What causes a voltage regulator to fail while being used in spec?

[Mp3]

I am just curious about this. I repaired a non working device by finding that 10V DC is split between two 7805's, each powering separate circuits, one of which is working as expected putting out nice 5.015V but the other (which was affixed to a very large heatsink) putting out 2.2V.

After simple swap  with a new 7805 everything is dandy, and testing the old one out of circuit confirms it simply won't put out any higher than 2.2v,

this is a roughly 30 year old JRC 7805A. I figured if some event such as wrong power supply damaged a regulator, both would be putting out incorrect voltage. When these things die, do they slowly climb down in output voltage or simply fail and one day it is much lower than usual? Or does it simply come down to "got unlucky, part went bad"?

[T3sl4co1l]

Who knows; you'd have to disassemble the device to see where or how it was damaged.

Most likely sites are the power transistor (overheating or overvoltage, fail shorted), and something in the control circuit (over or reverse voltage to GND, who knows; most likely low output voltage?).  Or I'm not sure what happens under reverse voltage (Vout > Vin), maybe that blows up the pull-up transistor (which drives the output Darlington).

And yeah, parts can fail due to age, storage, contamination/corrosion, etc.  More likely with early parts before they had encapsulation formulas pinned down.  I wouldn't think a 90s JRC part is a very likely culprit here; I mean like 60s and 70s.

Tim

[magic]

That's interesting. I thought that such things typically fail short circuit if overloaded, or perhaps open circuit if they get vaporized by the discharge of some large capacitor bank

The reference circuit seems well protected, it's only connected to the output through resistors.

What is its idle current consumption?

[bobbydazzler]

Large output capacitor on no protection diode?  Though I've seen some data sheets mention that's not a problem below 6v.  Otherwise I'd suspect some type of overvoltage (either once off/or regularly) to cause failure. 

[Rick Law]

I am just curious about this. I repaired a non working device by finding that 10V DC is split between two 7805's, each powering separate circuits, one of which is working as expected putting out nice 5.015V but the other (which was affixed to a very large heatsink) putting out 2.2V.
...

Curious about the way you put it: "...10V DC is split between two 7805's..."  They are probably not "split" (as in cutting the 10V in half giving one half to circuit A and the other half two circuit B), but instead both connected to the 10V DC in parallel.   Thus, each of the two 7805 individually regulates (drops) the 10V to 5V - individually and independently.

Dropping from 10V to 5V is within specs but is a lot to drop (hence the vary large heat sink you mentioned) and the waste goes to creating heat.  From 10V to 5V is 50%, 50% of the total power taken will be converted to waste heat by the 7805 and that heat needs to be dissipated effectively.

I'm not sure how much current/wattage it uses.  You did mention a very large heat sink, so effective dissipation was called for.  That being a given, I would check and see if the heat-sink compound is all dried out or is getting too dried.  If so, the 7805 may not be having good contact to the heat sink and consequently died of heat stroke...

[nigelwright7557]

It could have overheated but they have a thermal cut out so should have been ok.
Probably just old age from expanding and contracting over many years due to heat.

[bobbydazzler]

If it were any voltage/current spikes that wore the regulator down over time you could prevent it by putting a diode in series with the input and another one from output to input, and even another from ground to input.
I'm not sure it would've burnt out over time considering it was attached to a large heat sink and if it were outputting 500ma-1amp that's only 2.5-5watts, I don't think it would've been roasting away at 85c+.