[opensuse] How much difference does the right amount of thermal grease make? -- A LOT!
Listmates, I thought I would pass on the results of a (non-official) scientific test concerning the huge difference in cpu temperature just a small difference in the amount of thermal grease can make. The results below are on an AMD thoroughbred 2400+ using Manhattan silver grease. Test #1 (dab of grease on processor core with syringe, pencil eraser size, roughly 3mm high -- result - way too much) w83697hf-isa-0290 temp2: +62.5°C (high = +120°C, hyst = +115°C) sensor = transistor Test #2 (minimal amount of grease brushed on with Q-tip, very thin layer, much less than 1mm thick) w83697hf-isa-0290 temp2: +51.0°C (high = +120°C, hyst = +115°C) sensor = transistor Word to the wise, DON'T USE TOO MUCH GREASE! -- David C. Rankin, J.D., P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Mon, Mar 24, 2008 at 2:46 PM, David C. Rankin
Listmates,
I thought I would pass on the results of a (non-official) scientific test concerning the huge difference in cpu temperature just a small difference in the amount of thermal grease can make. The results below are on an AMD thoroughbred 2400+ using Manhattan silver grease.
Test #1 (dab of grease on processor core with syringe, pencil eraser size, roughly 3mm high -- result - way too much)
w83697hf-isa-0290 temp2: +62.5°C (high = +120°C, hyst = +115°C) sensor = transistor
Test #2 (minimal amount of grease brushed on with Q-tip, very thin layer, much less than 1mm thick)
w83697hf-isa-0290 temp2: +51.0°C (high = +120°C, hyst = +115°C) sensor = transistor
Word to the wise, DON'T USE TOO MUCH GREASE!
Do you think it was the amount, or the even-ness of the application? Seems to me that the brushed version would be more evenly distributed than relying on the "smash" effect of a deb in the middle... -- ----------JSA--------- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
John Andersen wrote:
On Mon, Mar 24, 2008 at 2:46 PM, David C. Rankin
wrote: Listmates,
I thought I would pass on the results of a (non-official) scientific test concerning the huge difference in cpu temperature just a small difference in the amount of thermal grease can make. The results below are on an AMD thoroughbred 2400+ using Manhattan silver grease. Word to the wise, DON'T USE TOO MUCH GREASE!
Do you think it was the amount, or the even-ness of the application? Seems to me that the brushed version would be more evenly distributed than relying on the "smash" effect of a deb in the middle...
I think it was the amount. From the appearance of the layer of goo that remained on the chip after the first attempt, I simply swabbed a little onto the Q-tip, finished cleaning the chip and heat sink and then used what was on the Q-tip to reapply. I think in this game measured in microns, the amount is definently the enemy. -- David C. Rankin, J.D., P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
David C. Rankin wrote:
John Andersen wrote:
On Mon, Mar 24, 2008 at 2:46 PM, David C. Rankin
wrote: Listmates,
I thought I would pass on the results of a (non-official) scientific test concerning the huge difference in cpu temperature just a small difference in the amount of thermal grease can make. The results below are on an AMD thoroughbred 2400+ using Manhattan silver grease. Word to the wise, DON'T USE TOO MUCH GREASE!
Do you think it was the amount, or the even-ness of the application? Seems to me that the brushed version would be more evenly distributed than relying on the "smash" effect of a deb in the middle...
I think it was the amount. From the appearance of the layer of goo that remained on the chip after the first attempt, I simply swabbed a little onto the Q-tip, finished cleaning the chip and heat sink and then used what was on the Q-tip to reapply. I think in this game measured in microns, the amount is definently the enemy.
Right. The thicker the layer, the more insulation (the grease is an insulator...just not as much of an insulator as air!) between the CPU and the heatsink. Uneveness can cause problems, too (air pockets!). -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On 24/03/2008, David C. Rankin
Test #2 (minimal amount of grease brushed on with Q-tip, very thin layer, much less than 1mm thick)
I would not use a qtip, you'll get fibres stuck in the thermal grease. Use a polythene bag or similar. And yes, the thermal grease should be as thin as possible. -- Benjamin Weber -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
David C. Rankin wrote:
Listmates,
I thought I would pass on the results of a (non-official) scientific test concerning the huge difference in cpu temperature just a small difference in the amount of thermal grease can make. The results below are on an AMD thoroughbred 2400+ using Manhattan silver grease.
Test #1 (dab of grease on processor core with syringe, pencil eraser size, roughly 3mm high -- result - way too much)
w83697hf-isa-0290 temp2: +62.5°C (high = +120°C, hyst = +115°C) sensor = transistor
Test #2 (minimal amount of grease brushed on with Q-tip, very thin layer, much less than 1mm thick)
w83697hf-isa-0290 temp2: +51.0°C (high = +120°C, hyst = +115°C) sensor = transistor
Word to the wise, DON'T USE TOO MUCH GREASE!
