Developing HV DC Pulses

I even have teflon standoffs! Any comments on the Cascode Voltage Ladder?

7KV, all the Voltages on the resistor divider will be 7 times higher, for the bottom one, would I adjust the value to get the 15V. Mikek

I got some feedback on the capacitance and resistance of the vessel with an emulsion in it. With just air, 2 to 3 pf, with RP over 200MΩ, the limit of my DE-5000. Plain water with acid was 120pf, with RP over 200MΩ. With an emulsion 18pf with RP over 200MΩ. Mikek

Air is cheaper.

That would work with kilovolt mosfets too.

IXYS has an (expensive) 4500 volt mosfet. 1700 volts is more affordable.

There are high voltage reed relays too.

Is there something you don't like about the 1400V transistor? At $1.03 each.

I'm thinking pulling say 5ma through the transformer secondary while the switch is closed, and then opening the switch, to say a 5ua load, might cause an inductive spike. Is that a concern? I did find a 25,000V tube for $10. Mikek

Would those work at 8hz? Mikek

Easily. Reeds are fast. Even regular relays work in milliseconds.

Digikey has some HV relays.

I'd expect that the process is basically AC coupled, so the brief pulses can be from quiescent HV and pulsed to ground. So a passive pullup won't work very hard.

A glass capillary is going to be a lot of megohms. But even 200M is a high impedance. Megohms of pullup would work.

I'm hoping to lead them into a different type vessel hopefully with less corona discharge, I would hope somewhere in this process to get a somewhat accurate current measurement. 18pf at 60hz is 147MΩ, at 7000V is about 48ua through the vessel. A dropping resistor of 10MΩ, drops about 500V before the load, a lot but easily adjusted at the variac.

7000V /10MΩ = 700ua 7000 x 700ua = 4.9 Watts, less if duty cycle is included. Mikek My son got his first shock today, says everything was off. The 18pf could hold a charge, but, not long at all. ¯\_(ツ)_/¯

They look fine. Stack them if you need to, but bipolars need base current, and mosfets don't.

I'd look for some small, low-capacitance, protected-gate mosfets.

Old color TVs sometimes had an HV shunt regulator tube, a tetrode? Can't remember the part number. A tube would be cool, maybe cascoded with a tiny mosfet.

Two issues: a 1000 volt transistor might hold off 1 kV, but not be able to turn ON with that voltage applied (like horizontal output transistors in old CRT circuitry, the HV rating only applies with zero collector current). And, a "HV resistor" is geometrically designed to not generate ions in air; regular old air is NOT a good insulator at high fields, you can get corona discharge by using thin wire, or pointy parts, or by a stray bit of lint... Real HV resistors are fat things, and the fields around them are not well controlled if you try to attach them to a PCB with either a ground plane, or even just a fingerprint, to distort field symmetry (generate ionizing hot spots). Those fat rings of metal around van de Graaff generator parts are IMPORTANT. You can't print a wire on a PCB with fat rings around it.

A 1-cent axial isn't fat enough to do real HV, unless you put it in a vacuum, or pot the assembly.

What use is a transistor if it can't turn on? Of course it can.

I suggested using several in series. That works fine.

7 KV is not very high voltage. People sell 20KV surface-mount resistors.

A CRT horizontal output has a 100V supply, switches ON, dropping it to near zero V, then OFF, and flyback peaks the collector voltage at 1 kV.

It is NOT rated at 1 kV for its switching operation, only for reverse B-E bias OFF conditions, during a flyback voltage excursion.

It isn't end-to-end arcing that's the issue, but local E field causing corona discharge (and chemical corrosion that can kill nearby components in a few months). Diameter, not heating, is the critical bit. Spark plug wires are never less than 8mm or so, and it isn't material properties of the insulator that sets that size, it's the air at radius 4mm around a cylindrical conductor.

I suspect they're potted assemblies, then. I've got some HV reed relays, THEY were potted. The fat rings on a van de Graaff are for higher voltages, of course; megavolts, sometimes.

Run a Spice simulation of turn off and turn on. Each of the collector base capacitances has to charge up or discharge through the biassing string, and all the base currents have to come from it - MOSFETs might be a better choice, since the gates don't drawn any current when they are on or off. You might need a "gate stopper" in series with each gate to prevent oscillation.

Whatever.

But doesn't provide much mechanical support.

The drain gate capacitance can be a problem, and it changes with applied voltage. Essentially you have to get fixed amount of charge into the drain-gate capacitance to turn the part on, and get it out again at turn-off. It's usually specified on the data sheet - with the usual production tolerance. The impedance of the divider chain is at a maximum in the middle and you run the risk of the MOSFETs turning on at different rates and getting excessive voltages over one or more of the MOSFETs during the process. Paralleling the divider chain with a string high voltage zeners - 270V is the highest I can find - could limit such turn-on and turn-off excursions.

I was building 12 KV supplies when I was a kid, and I used strings of series 1/4 watt carbon axial resistors, and they worked fine. A 1B3 makes a great series regulator.

Mid-air soldered junctions are great in other situations too. The old Radio Amateur Handbooks warned against unsupported soldered junctions, but Tek just did it and sold a lot of oscilloscopes.

How can you have usenet but no google?

I am sure glad that you don't design electronics.

John Larkin trots out his standard insult. The irony is that John seems to evolve his electronics rather than designing anything.

Quite why an observation about van de Graaff generators suggests to him that Whit3rd doesn't design electronics isn't entirely clear. Whit3rd does seems to be a physicist, and quite a few of them design electronics, not always all that well, but Whit3rd does post here which suggests enough interest in electronic design that he could do it well (as some physicists manage to do).

I was trained as a chemist and most of the chemists I worked with as graduate student were hopeless at electronics. Win Hill started off doing a Ph.D. in chemical physics and he turned out pretty well.

In the famous words of Rodney King, "Can't we just get along"

Tell me how to control a tube. Can I drive it with a 555? Do I need to run the 555 from plus 3V and minus 7V or some combination to get full off and full on? Rather that a pull down resistor scenario, can I just put the tube in series with the load and switch it? 18pf, 200MΩ+ load. That would be better for upscaling, eliminating a 100watt 10MΩ resistor.

Thanks, Mikek

First thing is to pick a tube and get its data sheet.

Depends on the tube. I'd suggest a mosfet-tube cascode so you'd only need 5 volts drive.

To drive a capacitive load, something has to pull up and something has to pull down, charge and discharge.

A resistor is a cheap, safe pullup, but you shouldn't need a 100 watt resistor.

7 KV across a 10 meg resistor is 5 watts. Pulsed at a low duty cycle you'd be below 1 watt. A string of 1/4 watt resistors would be fine.

Pick a tube (and a socket!) that you can get and we'll work out a circuit. Group project. That would be fun, to design something with a tube, like in olden times.

A tube thing would look cool, as a science project. The teachers and kids have probably never seen one.