Weak 5V line, Nexus RX-5300 530W modular PSU.

[PopcornMobo]

Since i'm kind of broke i had the brilliant idea to go and buy a defective PSU from the local PC parts guy and try to fix it.
I got this pretty nice looking RX-5300 with all the cables for 5€.
As soon as i got it i took it apart in search of bad caps (as you do) but much to my surprise they were all good.
Plugging the PSU into a known good mobo results in a no post, so i took the PSU to the bench and started measuring voltages.
No load 12V and 5V were all good but as soon as i put a 3R9 resistor on 5V it dipped down below 1V and the PSU started squealing and resetting, while it's doing that the voltage on the 5V rail slowly goes up until it kicks back to life and jumps to 5V steady.
Same resistor on 12V makes no difference and only results in a burning hot resistor.
All voltages come from one main transformer so my only guess is that the 5V diode is bad but i measured it at 0.1V, which seems all right for a high current shottky.

I'll go probe some more, meanwhile i'd appreciate any help.

[R_J]

Re: Weak 5V line, Nexus RX-5300 530W modular PSU.

My guess would be any small elec. caps in the primary, they will look fine but could be close to being open (no capacity)

[PopcornMobo]

Re: Weak 5V line, Nexus RX-5300 530W modular PSU.

My guess would be any small elec. caps in the primary, they will look fine but could be close to being open (no capacity)

Just checked all the caps on the primary side and they're all good.

One thing i noticed is that with a 15Ω resistor the 5V rail dips down to 4.2V but all the other rails stay unaffected, i might have to do some more testing.

The 5VSB rail that is created by a separate supply is always there and is not affected by any of the main power rails.

I'll go hook up a scope on the 5V rail to check the ripple.

[PopcornMobo]

Re: Weak 5V line, Nexus RX-5300 530W modular PSU.

Some more stuff i found out.
I gave the 12V rail a 2.3Ω load and quickly noticed smoke coming from the power supply.
At first it looked like it was coming from a capacitor but turned out there was a cheeky little shottky (SB2100) hiding under the huge output common mode choke.
When i took it out it was well smoked and completely shorted, i replaced it with some other shottky, not sure what the model was but it was about 4x bigger
Now the PSU seems to be working much better but it's not perfect.
Before i replaced the diode the PSU was making this clicky noise at the transformer but it's not doing that anymore.
It also no longer goes unstable when i "overload" the weak 5V rail.
The 5V rail is also stable but only when i have the 2.3Ω load on the 12V rail.
I'll have to see where the diode went, might be that my optocouplers are not working right.
Too bad the board is double sided and also a pain to work with because it's held in by all the wiring.

[momaka]

Re: Weak 5V line, Nexus RX-5300 530W modular PSU.

Would like to have seen some pictures with this.

Just because caps don't look bad doesn't mean that they aren't - at least that applies very much for the cheap non-Japanese brands. If your PSU had Japanese caps, then they are probably okay indeed.

Also, it might be worthwhile to try and figure out why that schottky diode blew. While parts can blow at random, it's also not uncommon for some models/designs to have design flaws.

[PopcornMobo]

Re: Weak 5V line, Nexus RX-5300 530W modular PSU.

Turns out the diode was the rectifier for the -12V rail. Another thing is that there's a WT751002 supervisor IC, which should take care of things but it doesn't really help much.
There are three optocouplers in total, one is for voltage feedback, other is for 5VSB feedback and the third is for fault feedback.
All are on when the PSU is running, if fault feedback opto is turned off then the PSU shuts down.
Since the PSU is running there are no detected faults by the supervisor, also the 5VSB is good so i can rule those two out.
The voltage feedback opto is controlled via the 12V rail but only it and nothing else.
There's some weirdness going on at the 5V and 3V3 rails, they both have these nearly identical circuits next to them consisting of a TL431 that controls a transistor, but from what i can gather this does not influence the voltage feedback.
I might just bodge a resistor divider from another PSU schematic to control the voltage feedback.

[PopcornMobo]

Re: Weak 5V line, Nexus RX-5300 530W modular PSU.

I took two pics, one of the supervisor IC area and one of the 5V and 3v3 areas.
Please excuse the fluxy mess
Edit: calculated what the TL431's trigger at, on the 5V side it's 5.194V and 3.435V on the 3V3 side.
I might just hook those TL431s to the voltage feedback opto i think.

[PopcornMobo]

Re: Weak 5V line, Nexus RX-5300 530W modular PSU.

I did some bodging but did not achieve much. I found a 400W PSU schematic and copied the resistor divider network from it to my PSU but it didn't work well.
5V was ok, 3V3 was ok as well but 12V rail was way too low so i took my bodge out and replaced the original resistor divider.
The PSU works as it did before the mods so no harm there.
Another important thing (i think) is that my power-good (gray wire) signal is low, and wikipedia says it should be high, so i'll have to investigate that.

[momaka]

Re: Weak 5V line, Nexus RX-5300 530W modular PSU.

There's some weirdness going on at the 5V and 3V3 rails, they both have these nearly identical circuits next to them consisting of a TL431 that controls a transistor, but from what i can gather this does not influence the voltage feedback.

Most likely, your 5V and 3.3V rails are mag-amp regulated. If you have separate toroid outputs for the 5V and 3.3V rails from the 12V rail, it's definitely that. Providing more picture of the entire PSU should allow me to tell you for sure, though.

As for fault feedback - there's no way the 3.3V, 5V, and -12V rails are not connected to the WT751002 supervisor IC. There should be thin traces from these rails going towards the WT751002 chip... unless there is another chip that they go to (but I doubt it.)

If your PG signal is low, that may be because of of the rails is too low or too high, but not enough to trip protection. Or just not detected properly by the WT51002 chip to issue a good PG signal. *OR*... some of the small caps surrounding the supervisor IC are bad. Again, you didn't tell us what brand of caps the unit has nor did you check them with an ESR meter. So there's no telling where the problem is.

If I were you, I wouldn't modify the PSU - i.e. the 3.3V and 5V rail circuits. After all, the PSU has to have functioned properly at one point in its life with these original circuits. Sure there may be design flaws. But first fix the original circuit. Then you can see if there's anything that can be done to improve it. I've done the same copying of circuit from one PSU to anther in my early repair years, and it's rarely worked properly.

*EDIT*
Just did a search on the PSU online and found some pictures. It is indeed a dual mag-amp regulated unit for the 3.3V and 5V rails. And it appears to have Teapo capacitors throughout. So I can bet you some of the very small Teapo caps are already going high-ESR and that could well be why you have bad PG signal. Time for an ESR check and possible recap my friend.

[PopcornMobo]

Re: Weak 5V line, Nexus RX-5300 530W modular PSU.

Fixed, i think....
As mentioned above i was not getting power good so i looked at the function table in the WT751002 datasheet.
PGI (power good in, pin1) had to be above 1.2V to send a PG signal, mine was at 0.06V
The PGI signal is derived from the +5V tap at the transformer and gets rectified through a tiny SMD diode (D19), in my case the diode was showing a 0.2V drop both ways and would only show 3V rectified so i replaced it with a SMD shottky diode i had.
After i replaced the diode the rectified voltage got up to 15V and PGI voltage was 1.256V.
The PG signal went to 5.16V so i assembled the PSU and put it into my PC.
Upon power up everything works as it should, i get a post screen and BIOS HW monitor shows 3.440V (3V3), 5.222V (5V) and 12.249V (12V) and the PC is running fine.
I think this concludes the repair.