looking for a high voltage (1kv) transistor

[ChrisA]

One question: Can someone recommend a high voltage NPN transistor?

I want to build a lab bench power supply for use with vacuum tube amp prototype design. I need variable voltage from about 500V to 100V DC I'd like to design for up to 1KV (2X safety factor) but I'd never use more than about 650V Current up to about 100 mA if I can

I don't need good regulation but I want continuos variable voltage. So my plan is to use an LM317 to compare output volts to a set point and control the base of this transistor so that the LM317's input end is about 5 to 10V higher than the set point. Very simple. The LM317 will shunt the last 10V

But the transistor needs to take 1Kv C to E at 100ma. That is 100W worst case.

Does such a transistor exist. If there are several I want the "most generic" and easy to source part.

Yes I can hunt parts catalogs but I want a common part that "everyone" knows and is easy to source in the US. Parametric searches don't give you that kind of info

[Mooly]

Look at BUT2520F although power dissipation may require you to parallel (with current sharing)
http://docs-europe.origin.electrocom...6b80027ad0.pdf

Apart from power rating there are many suitable devices.

Edit... why not use FET's such as 2SK1120 etc
http://www.datasheetcatalog.org/data...shiba/2998.pdf

[Ardee]

Why not use a FET and LR8 regulator? You might also want to have multiple taps on the transformer to reduce power dissipation when working at lower voltages.

[Elvee]

The problem with bipolars is their frighteningly small SOAR at high voltages: the BUT2520 in the example above only accepts less than 30mA Ic at 800V.
This is the case for most high voltage bipolars.
Some have a Vceo of 1.5KV; maybe they could do a little better, but I don't have detailed data:
2SD300, 2SC643, BU500, MJ12005...
The best solution IMHO is to use a MOS: this guy could easily handle the job:
http://www.st.com/stonline/products/...2/stw4n150.pdf

[Serge66]

Hi,

I think the IRFBG30 would do the job.

Some IGBTs are rated to 1200V.

Cheers,

Serge

[star882]

Have you considered building a flyback converter?

[infinia]

But the transistor needs to take 1Kv C to E at 100ma. That is 100W worst case.

I think 2x Vmax as a safety margin is unreasonable. ( a linear AC/DC converter it doesn't happen) I would agree with an added current and power requirement. Because the hardest part of a lab supply design will be current limiting the thing!

Offline Flybacks need 2x voltage, but it's almost impossible to design for 5 to 1 output voltage control. They like PWM for current/power control. You will probably need 2 stable loops under all conditions to even attempt it.

[multisync]

a very long time ago when I was young and careless (stupid) I drove a tube output transformer backwards with a square wave and a "H" switch. A "H" switch is just another name for a bridge. The bridge was driven by a square wave generator, ne555 timer, and appropriate current drivers. I used a single power supply +48 volts and a bridge on the output of the transformer with a small value high voltage cap. The bridge was a stack of 1n4007's. My VTVM ( vacuum tube voltmeter) could not read the output as it was was greater than 1.5KVdc.

since you want to vary the output DC you would have to use something like a high power audio amp to drive the transformer to vary the output DC. If you drive the transformer with a square wave, the output rectified DC would not need a lot of filtering. You could also use a higher frequency 1khz for example. Picture a square wave going thru a bridge rectifier, you end up almost with DC with very little ripple.

[multisync]

you could also use a low voltage transformer backwards. 220v to 12volts, you would have to check the maximum freq of the transformer. You need about 5x bandwidth to get something that looks like a square wave.

[jackinnj]

My suggestion -- regulate the primary -- you can use an opto-isolator to sense the output voltage and drive the error amp on the primary side. Back when I was a lab RAT we used this to derive very clean voltages for our NMR stuff.

It's not difficult -- you can use an amplifier and a sine source to drive the primary -- even a paralleled LM4780 type. Yes, you'll need a DC power supply, but You'll also burn less power this way -- just the power dissipated by the amplifier in Class AB.

Make sure that the insulation on the secondary can stand off the voltage.