[Solved] Sparking DC- pin of rectifier bridge

[nuno]

Troubleshooting another A/C controller board... this one sparks at the DC- pin of the rectifier bridge. Sparking has enlarged the PCB's pin hole by literally burning the PCB's substract (and through the pin's solder too; the pin has even melted a bit and formed a metal ball). There's copper on both sides of the board, connected by vias (and by the bridge pins soldered, of course); a bit beaten up, copper got loose from the PCB at one of the pins.
Bridge is fine. Before the bridge there's filtering and after there's a PFC circuit, basically a boost with an IGBT. IGBT is fine. No shorts ahead.
I've never seen such behavior. A couple of bridge's pins are very short and I resoldered to make sure it was all well soldered and not only "in contact" thus sparking. Any ideas on causes for such sparking?
Thanks!

[ArthurDent]

To have sparking (arcing) you have to have a gap instead of a solid connection. If everything else is o.k. you may have to trace out the runs and use wire to solidly connect the bridge pins to where they need to go.

[Gyro]

I've never seen such behavior. A couple of bridge's pins are very short and I resoldered to make sure it was all well soldered and not only "in contact" thus sparking. Any ideas on causes for such sparking?
Thanks!

Poor quality solder joints, weakened further with age and heat until they fail completely and start arcing.

You might want to replace the bridge as a precautionary measure - it will have seen very high temperatures on those pins. A replacement would also have decent length legs which would improve your chances of a decent lasting repair. You probably want to visually check the board for other suspect joints, where one lurks there may be others just waiting to fail.

[SeanB]

Dry joint on the one lead will do that. Replace the bridge rectifier, as it will have been heated up a lot from the arcing, and will fail short circuit sooner rather than later. new one with wire leads space 2mm above the board, so you can solder the lead to top and bottom copper, and if a SMD part you probably will have to put in some thin wire jumpers to bridge the burnt vias, after cleaning all the carbon away to bare board.

[james_s]

I've seen this problem lots of times, it's fairly common on the main relay in microwave ovens. It's caused by a bad solder joint as others have already mentioned.

[nuno]

Ok, thanks, makes sense. Yes, I already had in mind to drill and add a "via", those tracks are pretty bad.

After fixing that it was still not coming up. Some more debugging and the low voltage rails were all near zero Volt, then found a broken track in the PSU controller feedback loop. It was weird, because a little piece (1 or 2mm) of the track was missing, but no darkening of the area, I could barely see it (discovered it with the continuity test). Nothing else seems to have been affected. Everything came up after fixing the track.

I think the original problem is a mechanical design flaw. The bridge plus another "rigid pins" big power device (a 3 phase power IC) are rigidly soldered to the PCB and rigidly screwed to the heatsink (there are also other 2 TO-264 devices but the legs were bent in an arc to provide for some strain relief). Then, to make things worse, the plastic housing where the PCB is screwed is very tight on the PCB and keeps it constantly forced sideways, thus forcing the rigid legs of the power devices. This in a device where there is some vibration, and voilà...