Thermal paste on laptop CPU/GPUs (avoiding pumpouts)

https://www.keyence.com/ss/products/measure-sys/gd-and-t/form-tolerance/flatness.jsp

If you see the Lenovo T480 heat sink, you'll notice that overpressure is more likely than underpressure. It's just designed badly.

 

I have an undergrad in mechanical engineering, I know what GD&T is, you didn't answer my question. What are you measuring flatness with?

 

None of the notebooks regardless of brand, have heatsink with adequate or good enough pressure over the Cpu/Gpu. And you noticed overpressure. Nope.

Btw. Thermal Grizzly offer a new thermal paste... Kryonaut Extreme

 

A confocal laser CMM

Have you tested a modern Lenovo? Their obsessions over thinness comes at great costs. Ranging from keyboards touching the screen when the lid is closed, and inadequate thermal solutions.

 

And what was the difference in z-height, and which heatsink were you measuring? Also what about the test mounting? If you have a perfectly flat heatsink with good pressure, you shouldn't have pump out with any thermal paste. Also I'd love to see pictures of your CMM setup, I've owned a lot of laptops and I've yet to see a perfectly flat heatsink.

 

The heatsink for T480 on a 5mm thick piece of +/-2 micron aluminum all purpose jig vacuuming down points of interest. Flatness measurement is ~+/-68 microns which is more than "flat" enough for a laptop heatsink considering it gets at least a 1mm layer of thermal paste.

I don't think I have the clearance to publically post a picture of the lab.

 

I',m not sure what you mean with modern Lenovo. But have benched L460 and E480. The Cpu in the Lenovo E480 with the power gimped i5-8250U perform around or above an 45w H class chips ( i7-6700Hq).


And this one haven't the pump out of paste problem as your so called overpressure/perfect machined heatsink. What you think/belive is an perfect machined heatsink is only an illusjon. There is always same reasons for thermal paste pump out.... Bad heatsink fits combined with an soft and creamy thermal paste as Kryonaut.

 

Of course the "points of interest" are the contact points individually, not the whole thermal system including the heatpipes and etc.

 

And these points of interest include the CPU and GPU mounting surface and comparing between the two?

 

Something is very wrong if you have a 1mm layer of thermal paste...
And that also explains the pump out, thermal paste is not made to be stable with 1mm thickness..
Fix your pads and let your heatsink TOUCH the die..
Thermal paste is there to fill micro irregularities, not macro.

 

I want to apply a thermal paste on a U CPU (naked die, no IHS) Thinkpad. I am looking for a long lasting solution, these are some of the thermal pastes that I read are prevalent:

Themalright TFX or TF8
Phobya NanoGrease extreme
Kingpin KPx
IC7
IC24
CM MasterGel Maker (New)

There are also pads as the IC pad and the Carbonaut from TK.

I would have gone immediately to the Carbonaut, but I've seen threads where the pad is actually teared up upon inspection from usage after just a few months.

Phobya, I see it was very recommended and still is, however also not immune to the pump out effect.

For the IC Diamonds, they seem to be kings of long lasting, this could have been my choice, but I feel like I do not want scratches directly kn my die. I did read that being careful with it might prevent all of this, so it still is in my radar. I haven't researched for differences between the two models mentioned.

MasterGel from Cooler Master, seems like a less scratchy version of the IC Diamond, with skmilar viscosity iirc the thw compliation statistics, so maybe this is my top contender.

Thermalright, also seems goos, but barely any reviews and my research on is 0. So fill me in.

And here it comes the most esoteric to me, at least from community perspective, but the KPx. Seems like the perfect
paste, rated very high in terms of performance, highest in terms of viscosity, so great for laptops (but isnit too great?). It spreads hard, but aside of that, even if the performance in a notebook is not as great, why is this not the most reckmmended paste. It is also on the market for quite a while now, well if ~3 years is a long period of time. So why is it not much tested.

