The Test Plan
Now that I have enough data to work with, and knowing how my areas of interest operate, I can put together a list of things that need to get looked at and things that I need to keep in mind.
The first and most important point is that the primary side of the 5VSB and PWM sections are referenced to -170V or whatever the low-side capacitor happens to be charged at. Without a proper differential probe, this means I will only be able to make rough DC measurements. But for most of what I want to look at on the primary side, that should still be good enough: if there is trouble to be found, it won't be subtle.
In flyback converters with multiple outputs, energy gets dumped into outputs following the path of least impedance. When capacitors start to go bad, dropping capacitance and increasing ESR, more energy gets dumped into other outputs, ripple voltages and currents increase across all outputs, and regulation gets worse. This tends to wear them out fairly evenly until they start failing altogether. Since the 5VSB caps are definitely in poor shape, all other outputs from that flyback transformer become suspects.
With that in mind, what are my targets?
- The 5VSB output caps are definitely busted, so I will be swapping those out and see how much it helps. I doubt they'll do anything about the out-of-spec 5VSB voltage, but this should at least eliminate most of the switching noise.
- Voltage on the 5VSB's auxiliary output needs further investigation to see if the 47µF capacitor is still doing its job and what its actual operating voltage is. Changing this capacitor for a larger one should reduce the dips caused by the PWM section starting to draw current from the auxiliary output and make it less susceptible to whatever happens on the 5VSB output.
- Check voltage on the PWM chip's VCC pin to see if there might be an issue either with its bypass capacitor or the power it receives from the auxiliary output through the photocoupler and booster transistor that might not come from the the auxiliary supply itself.
- Check signals around the 5VSB's TL431 to see if incorrect voltage might be caused by a flaky IC, resistor or leaky capacitor
Those would be my four main suspects, and I am 90% confident I will not need to go beyond the third. Before digesting my PCB assessment, I considered investigating the PS_ON# path and the TL431 on the 3.3V output, but decided to toss those ideas in the improbable bin. Same goes for everything related to the main converter feedback loop, since outputs look perfectly fine when the converter manages to start.
Don't tell my boss, I've managed to convince him that I'm only an expert at running benchmarks and writing about the results :)
However i do hate to throw stuff earlier than it should, like you.
Still working good, but I get black soot on my hands every time I open up that case. It's a black case, so it's hard to clean it off properly.
Thanks for taking the time to document this repair.
That's really bad--I doubt it is even legal to sell a power supply wired like that. I've never bought a CWT (Channelwell) power supply, and based on this I wouldn't buy one, except perhaps for a high end model where you might gamble that the company would exercise a bit more care.