Thyratrons? Alternatives?

[Joerg]

Hey guys,

Seems like thyratrons have become quite rare these days and SCRs or GTOs
are just too freaking slow. Is there a readily available alternative
that doesn't cost an arm and a leg, meaning under $100? A kilovolt would
be nice and >50A.

 

[Spehro Pefhany]

Hey guys,

Seems like thyratrons have become quite rare these days and SCRs or GTOs
are just too freaking slow. Is there a readily available alternative
that doesn't cost an arm and a leg, meaning under $100? A kilovolt would
be nice and >50A.

IGBT? IIRC, 1-1.5kV is available.
Best regards,
Spehro Pefhany

 

[Rene Tschaggelar]

Hey guys,

Seems like thyratrons have become quite rare these days and SCRs or GTOs
are just too freaking slow. Is there a readily available alternative
that doesn't cost an arm and a leg, meaning under $100? A kilovolt would
be nice and >50A.

A thyratron is a switch as I recall. How about two
or three cascaded FETs ? The circuit is called Marx
stack I believe and achieves few ns out of ordinary
FETs by avalanching them. 50 Amps are hefty though.


Rene

 

[Joerg]

IGBT? IIRC, 1-1.5kV is available.

Those are great but even the "HiperFast" are still >250nsec turn-off
time. However, I'll go through the latest offerings again. Thanks.

 

[Joerg]

A thyratron is a switch as I recall. How about two
or three cascaded FETs ? The circuit is called Marx
stack I believe and achieves few ns out of ordinary
FETs by avalanching them. 50 Amps are hefty though.

Supertex has one that can do this current but the problem would be the
lifetime. I believe they quoted 4e11 avalanches. That's high but not
infinite and I'd blow through a set within days.

 

[Joerg]

How slow is "too freaking"? The Russians still make some nice hydrogen
thyratrons; Los Alamos used to buy them on the sly, maybe still do.

I need to switch within 100nsec or so. Do you remember the Russian
company? Svetlana or Sovtek? Boutique prices? A thyratron would be cool
but it'll have to be something that can still be bought a few years from
now.

You can get 1KV from one or two mosfets, tens of amps in a few ns. A
string of maybe 3 avalanche transistors, like the Zetex SOT-23's, will
output 30 amps or so at 1KV, for short pulses.

In my case a FET would work, that's what I am trying right now. But it
ain't ideal because it should turn off when a certain resonance has run
its course, not when a gate driver tells it to. In a pinch I can try
some nifty feedback for that, and maybe I have to.

I did one gadget that puts 1200 volt pulses into 50 ohms, 2400 volts
into a small capacitive load, with 3 ns pulse width, at up to 500 KHz.
It uses a drift step-recovery diode (another Russian invention) driven
by a couple of 400 volt mosfets.

http://www.highlandtechnology.com/DSS/T220DS.html

It was fun, but we didn't sell many.

Sure looks high-tech but we'd need a good order of magnitude higher PRF.
Do you think it can be spiffed up some more?

 

[JeffM]

Seems like thyratrons have become quite rare these days
and SCRs or GTOs are just too freaking slow.
Is there a readily available alternative[...]under $100?
A kilovolt would be nice and >50A.

Beefier than this starts to get slow.
http://www.google.com/search?q=Advanced.Power.Technology+1000V+28A+APT10035B2LL
IIRC, these are in the $30 range.

 

[Joerg]

Aha! Thanks, John. They even sell a RoHS version which is encouraging.

 

[Joerg]

---
I'm confused.

From what you've said, it seems you want to ring a bell and keep the
clapper stuck until the amplitude of the ring decays to less than some
reference threshold, then you want to release the clapper and start
the cycle over again.

Am I right?

Well, pretty much. Can't reveal details though.

 

[Joerg]

Seems like thyratrons have become quite rare these days
and SCRs or GTOs are just too freaking slow.
Is there a readily available alternative[...]under $100?
A kilovolt would be nice and >50A.

Beefier than this starts to get slow.
http://www.google.com/search?q=Advanced.Power.Technology+1000V+28A+APT10035B2LL
IIRC, these are in the $30 range.

Thanks, Jeff. If I decide to go the FET route those look like good
candidates. 100A pulse sound just like the ticket.

 

[MooseFET]

Hey guys,

Seems like thyratrons have become quite rare these days and SCRs or GTOs
are just too freaking slow. Is there a readily available alternative
that doesn't cost an arm and a leg, meaning under $100? A kilovolt would
be nice and >50A.

How about:

http://www.appliedpulsedpower.com/prod/switches.htm

 

[Ecnerwal]

Seems like thyratrons have become quite rare these days and SCRs or GTOs
are just too freaking slow. Is there a readily available alternative
that doesn't cost an arm and a leg, meaning under $100? A kilovolt would
be nice and >50A.

