How acurate are LCD displays with thermal probes

[marlin1856]

As my vid card has no temp moitoring feature I am forced to use an LCD display with temp probes. I have slid 2 probes under the VF900cu (1 either side) hard up to the gpu but not obstructing the HSF GPU contact at all. This gives me an idle temp of 23 and full load of 42, can any one please tell me how close this temp would be to actual core temp.

 

[Exeodus]

check what your temps are in the overdrive section of the catalyst drivers, and see how far off you are...... your temp probes probably arent 100% accurate, but at least it will give you a good idea.

 

[W2hCYK]

As my vid card has no temp moitoring feature...

He has no built in temp monitor for catalyst to detect.

I'd say the probes are fairly accurate. They should be detecting it very close to 100% if mounted properly. Put them on the core's corners to get the most accurate temp without losing thermal transfer capability.

 

[marlin1856]

Thanks for the replys, yes your right, my vid card has no LM63 temp mon kit onboard (Xpertvision cost cutting) but i will try to get the probes even closer to the corners of the core. Again thanks for the advice.

 

[ex_reven]

hey ... just a thought...although i have no experience or knowledge of probes

if you wanted to work out their accuracy why wouldnt you get a household thermometer (a mercury one) and stick it in a glass of water. When you get a stable temperature reading, stick your probes in there and look at the difference in readings?

or would this damage the probes?

 

[marlin1856]

Thats a good idea, thanks but its not the accuracy of the sensor i need, its the temp difference between the probe and the actual core temp or at least as close a guess as i can get eg.5,10,15 deg difference then i can add it to the probes temp and i should have a reasonably close reading.

 

[ex_reven]

hmm
if your heatsink was 100% effective at transferring heat (from core to the air via fan) then you would have no problems as you could mount the probe through contact with the heatsink.

But im guessing no heatsink is anywhere near 100% (imagine how expensive...) so you might have to make do with the reading you have now. If the probe is touching your heat sink or the base of the core i cant see why your reading would be far off. But there it probably still be at least 5 degrees off the actual temperature (with an accurate probe).

I dont know how far off youd be if your probe is cheaper.

The X850 is a good card, ive got a x800 pro and i have temperature readings on my card :S

 

[trog100]

the probes are fairly accurate.. but unless u mount it at exactly the same spot the card sensor is the reading isnt gonna mean much..

quite what bit of a card gets measured i dont know.. i can guess buts that all it would be a guess..

trog

 

[W2hCYK]

in extreme cases, i've seen people use thermal epoxy and a file, etch out a groove in the heatsink, place the probe in it, 'cement' the probe into the groove, and lap it smooth after it hardens. This will help you get near 100% readings. maybe it will be 1 or 2 degrees off.

If you feel like doing that, i'd recommend doing it on a practice heatsink first, with a practice "Probe" such as a regular piece of wire or something. Maybe give this process a shot.

 

[marlin1856]

Thats a cool idea, sounds like i may need to tinker a little bit more but i don't want to have to hack into my new VF900 unless its going to work but thanks to all for your input.

 

[ex_reven]

Thats a cool idea, sounds like i may need to tinker a little bit more but i don't want to have to hack into my new VF900 unless its going to work but thanks to all for your input.

shizer
i wouldnt even think of touching my mobo, especially not with something sharp ....

 

[Dia01]

I currently use a Cooldrive 6 with provided thermal probes (RTD's). By nature, as the temperature increases the resistance output decreases and vise versa, so unless the probe tip or lead has sustained any mechanical damage the only inaccuracy reading obtained will be a result of the monitoring device itself. The calibration of accuracy can and will almost change over time also. To be sure I advise to compare the readings of your probe against a known temperature and at the end of the day you are only after an indication, there will always be a tolerence of accuracy.

 

[marlin1856]

That all makes perfect sense, I did not think of it in that way so thanks I will take your advice on board and try to calibrate the probe with a known temp and run with that,Cheers.