Corshair RM750 - doesn't allow for board to turn on

[kotel studios]

Well, this is odd.

Even after sitting for more than 30 minutes unplugged there's still deadly voltage inside the primary cap (230VDC). The APFC should discharge the caps in less than 20 minutes, so maybe the more advanced could figure out what is happening here.

[stj]

apfc is only active at load and with the psu running.
a lot of pc psu's dont have discharge resistors either - i noticed that before.

[kotel studios]

I seem to have found the issue that made the capacitor not discharge (and potentially the thing that caused the FETs to burn out). I forgot to solder the schottky diode back
Just to be on the safe side I'll let the capacitor fully discharge for a few days before I go ahead and fix it up (hopefully for good).

[stj]

discharge them with a 470k resistor, .5w or more

you have me curious now, i need to grind down an FP packaged fet!

[kotel studios]

Finally replaced the MOSFETs again and soldered that schottky diode back in place.
So far the PSU does not blow up, generates the voltages proper and can spin up an 3.5" SATA HDD. Sadly I still cannot power on any boards with it. PG is there (5.08V) when I manually short PS_ON and GND. Powering up just an small MATX board without the CPU P4 connector (to minimize load, this always worked for me on all PSUs).

Any ideas?

[kotel studios]

Taking a look at one dead sacrificial board I used which had an screwed SIO which caused it to power on the moment power was applied, it is now completely dead. It still does power on right after power is applied, but I can't get it to POST by mashing power/reset buttons.This would cause the 5VSB is doing funky stuff.
Other higher quality boards don't have this issue. They just shut down (like my Asus in the opening post)
Any ideas on why the PSU still doesn't power on boards?

[stj]

missing "PG" signal from the supervisor chip
it's used to reset the motherboard once the psu is stable

[kotel studios]

Yeah, I also think that too, but wouldn't the board respond to RESET button? I know this one was a very dead board (was always on, can't shutdown) with other finnicky stuff. That's why I've also used it as a sacrificial board.
Let's not dive down on why that board doesn't work no more.

It seems like Corshair couldn't even figure out how to make the "quiet" fan solution (forum post here).
So, since PG signal is only responsible for resetting the board when voltages are stable, we can safely rule it out as the culprit for "no life" issue on boards.
My next assumption could be the fact that 5VSB/PS_ON generation isn't all perfect and it's causing the OCP on the boards to trip. This would've need to be checked with an oscilloscope, which I don't own (mine has it's screen broken).
Lastly, it might just be the OCP/UVP,OPP tripping falsely. This model has that corshair link stuff, although I don't know the pinout of it and I don't think the PSU sends any signals of it's state when it's off.


Done measurements on the power button on the sacrificial board no (MSI, just with ATX20 pin, no P4 connector). With this broken corshair, I get 2.45V on positive side of the switch! With a known good bequiet PSU I get 3.25V (isn't bat nor great, just enough to work). So now I guess this could be the UVP protection stopping the PSU from running? Or is it a bad supervisor IC?

[stj]

the 5v standby is independent and should always be present on a psu,
the motherboard controller should use it and a clock - usually 32.768khz to monitor stuff and power up the psu.
(and then monitor PG and maybe voltages to bring the system out of reset)

you mentioned a psu that switches itself on - thats usually a bad opto-isolator.

as for fan control,
i'm not a fan of that, the psu is usually responsible for keeping air flowing through the machine unless it's a gaming rig.
if you slow or stop the psu fan the motherboard or gpu could overheat.

[kotel studios]

PSU doesn't switch itself on. It was that dead asus board which did. PSU stays off until I force it to by shorting PS_ON with gnd.

Attaching supervisor board pictures. It is based on Weltrend WT7502. That solder blob on the back is factory. Don't know what corshair did.
PCB PSU model

D750E005L/D850E003L
9MC850E00FCTX5LF
REV.: 03