risetime). Not impossible, but difficult.

[Qmavam]

Hm, in that case maybe a cascoded BJT stack would work better for you.
https://commons.wikimedia.org/wiki/File:Cascode-voltage-ladder.png
BJTs over 1000V are not too expensive. eg https://nz.mouser.com/ProductDetail/STMicroelectronics/STN0214?qs=pmcUWs1B1IKOCzUPe9wAgg%3D%3D

The killer in your application might be off-state leakage current. I'd suggest using LTSPICE (or similar) to check before building.

The transformer approach is made difficult by the volt-second requirement, and the high bandwidth (short risetime). Not impossible, but difficult.

  That's pretty neat and something I could build, but, I wonder about inductance and if it could cause voltage spikes that would kill the transistors?
i.e. I would need to put it together properly.
 Thanks, Mikek

[Weston]

Is there a risetime or fall time requirement? What is the polarity (is the signal bipolar?)? What determines your duty cycle?

1Hz is pretty low, but at the higher end of pulse repetition/lower end of pulse width you could possibly use a NST or something similar as a pulse transformer.

If you need a fast rise or fall time you will need to directly gate a high voltage source. Any sort of switch that can do that is going to be complicated / expensive.

1Hz - 100Hz is slow enough where if you have a high-ish frequency AC source and a voltage multiplier you could just gate the drive for the AC source. The rise time would be determined by the current sourcing ability of the AC source and the fall time would be determined by the load resistance. You could add a parallel resistor to speed it up fall time if needed, but you would be burning more power and possibly impacting rise time.

[Twospoons]

Are you doing something a bit special with the electrophoresis? Most of the commercial units seem to be around 500V max, the highest I saw was 3kV. I didn't see any mention of pulsing at all.