Krytrons... again! :0)

[Majorana]

Hello, everybody! Just finished watching (for perhaps the 90th time...) "Frantic", it's broadcast at least twice a year, every year... Of course, the scene where Harrison Ford almost falls from the roof and then retrieves the krytron (INTACT!! ) from the broken statue is my favourite. And every time, the usual question comes back to my mind: why is a krytron so special? At the end, it's just another triggered spark gap! There are triggered spark gaps that can handle much higher currents than a krytron, that are not subjected to any special export regulations, ITAR etc... used every day to trigger lamps, spark igniters and all sort of things. As far as I can understand, they can do the same things than krytrons, but without a ton of stamped paper in trail... You zap a krytron, it fires, you zap a triggered spark gap, it fires too... and the latter can be even more sturdy. So why all this fuss around krytrons?? (...apart from being capable of remaining intact after falling on a concrete floor from a roof in Paris... )

 

[Borek]

My understanding was that their specialty is a very precise timing (they are very fast), allowing well synchronized ignition of charges. But I can be completely wrong.

 

[Baluncore]

Krytrons are exceptionally fast switches. They were originally used as the TR switch in radar systems, where one antenna is used for both transmit and receive. The high power transmit pulse causes the krytron to conduct, which "shorts out" the input, protecting the receiver, while the outgoing pulse is being transmitted. That radar application operated with quarter wavelength lines, so the krytron became an RF reflector after it fired, and did not absorb further transmit RF.

The TR switch is a performance critical part of a pulse radar set, hence the need to protect it.

 

[DaveE]

Very fast rise time switches with (sort of) high power. Most have a β emitter to "pre-ionize" the gas. It's a spark gap, but not a "normal" one, they are biased right up to the edge of conduction. We used them in a laser Q-Switch many years ago for a DoD contractor. Now days that's not a great application, although they worked fine for us. They have a short lifetime, are a PITA to drive compared to semiconductors, and are expensive and hard to buy. You won't find them in any lasers now, most favor AO Q-switches not EO.

KN-22 is the most common one (in my experience anyway):
https://lampes-et-tubes.info/rs/EG&G_KRYTRONS.pdf

 

[Baluncore]

So why all this fuss around krytrons??

Krytrons are known for their use in igniting exploding-bridgewire and slapper detonators in nuclear weapons.

 

[Majorana]

My understanding was that their specialty is a very precise timing (they are very fast), allowing well synchronized ignition of charges. But I can be completely wrong.

I too read something like that a long time ago, but I lacked (and still lack) the information to understand the special necessities in these applications...

 

[Majorana]

Krytrons are exceptionally fast switches. They were originally used as the TR switch in radar systems, where one antenna is used for both transmit and receive. The high power transmit pulse causes the krytron to conduct, which "shorts out" the input, protecting the receiver, while the outgoing pulse is being transmitted. That radar application operated with quarter wavelength lines, so the krytron became an RF reflector after it fired, and did not absorb further transmit RF.

The TR switch is a performance critical part of a pulse radar set, hence the need to protect it.

Hi Baluncore, thank you very much! I remember the "TR switch" and "ATR switch" from back when I studied the radar theory, but krytrons were never specifically mentioned in my books! As often happens, a well designed tube (vacuum or gas filled) can replace a lot of exotic semiconductors with elaborate driving circuitry...

 

[Baluncore]

I too read something like that a long time ago, but I lacked (and still lack) the information to understand the special necessities in these applications...

In the implosion type of nuclear weapon, explosives surround a sub-critical sized sphere of fission material. Synchronous detonation of the explosives compress the material, increasing its density, and so exceeding the critical mass. If the explosives are not detonated synchronously, the compression of the sphere does not reach the critical density, so the parts become separated before the chain of fission can grow exponentially.

 

[Majorana]

Very fast rise time switches with (sort of) high power. Most have a β emitter to "pre-ionize" the gas. It's a spark gap, but not a "normal" one, they are biased right up to the edge of conduction. We used them in a laser Q-Switch many years ago for a DoD contractor. Now days that's not a great application, although they worked fine for us. They have a short lifetime, are a PITA to drive compared to semiconductors, and are expensive and hard to buy. You won't find them in any lasers now, most favor AO Q-switches not EO.

KN-22 is the most common one (in my experience anyway):
https://lampes-et-tubes.info/rs/EG&G_KRYTRONS.pdf

Hi DaveE Yes, I have read somewhere (can't remember where, sorry ) that krytrons have a very small quantity of some radioisotope (⁶³Ni, if memory serves me) added to achieve that "pre-ionization" you mentioned. I remember that the same trick is adopted in the much more common and mundane 2-terminal (untriggered) spark gaps used as the switching element in the high-energy ignition units that fire the spark igniters in gas turbine engines. However, I have like a feeling - possibly wrong - that all the restrictions and "hush-hush" around krytrons are due to their use in nuclear weapons (again: a big ) rather than lasers or radars. En passant, I would not sleep well if I included a component that is constantly, by design, "on the verge of triggering" in the firing circuitry of a nuclear device! (...and think of all the vibrations an radiations going on in those things... brrrr!) I would choose something that stays firmly away from the threshold and would require a good, convinced and energetic kick to fire! En passant 2, I just can't understand the reason why they use such a restricted/controlled component in photocopiers! Once you install a restricted component in some equipment, the whole equipment becomes restricted as well, right?... A classified copy machine?!?

