Very weird Delta Powersupply issue

[Arkade]

I have 2 Delta's I bought for budget built Office machines that specifically require 500+W and an 80mm rear fan. I bought a Delta DPS-700-FB 1 and a Delta DPS-600-MB A

Both are BRICKS of psu's and are amazing, huge power output, nice cables, nice quality, very stable output, little ripple, and fans aren't even loud.

But then i dared to go into S3 sleep.... the computers 50% of the time will fall flat on their faces and fail to come out of S3 sleep, either just starting in the bios OR fans going to full speed and the management chip resets the system. Both computers have the same issue, tried both powersupplies in my main desktop and the same thing happens

Looking at the design of these PSU's i fail to understand how this could happen, S3 sleep involves storing data in RAM, which remains powered by the 5VSB rail, which should NEVER turn on, if it did, then the sleep LED should go out along with the 5VSB lights on the mainboard, however instead the system appears to lose or corrupt memory anyways. Putting a lower wattage PSU in the computers results in things working unless I put it under load which then it shuts off...


What could cause this and How can I remedy it? Is it the PG signal somehow or the 5VSB, PG reads 300-340MS each time. I'm about to try a 5000 farad capacitor across the 5VSB to ground and see if it maybe dips a little right before it shuts off, maybe that will help(should hookup a scope)

 

[EarthDog]

1. How did you test ripple to know its low?
2. What system are these on? Newer intels have a deeper sleep which were not compatible with some power supplies.
3. Id guess the lower wattage unit is newer and supports those sleep modes.

 

[Arkade]

the lower wattage psu is a 200w delta is from a pentium 4 478 system, the systems i'm putting these in are Core2Quad with a Q43 chipset, i tested ripple with a very slow polling DSO

nice to see you earthodog

 

[EarthDog]

Then that wasnt it...

 

[Arkade]

I stuffed a 5000uf Cap on the 5vSB rail, WOW what a difference, no more sleep issues, removing it, issues return, been trying to make it fail with the capacitor but i just.... can't.... It actually works very well with it. except now I need to make it a little more.... permanent, considering seeing if theres room inside the PSU OR just splicing the 5vSB and a random GND at the 24pin atx, crimping it on, then leaving the fugly huge capacitor to do it's work.

Think I could get away with smaller than 5000 uf? this thing is giant

5vSB rail has some not so good looking scope shots

 

[NiHaoMike]

Most likely the cap for that rail inside the PSU has failed and adding the external cap is just masking the problem.

 

[wingman99]

Wow it is great that you fixed it.

 

[Arkade]

It would be odd that it's bad in the PSU, i have MANY of these PSU's, all are behaving kinda goofy. They are Server PSU's, they seem to have almost a pulsed 5V output instead of a strong one, like it's almost trickling

 

[Arkade]

Hmmm, the 5Vsb is generated on the main high voltage PCB and the rest of the rails use the low voltage PCB so if I had to guess, it's noise from the high voltage stage, I see a 3300uf 10v capacitor, I can fit a 10000uf 10v cap barely, going to try that and if it works have a not stupid looking solution