Right, so after I started removing components, I found a broken transistor (q212 - ksp92), which I guess is part of the driver group. After replacing it, now I have a well working sub amp.
Still, I'm confused how tf the negative rail was getting that low and why the amp was producing sound at all. Yes, indeed the waveform (at least according to my cheap scope kit) had offset to the positive side, but yet a waveform good enough to produce sound.
Most likely the damage to the transistor was caused by me, when I was removing the black goo years ago.. and all that time it has been "working"...

Not having the -12V rails is no bueno. It is tapped into the transformer before D901.
At this point I can figure out only two reasons for not having the negative rails - dead short on one of the rails or the transformer winding for the negative rails is gone.
The protection circuit switches off the smps IC on the primary side by shorting VREF to DIS with opto Q955. Usually this is triggered by overvoltage, undervoltage, over-current and high temp.
One last try before focusing on the transformer - desolder d901, d912 and d911. This will disconnect the transformer windings from...

Check for bad caps on the power rails. Excessive ripples might cause the microcontrollers to freeze.

That's the smoothing caps on the primary getting charged.



Is the buzz coming from the transformer(s)?




Did it exceed 40-45V?
Manual says, according to the model, it should be 30 or 40V (+/- few volts)
On c902 the voltage should be same but with opposite polarity.
If there is no voltage on c902 you have the following options:
there is short on that rail
d901 is bad
the transformer is bad

It would be interesting to know if the 12v rails are present (c921/c911 and c922/c912)...

If the short is gone, you can try powering up the unit (without q502 and q901). Preferably through a dim-bulb tester, but even without, looks like the smps ic has a working protection and a test should not do more harm.
A good result would be stable power rails (stable voltage on c905, c902 c901 c911 c912).. from the schematics and the smps documentation, looks like the smps IC supports low and high power modes, so not really sure what should be the voltage.

Good day everyone.

I have on the bench an amp board from DALI SUB E-12F. It is class D amp with a smps on the same board.
The issue is that it powers off way before the rated power is reached (the protection circuit kicks in). It has an independent standby power supply, which stays on after the main power supply is turned off via a relay on the primary side (thus there is no reason to believe that the protection is triggered from the smps IC on the primary side).

The normal value of the positive and negative power rails is +/- 43V.
The positive rail stays...

Check again d901 in diode mode for short (I do not remember by heart the internal schematics of the bridge rectifiers, so I test each pin to each other pins, then switch the probes and again).
You can use IPA to find the hot component(s) - drop IPA on the component and see how fast evaporates.
Thermal print labels also can be used - cover area with a label and see where it will get black.

This power rail doesn't have too many common points to ground - q502, d516, q501, q434+q436... all other possible paths include too many components that have to be shorted. So try injecting...

so now you have q901 in place and again short on c902?
Last time did you replace q901 with a new one? Check the resistance in ohms between the 3 legs of q901.

If you have bench power supply, you can inject voltage (1v@2-3A should be ok) on c902 ( positive on the positive leg of the capacitor, which is ground for the board and negative on the negative leg of the capacitor which is the negative rail ) and see what is getting hot....

I guess .475 is voltage drop, which is fine value for a diode in circuit.



I'm starting to get lost here.

Let's summarize the things. Make a list with the following data:

Which components are removed while testing: (ex. qXXX, qXXY, uZZZ, etc. )

Measure and write down the resistance in OHMS >>ACROSS THE LEGS<<< of the following caps: (ex. c905 = 1.3kOhms and rising, or just "OK" if the value starts low and then rises to > kOhms or mOhms). Do not use "beep" or diode mode of the dmm.

...

Hmm, q901 is a 12V voltage regulator. Check c907 >>>>across its legs<<<< for short.

p.s. If you still have not soldered back Q502 DONT. Keep it desoldrered for the moment.

Alright. C902 is on the -B power rail. The most probable suspects on that rail are d901 and q502... remove them one by one and test when and if the short will disappear.

"C954, C955 and C963" are not grounded. They are on the primary side and on floating power rails (see https://en.wikipedia.org/wiki/Floating_ground ).
For instance C954's negative terminal is on the neutral (or live, depends on how you connect it to the power outlet) which is used for reference for + and - after the bridge rectifier. That said - do not touch anything on the primary side when the unit...