Don't know what the prices run (damned priceless websites), but Perkin
Elmer claims they still make them, and the baby of the bunch (HY-2)
would exceed your ratings. The one source one distributor setup probably
won't sit well with you, and the price may not be low enough either.

http://industrial.rell.com/et_Hthyratrons.asp

I used to use something similar to the HY-3192 on Nitrogen lasers, but I
think it was an EG&G part. Suppose it's possible one bought the other in
the couple of decades I haven't been doing that.

Google spits up two websites from China, not my idea of a good source,
especially if your hush-hush application is in your typical medical
field.

 

[Frithiof Andreas Jensen]

Hey guys,

Seems like thyratrons have become quite rare these days and SCRs or GTOs
are just too freaking slow. Is there a readily available alternative that
doesn't cost an arm and a leg, meaning under $100? A kilovolt would be
nice and >50A.

How about a Cascode with a high-voltage bipolar transistor "on top" and a
low-voltage MOSFET driving the Emitter of the bipolar?

Just remember that while the Bipolar storage time runs, the full collector
current is diverted to the Base so you have to have somewhere to put the
charge. After storage time the whole contraption blocks in about 10-20 ns
so, again, there will be transients. The MOSFET only need to be able to hold
the maximum B-E voltage so it will be cheap and efficient; the Bipolar does
not have to be particularly fast and since the Emitter is cut at turnoff
it's SOA becomes square, right to the VCEmax limit.

Hope that helps.

 

[Tim Williams]

Supertex has one that can do this current but the problem would be the
lifetime. I believe they quoted 4e11 avalanches. That's high but not
infinite and I'd blow through a set within days.

Somehow I don't think you'll manage to run a thyratron at 1.5MHz, Joerg.
Check your numbers...!?

(Conservative estimate, 4e11 / (7 days = 0.6M seconds) = 1.5MHz.)

Tim

 

[Joerg]

Somehow I don't think you'll manage to run a thyratron at 1.5MHz, Joerg.
Check your numbers...!?

There are other permutations of those but the problem is that many are
(most likely) still under export restriction so you can't use them in
designs that are to be sold worldwide.

(Conservative estimate, 4e11 / (7 days = 0.6M seconds) = 1.5MHz.)

Not that high but this one can't contain such "disposable" parts. :-(

 

[Joerg]

And a thyratron won't be cheap. 1000 volt fets and transistors are
cheap. At 100 ns, I'd thik that a horizontal output transistor might
work.

Yes, although the original market for those is rapidly drying up so one
has to be careful.

The Zetex avalanche things would certainly turn off, a lot like an scr
or a tyratron.

From the datasheet it looks like their number of avalanches is finite
even at low currents. That won't fly in this case.

PRF of 5 MHz? That excludes most devices, including thyratrons.

Yup :-(

OTOH if it was easy guys like us wouldn't be needed.

 

[Joerg]

How about:

http://www.appliedpulsedpower.com/prod/switches.htm

Thanks! Now that's some real muscle and speed.

 

[Joerg]

Don't know what the prices run (damned priceless websites), but Perkin
Elmer claims they still make them, and the baby of the bunch (HY-2)
would exceed your ratings. The one source one distributor setup probably
won't sit well with you, and the price may not be low enough either.

http://industrial.rell.com/et_Hthyratrons.asp

I used to use something similar to the HY-3192 on Nitrogen lasers, but I
think it was an EG&G part. Suppose it's possible one bought the other in
the couple of decades I haven't been doing that.

Unfortunately those won't give me a sufficient PRF.

Google spits up two websites from China, not my idea of a good source,
especially if your hush-hush application is in your typical medical
field.

Yes, that would be a concern. There are some good mfgs but info about
their corporate health is usually not publicly available.

 

[Joerg]

How much forward drop can you tolerate while it's on?

It doesn't really matter. 5-10V would be fine. The less the better, not
because of function but because of dissipation.

What kind of pulse, magnitude and duration will trigger it on?

It must be able to turn on within 100nsec and retriggerable after a
couple usec or so.

 

[Joerg]

How about a Cascode with a high-voltage bipolar transistor "on top" and a
low-voltage MOSFET driving the Emitter of the bipolar?

Just remember that while the Bipolar storage time runs, the full collector
current is diverted to the Base so you have to have somewhere to put the
charge. After storage time the whole contraption blocks in about 10-20 ns
so, again, there will be transients. The MOSFET only need to be able to hold
the maximum B-E voltage so it will be cheap and efficient; the Bipolar does
not have to be particularly fast and since the Emitter is cut at turnoff
it's SOA becomes square, right to the VCEmax limit.

Hope that helps.

I'll have to check that idea, thanks. The current diversion could be an
issue. It'll be tens of amps.