DaClan Review: Thermal Interface Shootout

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

sailer

Splendid
Argh... so then for me to ever really know what will do best on my system i'd have to find a review where they used my heatsink.. well if I was going to be picky at least.... it just sucks I ordered a tube of AS5 :(

I try to think of it this way. The AS5 dropped my temps about 5c over the stock factory junk. That's good. Now I know that the next time I need to buy some paste, there's something even better, so that's better then good.

We live, we learn, and if we're smart, we apply what we've learned so we make fewer mistakes the next time. Long ago, somebody called that wisdom.
 

ardisian

Distinguished
Jan 17, 2006
25
0
18,530
The ceramique is just that. You get it on and all taht, you're not gonna get the HSF back off. That's why it's meant for vid.cards or N/S bridges.
If you want that CPU,mobo, or HSF back off of there. Then use the AS5.
 
Well then, if you have both... do your own review. :wink:

All you need really is a rice sized grain of AS-5 and a smidgen of AS-C. Depedning on your cooler you use, the results could differ.
 

Mondoman

Splendid
Nice info! Is there a link to the methodology used? (e.g. number of runs, how measurements were taken, how compound was applied, environmental conditions, details of test systems, etc)
Any guesses on what might explain the performance differences? For example, could it be thickness of the final layer?
 
10 runs each.
Measurements were done by averaging the readings from TAT, a thermistor and CoreTemp.
The compounds were applied by spreading with a plastic knife, with the amounts being the recommended sizes (rice grains)
The environmental conditions were listed in the original post
Personally, I think it the make up of the actual TIM's makes up for the difference in performance. Silicon grease performs better than silver goops.
 

Mondoman

Splendid
Thanks! It would be cool if you would post the raw data, so we could calculate standard deviations and do other such calculations.

10 runs each.
Is that
10x applying compound, doing one run, removing compound, or
1x applying compound, 10x run, 1x removing compound.
The reason I'm asking is that I've long suspected that the application-to-application variability in layer thickness, etc, may make as big a difference as the compound choice. It's hard to get an idea of the relative contributions of the two factors with just a single application of compound.

The environmental conditions were listed in the original post
Sorry, I should have been more detailed. I noticed in the 24 hour runs that the temps in all cases seemed to be almost cyclical, with minimums near the beginning, peaks a bit after halfway, then dropping again. My guess would be that these are single runs (doing 10 of each would have you still nowhere near done!); could the cyclical temps be due to variations in ambient temp over the course of a day (cooler at night, warmer in the daytime)? If so, were the 10x "main" runs done at the same time of the day to minimize this effect?

Thanks!
(As you can see, I get into experimental design!)
 
The enviroment was a controlled lab. Therefore, I cold be absolutely sure that the temperature stayed the same thru out the experiment.

And by ten times I mean ten times running the tests. Reapplying, waiting to cure and having to deal with setting the heatsink (that damn stock cooler) would drive me insane... and would take for too long. 100 hrs curing time x 10 runs x 7 TIMs + (7 X 24Hrs) = 7168Hrs. That would take 298 days. o_O
 

Mondoman

Splendid
... 100 hrs curing time x 10 runs x 7 TIMs + (7 X 24Hrs) = 7168Hrs. That would take 298 days. o_O
Exactly! :lol:

Any ideas on what might account for the cyclical temp profiles on the 24 hour runs? The reason I was thinking ambient temp changes was that in my old lab, they turned the heating/ac down at night and back on at 6 AM. If that's not the case here, I wonder what else might lead to such a profile?
 

Lionhardt

Distinguished
Jul 3, 2006
581
0
18,980
how would you recommend applying thermal paste as a complete nub...
cause, see, i have this friend.... :roll:

this is using AS5 or SHIT-zu, as those would be my 2 choices...

but im 14, dont wanna mess up, money is tight, yah u got it...