Three TIMs Compared in Clevo P-170EM

I recently purchased a tube each of Gelid GC Extreme, ProlimaTech PK-1, and IC Diamond 7 to compare their effectiveness on the AMD 7970m in a Mythlogic built Clevo P-170EM. I also compared how adding foil tape by the end of the fan helps with the cooling. All credit for the foil mod goes to bn880 and his thread.

Furthermore, I will describe the fan profile that the Clevos use and show patterns in the temperature graphs provided in Furmark.

MATERIALS USED

One Tube Each of the 3 TIMs


Rubbing Alcohol and Soft Absorbent Material


Non Magnetic Screwdriver and Tweezers (to help grab the tiny screws)


Foil Tape and Scissors




REMOVAL and CLEANING

Unscrew the 4 screws and the heat sink will come off of the GPU. Use the soft gauze and rubbing alcohol to gently rub away the old TIM from the chip surface and copper heat sink. The chip should be shiny like a mirror. You should see your creepy eye staring back at you if you look straight down at it.



















APPLICATION and RE-ATTACHING HEAT SINK

I used a relatively small drop of TIM. With all three TIMs, the temperatures were unstable if you used too much or too little. A drop about the size of a rice grain, maybe a little bigger, seemed to work the best. After applying the TIM, replace the heat sink and tighten down the screws in an "X" pattern. I'd only tighten each screw a little at a time and made several passes through the "X" pattern before they were all tightened. After running a few burn in tests in Furmark, I'd go back in and apply the foil tape mod. If you look closely at where the fan meets the copper, there are some gaps where heated air can be recirculated back inside of the computer. The foil functions as baffling and maximizes the efficiency of the cooling system.

















TEMPERATURE COMPARISONS

For each test, I ran the Furmark burn in test in 720p for 16 minutes each. I made notes of when the fans increased in speed, when they decreased, max temperatures, and when certain temperature milestones were reached. I cut and pasted all the Furmark graphs and labeled them. The labels are estimates of where the temperature milestones were at, but the actual time and temperatures are accurate. I used HWMonitor as a backup to make sure the temperature readings in Furmark were correct. I will post the raw numbers at the end of the post. Room temperature was a constant 24C.

NO Foil Applied:



All three TIMs performed similarly as far as the max temp achieved. However, it took ProlimaTech PK-1 less time to reach max temps (89C) before the fan geared up to cool it down to 86C. Based of this, it would appear that ICD7 and Gelid GC Extreme performed slightly better here. I wish I could modify the fan profile to keep it at maximum rpm and see how well the cooling system performs. It's obvious that these Clevo systems are simply allowing the temps to get up to 89C before increasing the fan speed to bring it back down to 86C. You can see this happening over and over again in the ProlimaTech graph. I purposely applied the paste incorrectly (i.e. too much, or spreading with some plastic) and all three TIMs exhibited this phenomena, but much faster. I'd really like to see what the temp would max out at when the cooling system is on maximum output rather than porpoising up and down between 89C-86C.


Foil Applied:



The application of foil has obvious positive results. None of TIMs reached 89C so max fans never kicked in. Furthermore, it took much longer to reach certain temperature milestones. For example, it took 3 minutes for Prolima to reach 85C without foil but 6 minutes when foil was applied. ICD7 reached 89C at 14:32 without foil but never went above 84C with foil. The foil definitely increases the efficiency of the fans. On average the temperatures dropped 3C and temperature milestones took longer to reach. For whatever reason, ICD7 really liked the foil and had the most improvement. In the end, however, there really wasn't that huge of a difference. The temperature spread was 1C without foil and 3C with foil. It would seem that HOW I applied these TIMs is much more important than which one was used.


FAN PROFILE

~55C - 1st ramp up
~60C - 2nd ramp up
~65C - 3rd ramp up
~70C - 4th ramp up
~75C - 5th ramp up
~89C - 6th ramp up (reduces to 5th ramp after cooling to 86C)

The one thing that really stuck out to me is how there are roughly 5C temperature increases between fan speed increases and then a 14C increase in temperature before the fan increases speed after 75C. I don't know why it was designed this way, but it seems like the Clevo systems are simply allowing the temps to increase this high. It's obvious the cooling system can keep the temperature much lower, but as soon as 86C is obtained, the fans let off. I'd really like to see the max temperature reached with the fans running at a constant maximum rpm. I'd be able to much better compare the TIMs this way. As it stands, I had to rely on how long it took to reach temperature milestones for some of the comparisons because of the funky fan profile. I don't even know if the 6th ramp up is the maximum speed or not. The Fn + 1 function is disabled in my system so I wasn't able to test that.