The main key here is surface area. The greater percentage of the die surface covered in goo that actually touches the heatsink area, the better the heatsink will draw heat of the processor. I've always had the best success applying with my finger as little as possible that fully covers the die, and smoothing it as much as possible, though I do like Benji's idea of using a plastic bag or glove or something to avoid contaminating the goo with hand oils and such that probably decrease it's efficiency. It's like a bed of nails, it is actually quite easy to lay on a bed of many nails if you are careful, the total surface area of all the nail points actually becomes fairly significant. I want to see some guru lay on a bed of 4 nails, now that is mind over matter! --Jason -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Jason Craig wrote:
David C. Rankin wrote:
Listmates,
I thought I would pass on the results of a (non-official) scientific test concerning the huge difference in cpu temperature just a small difference in the amount of thermal grease can make. The results below are on an AMD thoroughbred 2400+ using Manhattan silver grease.
Test #1 (dab of grease on processor core with syringe, pencil eraser size, roughly 3mm high -- result - way too much)
w83697hf-isa-0290 temp2: +62.5°C (high = +120°C, hyst = +115°C) sensor = transistor
Test #2 (minimal amount of grease brushed on with Q-tip, very thin layer, much less than 1mm thick)
w83697hf-isa-0290 temp2: +51.0°C (high = +120°C, hyst = +115°C) sensor = transistor
Word to the wise, DON'T USE TOO MUCH GREASE!
The main key here is surface area. The greater percentage of the die surface covered in goo that actually touches the heatsink area, the better the heatsink will draw heat of the processor. I've always had the best success applying with my finger as little as possible that fully covers the die, and smoothing it as much as possible, though I do like Benji's idea of using a plastic bag or glove or something to avoid contaminating the goo with hand oils and such that probably decrease it's efficiency.
Efficiency? Oils are insulators. What makes thermal greases thermally conductive is all of the metal particles they mix into it.
It's like a bed of nails, it is actually quite easy to lay on a bed of many nails if you are careful, the total surface area of all the nail points actually becomes fairly significant. I want to see some guru lay on a bed of 4 nails, now that is mind over matter!
Well, laying on even a normal bed of nails is "mind over matter" in the fact that you have to get your mind over the idea that it's gonna hurt. That is what the "mind over matter" statement means... not a claim to the physically impossible. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
David C. Rankin wrote:
Listmates,
I thought I would pass on the results of a (non-official) scientific test concerning the huge difference in cpu temperature just a small difference in the amount of thermal grease can make. The results below are on an AMD thoroughbred 2400+ using Manhattan silver grease.
Test #1 (dab of grease on processor core with syringe, pencil eraser size, roughly 3mm high -- result - way too much)
w83697hf-isa-0290 temp2: +62.5°C (high = +120°C, hyst = +115°C) sensor = transistor
Test #2 (minimal amount of grease brushed on with Q-tip, very thin layer, much less than 1mm thick)
w83697hf-isa-0290 temp2: +51.0°C (high = +120°C, hyst = +115°C) sensor = transistor
Word to the wise, DON'T USE TOO MUCH GREASE!
Most greases are insulators (both thermal and electric), so yes, of course, the more grease, the more insulation there is between the CPU and the heat-sink. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
David C. Rankin wrote:
Listmates,
I thought I would pass on the results of a (non-official) scientific test concerning the huge difference in cpu temperature just a small difference in the amount of thermal grease can make. The results below are on an AMD thoroughbred 2400+ using Manhattan silver grease.
Test #1 (dab of grease on processor core with syringe, pencil eraser size, roughly 3mm high -- result - way too much)
w83697hf-isa-0290 temp2: +62.5°C (high = +120°C, hyst = +115°C) sensor = transistor
Test #2 (minimal amount of grease brushed on with Q-tip, very thin layer, much less than 1mm thick)
w83697hf-isa-0290 temp2: +51.0°C (high = +120°C, hyst = +115°C) sensor = transistor
Word to the wise, DON'T USE TOO MUCH GREASE! Even the <1mm is too much!
And don't use Q-tips! They will leave a fibre behind and all your hardwork will be wasted. To do a proper job, clean the contact surfaces (cpu and heatsink) with acetone (get a small bottle from the pharmacy) - but NOT the stuff women use to remove nail polish (it contains oil)! Don't touch the surfaces once this is done, and protect the surfaces from contamination from dust etc. Place a SMALL tab - about the size of matchhead on the little small engraved area on the cpu. Use your CREDIT CARD (make sure the edges are clean) and use it to spread (as) EVENLY (as possible) the silver grease across the surface of the cpu contact point. Put the cpu into its socket, making sure to keep dust etc off the cpu and then put the heatsink in place. After the heatsink is in place and clipped in, gently move the heatsink clockwise and anticlockwise to help spread the grease to better match the mated surfaces. Job is now done but also remember that it takes some hours/days of running the computer for the silver grease to become really effective - read the instructions which come with the grease. Ciao. -- If you really want to know, you won't ask me. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
participants (6)
-
Basil Chupin
-
Benji Weber
-
David C. Rankin
-
Jason Craig
-
John Andersen
-
Sam Clemens