Amp with high voltage output(+- 800V)

[Gunders]

This is not going to be an audio amplifier, but I’m anyway posting this thread here ‘cause I hope there’s someone here with experience who can help me.

As a project at school I have to build an amp which give out +-800V, this amplifier is going to be used in a generator simulator. This simulator is going to be used to calibrate instruments who measure currents, voltage, and the phase between two generators.

The real problem with this amp is the requirement to the output voltage.
I have tried to figure out how I can make an amp which satisfies this requirement. So far I have only figured out that I have to/should use a N-channel amp with a class G or H output stage.

I have also done some research to see which high voltage transistors which are available and their max voltage.
For the N-channel types I found FET’s with Vdss up to 1600V, but this was only for the power types.
But what about the transistors I have to use in the VAS or the phase splitter?
The VAS transistors have to have the same voltage requirements as the output transistors(don’t they?).
How should I solve this problem? Is there a way I can use a different configuration, or use some kind of circuit solutions so transistors with lower max-voltage can be used?

I’m also confused about how the rest of the amplifierdesign should be done. I have build both vacuum tube and solid state amps before, but not any solid state amps with supply rails higher than about 100V, and here I must use at least a bit over 800V.
It really gives me a headache.
I would like to hear all kinds of recommendations and tips of how I can build up this amp.

[azira]

What kind of input signals are you planning on using? Why not just a step up transformer with a FET follower stage?

[Gunders]

The signals to the amp come from a microcontroller, maks output voltage from the controller will be near 5V. The frequency should be variable from 35 to 75 Hz.

I may use an interstage transformer, but how should I apply miller compensation with the interstage transformer involved?

[crown300]

If you can purchase an amp, check out:

www.apexmicrotech.com

They make some chips with 900 and 1200 volt outputs.

[Gunders]

interesting, but the modules with 900V and 1400v output can't deliver much current. I need at least 6A output.

[bigparsnip]

That is a whole lot of power (3.4kW to be exact) are you sure you need the high curent and high voltage circuits at teh same time, as it would seem far more sensible to me to use one circuit for low voltage/ high curent testing and a diferent one for high voltage/low curent testing.

But, then if you doo need one that will give +/-800V at 6A you are looking at a VERY expensive (and dangerous) piece of kit.

[cabbagerat]

Don't think it's possible. Even if you only dissapate 5% of total power in the output devices then we are still talking 480W. To keep a junction at a decent temperature (30* above ambient at the heat sink) you would need a 0.06 K/W heatsink. Which would be huge.

I seriously doubt you wil find output devices that will take this kind of punishment. If you do manage to build such a monster I would love to see a photo.

[pjacobi]

Huh, big project!

Perhaps some power devices like these:
http://www.semikron.com/skcweb/e/pro...bte_1700v.html

Can be controlled by opto-isolated drivers like these:
http://literature.agilent.com/litweb...988-8713EN.pdf


But I assume this is all meant for switched mode. Do you really want to have some kW in linear mode?

For cooling, bolt it to the water pot => boiling water cooling.

Regards,
Peter Jacobi

[Gunders]

There's no direct correlation between output voltage and current.
The specifications says that this amp have to deliver up to 800V and 6A, but not simultaneously. The ouput voltage should also be adjustable.

I know there's gonna be a lot of power dissipation for this amp, but i don't know how much.

Class D is an alternative. This is a three phase simulator and we shall also vary the phase between the different outputs, the resolution should be 0,1 degrees. Don't know if class D amps is messing things up.
Unfortunately I don't know very much about class D amsp either.

But what if we concentrate us about the high voltage first, and take the dissipation later.

[pjacobi]

This all sounds like (brushless) motor control. Read some app-notes from the IGBT manufacturers. Use some inductors instead of the actual motor. This is all peanuts, as Deusche Bank would say, in a modern electric loco, 7.2MW is put through the IGBTs.