This was just a short term study I did over the past few days. I'm starting to do a longer term comparison to allow the TIMs to cure. Well, I have ambitions to do this study and hopefully I'll stay motivated enough.


RAW NUMBERS

GELID GC Extreme (No Foil)
70C :50 (note: this test was done first so the GPU was the coolest initially)
80C 3:03
85C 5:37
88C 12:00 (max temperature)

GELID GC Extreme (Foil)
70C :32
80C 2:24
85C 7:30
86C 13:30 (max temperature)

ProlimaTech PK-1 (no foil)
70C :26
80C 1:18
85C 3:00
89C 6:20
89C 7:00 (max fan? kicked in)
86C 7:20 (fan slows)
89C 9:09 (fan cycle continues)
86C 9:26
89C 10:44
86C 11:10
89C 12:04
86C 12:28
89C 13:29
86C 13:52
89C 14:53
86C 15:23

ProlimaTech PK-1 (Foil)
70C :28
80C 1:57
85C 6:00
86C 8:22
87C 11:40 (max temperature)

ICD7 (No Foil)
70C :31
80C 2:01
85C 4:50
86C 6:00
87C 8:00
88C 10:20
89C 14:10
89C 14:15 (max fan? kicks in)
86C 14:32

ICD7 (Foil)
70C :32
80C 2:31
84C 6:30 (max temperature)

 

nice review! + rep ... i just ordered some GC extreme to replace my ICD7 but you are giving me second thoughts lol. did you bend the heatsink arms back at all?

 

Another awesome review. Thanks for sharing!

 

GC Extreme performed similarly to ICD7 so it may not be worth it. I already had ICD7 and my temps were fine, but I wanted to do a comparison out of curiosity. If you are getting high temperatures then your ICD7 was probably applied wrong and a proper repaste with any of the popular TIMs should do the trick. IMO, you probably won't see drastic changes between any of these TIMs as long as it's applied correctly.

 

well i was going to take my heatsink off so that i could do the foil mod all the way around to completely enclose the cooling system. I needed some paste for when i put it back on and GC has some great reviews so i jumped on it.

 

ICD7 is the best for misapplication or not the best contact since it's so thick and just gets into all the crevice etc. Laptop heatsinks are not nearly as well made as desktop heatsinks or lapped that well. Few people on NBR who do lap their heatsinks reported that made more difference than any TIM. But I find ICD7 solves this issue. I went with PK1 this round just to try it and didn't want my 7970M or heatsink marked/stained/scratched. I'm not going to reapply with ICD7 considering it will only lower at most by 1C.

 

ICD7 was definitely thicker than the other two. I was worried about scratching, but using rubbing alcohol as a solvent and soft gauze helped. I didn't see any scratching at all and I re-applied/removed each of the TIMs multiple times. I was very careful, though, and if I were to skip using the alcohol I probably would have scratched it.

I did find that the ICD7 was the most forgiving when I was misapplying on purpose to see how the temps would react.

After running all these tests, I agree that it's not worth it to switch the TIM for a mere degree or two. If it was misapplied then it's worth reapplying any of the above mentioned TIMs and a few others.

 

Great write up! I'm thinking about doing the foil method to my Envy. There are HUGE gaps between the heatsink assembly and fan.

 

Just noticed, you and I have the same build dates for our laptops, though you have the big brother.

 

Awesome review, doubles as a guide for me so +rep for you sir!

You've been doing good work for the 170EM owners, keep it up mate.

 

Rep to you my friend !

 

good stuff. repped.

 

So thorough...

Question, so any of these compound recommend a cure time? Might want to take that into consideration, though it would make the test weeks each maybe.

 

Yeah, I considered that so for now I'll keep one in for a few weeks before changing again. I'll run furmark every few days to see if there are any trends.

 

Rep'd! Thanks for doing this and sharing.

