I made thermal paste roundup using liquid nitrogen.

Results are pretty interesting

 

 

 

Link to my test

 

It is in Polish, but you can put it to Google Translate. I checked translation to English. It is rather imperfect, but you will know what is it all about

 

Edited September 29, 201411 yr by ivanov

Always interesting, will read through. Thanks!

Hm, dat strange. Completely different to what I saw in der8auer's roundup - gelid GC-E was performing better in his tests.

Yeah, I know. I was also surprised.

Before the test began I was 100% sure, that GC-Extreme would win with ease.

 

Keep in mind that some of the samples were directly from vendors and some were just from regular shops. I know it is serious accusation, but maybe sometimes what we get in shop isn't exactly the same, what we get directly from vendor (there were several PSU issues like this, that I read about) But it is just my theory, I have no proof to back it up.

Edited September 29, 201411 yr by ivanov

Wow yes, did not see this coming at all. Noctua on top?? Gelid extreme close to bottom in every temp... I think I've got a tube of the noctua paste at home, will give it a try on compressor.

And one more thing. I know it sounds not modest, but I think Roman didn't conduct his test 100% properly (sorry Roman).

 

As it is said in the arcticle, his first mistake is, that he didn't go to -190 degrees C. Stopping at -150 makes the test not good, since at -150 some pastes just start to loose their thermal conductivity.

In my opinion second mistake is this thermal probe attached to the top of CPU IHS.

I also speaked with Xtreme Addict bout EVGA Frostbite. This paste shoud perform better in my opinion.

He said that it needs very high pressure to work well. I didn't test it, since I wanted to give each paste equal chance. So every single TIM was given exactly the same pressure (you can see in the pictures, that pot springs were compressed, but not extremely tight).

Ivanov, I've read the complete test but some of the translation is... funny (Ryba means fish lol? that made for some confusing sentences at start)

 

Can you explain here how did you measure cpu temp?

Yes, ryba means fish in Polish

 

There is no CPU in my test. I used dummy load (set of resistors), that is usually used to check perfomace of phase change cooling.

 

 

 

By the way, it seems, that AAB TG3 is rebrand of Phobya HeGrease Extreme (all parametrs are the same), although it is not official info.

Edited September 29, 201411 yr by ivanov

It can also be the case that a manufacturer varies components of the paste down the production line to minimise the costs while remaining in a given spec.

I also thought of that.

 

For example it could be like this, that what you can buy now in shop, as Gelid GC-Extreme is not exactly the same, as it used to be at the GC-Extreme introduction to the market.

 

But keep in mind - it is only my assumption.

Great roundup !!

The only thing I can see here that also could change results is in my testing I had a number if paste that could keep a very tight delta, the differances where what paste would hold together under massive load spikes. Gelid for me was the strongest at holding the load and not cracking losing all thermal transfer. Noctua paste was one I had tested that did not hold up well at all when tested with 3d06 CPU test at 6GHz.

Thanks

 

Well, that is a good point. In near future I intend to check some of the pastes using i7-3770K with full pot, just to see which one offers best CPU-Z frequency and which performs well with e.g 3DM 06 CPU test.

 

 

I also plan to check more TIMs. As you see in the picture below I have much more, then are included in the roundup right now

 

Edited September 29, 201411 yr by ivanov

I know extremeTIM, he's actually a good friend of mine.

 

 

 

PS keep a heads up for MSI's new line of BBQ, and BBQ accessories.

i only use noctua nt-h1 for CB chips(specially 4770k @ -100 to -130) but good to know it will be good on full pot too

nice work man

I know extremeTIM, he's actually a good friend of mine.

 

 

 

PS keep a heads up for MSI's new line of BBQ, and BBQ accessories.

 

I like this - I think I will most likely buy

 

 

 

@topic Valuable information, thanks for the test

Now I'm glad that my bro bought a Noctua HS.

Nice

 

How much was your dummy load ? 200w ?

It could be nice to see with 400-500w

 

Thank you for your works, it's not funny to test TIM, but it's reallllyyy interesting

 

something I notice during my test : for CPU (Fx8155), with EVGA or with Gelid, it can reach the same freq ( 8.4G) BUT, on VGA, between EVGA and Gelid.... it's 100% different

Maybe T° ( full pot for CPU and -130 for VGA),

or

surface ( big between cpu/pot and smaller between vga/pot. smaller surface -> TIM is more important)

or

Watt

 

If you have time, it will be awesome to test gelid vs "hot gelid". 1st put gelid, test, then put gelid, heat @80°C or more and put pot and test.

 

 

Anyway thanks for sharing

It was exactly 261 W load.

 

I must admit it was quite fun for me to make this test The only downside was, that I had to "waste" LN2 which could be used for OCing

 

I heated the pot to 50 oC (cured the paste) before pouring LN2.

noctua for me it's very special.....but for me cracks at -140....did you try compound cracks temp??? thanks

 

EDIT: i looked now the video, the two probe have 4° of difference in reading! the probe on lower of display is more warm about 4°!

 

and for tighten the mounting screws with the same force, you must to measure the height of the springs compressed(if all springs it's the same)!

Edited November 18, 201411 yr by barbonenet

EDIT: i looked now the video, the two probe have 4° of difference in reading! the probe on lower of display is more warm about 4°!

Think about it.

There is thermal probe in TIM, layer of copper (heater), layer of TIM, layer of copper (LN2 container), thermal probe in TIM.

 

When you turn the heater on, it would be quite strange if both thermal probes shown the same temperatures. It is rather obvious, that probes will show different temperatures.

 

 

 

Springs are equally compressed, threads are of different length.

 

I turned off and on the heater under -190 oC to simulate stress.

Edited December 1, 201411 yr by ivanov

Think about it.

There is thermal probe in TIM, layer of copper (heater), layer of TIM, layer of copper (LN2 container), thermal probe in TIM.

 

When you turn the heater on, it would quite strange if both thermal probes shown the same temperatures. It is rather obvious, that probes will show different temperatures.

 

 

 

Springs are equally compressed, threads are of different length.

 

I turned off and on the heater under -190 oC to simulate stress.

thanks for reply...very good test;)

 

can i ask you what's the compression of springs? 1-2mm or 3-4mm?? thanks;)