# Ultra high voltage op amp

I need to amplify a signal which is oscillating between 3V and -3V at a frequency of +- 10MHz, into the 10-20kVolts range.

I've been looking for an opamp or a transistor but the max supply voltage I've found is 1.7kV

Anyone knows of an opamp/transistor meeting those requirements? Or any other means to increase the voltage of a signal...

Thanks

Edit : Following different answers, here are some more informations. I'm trying to duplicate, with a few upgrades, a device that generate a wave at 1MHz, with an constant amplitude of 4kV. I've measured those data with my oscilloscope and a voltage divider quite easily. Example of such device : https://en.m.wikipedia.org/wiki/Electrosurgery @ "Electrosurgical waveforms"

I would like to be able to generate a carrier wave from 1MHz to 10 MHz and modulate its amplitude at will. And a bit more voltage is always nice ;-)

In order to have complete freedom on the waveform, I believe it is easier to generate the voltage digitally and use a DAC to convert it. This is the reason of the initial signal being between -3V and +3V. (Or between 0V and 5V, or 0V and 3V, ... In the micro-controller range)

I need to be able to change the signal on the fly (some software lags are acceptable but not powering down and changing a couple of inductances to adjust the resonance frequency)

Concerning the power output, less then 100 W is perfect.

Again, thanks a lot for your time and your inputs !!!

• That is an awful lot of skew and power, you will likely need an IOT or similar tube. What are you trying to do, run your own radio station? – PlasmaHH Nov 2 '16 at 9:56
• It's awesome slew rate, sure such technology exists? I know for sure there is no equipment to measure it, at least nothing you could just purchase – Gregory Kornblum Nov 2 '16 at 10:35
• You don't say how much power you will need to output at 10-20kV, that strongly affects the types of solutions that may be available. What does a frequency of +- 10MHz mean? 10MHz +/- a small tolerance like 10Hz, +/- a large variation like 3MHz, or something else? The frequency specification strongly affects the types of solutions that may be available. – Neil_UK Nov 2 '16 at 12:28
• XY problem - I suspect that you're probably going about this the wrong way. Unless there's something particularly special about the +3V/-3V signal you would probably be much better off generating a high-voltage signal from scratch rather than trying to amplify it. Can you give a bit more detail on what your overall plan is? – brhans Nov 2 '16 at 12:51
• How about gaining the signal up to ~100V p-p and putting it through a transformer? (I don't really know HV so that may not work.) – George Herold Nov 2 '16 at 19:59

## 3 Answers

I would consider using a resonant step up transformer (rather like what are used in a Tesla coil). You won't find any silicon amplifiers capable of running at this sort of voltage level so a magnetic solution seems your only viable option.

If you can drive 100 Vp-p into a step up transformer and get it to resonate at 10 MHz it should work AND be the likely easiest path to a solution BUT don't underestimate the skill, knowledge and perseverance in designing such a transformer. It will also be highly dangerous and it is likely you will need to oil dip the transformer to make it work.

• More specific, you need to oil-dip L2. L1 is an air coil around the base of L2. Most dangerous part of the circuit is C1. – Janka Nov 2 '16 at 19:17
• If I'm not mistaken, this circuit will only work at one given frequency. Changing it would require to modify the inductors/capacitors to adapt is resonance frequency. That is a bit to restrictive for plan A, but I'll keep it in mind for plan B ;) – Joachim Nov 2 '16 at 19:24
• Didn't you specify 10 MHz. Maybe you've changed your mind? – Andy aka Nov 2 '16 at 19:38
• Well I did say +/- 10 MHz, but with the opamp design in mind, not specifying, from 1MHz to 10MHz didn't seem a big issue... My bad – Joachim Nov 3 '16 at 7:07
• So now you have to get real mean with what you want to achieve because at the moment it seems like there is nothing that can achieve what you want. – Andy aka Nov 3 '16 at 9:07

Instead of trying to amplify a low voltage signal, it will be easier to create the high voltage signal directly. At that frequency and voltage, look at something called a Tesla coil. That's basically a resonant transformer with a high voltage secondary.

## This is what (transmitting) tubes are for

At the voltage levels you are dealing with (10kV+), there's really only one option left for serious RF amplification work, and that's the good old fashioned vacuum tube. A wideband linear (class-AB) output stage could easily work over the frequency range in question while generating the voltages necessary; however, you'll need transmitting tubes for this (i.e. the stuff Eimac still makes to this day). You'll need to have significant interlocks to protect the tube from improper drive in addition to providing HV safety, too.

• Especially as there are tubes out there that can generate a 10kV (and higher) swing at their output while running at 10+ MHz. +1 as this is really the only way to have cycle-by-cycle control at hobbyist prices (provided you shop on the surplus market). – Sam Dec 25 '17 at 21:26
• Thanks Olin, I have been looking a lot at tubes already but I find it difficult to find general informations about tubes. For example, you are talking about "transmitting" tube. What are they exactly, what's the difference with vacuum tube, valve, etc...? Thanks a lot. – Joachim Jan 1 '18 at 12:07
• @Joachim -- I'm not Olin, but a transmitting tube is simply one designed for RF power amplification (i.e. low stray capacitances and good heatsinking/cooling capacity). I added a link to Eimac's site on the post, they have some excellent appnotes in their library. ("Tube" is a US term btw, Brits call them "valves" instead.) – ThreePhaseEel Jan 1 '18 at 12:18
• Hi ThreePhaseEel, sorry I mismatch the name and the comment... Thanks a lot for the info, I'll look thoroughly into it ;) – Joachim Jan 4 '18 at 0:49