5V rail of PC only 4.7-4.8V ?

[MathWizard]

I noticed the 5V rail of my 850W silver rated PC PSU, is only 4.7-4.8V, (on a good DMM, 4.7xxxxV) I never looked on the scope yet. I think I might have noticed this 1-2 years ago. The PSU is maybe 3yo now.

So perhaps the ESR of the output caps is getting too high, which would be an easy fix, and maybe now reduce stress on the secondary rectifiers.

This PSU is under warranty, so by rights, it probably is eligbeable'sp '  at 4.7V. Maybe I should conact the corp

[fzabkar]

Where is the +12V rail sitting?

[amyk]

How much current are you drawing from the 5V?

A lot of the newer "high end" PC PSUs have most of their output power on the 12V rail, and thus possibly far less available on the 5V rail than older PSUs of lower total wattage.

Here's a 510W PSU with 40A on 5V:
https://www.pcstats.com/articleimages/200507/PCpowerTC510SLI_s4.jpg

Here's 1kW and 1.2kW PSUs with only 22A on 5V:
https://images.anandtech.com/doci/12702/SILVERSTONE_ST_PLATINUM_04.jpg

[edpalmer42]

The ATX spec allows +-5% so 4V75 is acceptable for the 5V line.  What did you measure and how much do you trust your meter?  What do your other rails measure?

[rs20]

What set-up did you make these measurements with? Was the PSU plugged into a running system at the time?

A lot of PSUs have regulation/feedback shared across multiple rails, so a lightly loaded 12V rail and a heavily loaded 5V rail will cause the 12V rail to increase in voltage, and the 5V rail to drop. And vice versa. Similarly, measuring a power supply with no load (aside from a multimeter) is questionable, and measuring while a significant load is on the 5V rail and no load is on the 12V rail is doubly likely to give these results.

So perhaps the ESR of the output caps is getting too high, which would be an easy fix, and maybe now reduce stress on the secondary rectifiers.

This makes no sense, the ESR of a capacitor is not along the delivery path so a high ESR would not manifest as a drop in output voltage (it manifests as a increase in ripple instead). I suppose I might be wrong though in the case that the spiky output confuses the voltage feedback though.

[james_s]

This makes no sense, the ESR of a capacitor is not along the delivery path so a high ESR would not manifest as a drop in output voltage (it manifests as a increase in ripple instead). I suppose I might be wrong though in the case that the spiky output confuses the voltage feedback though.

I kind of doubt it's the issue here, but high ESR can absolutely cause a low output voltage on a swithmode power supply. Typically the ripple goes way up and the average voltage drops. It behaves a bit like a PWM control.

[MathWizard]

The ATX spec allows +-5% so 4V75 is acceptable for the 5V line.  What did you measure and how much do you trust your meter?  What do your other rails measure?

So it's borderline at 5V -5%*5V =5V-0.25V

I think the 12V was good, and over 12V, I'll have to put this on ths scope.

I don't have much power usage overall, just some SSDs ATM, my near 10yo HDD gave out, but IDK if that was related

Its a 850W EVGA supernova B2 I think, and it's basically a Seasonic model, and Seasonic is top level PSU "name"

[rs20]

Yeah, that PSU is kinda designed to be powering a PC with 2 graphics cards or something borderline crazy like that -- A limit of 12V at 70A = 840W on the 12V rails alone (even though the total limit for the PSU is 850W). With your 12V consumption seemingly probably less than 10% of that, I'm not surprised that the 12V is over and the 5V is under. Would be interesting to test if putting 500W of load on the 12V rail would change things, but unless you have some very specific gear I think that might be a challenge to do.

[amyk]

This one? https://www.evga.com/products/specs/psu.aspx?pn=6c83d705-1398-48c1-bac2-329f94dec9e1

The ratings suggest it's a two-stage design: mains to 12V and then 12V to 5V and 3.3V. With that design, a high 12V is expected if you're not using much of its capacity, but a low 5V isn't.

Would be interesting to test if putting 500W of load on the 12V rail would change things, but unless you have some very specific gear I think that might be a challenge to do.

Automotive headlamps would make a good test load for that.

[fzabkar]

I'm guessing that the 12V rail would be sitting between 12.5V and 12.7V.