SMPS Troubleshooting - What's Making the Electrical Sound?

[DaJMasta]

I've been puttering around troubleshooting the main SMPS for an Agilent 4UHV Ion Pump Controller - the first stage after the AC input (not the high voltage stuff, that's unplugged for troubleshooting) and while I think my understanding of SMPS is below most of my repair skills, I've replaced a few parts and changed behavior a bit but haven't yet succeeded.  While not critical to my question, it's based around the UC3844A controller, using a very distinctly non-reference design, and has a separate chip doing active PFC.  The issue I'm seeing is in startup, I've gotten it to actually run the backlight on the display, but it's not starting sufficiently to actually boot and it's making a horrible electrical hum - even with the front panel board disconnected.  The power supply board has the rectifier, filters, this controller and associated PFC controller, the secondary transformer, and a handful of parts and ICs near the output connectors, some of which are just the output rails, some optoisolators, an EEPROM, and I think there's something else too - it's been a bit since I last looked through.

In any case, the cause of the hum is my question - I think IDing it will be helpful in figuring out what may actually be going on, and it's been difficult to trace down so far.  I've used a plastic probe to touch every electrolytic and inductor/transformer on the board and it doesn't change the hum even a little.  It's quite audible when on, like close to normal speaking volume.  When you turn off the power switch (physically cutting mains input), you still hear it for several seconds until you hear a faster dropoff.  I've had a look with a thermal camera at the board and nothing's cooking itself - the hottest thing are a couple of what appear to be low power load resistors connected to the secondary side.

So what could cause this significant noise?  Is there a methodology or order of operations you prefer when trying to diagnose issues with SMPS where you don't fully know/understand the topology?

[tautech]

One would need to suspect caps.....but how they might pull switching frequencies down to audible levels is another thing. 

I'd be pulling some secondary side ones and testing them....remember they should be Low ESR types.

[DaJMasta]

Thanks, I checked them but while some voltages were absent, I wasn't seeing ripple, really, and I wasn't able to find one making noise.

That said, I did use a sound meter to trace down the sound, and then eventually found the cause and fixed it!

The sound was coming from the transformer, and the sound was generated deep enough inside that touching the tape around the windings or core itself simply did not contact the part making noise.  I scoped around for a bit and identified that the power for the controller and PFC controller was low - and when viewed at the main filter cap after the diodes for it - was normal looking except for brief, regular excursions to zero.  All the way to zero, with very fast edges for the ~8.5ms period they were happening at, and their width was so narrow, it wasn't showing as AC activity when I had my meter on DC mode using it - it only showed on the scope.

After some tracing, removing and checking parts, and narrowing down, I found that the same drop was present at the switching controller, except that the downward spike actually went negative (this was being blocked from the rest of the rail by a series diode), and there was no negative signal or inductor (aside from the transformer) for it to be just switching to.  After some checking, it turned out a little extra solder between two pins had formed a very small bridge that really didn't look it under magnification, and that getting rid of this meant no noise and full startup.

The unit had been broken when I got it, and I had replaced the controller IC, so it was my fault in the end, but I think replacing that, the PFC controller, and a diode or two actually would have fixed it months ago when I made the first attempt, would I have been slightly more sparing with my solder application.