Actual temp readings?

[Harold7]

On the ASUS A7V KT133A board, how can I tell what my actual CPU temperature is?

For a test I left the cover off overnight to equalize the internal temperature of the computer to ambient temp.

After putting the cover back on, I took temp readings using MBM 5 at startup--- Ambient air 68 F, CPU 88 F, motherbard 68 F.

Right now after running for 1 hour temp readings are: Ambient 69 F, CPU 105 F, motherboard 79 F

Shouldn't CPU and ambient temps be essentially identical at startup?

If I use the 20 F temperature differential between ambient and CPU temperature at startup as an offset, does that mean that the present reading of 105 F for the CPU is actually 85 F?

Is there any hardware sensor and display I can get which will show me the actual CPU temp?

Given the wide range of CPU operating temperatures reported by people using approximately the same HS and fan combos, it appears that reliance upon the ASUS CPU temp probe for any consistently accurate temperature reading is a waste of time... am I all wet or am I reaching valid conclusions with these observations? http://www.sysopt.com/forum/smile.gif

[This message has been edited by Harold7 (edited 05-12-2001).]

[ragtop]

I don't know about the A7V, but if I do the same thing with my MSI K7T Pro 2A I do get a processor temp reading pretty close to ambient temperature immediately upon startup. You have to check it quick though, those Durons/Athlons heat up real fast.

[Bovon]

As ragtop said, Athlons heat up very fast. Without any cooling whatsoever, they last about 5 to 8 seconds. The thermistor under the cpu in the socket is placed differently from mainboard to mainboard. For example. On the Abit boards, the thremistor will touch the bottom of the cpu if not adjusted down a little bit. Other boards like the Iwlll dosn't have long enough leads for the thermistor to touch the cpu...so, you can see here where the temps between an Abit and an Iwill will be different. I have read where the thermistor is not designed to actually touch the cpu, but to read the temps in the open area under the cpu...but, some ppl dissagree...and I have not found an answer.

Also, Durons mostly are made in Austin Texas...they fab Durons from aluminum and they run quite a bit hotter than the Tbirds that are now made in Dresden Germany from copper. This is only recently come about...sence about June of last year from what I have read...before that, Dresden may have fabbed Durons and Tbirds as well as Austin. Austin only fabs from aluminum and Dresdin fabs from copper...so, if you have an aluminum core cpu...it probably will run hotter, quicker than its copper counterpart.

[Harold7]

I have a T-Bird 1.2 GHZ/266 FSB.

My CPU and ambient temperature readings differ by 20 F at startup while the motherboard reading shows ambient at startup... shouldn't all three agree fairly closely, at least at startup?

Is there a hardware temperature sensor and display I can install which will show the true CPU temperature? http://www.sysopt.com/forum/smile.gif

[Fingers]

Maybe these will be of some help.

How to Take Your (computer's) Temperature: (http://www.burning-issues.co.uk/how_to/TakingaTemp/takingatemp01b.htm)

Calibrating Motherboard Monitor(s) (http://www.burning-issues.co.uk/hardware/calibrating/calibrating.htm)

[Harold7]

Let me put my hypothesis a different way... in order to evaluate the accuracy of any temperature readings you get in your computer, don't you first have to establish a standard temperature baseline from which all readings are compared?

I believe that, using ambient temperature readings as that baseline, after a long power-off period all components inside your computer should show essentially the same temperature reading at system startup.

Under this scenario, temperature probe placement is irrelevant, whether the probe is actually touching the CPU core or sampling the temperature of the air above the CPU, at startup at least, all temperature readings should agree with ambient temperature.

Since my CPU temp was consistently reading + 20F from ambient at startup, using MBProbe, over the course of several days, I let my system cool for at least 8 hours, recorded ambient temps from an external sensor, recorded MBProbe CPU and motherboard temperatures and applied correction factors in MBProbe until at startup all three temperature readings agreed.

Now at startup, my system agrees with ambient temperature, the motherboard temp slowly climbs to approximately ambient +8F and my CPU in normal operation runs at ambient +17F and ambient + 25F at 100 % usage, when I drop back to normal usage temps quickly return to the previous value.

Am I wrong in assuming that this procedure will work for any temperature analysis of computer components and cooling? http://www.sysopt.com/forum/smile.gif


[This message has been edited by Harold7 (edited 05-13-2001).]

[oksenior]

Harold7

I've placed an indoor/outdoor digital thermometer with the outdoor sensor attached to my cpu, or meybe the heatsink.

this thermometer stores the min and max temps as well as real time temperatures.

I got it from radio shack but don't get a wireless because I suspect the interfence from which we are shielded would render it useless.

oksr.

[Harold7]

I'll be interested in seeing your results tazman.

Be aware that the process of accumulating temperature data and applying compensation values to MBM or MBProbe may take several days since you must open the case up and just let it sit overnight to allow all internal components to reach ambient temperature so that when you do your initial temperature check after applying corrections you are operating from the ambient temperature baseline.

On my first attempt, I only corrected MBProbe -8C and when I ran another test the next day, I was still showing a + 2C differential between ambient and CPU... I corrected another - 2C and the next day, tested again and ambient, CPU and motherboard all agreed within 0.5C, so you might want to start out correcting a full - 10C, let it cool down overnight and then see how close all your readings are... I suspect -10C will be enough for you as well. http://www.sysopt.com/forum/smile.gif

[This message has been edited by Harold7 (edited 05-14-2001).]

[Harold7]

oksenior
I'm sure an accurate sensor setup with internal and external sensors with digital readout would work really well to get a reliable, real-time indication of CPU temperatures.

I've seen them advertised before, I just can't remember where, however, the procedure I outlined above I believe will factor out the + 20F error that appears to be present in most, if not all, ASUS CPU temp probes.

If the temperature differential I see at startup between ambient and CPU temps is linear and remains that way as temps rise, then compensating for that error in MBProbe or MBM should give a valid, and I believe, much lower CPU temperature reading for most people who see their T-Birds running in the 50-60 C range even though they seem to have adequate HSF setups and are experiencing no problems.

Maybe someone else who thinks their T-Bird is running too hot will try the same thing I did to see if their CPU temperature readings are significantly higher than ambient at startup also. http://www.sysopt.com/forum/smile.gif

[tazman]

Harold7 -

Great topic. I built an 800 Athalon on an Aopen AK-33 mobo and the temps are being reported at 131f by Sandra on both the mobo and cpu. I got paste, etc. with an excellent hs/fan. I'll follow your steps and get back with the results - I myself am really curious too.