Reality of over voltage protection.

[davelectronic]

After having put together a few large low voltage power supplys, and having added a separate crowbar protection circuit. What are the chances of a BJT failing short circuit and in turn putting out the full unregulated voltage ? I've never had a BJT fail in 16 years of tinkering with electronic power supply circuits. I know they can and do fail, but is it a common occurrence with BJT transistors. It goes with out saying i use short circuit protection, and earth bonding on the primary side of such power supply circuits. Is a crowbar the best option for this type of over voltage protection, or is there something better that can be implemented. All this is referring to linear psu circuits with the above in mind. Your thoughts would be greatly apricated in expanding this topic of over voltage protection. Thanks for reading.   

[Whales]

How large is large?  A few amps or hundreds of amps?

If the hazards are massive (hundreds of amps -> fire & explosion) then it would probably be worth doing active monitoring of the parts most likely to fail.  There is probably a clever way you could use a comparator to detect when a BJT fails short; then use this to shut down the entire unit and latch a fault LED.  EDIT: or just detect if the output voltage is above what's expected, eg +1V above the setpoint.

[David Hess]

Power pass elements, bipolar or MOSFET, usually fail due to overheating and a crowbar on the output combined with a fuse on the input is the way to protect other circuits.  More rarely a bipolar power pass element will fail because of a higher voltage applied to the output causing reverse breakdown.

[TimFox]

A good crowbar short-circuit will activate very quickly, and protect the load long enough for the slower fuse to blow.

[wizard69]

I suspect that when a transistor goes open it has shorted shortly before hand.   As for crowbar protection It is the big hammer approach and you have to judge if it is important to protect down stream stuff by blowing up the  supply.   Fuses are not always going to react quick enough to keep the magic smoke in when the supply does fail.

As for power supply failures, I suspect you are right about the low occurrence of a pass transistor failure.   However in most supplies there is a lot more that can go wrong.   If you need to protect against high voltage just keep that in mind.

[davelectronic]

Thank you for your replies. No not hundreds of Amps, the most would be 30 Amps at 12.00 Volts to 13.80 Volts. Might go as far as 50 Amps at that same voltage in the future, but it's only a maybe. Some years ago I had a retail linear power supply 13.80 Volts DC go high voltage, the potentiometer for voltage adjustment failed letting the full unregulated voltage through, about 22.00 Volts from memory. I was lucky nothing was connected to this 3 to 5 Amp linear power supply. What I use these homebrew power supplys for, is mainly HF radio equipment, mostly CB radio gear. I have a 12.00 Volts DC 15 Amp PSU and put together a separate crowbar circuit for it in a small project box. I have tested it with over voltage scenario, and it does fire and blow the fuse very quickly, or it seems quick to me. After replacing the fuse the crowbar is good to go again. What I would like is to incorporate the crowbar circuit as part of the PSU circuit on one board. But I have constraints with space. I've got as far as a 30 Amp power supply using a L7812 with 6 X TIP36C PNP power transistors. And it's very impressive with regard to power and stability. I'm tempted to use it with just primary and secondary short circuit protection, but there's that niggle, what if  a transistor failed and it put out the 21.00 Volts rectified and filtered unregulated voltage. I expect I will be safe and use a crowbar circuit, just a though, could the crowbar circuit cartridge fuse be replaced with a thermal resettable circuit breaker ? Or is one of those to slow to act upon a fault.

[Terry Bites]

The crowbar is brutal but effective. How about something a litte more contemporary, eg the LTC4381 electronic circuit breaker?
https://www.analog.com/en/parametricsearch/11394#/

[Neomys Sapiens]

In many power supplies that sport a crowbar circuit, it was not primarily intended to protect against a shorted pass element, but to protect against programming (externally set value) errors or related problems, like a sense lead becoming accidentally disconnected. The idea was that the crowbar could be set to a value, which would appear never on the output if the controller setting the PS did work properly, but which were possible to generate because of the input control range.

[davelectronic]

Interesting IC, I will read up on it to learn more about it. I never knew the crowbar circuit was used for PSU data applications. I'm surprised there aren't more solutions, although there probably are through descreate circuits constructed for special applications. I've heard of something called " foldback" protection, I'm not sure how that's achieved. Mostly I've seen that protection type mentioned in a few retail bench power supplys.