lapped q6600 - strange behaviour

[panheimer]

Last week i got a Q6600 (can't remember batch..), vid 1.225 wich is strangely really hot. At a very low oc of 3.0 Ghz, wich 1.25 v and cooled by a lapped TRUE with 2xscythe s-flex 1200rpm i've got an idle temp of 42°C and load temp of 65° (with 31°C room temp).
So i popped the cpu out and lapped it. Well after the lapping it resulted out an idle temp of 47°C!!!!! Checked and rechecked the mounting, the paste, the fan but the same results! An very high temp of 47°C! The strange thing is that now i've got a full load temp of 60-61°C, so the lapping lowered my load temps by some degrees, but still can't figure out why idle temps got higher!
Any clue?

[it_burns_when_i_pee]

TRUE usually wins at load temps, the number of heatpipes are efficient at removing heat

therefore, if you have a bad mount or paste is un-evenly spread to one side, it could be enough to affect idle one way and load the other

i had this problem once with the TRUE and with both HR03+ and HR03gt. in all my instances it was the paste spreading un-evenly.

lately i've turned to putting 2 blobs of paste so it spreads to the middle and out.

pull the TRUE off and inspect it like jailbait

[lowtcell]

Low VID chips tend to clock higher but run hotter. Are you using Realtemp or Coretemp to monitor temperature?

[MadHacker]

I also lapped my Q6600 and my water block,
when i apply paste i apply a very thin layer to the entire CPU,
since both the block and the CPU are supposed to be perfectly flat I don't think you would want a large blob in 1 or 2 places.
then the paste is making most of the contact between the block and the CPU.
You want the metal to make most of the contact and the paste just to fill in the micro cracks that are left over after lapping it.

my 2¢

[panheimer]

i'm reading temps from everst, wich reports temps exactly like coretemp, while realtemp reports 5°c lower (as it use 95°c tj).
i've also checked the paste footprint and it looks good to me:it is even spreaded, and i use very little of mx-2.
i've used the single drop method and the stright line accross the cores ones but results still remain the same.
i've also tried to put a little more paste, but of course the temps rised by 1-2°C so i guess that the quantity put before were right.

i've heard that low vid chips runs hotter but never tought SO hotter! The Q6700 (vid 1.26) i've installed on my primary rig idles at 41-42° C with 1.4375 at 3.6 Ghz and loads at 66°C, with another lapped TRUE and 2xnanoxia fan @1600 rpm.

i must say i like my load temps of 60°C, and want to try higher clocks with this q6600, but idles temps worries me a little bit. i find not comfortable knowing it idles near 50° even if load temps are low. what do you think?

[Sparky]

I don't care much about idle temps, load temps is where it is at. Especially since when folding my PCs are almost never at idle anyway :D

[YMAA]

To what grit did you lap your Q6600? How long did it take? It's possible that a bad lapjob can do more harm than good. I lapped my Q6700 this weekend and it brought down idle and load temps by a few degrees (also using a lapped Ultima-90). I may lap it a little more with 800grit this week.

One more thing to keep in mind is that lapping removes material from the contact surface. It's possible that you're not getting as much mounting pressure now that your CPU isn't as tall, which could lead to higher temps.

[it_burns_when_i_pee]

I also lapped my Q6600 and my water block,
when i apply paste i apply a very thin layer to the entire CPU,
since both the block and the CPU are supposed to be perfectly flat I don't think you would want a large blob in 1 or 2 places.
then the paste is making most of the contact between the block and the CPU.
You want the metal to make most of the contact and the paste just to fill in the micro cracks that are left over after lapping it.

my 2¢

blob, 2 blobs or line are far better than covering the whole chip.

with a blob or two, when the waterblock or cooler are installed the pressure pushes and spreads the paste outward. but when you cover the entire area; you trap air. i take it you are aware that air is a poor conductor of heat, hence the job of TIM - bridge the tiny gaps between the 2 metals to achieve maximum thermal conductivity.

this topic is covered well over the web. but if you dont want to take forum advice on the matter, check out arctic silver and arctic cooling's websites. they give detailed instructions. more and more are switching to 2 blob method. just think about how it spreads under pressure

issue with one blob is that some coolers are installed downwarrd which can push the paste un-evenly. even clipping or screwing (whichever the cooler's mounting mechanism) can cause the spread to be un-even.

[MadHacker]

blob, 2 blobs or line are far better than covering the whole chip.

with a blob or two, when the waterblock or cooler are installed the pressure pushes and spreads the paste outward. but when you cover the entire area; you trap air. i take it you are aware that air is a poor conductor of heat, hence the job of TIM - bridge the tiny gaps between the 2 metals to achieve maximum thermal conductivity.

this topic is covered well over the web. but if you dont want to take forum advice on the matter, check out arctic silver and arctic cooling's websites. they give detailed instructions. more and more are switching to 2 blob method. just think about how it spreads under pressure

issue with one blob is that some coolers are installed downwarrd which can push the paste un-evenly. even clipping or screwing (whichever the cooler's mounting mechanism) can cause the spread to be un-even.

I never though of air getting trapped...
always considered the TIM being to thick to let the metal (heatsink to CPU) touch each other directly. after all you want the metal touching each other as much as possible and a glob of TIM may decrease that a bit, especially when both heat sink and CPU heat spreader are lapped.
things for me to ponder...

[Tonucci]

Strange results... Possibly due to unflat TRUE base. If you are not going to lap it, you could try it_burns_when_i_pee thermal paste aplication method, in order to fill the empty gap caused by the concave base.

[panheimer]

The TRUE is lapped and it works well. I've lapped both to 1200 grit - can't find an higher one here. For the CPU i've started from 600-800-1000-1200 and tooks me about 2 hours of work. With TRUE i've started from 300, then 400 and then 600-800-1000-1200. Tooks me a whole afternoon. They're both not mirror finished, but sure pretty flat: tested with a glass and a drop of water, the water was pushed out of the base, leaving very little on the base.
Also the mounting is pretty tight and the TRUE won't move nowhere, can't slide or rotate it - maybe is it the problem? Would i have to try lessen the mounting pressure?

[it_burns_when_i_pee]

I never though of air getting trapped...
always considered the TIM being to thick to let the metal (heatsink to CPU) touch each other directly. after all you want the metal touching each other as much as possible and a glob of TIM may decrease that a bit, especially when both heat sink and CPU heat spreader are lapped.
things for me to ponder...

i cant take credit for the info i gave. i read it on another forum one day. as i said i've had a few paste spreading / mounting issues in the past. but so far the 2 blobs method has done the trick every time

intel put 4 or 5 lines of paste on the bottom of their retail coolers. obvious there were already aware of this.

[MadHacker]

guess i'll try it next time a tear my machine down... that will be soon as welll...

[Loser777]

I've gotten two Q6600s (1.2125v VID chips) that I did a solid mount job on, both were around 26C Idle BIOS @ 1.38vish and up to around 72C Load.

The one 1.3125v VID Q6600 I got only hit 67C @ 1.448v.