Logitech z906 power boards | transformer clicking

[vidi]

Hello folks,

I'm trying to repair a Logitech z906 and I'm kindly asking for any advice/tips/ideas. Let's story begin.

1) The primary issue was a popping sound, after cca 40-60 minutes of playing, exactly as you can hear on (someone else) video
 

I wired all power rails (Power board: +43HV, +12V, +5V) (audio board: +3.3V, +1.8V) to several multimeters and 5V LDO output was dropping to 4.1V, which causes 3.3V regulator to drop to 2.9V, which causes CPU to restart. So I replaced the LDO with exactly the same type and it seems to work fine without any issues.

2) Second issue (or feature?) is that I can hear a clicking noise on power board when the whole system is in iddle state - mean everything is "off" - just plugged into a power outlet. According to thermal camera it looks like this:

https://drive.google.com/open?id=1UKfCatTGDgBb1rMwRU7lwsqOfZ2te56b

and what shines is a start bias (Q6 mosfet, R111, R77, R78). The power consumption of the iddle state jumps between 0.2 and 8 Watts, measured by cheap power meter.

When the system is "on", start bias gets colder (means not used anymore, DC/DC is powered from transformer coil) and there is no clicking sound. Power consumption is stable on 22.8 Watts.

My assumption is that nobody sane would design the idle power state this way - using OV/UV inputs - and what I see is a failure.

In idle state OV input of ISL6721 DC/DC convertor has following signal (left part):
https://drive.google.com/open?id=1Z9DnGQM7Ke8OmWiCHwAPZl5TqJIzJD0w

OV input is wired to output of optocoupler U3, powered from section above TP59. The power for U3 has waveform (left part):
https://drive.google.com/open?id=1o5Ki5piYNZA0SEfn1Y4XeDeHwvbGKzFT

My question is:
How should I proceed further with troubleshooting, without damaging the board.
I mean, the feedback for OV might be the issue. But if I remove the U3 to confirm my assumption
and the OV feedback was correct, I might damage some other parts.

Photo of board https://drive.google.com/open?id=13AdM41LgOiR9WiiFkCbGh4ff3eWNYfVr

[ciine]

In this case you have an optocoupler for current (protection) and the other for voltage.
It's not good to cut to nothing
Check esr at the filter capacitors for the 5v voltage,
there might be your problem

[vidi]

Hi. Thanks for response.

Don't worry. I've kept my sanity and I didn't remove the optocoupler , even though I'm still curious what would happen if I did.

The ESR and capacity of all caps on power-board were fine, including the 5V-rail-one. Caps were tested with low voltage - using DE-5000. Currently I don't own anything like Mr. Carlson's Leakage tester for higher voltage.

I've spent quite a lot of time on this nonsense and bothering in person 4 analog designers for advises. And the result ? All of them claim that it may be designed this way to maintain the low-power mode. Even if it is weird to me, since the power factor I've measured has value 0.24. Unless I'll get a chance to disassemble and measure some functional Logitech, I'm not able to tell if it should work this way.

Changing the LDO and adding an extra (1 mm) thermal pad between the outer aluminium cover and the 5V LDO ( since the LDO itself dissipated heat only into the power board ) seems to be enough for fix. After 2 months of daily use, the owner claims everything works fine.

Anyway, here is a connection scheme I've did in Inkscape, perhaps it may help someone. I've learned a lot while trying to fix this device.

(exported .png) https://drive.google.com/open?id=19qbhhGN-kQAjGvCY-5xtQrQH1vmN4BjF
(Inskcape .svg) https://drive.google.com/open?id=1XWBmyK9vNP4qce2RwMOwIh7a55EPO18O