TP600X just installed 850 MHz CPU, heatsink compound needed?

[wb0gaz]

Just installed a 850 MHz speedstep CPU in my 600X; seems to be working fine, but "mobimeter" shows climbing temp (>90C after a while).

Here's my question - should there be any thermal compound between the CPU module and the heatsink assembly? None was on the old (orig.) CPU, so didn't add any.

Tks,

Dave
wb0gaz@hotmail.com

[joester]

First question, what condition is the thermal pad in?

second question, are you sure you have adequate contact between the chip and heatsink?

Third is a reminder... More speed usually means more heat. Heatsinks can only dissapate so much heat. you may be close to maxing out the cooling ability of the heatsink.

I would apply a light layer of Arctic Silver or equivalent, and see what happens.

Joe

[wb0gaz]

Hi Joe, this is my first attempt at 600X surgery, so I appreciate your quick reply!

If there was a thermal pad, it remained on the heatsink (I didn't look at the time), as the removed CPU and the newly installed one each didn't have a thermal pad visible. I should have looked, but didn't. Tells me I get to dig into it again! I do have some of that arctic silver stuff and will apply if needed, along with more-carefully checking the positioning and tension on the heatsink assembly.

Also, very good point re the total thermal capacity of the system, although I'd expect the fan would get into high gear before giving up, and so far this didn't seem to be happening. In any event, some careful testing called for before I commit anything critical to the machine.

Again, very much thanks for the quick reply!

Dave

[ms5133]

I have done the same mod.

I did not use any thermal paste..I just replaced the pad from the old CPU to the new (850).
My temps with mobil meter are :Idle 43-46
Browsing 47-51
Video editing 66-69

All degrees celsius.

I have never had a themal shutdown situation..90C seems very high .

[wb0gaz]

Thanks for the info --- it appears my results are much more problematic - I disassembled again, found that the thermal pad was in place, but not in very good shape, so I removed and replaced by thermal compound. I removed heatsink twice (to confirm that the terminal compound was covering the entire heat transfer area.) My temp indications are far higher than yours, idling in the high 50s and quickly going overtemp when I run something that wants high CPU usage (a digital signal processing program.) It appears thermal transfer from CPU to the heatpipe/sink assembly is very inefficient (although it was interesting to see the CPU speed being throttled back from 866 down to 700 as it hit the low 90s, and that it would try to work it's way back up to 866 as CPU temp dropped off. FYI I'm in a cool ambient temp (23C) and the heatsink temp is slowly rising into the high 30s, at which point the heatsink fan comes on normally.

Anyway, perhaps someone else has had similar struggles and can advise, because most other experiences doing this seem much more positive.

Tks,

Dave

[wb0gaz]

Another few data points ---

Switching to 700 MHz ("battery optimized", even though I'm on AC power) and running the DSP application, which poses continuous CPU load (and with mobimeter also, so total CPU utilization is about 90%) the CPU temp seems to stabilize in the mid 70s. Switching to 866 MHz, utilization drops off a little and the CPU temp climbs steadily and fairly quickly to the point of shutdown. Just running mobimeter at 866 MHz produces CPU load of about 30% and CPU temp stabilizes at about 70C.

Any other thoughts/suggestions appreciated,

Tks,

Dave

[Dark Savant0]

You could try removing the fan grill on the side so there is more air cooling?

[whizkid]

Another data point: I upped my 600X to 800MHz (850 was to pricey at the time) and have not had any heat issues. It was a while ago so I don't remember the exact steps, but I'm certain I didn't use any heat transfer goo.

[Grey Area]

I'm using a 850MHz-CPU in my 600X, and the maximum heat I have been able to achieve was near 70 C. I still use the original thermal pad, it was not very sticky on the CPU side. I did not put any thermal paste on.

The cooler has four small metal ridges "framing" the thermal pad, so if the pad is removed, then the CPU will only have contact to those thin ridges. I don't think thermal paste can cover that gap - at least not when sensible amounts of paste are used. I have read about people filing off the ridges, so that the CPU gets direct contact to the cooler, but I'm a bit skeptical - won't the assembly get bent that way, if it is designed to have a two millimetre gap for the pad?

[u.mac]

I've done those things...
cut small metal ridges, polish the plate
then fasten the 2 screws from cooler to processoboard and - oups - there was no contact between cooler an processor... ~1mm is missing
very carefully(!) drilling deeper the coolerscrewholes... contact
put 2 pieces "Collabratory Liquid MetalPad" between processor and cooler
using metalrings for more space on one of the screws, because screw now ist too long , second screw has enough place

600x @ 850MHz WinXP + MobileMeter presents... <40°C idle, <50°C "normal work" - but calculate Pi... >90°C.
@ 700MHz the temperature is <38°C idle, <45°C "normal" and max. 85°C.

I prefer 700MHz, because my non-SpeedStep-Board runs >7 hours with both batteries and on powerline the vent is very much longer still.

btw. is there any tool for fancontrol under XP? Not meaning sleep+awake