Intermitted short inside hermetic package, tin whiskers? Any remedies?

[mzzj]

I have been working on a real puzzle: Infrared temperature scanner that works sometimes for 1 hour and sometimes 2 months.

A real pleasure to debug  when the fault shows up after 2 months but I think I got it figured down to IR detector itself.
When the fault is present the detector pins measure ~100 to ~200 ohms resistance, when the unit happens to work the measured resistance is ~500ohms. 

Any great ideas if the internal "short" can be blasted to sky-high?. Higher than usual voltage applied to detector? Ultrasonic washer to vibrate it "somewhere else"?   

New detector would be obvious solution but these are about rare as hens teeth and priced accordingly. Somewhat similar TE cooled IR detector was ~3000 USD so these are not exactly cheap.
And I haven't been able to figure out exaxt model... close but no.
This gives general idea what sort of animal we are talking about:
https://www.hamamatsu.com/eu/en/product/optical-sensors/infrared-detector/insb-photoconductive-detector/index.html

[mzzj]

Oh well, what do we have here:





Guess which pin out of the 9 pins in the package the one shown is... yes. The detector pin   

[Gyro]

Sneaky little bu**ers aren't they, it's almost as if they're alive!

Nice photos!


P.S. Given the package type, I wonder if you could get it taken off completely with a laser, just a wild thought.... Or apply significant heat to just that pin and melt it off?

[mzzj]

Sneaky little bu**ers aren't they, it's almost as if they're alive!

Nice photos!


P.S. Given the package type, I wonder if you could get it taken off completely with a laser, just a wild thought.... Or apply significant heat to just that pin and melt it off?

I got rid of the short... kinda.
Various NASA papers on whiskers stated that they fuse with "tens of milliamps". So I desoldered the detector from the board and the fault was still "active" ie  low resistance so I thought that maybe the 150 ohm whisker would melt before the 500 ohm sensor (of unknown type)
47uF capacitor charged to 50 volts evaporated the short.. and the sensor itself. 

Oh well, it was just 17 grand experiment    

Whisker shown in earlier photos is what I was able to still find after the fusing experiment.
Since I found one whisker I think there is more than that. 
I'm unable to see some parts of the detector under the black radiation shield and smallest whiskers don't necessarily show up with the ADSM302 microscope resolution.

[Berni]

Ouch, that's not nice.

Tho 47uF at 50V is quite a bit of energy for taking out a whisker. This sort of oomph would likley blow up a most ICs. I think a better choice would be something like 100uF at 5 to 15V. Less voltage but longer current impulse.

I often use this brute force method to track down power rail shorts on PCBs, be it on a new board i just soldered and happened to get a solder ball wedged in the wrong spot, a old board where an IC or cap blew and hard shorted a rail or any other case where there is no obvious cause for a power rail short while there being way too many components to easily track down. For that i just set my PSU to the rated supply voltage such as 3.3V but gradually crank up the current to many amps. Sometimes the guilty components just gets warm, sometimes it catches fire and sometimes it blows the short open and the board magically starts working again. Its a very effective strategy but the trick is to not turn the voltage too high.

[mzzj]

Ouch, that's not nice.

Tho 47uF at 50V is quite a bit of energy for taking out a whisker. This sort of oomph would likley blow up a most ICs. I think a better choice would be something like 100uF at 5 to 15V. Less voltage but longer current impulse.

I often use this brute force method to track down power rail shorts on PCBs, be it on a new board i just soldered and happened to get a solder ball wedged in the wrong spot, a old board where an IC or cap blew and hard shorted a rail or any other case where there is no obvious cause for a power rail short while there being way too many components to easily track down. For that i just set my PSU to the rated supply voltage such as 3.3V but gradually crank up the current to many amps. Sometimes the guilty components just gets warm, sometimes it catches fire and sometimes it blows the short open and the board magically starts working again. Its a very effective strategy but the trick is to not turn the voltage too high.

9 volts cleared the whisker previously but only for about 5 days.

[SilverSolder]

An expensive haircut!

[Bud]

Could have tried an ultrasonic bath to shake it off ?