Proper test of thermal pastes

[Charles Wirth]

Jon, have time to do a spring loaded test?

Contact pressure is important and may skew results a bit as thickness of the TIM may vary. .

Side note:
IT outlet in Taiwan has more TIM variation than we could ever name.

[Hondacity]

Jon, have time to do a spring loaded test?

Contact pressure is important and may skew results a bit as thickness of the TIM may vary. .

Side note:
IT outlet in Taiwan has more TIM variation than we could ever name.

sheeet i didn't see its mounted....good eye boss

some tims expand when going too cold...

i think the spike or the sudden increase in temperature indicates the pot tilted a little and that contact isn't perfect anymore...

[miahallen]

seems to be plenty of freeze at the OCZ Gear store:
http://gear.ocztechnology.com/produc...und/index.html

[Rotaryloop]

what about ic diamond.

[KanGaXx]

diamond based like ICD7 ? how they works ?

[tastymannatees]

Jon, have time to do a spring loaded test?

Contact pressure is important and may skew results a bit as thickness of the TIM may vary. .

Side note:
IT outlet in Taiwan has more TIM variation than we could ever name.

Spring load is a great point as continuous pressure helps the compound resolve down to it's bond line thickness or average particle size.

With screws alone you hit a point where it feels secure - the screws are set at a position that does not move and now when you apply heat the paste continues to flow leaving a light contact or gap and should be re-torqued like the head gasket on old cars to get the proper contact.

Here is one where the end user used a torque wrench to get a calibrated pressure. Great way to maximize temp performance

http://www.evga.com/forums/tm.aspx?m=681429

Also you can do a room temperature test to indicate whether a compound has reached it's Bond Line Thickness (BLT) as a general guide By taking an IHS Sized piece of metal between 2 pieces of glass, apply the compound with 4 binder clips @ approx 50 -60 psi and use a micrometer at timed intervals 1 min, 5 min, 10 min, 30 min, 1 hr, 2 hr. to mic the thickness and time to spread or resolve down to it optimal limit.

I found that 2hours @ cold or ambient room had most compounds at near the BLT in 2 -3 hours with maybe a minor improvement overnight with constant pressure @ 60 psi. If your are subzeroing compounds a pre cure would be in order otherwise the compound would possibly harden at a thicker BLT if it is cooled before it has a chance to spread.

[RejZoR]

FrostyTech is using peltier element to simulate selected heat loads. Though they test coolers, not thermal pastes.

[K404]

http://www.maplin.co.uk/Module.aspx?ModuleNo=351527

Anyone got any opinion on this stuff?

[dinos22]

nice elmor, Dino and I have been banging on about this for a few months now, it is a very interesting subject,

I have been collecting data on 10 thermal pastes over the last month,

Ceramique delta I have found to be very good over repeated tests but it fails miserably at -120, but if you have a -100 chip the delta can be nothing..

I have tested Antec Formula7, Ceramique, Arctic Silver, Noctua, Thermalright, Coolaboratory, OCZ Freeze, some others I can't remember LOL

If you check the specifications of all these products, the lowest any of them claim to be rated is -50.. But we are all using lower than this right?

oh guys OCZ freeze is EOL now..

^ this

when this info comes out fellas it will put this subject to bed forever and I wouldnt go out of my way looking for any paste just yet if i were you

[elmor]

Jon, have time to do a spring loaded test?

Contact pressure is important and may skew results a bit as thickness of the TIM may vary. .

Side note:
IT outlet in Taiwan has more TIM variation than we could ever name.

sheeet i didn't see its mounted....good eye boss

some tims expand when going too cold...

i think the spike or the sudden increase in temperature indicates the pot tilted a little and that contact isn't perfect anymore...

Testing with higher mounting pressure would of course yield better results, however I've repeated these tests atleast twice with same results. It will still show how much of an insulator the different TIMs becomes at these temps. The spike is not because of some tilt, I've seen this many times when benching (at a certain temp the benchmark crashes and you can't run the same frequencies again). For next round I'll try to make some proper mounting.

Spring load is a great point as continuous pressure helps the compound resolve down to it's bond line thickness or average particle size.

With screws alone you hit a point where it feels secure - the screws are set at a position that does not move and now when you apply heat the paste continues to flow leaving a light contact or gap and should be re-torqued like the head gasket on old cars to get the proper contact.

Here is one where the end user used a torque wrench to get a calibrated pressure. Great way to maximize temp performance

http://www.evga.com/forums/tm.aspx?m=681429

Also you can do a room temperature test to indicate whether a compound has reached it's Bond Line Thickness (BLT) as a general guide By taking an IHS Sized piece of metal between 2 pieces of glass, apply the compound with 4 binder clips @ approx 50 -60 psi and use a micrometer at timed intervals 1 min, 5 min, 10 min, 30 min, 1 hr, 2 hr. to mic the thickness and time to spread or resolve down to it optimal limit.

