Is this 'normal' - CPU core to evaporator head delta

[Arseface]

I'm running an FX55 at ~3.05Ghz with an XEII. The chip doesn't like much more than this no matter the voltage, which is dissapointing, but that's another story.

Anyway - I was wondering whether a 40C delta between the CPU core temeprature as read by SmartGuardian and the evaporator head was 'normal'? TBH, I was expecting less than 30C, at the most...

http://img167.exs.cx/img167/2900/phasedelta3am.jpg


The CPU core temperature will drop back below zero (where smartgaurdian can't read, so reports back 200+) almost instantaneously after the load is cut - less than 1s. Obviously the evaporator head cools down much slower than the CPU core after load is cut, taking around 30s to get back down to it's idle temp (~-32C)

The evaporator head temps seems reasonable for an XEII, but I'm not too happy about the CPU core temps. I seem to be getting decent contact between the CPU and the evaporator head, and I'm using arctic ceramique as thermal compound.

What results are others getting? I know most people use an LS or a MachII GT, but I'd still be interested in seeing CPU core temps under full load, assumming you can get them to rise above zero of course!

[Epsilon]

The CPU temp probes are VERY VERY inaccurate.

So is the mach2 sensor.

As long as you have a good contact (look at the inprint) everything is fine.

[berkut]

a 20C delta is normal with -40C evap temps and below

[Aphex_Tom_9]

The CPU temp probes are VERY VERY inaccurate.

So is the mach2 sensor.

As long as you have a good contact (look at the inprint) everything is fine.
that's the philosophy i go by, it works well enough :)

[saratoga]

Thermistor type probes in PCs are generally not meant to read much if any below 0C, so they don't report accurately.

Likewise diodes have extremely nonlinear effects at lower temperatures, which makes them hard to read unless they're specifically calibrated for the range in question.

Most likely your temp probe just doesn't work in the range in question.

[Ssilencer]

The XEII does not hold temps well at the wattage the FX55 is given.
The XEII probe is as inacurate as the probes in the mach2 (maybe this was what Epsilon was trying to tell :D)
So, you need an external temp sensor if you want to know really what the evap temp is.
And you need to check if you have good contact (Epsilon said "As long as you have a good contact (look at the inprint) everything is fine.")

I'm allmost sure that the XEII head is in the -10/0c range...

[Arseface]

Okay, thanks for the info.

Yeah, contact is pretty good. I've switched motherboards a couple of times with this setup in the past few weeks, and I've got good imprints so far (and similar performance to this), so I'll guess it's all good.


I'd still be interested to see what other peoples CPU core temp sensors read out under full load though... :)

[saratoga]

Hmm just noticed tonight that Maxim sells a diode reader, the 6658, that will read down to -55C.