2
\$\begingroup\$

I need to switch an inductive load on and off very quickly. Turning an inductive load on and off very quickly implies the supply voltage is large, several kV. Peak current through the load is several kA. It's common to use SCRs to do this, but the problem is that the supply is DC and you wind up buying a lot of transistors anyway to turn the SCR's off. (Parallel IGBT to turn off SCR needs to accept the same reverse voltage as SCR and just as much current, albeit for a shorter period of time.)

Here is my question:

Can a vacuum tube switch DC on and off very quickly?

I want the tube to turn on in < 1 us, stay on for 3 ms, and then turn off in < 1 us. After that the circuit is off for > 30 ms.

Is this a thing that a tube can do?

If yes, how do you model this mathematically? For example, is a tube a voltage-to-current converter like a MOSFET? Current gain like a BJT? Something else?

As for tube type, I'm thinking triode only because it's one of the simpler kinds. Obviously I'm new to tubes and I'm open to other tube types as well.

\$\endgroup\$
2
  • \$\begingroup\$ At this rate, consider a physically very long, pneumatically-actuated blade switch. Or large rotating disc switch. Speed proportional to desired turn-off rate. Inductor can create megavolts at turn-off; exotic insulation measures needed. \$\endgroup\$
    – rdtsc
    Commented Feb 24, 2023 at 19:49
  • 1
    \$\begingroup\$ James, your load is in the low milliHenries, it appears (a few kV times 3 ms divided by a kA.) To maximize power delivery (this says nothing about the energy you will waste getting into a given state at the capacitor, but since this is a repetitive process that initial cost may be affordable) you will likely want to at least consider the ideas related to a capacitive discharge into a transformer with your inductor tied into the secondary as a load. Here, semiconductors can be your friend. But I see a transformer in the picture, too, to match up loads better and simplify the project. \$\endgroup\$ Commented Feb 24, 2023 at 21:24

2 Answers 2

5
\$\begingroup\$

Vacuum tubes in general are poorly suited to switching kA due to high internal resistance. Even huge ones like this one are only good for amperes, albeit at high voltage.

The appropriate tube technology would probably be hydrogen (or deuterium) thyratron, some of which can switch kA and kV (amounting to many megawatts). However those tubes have a similar behavior to SCRs in that they don't turn off until the current drops to a low value (and they have a fairly long recovery time, limiting the pulse repetition rate).

Other than that, thyristors are probably the best bet unless you want to put a whole bunch of smaller MOSFETs or IGBTs in parallel. MOSFETs are more poorly suited since the die area increases greatly with Vds limit.

Commutation of thyristors can be achieved with secondary thyristors, this is old technology.

\$\endgroup\$
6
  • \$\begingroup\$ Can you tell me more about these secondary thyristors? Can you share a link to an example circuit? \$\endgroup\$ Commented Feb 24, 2023 at 17:20
  • \$\begingroup\$ Find a copy of the 1970s-era GE SCR manual online (it will be easy) and look at their example circuits. This stuff has not changed much in 50 or 60 years. The basic idea is to capacitively couple a turn-off pulse. \$\endgroup\$ Commented Feb 24, 2023 at 17:28
  • \$\begingroup\$ Shades of: electronics.stackexchange.com/a/654647/311631 you've been on a bit of a thyristor rip lately it seems :^) \$\endgroup\$ Commented Feb 24, 2023 at 17:53
  • 1
    \$\begingroup\$ @TimWilliams They are mature technology so they don't get much attention except by specialists. For example, the UHT DC transmission conversion technology uses terminology that's evolved in that domain. Given that they have "halls" of banks of thyristor modules, I guess that's understandable. There are still large numbers of mercury arc modules in use in some former Soviet installations, though they are gradually being replaced where that is practical. \$\endgroup\$ Commented Feb 24, 2023 at 17:56
  • \$\begingroup\$ Spehro, would those be anything like MARX switches? ;^) \$\endgroup\$
    – glen_geek
    Commented Feb 24, 2023 at 18:10
3
\$\begingroup\$

Yes, it can. Consider that in the old days high-power broadcast transmitters used tubes, even for UHF.

But, this is 2023: use an IGBT or a SiC MOSFET.

Tubes are getting rarer (and are often made in Russia).

\$\endgroup\$
4
  • 2
    \$\begingroup\$ kV yes, kA no. One main TV transmitter site I visited fed its UHF klystrons with 22 kV at 8 A (for 40 kW of RF, 1 MW eirp). \$\endgroup\$
    – Graham Nye
    Commented Feb 24, 2023 at 19:31
  • 1
    \$\begingroup\$ True. But also: I'm 99 % sure that @James Strieter is in no position to actually need turn on a 10 MW load for 30 ms. This feels like a hypothetical question, not a practical one. \$\endgroup\$ Commented Feb 24, 2023 at 20:29
  • 3
    \$\begingroup\$ IGBTs are also going out of style, with SiC MOSFETs becoming more affordable. \$\endgroup\$
    – Hearth
    Commented Feb 24, 2023 at 21:12
  • \$\begingroup\$ Thank you. I edited my answer accordingly. \$\endgroup\$ Commented Feb 24, 2023 at 23:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.