I found that 2hours @ cold or ambient room had most compounds at near the BLT in 2 -3 hours with maybe a minor improvement overnight with constant pressure @ 60 psi. If your are subzeroing compounds a pre cure would be in order otherwise the compound would possibly harden at a thicker BLT if it is cooled before it has a chance to spread.

Very interesting points, as I said more perfect results can be achieved but from my real world experiences these numbers match very well.

[Tomasis]

how about tests for aircooling like subzero air?

nice test however

[Kunaak]

I love posts like this, I bet they have some accountant at OCZ scratching thier head, wondering why freeze sales have just jumped by 900%.

[ME4ME]

did some testing too..



as you can see, with Formula7 i was able to test at -140 (CB at -149), while with OCZ Freeze, the CPU coldbugs at -133..

in my oppinion clear evidance of bad heat transfer of the thermalpaste..

CPU-pot used was the SF3D pot, mounted firmly.
(CPU = 980X at 1.85v)

[Leeghoofd]

I smell thermal paste sharing to be banned on Hwbot soon !!!

[ME4ME]

You of all shouldn't be talking about hardware sharing Albrecht, after last night's livestream your suspect of serious crime

[M.Beier]

Good stuff Jon, keep it coming!
Pressure, hmm.... 2kg of a F1EE is some pressure, isnt it? I dont see why he should add further, gives a good idea bout the figures anyway...

You of all shouldn't be talking about hardware sharing Albrecht, after last night's livestream your suspect of serious crime

Is that about K404's sister?

[Leeghoofd]

I clearly explained our thoughts on the matter Henry. HW sharing is a no go ! We have all the cards don't worry (piccies have been taken)

Only things shared for 3D 01/03 was Gamers REX board, my 8600 CPU and CSX rams for the S775 scores... The other 3D scores with the I7 : my and Rogers REX III and GTX2 rams and Gamers now dead GT CPU.

GPU's were all all over the place ( in fact I still needed to do about 9 cards (6 Nvidias and 3 ATI cards) but ran out of time... think the other still needed to bench a lot of cards too... my 3 S775 OSses were so fubarred with the multitude of ATI driver installs, that in the end 3 XP installs refused to even recognise an ATI card... what a mess

And we like to mess with our viewers, a whopping 5 fans lol !!! talk about pressure on our shoulders...

And Marc if there's one thing that I don't share : it's my woman !!! Massman I kindly share :p

[elmor]

did some testing too..



as you can see, with Formula7 i was able to test at -140 (CB at -149), while with OCZ Freeze, the CPU coldbugs at -133..

in my oppinion clear evidance of bad heat transfer of the thermalpaste..

CPU-pot used was the SF3D pot, mounted firmly.
(CPU = 980X at 1.85v)

Thanks for the additional results. My Ceramique is not exactly fresh from the store so that might be what's giving me such bad results with it.

[ZenEffect]

has anybody tried apeizon n? the stuff is way expensive for thermal paste (like +$100)

http://www.apiezon.com/uploads/apiez...onductance.pdf

looks to be pretty good from what i can tell.

[Hondacity]

still trying it...

i just got newer stuff from rosewill too...

[ZenEffect]

still trying it...

i just got newer stuff from rosewill too...

still trying apiezon n? i was interested in its pliability at room temps since every description of it says "rubber" but its a "grease" so i cant really imagine how liquidy it really is. its thermal properties around lhe is pretty amazing. at 77k like nitrogen has it should be over the 100 w/m mark

[Hondacity]

i had a higher cb with it...i was expecting lower...

[ZenEffect]

i had a higher cb with it...i was expecting lower...

if thermal conductivity is higher at lower temps... wouldnt you expect higher cb then?

[NEOAethyr]

Would be nice to see more results of diff tim's.

I use as ceramic at the moment.
But I'm getting tired of it.
As5 gets me 2-3c lower if applied right, but it's not worth it.
As ceramic lasts 2-3x longer then as5, for me anyways, that's why I use it.
But it still leaks/seperates and makes a mess.
Heck my tube of it is starting to seperate now , started doing that after I took it over a cousin's house this month, apperently it got to cold for the last time.

So I'm interested in seeing if there's any possible replacement.
I'm sick of tim's that leak or seperate and make a mess, and I'm sick of replacing it all the time.
2 weeks (super thin layer of as5).
As5 normal for me is a month, or a month and a half at best before I am forced to replace it from overheating.
3 months+ for as ceramic, from mild temp increases.

As2 or was it 3?, whatever it was before 5.
Seemed to last a bit longer then as5 did, for me anyways.
But that was so long ago, maybe I just didn't care about 3-10c back then.

Shin etsu, the good types of it are so hard to find, plus it's never a reasonable price.
Unlike 22g of as ceramic, which is affordable :\.

[Hondacity]

if thermal conductivity is higher at lower temps... wouldnt you expect higher cb then?

their claims don't reflect from my testing...

we tested celerons that didn't have cold bugs ..cele was putting out something like 200watts.

at ln2 temps...the thinned out icd was doing better.. with the cb free celeron..

here's the result of icd

http://hwbot.org/community/submissio...60_7932.67_mhz

the ap-n can't do that..