PSU test. Software vs multimeter results

So I went out and bought a multimeter after I discovered that my +12 volt rail was fluctuating in hwmonitor when I played video games. Ill show you the before and after via OCCT.

This is idle:


This is under load:


I know the 11.84 is still above specs, but I was told the software methods arent very accurate. So lets see if thats true!

These are under load tests.





As you can see the reading is way off in software. I tested every connection I could and they always had the same 12.08 result under load


EDIT: Just tried this on my PSU from my HTPC and my results on the 12 volt was 11.58 via software and 12.01 via multimeter under load.

The reason for the differences you are seeing, is because the software reads sensors on the mobo, which can be very highly innaccurate. I keep a DMM in my toolkit, for this same reason. Most of the local stores have this problem of using junk PSU's in their clients rigs, so I tend to have to fix their problems. It's kind of insane that these local guys think that because it says 680 watts on it, that it is sufficient to run an i7 with a GTX295 with only 22A on the 12V rail.

deadlysyn said:

The reason for the differences you are seeing, is because the software reads sensors on the mobo, which can be very highly innaccurate. I keep a DMM in my toolkit, for this same reason. Most of the local stores have this problem of using junk PSU's in their clients rigs, so I tend to have to fix their problems. It's kind of insane that these local guys think that because it says 680 watts on it, that it is sufficient to run an i7 with a GTX295 with only 22A on the 12V rail.

Click to expand...

Yea I knew that the software method was off but I just had to be sure. Also I figured if others relied on software that they could see how big the actual difference is between the two.

Hmm, how long before a mobo maker includes a built in highly accurate DMM rather than the standard sensors?

ratbuddy said:

Hmm, how long before a mobo maker includes a built in highly accurate DMM rather than the standard sensors?

Click to expand...

How about just using accurate sensors that report accurately to the software That would be the desired end result IMO...

DMM's are handy to have for sure! I keep a DMM and a Kill-A-Watt in my aresnal for troubleshooting...

I am not sure what they use on motherboards, but modern $2 ADCs (voltage to "number" converters) can provide at least 10 bits accuracy. That's about +-5mV for 5V.

So I would guess the errors came from elsewhere. Or maybe they don't have a sensor at all and just guess the number somehow.

EDIT: or perhaps you are looking at different rails?