Thermal range of an MCU

I assume you mean a crystal oscillator. Circuits should be fine if cold.

It wouldn't surprise me if the manufacturer refused to insure the part actually works in space. But satellite manufacturers probably still have component engineers. None of this JIT and trust the manufacturer. No GBD specs.

There is the school of thought that you can't "test in quaility". But I assure you burn-in does wonders. If you look at disk drive defects, you could certainly "test in" quality if you wanted. I think the whole philosophy that you can't test in quality is that nobody wants to pay for it. That is why statistical process control became the mantra.

Burn-in and testing is a procedure. Statistical process control is a plan. If failure is not an option, I vote for burn-in.

Deming's tumtum about testing-in quality doesn't really apply to burn-in, I agree. But once you get to what you hope is the flat part of the bathtub curve, it really does apply.

Cheers

Phil Hobbs

Anecdotally, I think some of the parts may be burned in for rad-hardness. My source found they had been pre-damaged.

--sp

MCU

-40C..125C.

specs

what

be

is to

replacement.

grade

for

plan.

I have seen that blow up in real cases. One instance was a limited life high energy density capacitor. They needed 250 full power operating hours life. The best caps that could be bought, had about 320 hours life. So they chose 168 hours burn-in. The very predictable result was far more "early" field failures. The killer is they would not accept that they were cutting their own throat with the overlong burn-in.

?-)

I recall reading a story that early (GE?) point-contact transistors were being shipped to military customers. Who naturally performed incoming testing. And got early field failures. The vibe tests were screwing up the contacts. Manufacturer came back, "Don't do that!"

Tim