 

Thanks for the tests!
Does the foil stays in place even if it's not adhesive, or you used some scotch tape to keep it together?
I want to try this on my MSI as well, just wanted to know before attempting it.
Thanks again.
Also +rep obviously

 

Why did you decide to cover the entire heatsink instead of just the problem areas like the original post? Do you believe covering the entire thing with tape has lowered the temperatures even further?

 

Its foil tape which is adhesive to 250°

 

Nice experiment misterHobbs! rep+

You should also go out and get a cheap $5 tube of ancient AS5 and compare. I have a feeling that it will also do pretty good

EDIT: Misterhobbs, another thing that you might want to try is to remove that black tape or whatever it is around the GPU. We dont have this in alienware...and I wonder if this contributes to higher temps in the Clevo

 

You should chose arctic silver for this test.
Its very popular thermal paste.

I repaste my laptop to because 90C in furmark (I run at 1080p)
I use OCZ Silver (99% silver) and no difference to stock
Tomorrow will try with Arctic silver.

 

curing is a myth.

 

lol i had put some artic silver, which is not very good btw compared to icd 7, and after three weeks of playing games about 3 hours a day and having it on all day, it was still just as liquidy as when i had applied it. no curing. although the ic diamond that i had removed was sort of caked and dry.

 

I got sick of cutting little slices and it was easier to cut larger ones. I didn't notice a difference either way as long as that gap was covered.

I was curious about that black rubbery material as well and wasn't sure of its purpose. This weekend if I have enough time to try another thermal compound I'll check it out. My fiance is starting to get jealous of my computer and feeling left out.

 

Nice work Hobbs. Thanks and +1!

 

i think that black rubbery material is to keep excess paste from going on the chip...however, as long as you don't put too much on and use a decent non-conductive paste, you have nothing to worry about. I bet it is contributing to a few degrees extra

Misterhobbs, as soon as you get married, your wife will love for you to get some hobbies so she can have some free time to herself...it is the irony of life lol

 

I took a closer look at that rubber pad and it's attached onto the card pretty good. I didn't want to force it too much so I left it in place for now.

I think you're right about it being there to prevent excess paste from getting all over the card.

 

Nice work indeed! But going by his name I think he works +.1 at a time.

(IIRC your previous avatar was an Orion right?)

 

So you're saying that no paste will see any kind of thermal improvement over time?

Proof?

 

Thanks for doing this Will rep when I can again

 

Not all foil tape is the same. Some are only rated to 80C, others can be 100C, the majority are under 150C, the one I am using is rated to 315C which is the highest I have found.

 

Yep, it was a P-3.

I'm doing a longer term comparison right now to see if I notice any differences. It will take a few weeks but I'll post again when complete.

 

Good stuff, thanks. Bit surprised with the highish PK-1 results without foil. Hmm. But recovered greatly when foil was added.

I'm running Gelid right now and even with foil it's cycling between 89 and 92 degrees after 10 mins. Interesting, will have to try some more pressure mods.

 

By the way, I had AS5 prior to Gelid on my setup, it meant +1 fan speed. So it's much worse than GELID for this GPU/heatsink combo. Don't waste your time with AS5 if you have IC7 or Gelid installed.

 

I'll reapply the pk-1 a few more times when I have time to see if I get any different results. I'd like to wait about a week or two between pasting to see if the temperatures stay the same.

Even though it was a bit higher, it wasn't significantly higher than the other two to cause any alarm. I really wish I could create a custom fan profile so I could max out the fan from the get go and see how high each TIM gets.

 

Great review..

Thnx
++ rep

 

Would it be possible to associate actual RPM values for the fans?

The latest beta of HWinfo64 may be able to read RPM fan values for some models:
HWiNFO32 and HWiNFO64 v4.01-1690 Beta released

If it doesn't work, perhaps correspond with the author of HWinfo64 to see if your model can be added?
Monitor CPU and GPU fan RPM on Clevo laptops

Well presented experiment!

 

I just purchased Tuniq TX-4. My main prior installation was both GPU and CPU with ICD. I am heavilly modded but the main thing is my GPUeatsink is well over saturated by the TDP of the q9200. But I thought I would give it a try.

