I know this is a huge subject and the general answer is ‘don’t try to reform a cap’ (with exceptions yada yada yada).
Mr Carlson has explained the ramifications in different ways. (And I have building his cool low-volt cap tester on my todo list.)
Here’s the deal: I’ve inherited two, rather beefy Dell server PSUs, each capable of max 50A @ +/- 12 and about 30A @ 5. Sustained is about %80 of that. I don’t know when they were fired up last… :/ Their innards are clean, dust free – which doesn’t necessarily mean anything if they were in a datacenter, of course.
Since I don’t have the cap tester yet, I’m not going to bother checking, much less try to reform, the several mambo caps in each unit. I’m just gonna yank and replace them and move on, saving them for later testing.
On linear PSUs, the lightbulb and variac method seems to work fine. But on a switcher, the moment it’s energized, that huge startup load hits the big caps. (And, yes, I’ve had one go boom … lucky I was expecting it.) And switchers freak out when undervolted.
So, the linears have a real advantage in that regard; there doesn’t seem to be a way to get the caps in, and downstream from, a switcher to ‘naturally’ reform by slowly increasing the voltage. That got me wondering about something I’ve never seen mentioned.
What would happen, if I had a turn-on-turn-off gadget that could turn a switching power supply on for say 5 to 50 milliseconds, then turn it off for how-many-ever seconds, repeating the cycle. Then after some predetermined (or sensor-evaluated) time, it turned on again for say, 75 msec. Wash, rinse, repeat…
Is this an unworkable idea? Will majik smoke come out?
Is there a way to deal with this problem, other just recapping for safety-sake?
Looking forward to your thoughts!
Thanks, and cheers,
– HP