New PSU is keeping Vcore and +12v to high.

[Mr_Perfect]

A while ago I replaced a suspect 300watt Antec power supply with a new 400watt Antec. At the same time I also upgraded my mobo from a dead A7M266 to a A7V333 and traded Windows 98SE in for Windows 2000. After getting the machine up and running I noticed that my 1.33 T-bird was idling at 48+ rather then 42-43. The rest of the machine is basicly the same, so I figured it was a difference in the two mobo's temperature reporting.

Recently I took a look at my Voltages and was a bit surprised
+12 is at 12.992
+5 is at 4.972
+3.3 is at 3.264
Vcore fluctuates betwean 1.792 and 1.808

I checked my BIOS settings, and the Vcore was automaticly set to 1.75volts. Setting it to 1.75 manually didn't change the voltage it was coming up as.

Tommorow I'm going to try setting the Vcore with the jumpers, but I have a fealing that 1.75 is going to keep turning up as 1.8. Is there some way to curb the output of the PSU? The machine is getting unusaly hot during gaming(54c max, it used to be 47).

Mr. Perfect <i>a name fraught with peril</i>

Descent, because gravity is highly over-rated.

 

[Mr_Perfect]

I messed around with the jumpers just now, and not only are all the voltage limiters on, but manually setting the Vcore to 1.75 still leaves me with 1.79 - 1.8 and no effect on the +12.

Oh hell. I just spent some time poking around in everythread PSU, and found this.

<b>You should replace your power supply. The voltage regulator must be failing if the reading is too high. If you were simply overloading it, the reading would be too low.</b>

So the PSU is probably damaged, huh?

Mr. Perfect <i>a name fraught with peril</i>

Descent, because gravity is highly over-rated.<P ID="edit"><FONT SIZE=-1><EM>Edited by Mr_Perfect on 10/22/02 04:48 PM.</EM></FONT></P>

 

[HammerBot]

These voltages are sufficiently accurate. Anything within +/- 5% shouldnt cause any concern. Further, its better to have the voltage a bit too high than too low, since the voltage monitoring device doesnt measure the voltage exactly where it is being used. There is an additional voltage drop across the PCB tracks that will cause the voltage at the components to be a bit lower.

 

[Crashman]

Your readings are almost perfectly where I like to see them. Due to stability issues, I like to see them as high as reasonable, which would be 12.5-13.8v for the 12v line! (most 12 devices are made to run at up to 13.8v because this is the voltage of a fully charged 12v car battery).

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>

 

[ejsmith2]

I've got some old 6.4 and 4.2gig hard disks that go into thermal recail., left and right, at 12.6v.

But all my newer stuff runs awesome.

[Jedi mind trick] You LOVE Palladium. [/mind trick]

 

[Mr_Perfect]

Just so long as my components are being damaged.

Would slightly elevated voltages cause a 6 degree jump in the CPU temps like that? I wish I could remeber what the voltages where at when it ran cooler...

Mr. Perfect <i>a name fraught with peril</i>

Descent, because gravity is highly over-rated.

 

[lhgpoobaa]

QUESTION!
What are you using to report these voltages?
If its asus probe, please use something else like motherboard monitor.
Asusprobe ain't too reliable.

<b>And if you gaze for long into Toms Hardware Forums, The Forum gazes also into you! </b>

 

[Mr_Perfect]

Heh, well, both Asus Prope and the BIOS. Though the BIOS reports the +12 at 12.45 rather then 12.9.

Mr. Perfect <i>a name fraught with peril</i>

Descent, because gravity is highly over-rated.

 

[lhgpoobaa]

well 12.45 is ok, as its within the 5% error margin that seems to be universally applied.
Get rid of asus probe. Get Motherboard monitor that actually reports the same values that your bios does!

<b>And if you gaze for long into Toms Hardware Forums, The Forum gazes also into you! </b>

 

[ejsmith2]

That really, really screws me up that the processor pulls off the 12v rail. You've got to bump that voltage down to 1.75, which means you've got to have regulators. Most boards have 3 or 6 regulators on them, but they should be putting out a *TREMENDOUS* amount of heat having to pull down 7amps of current. I sure wish someone would tell me why they function without heatsinks.

[Jedi mind trick] You LOVE Palladium. [/mind trick]

 

[lhgpoobaa]

Well i know for a fact that the power regulation chips and mofsets get quite warm.

I remember reading a report someone wrote about modding their epox board. basically he cut up a few heatsinks and thermal glued them onto all the mofsets/chips and got considerably better voltages as a result.

<b>Cogposto tomsa, ergo sum - <i>Descartes</i>
</b>Translation:<b> I post at Toms Hardware, Therefor I am. </b>

 

[HammerBot]

The regulators are not of the linear type. They are switchmode buck regulators, and they convert power with an efficiency of 90-95%. So if your CPU uses 60W of power, you will loose about 3W-6W in the regulators. Since there are several of these regulators working together each will dissipate about 1W. 1W wasted in such a small regulator will make it hot. But not a lot.


<i><b>Artificial intelligence will never be a match for natural stupidity</b></i>

 

[scriptasylum]

"I sure wish someone would tell me why they function without heatsinks."

They must be switching regulators other wise you are right, they would heat up really bad. Take that 10.25 (12-1.75) volt drop and multiply it times whatever current Vcore needs: thats the heat, in watts, generated by the regulators IF they were linear regulators. If what you say is right (7 amps), that IS a lot of heat: 71.75watts (more than some CPUs!). But, me thinks they are switching regulators, so there are a LOT less losses.

Why do CPU's (if true) use the 12 volt supply and not the 3.3? That would be MUCH easier to regulate.

LOL, looks like I type too slow and Hammerbot beat me to the punch


<P ID="edit"><FONT SIZE=-1><EM>Edited by scriptasylum on 10/25/02 03:07 AM.</EM></FONT></P>

 

[HammerBot]

Why do CPU's (if true) use the 12 volt supply and not the 3.3? That would be MUCH easier to regulate.

No, not necessarily. It depends on the switching topology. The advantage of bucking down from 12V is that your primary current is much smaller.
E.g. to get 70W, the current drawn from the 3.3V rail would exceed 21A. From the 12V rail it would only be about 6A. For a traditional BUCK-type switching regulator this significantly relaxes the requirement on the switching element (the mosfet) and improves efficiency.


<i><b>Artificial intelligence will never be a match for natural stupidity</b></i>

 

[ejsmith2]

Hmmmmmmmmm.

This is *highly* interesting.

I may have to make a run on Radio Smack this weekend.

[Jedi mind trick] You LOVE Palladium. [/mind trick]

 

[HammerBot]

I have made several switch mode power supply designs. So if you have some additional questions just post them.

<i><b>Artificial intelligence will never be a match for natural stupidity</b></i>

 

[ejsmith2]

Actually, I was thinking more along the lines of a voltage regulator for my fans.

Right now I'm using pots for stuff. But I'd have no qualms about using a three-pole switch to do it. Hell, I might even go so far as to use two switches in tandem, for 4 settings.

Figure 12v, 9v, 7v, and 5v. That would rock hard.

[Jedi mind trick] You LOVE Palladium. [/mind trick]