Proper test of thermal pastes

[chew*]

Not a bad little test, but I'd like to see the AS Ceramique results after its classic curing time (200hrs). I'd bet it would fair much better in your tests then.

I have to agree with Ket here.
Elmor, have you become a noob? Dont you usually run your LN2 for 200 hours straight?

[dinos22]

Ket making another valuable contribution to the community I see

[Leeghoofd]

I have to agree with Ket here.
Elmor, have you become a noob? Dont you usually run your LN2 for 200 hours straight?

+100 Best remark of the year award goes to Marc Beier !!

Jon is slacking somehow

[elmor]

Hey Jon, I appreciate the effort, but the point of TIM is to facilitate the thermal transfer between IHS or nucleus and POT. Synthetic loads can barely be called relevant, not so say proper, same goes for synthetic test surfaces...

This thread is kinda funny and elmor i do respect your testing but OCZ doesn't make paste.........they just bottle an OEM's paste..........

So how much are you guys paying for a tiny ass tube?


My pastes are straight from OEM's

The tube on right just some FYI is small least for me and amount of cpu swaps i do daily....., only 1fl oz.........cost $6.99

Valid point but those are usually hard to get a hold of and how to know which oem for which brand?

Hey Jon, I appreciate the effort, but the point of TIM is to facilitate the thermal transfer between IHS or nucleus and POT. Synthetic loads can barely be called relevant, not so say proper, same goes for synthetic test surfaces...

The way Henry has tested is probably the best but it's very hard to simulate the exact same conditions every time and a lot more work. I figured this way would be interesting as you get some real numbers out of it and can see how big the temperature delta might actually be.

I have a couple of points.

How did you secure your dummyloads to the hotplate? Thermal paste is needed between the resistors and the hotplate and you need sufficient clamping force to insure a nice thermal bond.

Which resistor were you measuring and where were you measuring from? If you were measuring from a resistor that wasn't directly underneath the CPU pot it may not be receiving adequate cooling as your thin aluminum hot plate won't be the best thermal transfer medium.

Some of the thicker pastes require a thermal cycling in order to thin them. With Ceramique I get consistently tissue paper thin mounts by heating the pot to 70-80C, rotating and pushing the pot against the CPU and then pulling the pot down to my running temperatures. My Ceramique tube must be a year old and requires an excessive amount of force to squeeze out of the gigantic tube I use but the thermal cycling does wonders for creating a thin/optimal mount.

I'll have to give this OCZ Freeze a shot. Anyone know which product OCZ was rebranding?

Note- I'm not snubbing Elmor's results but just pointing out several ways I can see to achieve more accurate results. Testing like this is important as it gets us thinking.

Again, valid points with some buts.

How the heat resistors are mounted and where temperature is measured from doesn't make a difference as long as you're comparing to other pastes as conditions will be the same. It'll just alter the scale. For reference they are mounted with Zalmans ad-hesive TIM glue and temperature measured directly from the middle mounted resistor.

Thermal cycling might help some pastes for sure, but I really doubt it'd make a big difference at low temperatures. The main point of this test was to see how thermal conductivity changed at different temperatures (ie how suitible they are for low temps). I might try that next time I do some testing to see how much of a difference it really makes.

Not a bad little test, but I'd like to see the AS Ceramique results after its classic curing time (200hrs). I'd bet it would fair much better in your tests then.

If you have 200 hours to spare you're more than welcome to test yourself Also same point as before, doubt it makes a huge difference at low temps.

Btw, I'm all ears for feedback on how to better compare different thermal pastes so please continue pointing out what can be improved. They way I figured it the way I did it was quick, simple and good enough to compare results (if differences would've been ~5-10 degrees that might've called for some retesting but they are quite alot more than that ).

[Gautam]

I get the feeling that using no mounting pressure (other than the container's weight) might have "punished" a thick paste like Ceramique. I don't think its fair to say that the mounting doesn't matter as long as its the same for all the pastes. Without mounting pressure, some pastes might have unrealistically high thermal resistance in a test like this when compared to the real world.

[zalbard]

I have to agree with Ket here.
Elmor, have you become a noob? Dont you usually run your LN2 for 200 hours straight?

I get the feeling that using no mounting pressure (other than the container's weight) might have "punished" a thick paste like Ceramique. I don't think its fair to say that the mounting doesn't matter as long as its the same for all the pastes. Without mounting pressure, some pastes might have unrealistically high thermal resistance in a test like this when compared to the real world.

It's very possible. Can still use this dummy setup, though, just need to increase the pressure (I don't know the exact pressure of a proper F1EE mount, but perhaps squeezing the TIM during the pot mounting procedure [just push on top of the pot, you get what I mean] and then adding 1kg of weight to it would be enough).

[Buckeye]

It's very possible. Can still use this dummy setup, though, just need to increase the pressure (I don't know the exact pressure of a proper F1EE mount, but perhaps squeezing the TIM during the pot mounting procedure [just push on top of the pot, you get what I mean] and then adding 1kg of weight to it would be enough).

Best way I have found is to apply TIM, mount the pot. Turn on computer and bring it to the BIOS. Watch temps and when probe says ~60c then shut off the comp.

Give the pot a few twists and turns and tighten down the screws just a bit more.

Then go cold

It gets such a tight mounting that often I have to do the reverse to remove the pot when I am all done, even when the pot has warmed back to room temps.

[Ket]

If you have 200 hours to spare you're more than welcome to test yourself Also same point as before, doubt it makes a huge difference at low temps..

I might do that. Just to satisfy my own curiosity really. Maybe when I get my 1075T.

[K404]

I always do a thermal cycle as well, before going cold. Not just so the thermal paste can spread out and make a nice uniform covering, but to make sure I have something resembling contact at all

Lets be honest... we've all tried a session and realised that max MHz have bombed because pot contact wasn't right. Giving a thermal cycle lets us check... even if the contact is terrible, hopefully we have enough time to power off before something dies


I never twist the pot once things have warmed up though, feels like im disturbing something thats looking after itself quite nicely...