PDA

View Full Version : Thermal compounds we're all using to replace the stock ones?



MaxCioco
01-28-2016, 08:44 PM
Let's talk about what thermal compounds/pastes/gels/etc. we're all using to replace the stoc ones in our laptops.

I myself used 4 types, with more or less good results.

I'm going to begin with the clasic Arctic Silver 5: behaves well and it's capacitive properties did not seem to cause problems as long as it was only on chips, not VRMs. At the time I used this paste I used a no-name brand for the VRMs. They used to run like ~5 degrees hoter. Repasted with it 2 times and never found it dry or underperforming. But being affraid that in time as it dies might show it's conductivity I gave up on it.

Idle cpu: ~40; Load cpu: ~70
Idle gpu: ~45;Load gpu: ~80

Next on my list was Arctic Silver Ceramique: one thing it's sure, that it's not made for high temps. Maybe I used it wrong or anything, but I found out that after 1 months of high temps usage it started to give up it's performance and when removed it was dry. Used it once and never again will I use it on my g750 laptop cpu&gpu, though on my lower power/super portable laptop it rocks. One year and counting and it still goes. I'm going to oped it up just to see how it looks.
I mention I applied it to the VRMs and it held between repastes (usualy anywhere from 2 to 4 months).

Idle cpu: ~44; Load cpu: ~80
Idle gpu: ~45;Load gpu: never got it to run on real load so no info here

The third is Noctua NT-H1: mostly same behaviour as AS5, but being not affraid of conductivity I used it on everything. I find it's price way to high for what it offers; maybe that's the case here in Romania, where 1,4ml is now 12euro.

Idle cpu: ~36; Load cpu: ~70
Idle gpu: ~40;Load gpu: ~77

The last on my list is an industial grade mercury based compound. I don't know the brand as I scraped it off a broken pulse generator for silicon analysis. For 1 week it's on (cpu and gpu only, the other parts are with noctua nt h1) and I can say it behaves just like the noctua. Pain in the *backside* to apply, as I had to use multiple protection measures: gloves, spatula (it's thick), grease on the sides so if it spils won't contaminate the whole mb and had to tint the heatsinks for 10 mins each, not to mention I used by mistake a clean cloth and it all came of in just 2 rubs.

I don't find my excel with all the temps and ambiental temps and workloads. Because I found that 100% on cpu gives out differite results depending on what the task is.
The temps I said are at somewhere 20 degrees celsius.
When I run it at +30celsius both the cpu and gpu throttle at their maximum temps. I usualy avoid that by canceling turbo boost on cpu and lowering the frecv. on the gpu.

So that's wraps it up for me. How about you guys? What do you use (used), and how did it go?
Do you use something special for the vrms and all the other stuff aside from the chips?

Note: did any of you modded the cooling system? If yes, how, if not, have any ideeas? Mine is to mod the optical drive into a intake fan with dust filter, using a low noise, low profile, low rpm fan so the cpu and gpu fans work less. Maybe more airflow will help. Still in the drawing board phase as I don't have time now.

Korth
01-28-2016, 09:26 PM
I favour Arctic Silver 5 and Prolimatech PK-2 - mostly because I have huge jars of them both, it's taking forever to use 'em up, lol.
And Gelid GC Extreme is good. I also hear that Phobya HeGrease is very good. Gelid is probably the most popular product because it apparently performs very well across a wide temperature range, especially important for exotic subzero and LN2 overclockers. AS5 is probably the next most popular, simply because it's cheap and commonly available.

Basically any premium TIM is better than any cheap generic TIM - and the worst premium TIM will easily outperform that nasty pink goop Asus uses on everything they build.

I'm of the opinion that there is no "perfect" TIM for every application. Some TIMs are better at covering some sorts of mating surfaces and part geometries. Viscosity and consistency of the TIM matters more to me than (biased) benchmark scores/reviews, I even mix TIMs together to achieve desired results. Any paste TIM you use will require a little cure time for best results and will cook off after a year or two at most.

Properly applied metal-gap TIMs like Indigo Xtreme and Coollaboratory Ultra Pro will always outperform pastes. Thermal conductivity of metals is one or two orders of magnitude superior to any metal-bearing liquid at room temps. But they may require multiple applications to set right, they are a real pain to remove, the branded products are ridiculously expensive per application, and they aren't suitable for all part applications.

You'll find lots of detailed discussions about TIMs if you search these forums.

