PDA

View Full Version : Why do motherboards suck at reading voltages?


GilmourD
12-08-2006, 10:01 PM
Well, I just got my new Silverstone ST50EF-Plus. Definitely happy. :) However, I have to wonder why motherboards typically suck at reading voltages. Right now my 12V reads as 12.52V, my 5V reads as 4.83V, and my 3.3V reads as 3.14V. However, when I put my Fluke 111 on a molex and SATA connector, the 12V is 12.19V, the 5V is 4.97V, and the 3.3V reads as 3.37V. Why do they even bother having the motherboard read the voltages if they're going to be so off?

jonnyGURU
12-09-2006, 07:47 AM
Well...

First off, it's a sample from only one lea. One lead that is providing power to other aspects of the motherboard and therefore has enough resistance to show a drop in voltage.

That lead is also going through traces, which drops the voltage even more.

So then why would the voltage read high because of low? Like the old CPU thermal probes, compensation for inaccuracies are taken into account and the number is adjusted accordingly. Unfortunately, when you do something like this, the number is going to be off exponentially the farther from mean you get.

Baron
12-09-2006, 09:46 AM
Silly head on........
Doesn't that make overclocking etc a little tricky when voltages for the cpu, memory etc are taken from BIOS options ?
Or do the overclocking experts use a DMM to check volts so that they don't overcook things ?

GilmourD
12-09-2006, 10:50 AM
Silly head on........
Doesn't that make overclocking etc a little tricky when voltages for the cpu, memory etc are taken from BIOS options ?
Or do the overclocking experts use a DMM to check volts so that they don't overcook things ?

That's exactly what I did. I've used the same PSU before on different motherboards and each board gave a different reading. Usually, my Fluke gives the same reading on the same PSU regardless of the motherboard.

GilmourD
12-09-2006, 11:11 AM
Well...

First off, it's a sample from only one lea. One lead that is providing power to other aspects of the motherboard and therefore has enough resistance to show a drop in voltage.

That lead is also going through traces, which drops the voltage even more.

So then why would the voltage read high because of low? Like the old CPU thermal probes, compensation for inaccuracies are taken into account and the number is adjusted accordingly. Unfortunately, when you do something like this, the number is going to be off exponentially the farther from mean you get.

I haven't seen anything, but do they make any products that connect to your PSU connectors and maybe display them in a drive bay mount or something? :)

jonnyGURU
12-09-2006, 02:18 PM
Yes. SevenTeam has a Power Meter that works just like that, but it doesn't have PCI-e connectors and doesn't support multiple +12V rails. :(

GilmourD
12-09-2006, 02:38 PM
Yes. SevenTeam has a Power Meter that works just like that, but it doesn't have PCI-e connectors and doesn't support multiple +12V rails. :(

Sounds like we need to petition an update. :)

madmat
12-09-2006, 02:43 PM
Well...

First off, it's a sample from only one lea. One lead that is providing power to other aspects of the motherboard and therefore has enough resistance to show a drop in voltage.

That lead is also going through traces, which drops the voltage even more.

So then why would the voltage read high because of low? Like the old CPU thermal probes, compensation for inaccuracies are taken into account and the number is adjusted accordingly. Unfortunately, when you do something like this, the number is going to be off exponentially the farther from mean you get.

How do they get high voltages? I take mobo "aa" as a sample of the "a" line, test the voltages against the bios readings then tweak the bios to reflect the true readings. Then I program the bios for every board to reflect this. Unless every other board is 100% identical to this there will be variances. Something as simple as a bit more solder or a change in the solder such as a different ratio of zink to tin and you've got higher or lower resistance in the circuit, boom the calibration is shot to hell.

Sometimes you get a mobo that does rerport the proper voltages, the 9CJS Zenith I had was on by .01V on all the PSU voltages, sadly it can't do the new gear such as PCI-e so it's going to be demoted to HTPC duties.