Re: P5WD2 deluxe + Pentium D 830
- From: nospam@xxxxxxxxxx (Paul)
- Date: Wed, 25 Jan 2006 20:47:09 GMT
In article <1138210692.978617.91990@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
"Viktor" <News@xxxxxxxxxxxxxxx> wrote:
> Hi!
>
> Has anyone tried out this combination with the latest Bios and has
> experienced false Temp-readings for the processor?
> I am using this combination with a proprietary cooler from silentmaxx,
> which look very convenient to me (and has won several tests for its
> cooling performance). However, under full load, the Asus temp tool
> reads up to 75=B0C and thereby exeeds the max temp by nearly 6=B0C.
> However, the manufacturer of the cooler took much effort to prove to
> me, that the temp read by the motherboard must be false.
> Under full load, he measured the temp of the plate between the
> processor and the cooler with a laser-thermometer, which was always
> below 50=B0C. He also installed a small temp-diode between the processor
> and the plate of the cooler, and showed me, that temp stayed below
> 50=B0C while the Asus tool was reading above 75=B0C. When taking the load
> of the processor, the temperature read by the Asus tool (PC-probe II)
> dropped below 50=B0C within a second while the temp-diode slowly went
> down, as one would expect it to do. Last, by using some third-party
> tool, they showed me, that the processor did not step down into
> power-savement mode, which it should, if it was really exceeding the
> maximum allowed temp that much.
>
> Since all those tests were done in my presents, I am quite convinced,
> that there must be an issue with the mobo-processor combination.
> Anyhow, maybe my board has a bug others don't have, so I am quite
> curious to know, whether others have the same problem or not.
>
> Greets,
> Viktor
False temperature readings are common these days, whether
working with Athlon64 or with Pentium.
PDF page 31 shows a typical measurement circuit.
http://www.winbond-usa.com/products/winbond_products/pdfs/PCIC/W83627EHF_EHG.pdf
Pg.83 here defines the diode formula.
http://download.intel.com/design/Pentium4/datashts/30235104.pdf
qVd/nkT
Ifw = Is*(e - 1)
n - ideality, a number close to 1.008 or so.
k = Boltzmanns constant (http://en.wikipedia.org/wiki/Boltzmanns_constant)
q = electronic charge (value is also on the Boltzmanns_constant page)
Is = saturation current
Ifw = the 100.0uA or 10.0uA applied by the ohms DMM
T = temperature in kelvin
This application note is proof that a diode does not have to be
a terrible solution for measuring temperature. This note explains
how the "two current method" can be used to give a value for T in
Kelvin. Note that no motherboard monitors use the "two current
method", and use a less accurate single current method (and the
current isn't exactly constant either).
http://www.smsc.com/main/anpdf/an1214.pdf
AMD quotes their temperature accuracy as about +/- 7C. You would
have to add to that, the error caused by the non-ideal circuit used
on the motherboard.
Another way to measure temperature, is to see if the processor
is throttling. A recent Intel processor will reduce internal
processing rate when it reaches about 65C to 70C, and will shut the
computer down at about 90C or so.
If you place a computing load on the CPU (Prime95 etc), then
adjust the CPU fan speed with a Zalman FanMate II or other
rheobus fan speed adjuster, you should be able to adjust the
CPU speed until you see throttling. At that point, you know the
CPU thinks it is at 70C. Then, compare to the temperature as
measured by the monitor chip. Now you know the delta_T between
temperature as measured with the two methods. You could also
calibrate your "scientist friend's" equipment, by seeing what
he reads when throttling (CPU slowdown) occurs. References on
throttling are here:
http://www.xbitlabs.com/articles/cpu/display/oc-guide_7.html
http://www.panopsys.com/Downloads.html (Throttlewatch download)
http://cpu.rightmark.org/products/rmclock.shtml (Rightmark Rmclock)
I have no knowledge of how Intel sets the threshold on their
processors. How ever they are doing it, the throttle temp is
probably set more accurately than the motherboard measurement
circuit.
HTH,
Paul
.
- Follow-Ups:
- Re: P5WD2 deluxe + Pentium D 830
- From: Paul
- Re: P5WD2 deluxe + Pentium D 830
- References:
- P5WD2 deluxe + Pentium D 830
- From: Viktor
- P5WD2 deluxe + Pentium D 830
- Prev by Date: Re: A8N-E Voltage Questions and Thoughts of PS
- Next by Date: Re: Asus board
- Previous by thread: P5WD2 deluxe + Pentium D 830
- Next by thread: Re: P5WD2 deluxe + Pentium D 830
- Index(es):
Relevant Pages
|