MaxCioco
01-28-2016, 11:18 PM
In terms of viscosity, what type of compound do you advise me to use for the vrms and other parts beside the cpu?
For the cpu + gpu I find that a medium (maybe a bit towards watery, but not too much) consistency works best. Doesn't last that much, but better performance. As for the other things, imo something towards high viscosity. But, what if it affects the heat transfer? Especially because the parts are not designed to vent out heat asap.

Do you recon it would be a problem if I used some AS5 on the vrms? Because it's a little bit capacitive and conductive.

Korth
01-28-2016, 11:38 PM
Depends on the surface areas and geometries of your VRM parts. If there's a big gap then a thermal pad or goopy thermal adhesive might be better than a thermal paste. If there's a very smooth and tight gap or an irregular interlocking mating surface then a thin and runny TIM (like Gelid) will flow into the gaps and provide better overall coverage. Good old AS5 is pretty good stuff, too.

You might see improved temps right away, but don't expect best thermal performance for a week or so - any liquid-paste-grease TIM you use will need some curing or dry-out time before it really starts working, this can range from hours to days.

I clean my part surfaces thoroughly, of course (as in, I obsess about them being absolutely clean, mechanically and chemically, no oils, no debris, no residue) and I scrape the TIM on with a (clean) razor blade - but so-called "pea" and "X" and "rice" and "dot" methods are all acceptable on larger parts like CPUs and GPUs. The key is to use the *minimum* amount of TIM you can to achieve full thermal contact, it transfers heat better than air/bubbles/voids but not as well as the metal heatsinks themselves - most people think more is better so they apply too much and see only marginal improvements in temps.

Noctua's TIM is generally not considered all that great. It's probably just some branded cheap stuff they make so that people have everything they need in the box when installing their new Noctua coolers.

You'll find that TIMs are a controversial topic, many people will champion their chosen TIM product zealously, many will promote or condemn TIM products based on reputation alone. All I can say is that I've used plenty of TIMs in plenty of devices, and the stuff sold for PCs is really no different than the stuff sold for other electrical devices.

Darnassus
01-28-2016, 11:48 PM
Don't even bother discussing thermal compounds when it comes to the G750 line, the way the sink is attached to the board doesn't put any proper pressure against the die. It's a very bad design flaw I find. You'll always see 80*C and dry paste within 4-6 months.

MaxCioco
01-29-2016, 07:55 AM
That much I figured out myself, that's why I'm going for mercury based compound. Maybe it'll spare me from all the work of opening a strangely designed laptop.

As for the Noctua, I can't say it's worse or better than any, just a little better behaved at pasting if preheated at 45C; nothing more, nothing less.

I thought that if I added a bit of air pressure to the flow with a third fan I would further cool that big plate of the cpu ansamble. I am pretty impressed with the performance and silence of the fans Clevo uses on their current laptops (I have to say that I never saw an older one), not to mention I thinks I might get it in with the least ammount of trouble. My only problem it's how I'm going to controll it. Power from the sata connector, but rpm? First I though that I could get it from the cpu, but then they would run in the same time; not exactly what I desire.

Does any one know how to controll sata pins? I know how to code a bit of code, but for that port I did not fine no info. All that I can think of is sacrificeing a usb port and controll it from there, but... not how I want it. Maybe from the card reader, but again I don't know how to access it.
And no, an external potentiometer is out of the question. I want it run on itself.

Could use an arduino or a raspberry pi 0, but I don't think I have where to put all that in the space of that slim unit.

Edit: so you guys recon AS5 safe for usage on everything?

Nate152
01-29-2016, 12:00 PM
This is a good thermal compound with a thermal conductivity of 12.5 W/mk.

http://www.performance-pcs.com/misc-tim/thermal-grizzly-kryonaut-hgh-performance-thermal-grease-1g.html

Darnassus
02-14-2016, 05:18 AM
I'm going to repaste again today, reaching 84*C again.. I think it's been.. roughly.. 6 months? The weather currently is err.. 38*C?

Darnassus
02-20-2016, 10:04 AM
Back, did a repaste after 6 months I believe. Went back down from 77*C at boot, to 53*C.

I guess if we want to stay away from the horrid temps, we need to repaste regardless of what we use.. unless we use the liquid ultra, but I'm not ballsy for that.

MaxCioco
02-20-2016, 09:20 PM
38C degrees inside? That is alot! Do you live in the african desert or australia?

As an update to my mercury based compound: the temps have dropped by 1-2 degrees on the gpu, but only under high load. Opent it up and the compound on the cpu looks just as when I applied it.

Here is something funny I found: the heat sink on the cpu is not made of copper, it's only copper plated. I tried to scratch a corner to see if the compound is penetrating through the hearsink and when I scratched more I found aluminum. The copper is just a thin layer. This is dissapointing, as the cooling sistem was marketed as "copper". From my technical knowledge there isn't any significant difference in their thermal conductivity, and price related the difference is not that big either. I don't get it why they did it like that.

