"breaking in" electronics?

[ftransform]

So, before a calibration of a precison circuit happens, what should be done to it to ensure the lack of drift? I have read that various things are done, for instance leaving it on for a number of hours and leaving it on for a number of hours at a elevated temperature.

What do you guys do for your circuits? Some of my op-amps I have used have figures in the datasheet for long term drift, so that is something to go by I suppose, but some of my parts do not.

Is there some kind of a translation between leaving it on at a elevated temperature vs regular temperature for long term drift? Like, 1 hour at 2x the working temperature = 3 hours at design temperature?

I have also seen circuits that were powered by higher voltages, i.e. to "burn in" a op amp that might run at 3V they run it at 18V, I assume this is just to increase the temperature and the same result could be had from a thermal chamber?

[Mechatrommer]

So, before a calibration of a precison circuit happens, what should be done to it to ensure the lack of drift?

design it properly and then characterize its drift behaviour?

[robrenz]

Power Designs precision series power supplies were burned in for 100 hours at max loading prior to calibration.  I have done some drift testing on these suppies. This is a trend plot from my 8846A of the power supply unloaded at 10V output and the meter zeroed to just show variation of the 10V. That is 888.43µV total vertical span and the trace is 3 days 17hrs 26min long. It is interesting to see how long it takes to get close to leveling out and 100 hours may not be enough. That logarithmic shape is also present when the PS is loaded and takes just as long.

[Paul Price]

Everything changes with time and stress conditions of temperature and voltage. You may have to wait for years to see some changes in operation. Accelerated stress can damage the DUT. Read the spec sheet.

[ftransform]

no general rules of thumb?

[senso]

Breaking in in electronics always reminds me of audifool things, in a precision thingy, stressing the components doesn't seem like a good recipe for long term stability I think..

[alm]

Parts like voltage references and crystals often have a much higher drift rate in the first 100h or so then the rest of their service life, so it makes sense to age them before shipping them to a customer with a guaranteed tolerance. This is what companies like Fluke and Agilent do for their precision instruments. Nothing audiophoolery about this, you can just observe the value settle, kind of like the plot robrenz posted.

A power supply should be able to maintain max power indefinitely, so it shouldn't make stability or reliability any worse. I can also see higher temperatures reducing stresses and speeding up relaxation processes, but this will depend on the exact physics behind the drift. I would never exceed the maximum specs, however.