How to find intermittent shorts in complex circuits

[Haywire]

I have a circuit that works flawlessly 90% of the time, but then it throws a random short.
And, naturally, it never happens when I’m testing it with a meter!
What’s your strategy for hunting down these pesky ghost shorts before I drive myself crazy chasing something that only acts up when it wants to?

 

[TheOhmBreaker]

Thermal cycling can show intermittent shorts. Use a heat gun on suspect spots while keeping an eye on the current. When it shorts, you’ll pinpoint which component is temperature sensitive. Gently flex the board while it’s on, cracked traces or bad solder joints will reveal themselves under stress.

 

[Ohmwardbound]

Consider adding an inline fuse or a current-limited power supply. This prevents damage when it shorts. While it’s running, wiggle every wire and connector to track down the fault, mechanical stress often brings out intermittent issues.

 

[ohmsweetohm761]

Looking under a magnifying glass can help you spot issues, check for solder bridges, metal shavings, or damaged insulation that you might miss otherwise. Also, look for loose screws or hardware that could cause intermittent contact. Sometimes, it’s just a stray wire strand causing trouble when things vibrate.

 

[barfle]

I would first check for intermittent connections, using the techniques described above. It can be tedious for sure, but you’ve already done the easy and the obvious.

 

[Alan]

Fantastic tips. Intermittent shorts can be a bit of a puzzle, but with thermal cycling, some gentle flexing, and thorough visual check usually lead you right to the culprit.

 

[oldgeek]

It's amazing what you can find with a thorough inspection of a board. I always start with my eyes, these days with the board under a scope. 10-20X is good for surface mount. look for spots that have been hot: discolored, loose track; "spider-web shorts"--those fine little bits of solder that sometimes become intermittent shorts--; and birdcaged wires. (Birdcaging happens when stranded wire gets flexed too much near the solder joint--individual strands break and can short to adjacent pins.) Some lead-free solders can grow solder "webs" to adjacent connections. Conformal coating helps with this problem, but it is not totally effective. I don't mess with lead-free, but I am old enough now that something is gonna kill me and it will be pretty soon, so I don't sweat the lead. For those of you under fifty, maybe gloves aren't such a bad idea if you work on old stuff or use leaded solder. Just a suggestion...

 

[TheOhmBreaker]

It's amazing what you can find with a thorough inspection of a board. I always start with my eyes, these days with the board under a scope. 10-20X is good for surface mount. look for spots that have been hot: discolored, loose track; "spider-web shorts"--those fine little bits of solder that sometimes become intermittent shorts--; and birdcaged wires. (Birdcaging happens when stranded wire gets flexed too much near the solder joint--individual strands break and can short to adjacent pins.) Some lead-free solders can grow solder "webs" to adjacent connections. Conformal coating helps with this problem, but it is not totally effective. I don't mess with lead-free, but I am old enough now that something is gonna kill me and it will be pretty soon, so I don't sweat the lead. For those of you under fifty, maybe gloves aren't such a bad idea if you work on old stuff or use leaded solder. Just a suggestion...

I completely agree, your eyes (and a good scope) notice things that meters often overlook.
Those spider-web shorts and birdcaging can be tricky little problems. And honestly, wearing gloves is a smart move.
Whether there’s lead or not, taking more time to fix additional boards sounds like a solid approach.