Ntwlf
02-21-2016, 01:11 AM
Hey MaxCioco,

In regards to copper and aluminum used for heat sinks. Both metals perform very well at conducting heat. But there's more to it, because there is a difference in their thermal properties/characteristics. And when combined together give the best performance with these 2 metals. This explains why many performance heat sinks combine both metals.

I'll put it in layman's terms to explain since I'm not a scientist, so I can't go in to further detail. I have learned this from reading and hands-on experience. Copper performs better at absorbing heat then aluminum, but aluminum releases heat better than copper. Both metals are close at spreading the heat through itself evenly and quickly.

Here's a little experiment that anyone can try. You'll need (3) metal wires/rods of the same dimensions, one each of copper, aluminum and finally steel for reference. An example would be, for steel - a coat hanger, for copper and aluminum - use sections of electrical wiring commonly found in homes and buildings. This wire is usually the same diameter as coat hanger wire. Now have them all cut to the same length (6 inches). You will need a heat source such as a candle as well. Taking one wire at a time, hold one end in your fingers and place the other end into the candle flame and make note of how long it takes for you to feel the heat move thru the length of wire to your fingers. Now remove the wire from the candle flame and make note of how long it takes to cool off.

Results for those that can't source the materials or take the time to try it: You will find in comparisons between copper and aluminum, that the copper wire warms the fastest and cools the slowest. Aluminum cools the fastest but warms slower. Now for the reference "steel coat hanger wire"...I suspect you'll still be holding the end in the candle flame until your arm falls asleep, since the end your fingers are holding never gets warm. Thus why steel is not used in heat sink's.

Conclusion: you get the best performance of copper and aluminum when combined. Copper removing heat from the component it is attached to with aluminum quickly releasing it into air.

I hope you all find this interesting and educational.

An additional note to make: certain metal based TIM's are not recommended for aluminum heat sink's as it will eat into aluminum so to speak. But with an adequate/thick enough copper plating this probably will not be a concern.

Darnassus
02-22-2016, 10:13 AM
The latter, Max. ;d

Yep, 38*C to 45*C during some lucky days..

The copper 'plating' is because I believe copper takes heat in better, as it's more conductive as a surface.. and once the copper square plate takes all the heat, the surrounding aluminium (which is better at actually moving the heat across itself, rather than respond to the heat in the first place) then gets hot and takes it through the pipe to the fins to be cooled.

If the CPU die was larger, or had an IHS on it (with a damn good paste underneath that IHS) then perhaps aluminium would be fine to use.

The copper coating on the rods I suppose is also to retain the heat inside as the aluminum carries it to the fins.

THIS IS ALL MY THEORY, I MAY BE WRONG.. ;d

Edit: Go away Ntwlf it's my spotlight! x'{

Also yes I forgot, as Ntwlf states, some pastes do eat into aluminium.. a prime mention of such paste would be the famous Coolabratory's Liquid Ultra.

Ntwlf
02-22-2016, 07:26 PM
Hi Darnassus, I wasn't trying to take the spotlight;). Just attempting to clear misconceptions.

Here's an explanation/read on heatpipe construction and principle...https://en.wikipedia.org/wiki/Heat_pipe.

Also, here's a good read on Thermal Paste comparison...http://www.tomshardware.com/reviews/thermal-paste-performance-benchmark,3616.html.

as Snagglepuss would say: SO Exit . . . Stage Left:rolleyes:.
https://www.youtube.com/watch?v=Q3-a4qWCtIg

Darnassus
02-24-2016, 07:11 AM
Hi Darnassus, I wasn't trying to take the spotlight;). Just attempting to clear misconceptions.

Here's an explanation/read on heatpipe construction and principle...https://en.wikipedia.org/wiki/Heat_pipe.

Also, here's a good read on Thermal Paste comparison...http://www.tomshardware.com/reviews/thermal-paste-performance-benchmark,3616.html.

as Snagglepuss would say: SO Exit . . . Stage Left:rolleyes:.
https://www.youtube.com/watch?v=Q3-a4qWCtIg

Heavens to murgatroyd! A snagglepuss fan, even!

Ntwlf
02-24-2016, 08:21 AM
Heavens to murgatroyd! A snagglepuss fan, even!

Oh my goodness! . . . someone else knows Snagglepuss, even.

Heckle and Jeckle, the Pink Panther and grape ape are also old favorites, . . . grape ape,grape ape!