There are the IC thermal pads, which to be quite frank I forgot why I concluded theybare not much good. I do know Carbonaut performed better and that they can be bought cheaper as a technical OEM.

 

I've tested numerous pastes myself and come to the conclusion that the best overall paste is new mastergel maker. KPX is another strong contender and the longest lasting paste out there but its price and availablity are what makes it a rare reccomendation.

 

Xigmatek explicitly says their PTI-G4512 is pump-out resistant, and it's a somewhat cheap and easy to find compount. I've used it in the past for GPUs and laptops and it works fine even after years.

 

So from reading this thread, I am interested in learning more about performance programming since I do work with Java and front-end JavaScript development for a popular CRM for work and want to get into more sophisticated programming.

However back to the point here, is the two most “set it and forget it” pastes IC diamond and Mastergel?

Because, when running prime 95 with turbo boost immediately spikes the temperatures up to 90*C from 40*C idle, I’m thinking it’s time to remove the ineffective pumped out MX-4 and replace it with something thicker.

 

Kryonaut did not pump out after 6+ months. However its performance dropped drastically from first 2 weeks and on. Max temperature on max load in the first two weeks stayed below 92C and that changed to 98C on third week. This is on my Lenovo T480. No drying out, nothing wrong whatsoever other than a noticeable incremental performance drop over time (at fixed fan speed and in similar ambient conditions). I only replaced it because I got a batch of hydronaut with a new cooler for my desktop.

Now trying Hydronaut. It's been underperforming compared to kryonaut since day 1. Max temperature already at 97C.

I have no clue what people mean by that Kryonaut is harder to spread because it does come out of the syringe a bit strangely but in terms of ease of application, it's average and as easy to work with as hydronaut.

So far, Kryonaut is the best performing paste without pump out effect across all of my laptops. It's especially impressive on my Alienware 15 keeping the CPU and GPU under 70C while on max load (with tweaked fan speeds). I think I am going to keep using kryonaut. I don't think I am going to experiment with more pastes until I find a good reason to look for a better one. If anyone has success with pastes that do not pump out other than Grizzly Kryonaut and Hydronaut based on personal experience on laptop please let me know. I'll keep them on the list if I find something wrong with Hydronaut/Kryonaut.

To respond to Senso and others who do not understand the concept of figure of speech. I don't actually apply 1mm of the paste. Based on my experience in this thread, I will not waste anyone's time trying to explain further than that. I will have a better chance at explaining what gravity is to a flat earther.

 

If it degrades quickly, it fails by my book. That's why I test longevity, not pump out which is a meaningless metric.

Doesn't matter if it doesn't pump out if its performance sucks.

 

By the time I replaced Kryonaut, it still performed better than Hydronaut on new application. I can't really call it failed because it still outperformed all other pastes and did not pump out. It was just really visible to me that it aged.

Whereas MX4 for example worked really well for over a month then dropped off the cliff from pumping out. I would consider that a fail.

 

Different standards and expectations I suppose. When a paste rises 5C above fresh application I consider it degraded.

 

Kryonaut degrades if exposed to temperatures above 80 °C for prolonged periods of time unfortunately. It doesn't pump out though which is good.

Phobya Nanogrease can work in your laptop. It's more thermally conductive than Kryonaut, but is prone to pumping out. It's really a paste for laptops with high quality contact plates on their heatsinks. It needs a smooth surface to not pump out, although increasing the mounting pressure can remedy this a bit.

I have a tube of both Kryonaut and Nanogrease and decided to do an experiment. I applied Nanogrease directly to my CPU die and put a ring of Kryonaut around it to see if it would act as a seal to keep the Nanogrease from pumping out.

So far so good. It's been 2 weeks and temps have not degraded yet, which means the Nanogrease has not pumped out. It usually pumps out withing 2 weeks after an application on my Ranger. I'm not going to say the experiment was a success just yet though. I'm going to give this more time.

 

Thermalright TFX is best paste.

