crowbar circuit - picking the correct thryistor

[sal_park]

Hi Guys,

I'm trying to build a crowbar circuit and I'm having a hard time understanding what spec thyristor I need. This crowbar circuit will be powered from a 36V 10Ha battery capable of delivering at least 20A and all the PCT spec's I can find (this is the one I'm thinking of using) talk about a max time to trip of 4 seconds - should I be expecting 20A+ to flow for 4 seconds ? Or is the response of the PCT so nonlinear that the max current only flows for a few milliseconds ?

So, how do I pick the thyristor ? Do I need to select one with an "On State RMS Current IT" > 20A ? How suitable is something like this ? Or is it overkill ?


sp

[SeanB]

A few misconceptions on your side. The battery will deliver considerably more than 20A into a short circuit, and thus you need to plan accordingly. The PTC device does not fully interrupt current, it limits it to a holding current, which will probably be enough to hold the thyristor on until the battery is completely discharged and the PTC is cooked. Better is to use a thermal breaker or another mechanical circuit breaker, or a replaceable fuse element.

As to the thyristor, it has to be rated for a single cycle surge current larger than the current the battery can supply into the impedance of the circuit. You need to consider the internal resistance of the battery when fully charged and the resistance of the wiring and the current breaker as part of this resistance, and select the device to be able  to handle more than this. you may need to add extra resistance ( a low value wirewound resistor capable of handling this current) to get it low enough, or select a massively overrated device so as to have the single cycle surge capacity.

[Kremmen]

+1 to everything SeanB wrote.
By definition a crowbar will create a hard short circuit across the supply lines, thereby positively interrupting voltage and current to the rest of the circuit. The circuit is now protected from overvoltage and current but unless the short is cleared, something will burn.
My sugestion: Whatever other protection measures you specify, insert a sturdy fuse in the supply input. When the crowbar thyristor fires the current through this fuse should be many times (at least 10 times) the fuse nominal current. At very high overcurrents the fuse ruptures almost instantaneously, thus clearing the circuit. If you select a robust enough thyristor, you may not need a limiter resistor at all, but make sure the thyristor has sufficient single-cycle current ratng to survive until the fuse blows.

[SeanB]

As well I will add a crowbar is often a very bad idea, as it can fail silently open circuit and have absolutely no indication of this failure unless you test it occasionally, which can also cause it to fail ( you lose in each case) either open or short circuit.

I have had them silently fail and when it was needed it was inoperative, allowing a whole computer to get the unregulated supply rail applied to it instead of 5V for a few hours. As a testament to the tolerance of 5V LSTTL and MSI TTL only one failed in a box of over 300 devices, and it was an open collector device rated for 30v on the open collector.