C2D temperature woes

[Babrbarossa]

I've been wondering what's behind my idle-temperature fluctuations and sswilson suggested that it might be the psu (couldn't find the thread where we discussed it so I started a new one). I tested the 5 and 12 volt rails today using a molex connector and found that my 5V stayed between 5.02 and 5.03 during boot, idle, and load, while the 12V stayed between 12.38 and 12.40 during boot, idle and load so I guess there's nothing out of the ordinary there- One strange thing I did notice was that there was 1.17V running through the 12V when the computer was shut off- does this have something to do with the cmos battery?

[sswilson]

You'll need to check the specs and see what 12v rails everything is on.

Your molex connectors are probably not on the same rail as your cpu so it's possible that the cpu rail is low while the other ones are not being loaded.

[SKYMTL]

Originally Posted by sswilson

You'll need to check the specs and see what 12v rails everything is on.

Your molex connectors are probably not on the same rail as your cpu so it's possible that the cpu rail is low while the other ones are not being loaded.

Bingo. On the GameXstream, if you are measuring the voltages off of a Molex connector you are taking readings from the +12V3 rail which powers your accessories and not your CPU.

If your +12V1 (CPU) rail was dropping that much below spec, you would have alot more issues other than a few idle temperature drops.

[Babrbarossa]

I checked three different molex connectors and they were all the same- just now I checked the atx main connector and it was a little lower but still ok - 12.37 on boot and idle, and 12.35 under load.

Today's idle temps are:

Program core 0 core 1
TAT 47 45
Speedfan 52 na
CoreTemp 46 45

friday they were in the 30's on TAT

As far as other symptoms- like I was saying in the other thread to wilson (wherever it went)- I had some freeze-ups last week with loud beeping- my system had been stable for a month unchanged before that.

Also- I find that one of the molex connectors seems to underpower things- fans spin too slow, gas tube light only lights half-way when it was pulgged into that molex (last I tried anyway)- that connector tested fine for voltage. Anyway- I'm more concerned with why the temos are changing daily- this may not have anything to do with the power supply, but it's as wilson said- it's a good thing to check, I guess.

[SKYMTL]

Personally, I think it is just a malfunctioning temperature probe.

[Babrbarossa]

That was the first thing that crossed my mind. I have the newest version of speedfan v4.32 and it measures both Tjunction and TCase- typically Tjunction is supposed to be about 15 degrees hotter than TCase and mine is measuring 7 or 8 degrees lower.

I'm a little concerned because if this is so, then I have been running my cpu at about 80 degrees Tjunction under load. It would mean that TAT, which I have been relying on has been reading temps that were 23 degrees too low! I realize it can't be that bad, or it wouldn't be stable, but It's pretty annoying not knowing what temperture to trust.

[Babrbarossa]

I've been monitoring with speedfan and core0 is showing strange temperature spikes at fairly regular intervals that don't seem to be related to any processes. The top green line is Tcase which should be lower than Tjunction apparently.

[Babrbarossa]

Well- both the Tjunction and Tcase dropped ten full degrees for no apparent reason- it's consistent on all the temo monitors- bizarre!

[enaberif]

Originally Posted by Babrbarossa

Well- both the Tjunction and Tcase dropped ten full degrees for no apparent reason- it's consistent on all the temo monitors- bizarre!

Speedfan is the WORST program to monitor cpu temperatures or any temperature for that fact.

[Babrbarossa]

I usually rely on TAT or Coretemp- - the reason I'm using Speeedfan today is that the new version tracks both type of temp sensor, and I wanted a 'second opinion'

I'm positive that speedfan 'sucking' isn't the reason my temps are fluctuating so wildly.