Hi,
I am driving a relay and it is driven by a NPN transistor, I want a circuit which can cut-off the relay supply when the emitter and collector of NPN is shorted accidentally, so that relay is not ON permanently.

 

I guess one would ask the question - How would you distinguish between a transistor that has failed short circuit (C-E) and one that is simply switched on in saturation?

 

Do you

I guess one would ask the question - How would you distinguish between a transistor that has failed short circuit (C-E) and one that is simply switched on in saturation?

Do you have any solutions

 

That's a pretty peremptory 'request' , are you on a smartphone?

It would be better to attack the problem from another direction.
Why does it matter if the relay in on all the time?

 

That's a pretty peremptory 'request' , are you on a smartphone?

It would be better to attack the problem from another direction.
Why does it matter if the relay in on all the time?

I am controlling critical signal from relay

 

That doesn't answer my question or tnk's.

If you genuinely want answers you need to put in as well, and that means providing a proper engineering description of the problem.

It may be possible to reconfigure your driver circuit to monitor the transistor, even if this means an extra power supply.
However, if the final output is that critical, then failure of the relay in the closed position (on) is much more likely than failure of the transistor and monitoring the transistor alone will not protect against this.

Again another strategy in mission critical components is timed replacement, well before MTBF indcators.

 

As Studiot says, relay failure is a possibility. It's quite common for relay contacts to 'stick' closed. Have you considered how you will detect/overcome that? If you have a solution for that, then you will not need to monitor the switching NPN transistor.

 

A simple comparator can detect when the transistor is “ON” when the drive signal is “OFF”. That should work as long as you can assume a single point failure. Such assumptions remain valid IF all the possible single point failures can be somehow tested.

The next question is how you will detect stuck contact welded into the ON position?

 

One common way to improve reliability is to use redundant circuits. You could use two separate transistor-relay circuits, both operated from the same signal, and connect the two relay's contacts in series. That way, even if one is stuck on, the other will still turn the output on and off.