 

[Majorana]

In the implosion type of nuclear weapon, explosives surround a sub-critical sized sphere of fission material. Synchronous detonation of the explosives compress the material, increasing its density, and so exceeding the critical mass. If the explosives are not detonated synchronously, the compression of the sphere does not reach the critical density, so the parts become separated before the chain of fission can grow exponentially.

Err... ok, I am sort of hesitant here. I'll try to explain. My natural curiosity is boundless, to the point that some friends tell me that I "look too much and ask too much" (and no, they aren't mafia members! ). But when a few years ago, before the covid thing, I asked here on PF in order to shed some light on my doubts about these matters, I got sort of friendly warning - can't remember from what user - that said (moreless) "see? none of the top specialist in these questions answered nor will answer". Something that I interpreted like I had asked a forbidden question... Okay, having said that, let's go on. Of course I have read both technical and nontechnical (Tom Clancy's "The sum of all fears") books, and watched a few movies on the subject ("Fat man and Little Boy", "Oppenheimer"). I know the concept of compression by implosion: a sphere of fissionable material, entirely enclosed into another sphere made of a very fast high explosive. To achieve the job, the outer explosive sphere must have a relatively large number of detonators fitted in carefully calculated positions, that of course must be fired all at the same instant. But where's the problem? If you connect all the detonators together (electrically in parallel, or series) you end up with just 2 wires... You provide a suitable high-energy pulse to these 2 wires, and the detonators will go off all at the same time. It's the same technique used every day in controlled demolitions, quarry blasting etc... they have a large number of blasting caps connected to ONE blasting machine to fire all of them at once. No need to use any krytron. The blasting machine has a high voltage capacitor and a heavy duty MECHANICAL switch: they charge the capacitor, throw the switch and kaboom, all the blasting caps get theirat exactly the same instant (in a parallel connection you could cut all the wires at the same lenght in order to achieve the same propagation delay). Consider that a blasting cap, or detonator, has its tolerances: you cannot build two or more having exactly (down to nanosecond-level) the same delay from the instant the pulse arrives to the build up of the shockwave and thermal gradient. There will be differences, it's unavoidable. And, as Tom Clancy wrote in The sum of all fears, "10 nanoseconds is thousands of times too short for any chemical explosive to do anything" (from memory, the original English may be slightly different). So at this point you can see where my perplexity comes from: why a krytron?...

 

[Baluncore]

But where's the problem? If you connect all the detonators together (electrically in parallel, or series) you end up with just 2 wires... You provide a suitable high-energy pulse to these 2 wires, and the detonators will go off all at the same time. It's the same technique used every day in controlled demolitions, quarry blasting etc... they have a large number of blasting caps connected to ONE blasting machine to fire all of them at once.

I have done some blasting, and know that if you connect electric detonators in series, or parallel, they will fire at different times. In series, some will fail to fire, while in parallel, they take it in turns.

In parallel, detonators will have a jitter of tens of microseconds, while better than 10 nanoseconds is needed for critical implosions, and the length of the transmission lines comes into the problem, so all the lines must be cut to the same length from the same batch of cable.

One krytron is needed per detonator, with all krytrons synchronously triggered in the same place, each with an independent charged capacitor close by, and a fixed length transmission line to the detonator.

In blasting, multiple charges are usually triggered by detonator cord, with junctions tied together, without wires or electric dets.
https://en.wikipedia.org/wiki/Detonating_cord

For accurate delay timing with electric dets, to throw rock in a particular direction, such as when tunnelling, the sequence can be programmed into a multichannel electrical exploder.

 

[Majorana]

I have done some blasting, and know that if you connect electric detonators in series, or parallel, they will fire at different times. In series, some will fail to fire, while in parallel, they take it in turns.

In parallel, detonators will have a jitter of tens of microseconds, while better than 10 nanoseconds is needed for critical implosions, and the length of the transmission lines comes into the problem, so all the lines must be cut to the same length from the same batch of cable.

One krytron is needed per detonator, with all krytrons synchronously triggered in the same place, each with an independent charged capacitor close by, and a fixed length transmission line to the detonator.

In blasting, multiple charges are usually triggered by detonator cord, with junctions tied together, without wires or electric dets.
https://en.wikipedia.org/wiki/Detonating_cord

For accurate delay timing with electric dets, to throw rock in a particular direction, such as when tunnelling, the sequence can be programmed into a multichannel electrical exploder.

(sorry for the delay in answering, the usual notification Email hasn't arrived this time! )
Thank you for your answer... I read and re-read it carefully, but I'm still missing the focal point that remains unclear in my mind... Don't worry, I won't go further ahead with this thread, evidently there is something wrong in my initial assumptions that prevents me from grabbing the explanation. Thank you SO much to all the contributors for your patience!