I'm wondering whether I can take a 1 Vpp sine wave (frequency = 100 kHz), and step it up to a 2000 Vpp sine wave using a transformer. The load would basically be a capacitor (literally two thin electrodes spaced 5 mm apart, with a nonconducting dielectric crystal in between).

The application is this: I have a "reference" sine wave from a commercial lock-in amplifier. It is 1 Vpp at 100 kHz. I would like to use it to switch a Pockels cell, to modulate the amplitude of a laser beam. The Pockels cell needs 2 kV. It acts as a high impedance load.

Can this be done? What kind of transformer would I be looking for? (keywords?)

I spent some time searching for transformers. I looked at power, audio, and pulse transformers. They all seemed to be tailored to other applications, such as stepping up/down line voltages or high power applications.

Note: If my specs are an issue, I could possibly do with stepping 10Vpp up to 2kVpp at 10kHz instead.

 

What about using your capacitive load with a series inductor so that it is at resonance? This will then be a series tuned circuit. If the Q is high enough you will get 2000V without a transformer.

If you can't get the Q high enough then a combination of a loosely-coupled air-cored transformer and a tuned secondary with your load as the capacitor might do it.

A few hundred turns on a toilet roll tube might be good place to start!

A more sophisticated design might use a ferrite (probably a pair of U shaped cores) with a large air gap.

Edit: I have attached a drawing showing some possibilities. If the load is a very low value of capacitance, you might want to add some extra in parallel so that the series inductance is a sensible value and stray or changing capacitance of the load is swamped out.

 

JDT - Good idea. I have a few capacitors rated to 10kV; I have to check, but I think they are around 0.1-1nF. I also have several toroidal ferrite cores. I'll see if I can play around with them today and respond with my findings. Thanks!

 

How about a two-step process? Take the 1V signal, pass it through a "power amp" (NFET?) to boost the signal to say, 50Vpp, and then use that to excite the primary of an aircore transformer with a more reasonable turns ratio.

 

How about a two-step process? Take the 1V signal, pass it through a "power amp" (NFET?) to boost the signal to say, 50Vpp, and then use that to excite the primary of an aircore transformer with a more reasonable turns ratio.

For the preamp ("power amp"), could you imagine an N-type MOSFET doing this job? I have many 100V (and higher) N-channel MOSFETs lying around. I'm used to using them for switching, but here I imagine one would need to be operated in the "linear region" to get a sine wave output. I just haven't done that before.

Here's an example of a linear JFET amplifier, but a MOSFET could be used, too (source):

This could then drive a step-up transformer and/or high-Q resonant LC circuit.

What about using your capacitive load with a series inductor so that it is at resonance? This will then be a series tuned circuit. If the Q is high enough you will get 2000V without a transformer.

If you can't get the Q high enough then a combination of a loosely-coupled air-cored transformer and a tuned secondary with your load as the capacitor might do it.

A few hundred turns on a toilet roll tube might be good place to start!

A more sophisticated design might use a ferrite (probably a pair of U shaped cores) with a large air gap.

Edit: I have attached a drawing showing some possibilities. If the load is a very low value of capacitance, you might want to add some extra in parallel so that the series inductance is a sensible value and stray or changing capacitance of the load is swamped out.

Thinking more about this, I think the series inductor is important. Without it, the primary's voltage would rise in response to the amplification from the high-Q resonator on the secondary. That would expose the function generator to a high voltage. The series inductor (assuming it's rated to HV) would protect the function generator.

My initial check suggests that the Pockels cell has a capacitance of around 10pF (measured with an Agilent 34405A Digital Multimeter and 8inch leads). I'm going to see if I can improve that measurement.

(Also - there's a chance that I won't see this through. I'm considering another path if the time investment for this project gets too great. We'll see.)

 

And the resonant series inductor has another benefit: If the load changes - goes short-circuit, or if you touch it, the circuit goes out of resonance and the voltage collapses. So short-circuit proof and safer!

 

That depends on how quickly the circuit rings dowm from its 2kV resonant voltage to the input voltage or below, you might still get a nasty RF burn if you touch the wrong part.