High voltage, high frequency discrete opamp

Your requirements look a lot like those for CRT video amplifiers; consider topologies from there. For discrete solutions, typically, the output stage is cascode and pnp/npn follower after that.

Perhaps, you can use a single video IC solution. If your signal is AC sine wave, you don't have to worry about level shifting. However who does CRT ICs in our days?

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

Hmm, 6MHz on 30pF is 884 ohms reactance, so the peak current will be 170mA.

The average CRT driver parts are 1GHz, 100V and 100mA, which falls a little short. Might be able to find some in the 300V, 200mA range.

I've heard of fair bandwidths from fairly pedestrian parts, e.g. MJE350. If you drive the piss out of them (possibly more drive than cascode), you might get something.

Tim

-- Deep Friar: a very philosophical monk. Website:

Sounds like some sort of mass spectroscopy thing. FTMS? FTMS is interesting, circuit-wise, sort of like radar as regards the t/r issues.

It would be interesting to transformer couple into the gates of a pair of totem-pole nfets. Do any required dc bias with dc/dc converters or better yet PV couplers. That has some huge drive advantages... a grounded medium-power opamp becomes the low-impedance driver for each fet. You may be able to run almost open-loop, with overall DC feedback and little or no AC feedback. Or make the feedback local to the power fets themselves, with just a couple of resistors.

A complementary output stage can also benefit hugely from transformer coupling the drive. You could even float some gate drive electronics. I'd scribble something but I've gotta hit the road. Maybe later.

John

For fast slewing, a cascode with the high voltage on a bipolar transistor is good (MOSFETs that take high voltage don't have as much transconductance). You might also consider using low-voltage power amp and a transformer to boost the output. You lose DC performance, of course, but there are ways to put a separate (slower) DC amp alongside; just drive the output winding end-number-two instead of grounding it.

Vacuum tubes ARE still available, and might be useful here...

Maybe another approach: wideband low(er) voltage amplification but more power to drive a step-up transformer which drives the load.

If you can tolerate a transformer-based solution, that might be good.

Otherwise... Electrostatic deflection amplifiers (as were used in oscilloscopes) didn't generally use emitter followers since their load capacitances were typically below 10pF - EFs would add as much. Most MOSFETs have atrocious capacitances - it's for that reason, more than their gm, that makes HV bipolars more likely to succeed. Along that vein - with some effort you can get up to ~2.7x impedance-bandwidth improvement with judicious inductive coupling, more if you can tolerate some overshoot in the time domain response. See, for example

-f (designed a few discrete deflection amps for Tek long ago)

Sounds like the vertical deflection amp from a hobby grade oscilloscope from 25+ years ago. DC to 10 MHz and a couple hundred volts differential output into a capacitive load.

Hans,

If you are still out there, there are two used high voltage excite amps on Ebay in the US. Look at item 290360425190 and item 350159383389, each listed for $999. These are from the Nicolet-designed FTMS 2000 system (Nicolet, Extrel, Finnigan, Thermo, whatever). They have two independent channels with differential output, gain of 100, max output of +/- 100 Vpp, and frequency response starting at about 5 kHz and flat past about 6 MHz, with response going up to at least 13-14 MHz at somewhat reduced output. Not quite the output voltage you wanted but having used a few of these over many years they are great pieces and at that price having something up and working almost immediately has a lot of attraction no matter what you may wind up building :-). The second vendor, craig4076, had an almost mechanically complete (no data system) FTMS 2000 on Ebay a couple of years ago, with excite amp, preamp, vacuum chamber with dual cells, and 3 tesla magnet. It looks like they are parting it out (they have a couple of chamber pieces and the old vacuum control computer listed now along with the excite amp). Might be worth asking if they still have the dual preamp if you need one of those, as well. (JL, if you are reading this close your eyes :-).)

----- Regards, Carl Ijames