Now I am on short term so far but a promising result. Without throttle stop and stock vcore and 2.4GHz. Before with all my mods it would run prime95 at the most 4-5 minutes without throttling the CPU to keep it cool. With the TX-4 it ran 7 minutes under prime95. Now before it would throttle and HW monitor had the highest core at 85C now the highest core now with the TX-4 was at 87C.

So far I am impressed. It was harder to apply as it has to be spread like a normal TIM. But the results seem promising at least for the short term.

 

Glad that while you did not do what I call a proper test (as outlined in TIM conversations before over the years in the forum) that you acknowledged the same problem I have pointed out to everybody testing TIM on a laptop.

The cooling system is variable, and if one TIM performs worse than another that it would not show as the system would simply ramp up the fans more to make up for it.

Something like my MSI laptop has a 100% fan override, but I feel to do a true test you need all static conditions, that means a thermal probe that produces a set amount of heat, a non software based heat reading unit as software time and time again has been proven to not always be real life results, and of course a static fan/heatsink setup.

 

Obviously, I'm not in much of a position to do a proper test as I only have one test system with one GPU and not enough time to do a multitude of tests. Also, since I can't manually control fan speed, a very important variable is taken out of my control, and to do a proper test you need to be able to control as many variables as possible. If I were to be able to set the same max fan speed at the same temperature I'd get more definitive results. All I could really do is create a general comparison of rates, estimated fan speeds, and indicated temperatures. I worked with what I had.

It would seem from my observations that there really weren't drastic differences between the TIMs when they were applied correctly. If and when I gain the ability to do a better test I'll do it, but being that my job often takes precedence, I may never be able to, unfortunately.

 

Based on fan rpm monitoring that I've done in the past, it seems the actual fan speed can settle in a range that I'm pretty sure could introduce an uncertainlty of ±2°C (prehaps worse depending on fan opearting level). The EC doesn't necessarily set a constant voltage on the fan. Instead it may apply a large voltage to get the fan going. Given the feedback it gets from the fan tachometer, the EC will then stop adjusting voltage once it reads an RPM value within some range. It will then keep the RPM steady unless it reads a value that seem off the rails.

For example, "maximum fan rpm" may (randomly) be anywhere between 2800±60rpm, when the temperature falls within 85°C...90°C.

So there are discrete steps for fan speeds that can be somewhat distinguished by ear, but I think there is also a non-negligible random variation of fan speed for each level that cannot be distinguished by ear.

In some scenarios, you will see the EC throttle the fan between two levels, as the steady-state cooling temperature lies between two working fan levels. Basically, the fan controller is unstable. (some are annoyed when this duty cycle is short)

In another scenario, and assuming the same given load temperature condition from the previous scenario, the fan will spin at a constant RPM. The fan settles at a speed high enough to prevent temperature from crossing the upper temperature threshold, or the RPM is just low enough to prevent temperature from crossing the lower temperature threshold. Also, in this case, if the cooling requirement is for a fan speed that lies right in the center of the spread, then one test may report 1-2°C higher when fan settles at lower rpm and 1-2°C higher when fan settles at lower rpm.

Whether one or the other scenario manifests is due to this uncertainty in the fan rpm values that can seem to be quite broad and random of course. It would be interesting to see how this varies from one system to another.

But this can probably be ruled-out as a significant effect if steady state temperatures don't deviate by much over many repeated tests.

Personally, even if just for 1-2°C difference, I would like to know which TIM to select for my next paste job. With the troublesome fan management implementation in these systems, a 2°C makes all the difference in the world.

 

Very very nice!

REP when I can thanks dude really appreciate it

 

edit: something went wrong, double post

 

Thanks! I'm thinking about playing some more around this weekend since I might have some much needed free time. I'll probably just end up actually gaming on it, though.

 

what about using adhesive copper foil ? or should i go with aluminium
i have this pretty cheap copper thing lying around that is for snails
specs says temperature can reach to 155C
it is very thin btw

self adhesive Copper slug repellent tape, copper foil, sold in 1M lengths | eBay

 

My concern is the adhesive itself is thermally conductive.

 

great review

 

Good stuff, +rep added.

 

I have some old lenovo laptops with the thermal tape on the heatsinks, I think I'll pulll it off and try it on my e7240.