Dunno, Chris seems to be into directly generated HV sq waves, though I think in his app that sines would probably do -- both have been used in quadrupole mass spectrometers with about equal results.
I'm not as sure
in my app. I'm of course hoping I can do what I want with primarily sine-like waveforms, but not sure yet, so for the moment I'm thinking (multiple) 6DK6's for an output stage, which can tolerate 1-2kv DC plate supply no problems. However, once I get past stage 1a in this, I may find sines will do, and I'd use these fets for that in a heartbeat. I'd rather skip the hot filaments if I can (especially the big transmitting tubes). But on the other hand, it's real hard to just fry a tube while you're making mistakes, compared to most any semiconductor.
The issue being, of course, that it's really hard to get to kilovolts with a broadband kind of thing from a low voltage/high current input as these would provide -- the required magnetics are a pretty black art I've not yet learned. If I find I don't need instantaneous bandwidth (shaped pulses and such) then you bet -- the fets rule.
That's one of the PITA issues in research -- till you are really sure what you need/want, you need gear that "does it all" which is difficult and expensive. Then once you know, more often than not doing it becomes simple.
I think such is the case here. If narrowband sines are usable, then heck, tuned circuits handle getting to voltage no problems. But if you don't start out knowing the frequency....that's a major hassle to even do a sweep of the parameter space.
I believe this is one reason most mass specs are made as they are -- fixed F, sweep everything else. "The Book" says so, anyway.
That would be Peter Dawson's Quadrupole Mass Spectrometry and its Applications. ISBN 1-56396-455-4
I'm still trying to get a feel for the math there. In theory, it should let me predict the conditions I want, though they are pretty way out there compared to schlepping low energy ions around in a mass spec. In practice, the theory is abstruse enough so that I can't yet manage that, and even there it looks like a lot of empirical work had to be done to learn how to really make things like mass-selective ion traps with low space charge density -- my problem is a lot harder than that one. So for now, my best back of envelope guesses contain fudge factors in the small single digit integers, which doesn't narrow things down a heck of a lot. I know I need MHz and Kv....but for bunching in the presence of space charge a bunch of unrelated terms get into play, and even that fancy math can't handle it. They are perfectly happy that in the normal operation of a ion trap, ions all avoid one another due to space charge and spread out in space over time. I need better than that. What I want to do is gather and control a diffuse ion cloud in a kinda-predictable orbit, then bunch and accelerate the mess (while focusing) onto a spot. That means most of the time, the space charge can "do what it wants and I don't care much" and I can get some charge built up. Only for a very short time-space do I need or want them to all be together, which theoretically is a lot easier (but not simple to predict how I'll get there).
That price isn't so bad if you're not going to let the smoke out of them. Big transmitting tubes aren't cheap either.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.