 

Very interesting approach. Never heard of this method.

 

Dropping from 92C to 98C from week 2 -6 months is really bad in my eyes. Kyronaut has reasonable temperatures initially, I can agree with my tests. I can't comment on longevity because I don't test but you are not the first user who reported a performance loss from Kyronaut. I believe the temps are too high for the Kyronaut, it doesn't like over 80 degrees over a longer period. For a laptop not the best choice in the long run.

 

It was inspired by the foam dam barriers and electrical tape barriers people on these forums used to keep liquid metal from spilling out. I figured I could to something like that for thermal paste to keep it from pumping out as well. I used Kryonaut for a while, but it kept burning up, so I switched to Nanogrease, which offers superior heat transfer, but is annoying since it pumps out every week or two.

I didn't have any of the right foam to make a dam, and I didn't have any electrical tape on hand, but I have 2 syringes of thermal paste on hand. One is Nanogrease and the other is Kryonaut. I knew that Kryonaut strongly resists pumping out, so I wanted to see if putting a ring of it around my application of Nanogrease would lock the Nanogrease in place.

I don't care if the Kryonaut burns up since none of it is on my CPU die, only Nanogrease is. The Kryonaut just needs to be a seal to push back against the pressure from the Nanogrease when it wants to pump out. If this experiment turns out to be a success in the long term, I will open up my Ranger and share some screenshots of the paste job.

For those who want to know exactly what I did, I applied some Nanogrease to the CPU die and then applied a thin ring of Kryonaut surrounding the die on the PCB next to it.

 

Welp the experiment failed, but not completely.

The Kryonaut did in fact slow down the pump out effect, but the Nanogrease still pumped out. I'm going to try Thermalright TFX next because it is extremely viscous and strongly resists pumping out like Kryonaut, and it doesn't burn up at high temperatures.

 

Extremely would be an exaggeration. It's slightly more viscous than average but thats it.

Have you spread old KPX before? Its like spreading blue tack if you don't warm it up prior to application.

 

My TFX is viscous, almost like putty, not sure if extremely is the right word but it definitely has a very high viscosity and surely the highest viscosity and hardest to spread from all the pastes I've tried.

 

I have not. I was under the impression that TFX had a very high viscosity.

 

I'm a little late to the party but I've got some suggestions that might help. I highly recommend ICD as a thermal paste even though there is a cosmetic hazard in scratching the die surface on application or removal. I've used it several times and on multiple laptops; the stuff is incredibly thick and does a very good job of dealing with uneven heatsink seating and addresses pump-out due to how it cures over time. It becomes fairly immobile once set and takes a long time to dry out. I have a 7-year-old laptop (Asus G75VX) that has never had the thermal paste changed and is still running within 5 degrees of when I purchased it.

Also of importance is thermal pad selection. As I'm currently using an Alienware laptop, I can't speak to what arrangement you have in your ThinkPad, but thermal pad thickness can easily disturb clamping pressure, even lifting up a corner of the heatsink enough to start the pump-out process. I've found finding the correct thickness pad to be an undertaking given those measurements aren't available to the end-user, at least not through Dell.

A potential solution is this product: http://www.tglobaltechnology.com/uploads/files/tds/TG-PP-10.pdf

It has a 10 W/mk rating, while not as high as Fujipoly at 17 W/mk, it is both a putty and reusable. I've found Fujipoly pads to be extremely delicate, often tearing when cutting with brand new X-Acto blades.

Additionally, I've stumbled onto this product as well: https://www.coollaboratory.com/product/coollaboratory-liquid-metalpad/

I currently have liquid metal on my CPU and have had noticeable success using it. However, there are hazards (which I'm sure you're aware of) in using a liquid, highly conductive material in an expensive electrical device. If you've done your prep-work, there's little to no chance of a problem. From what I've read, the liquid metal pad solves two problems that normally face users of traditional liquid metal; spillage due to pump out and hazards with the TIM coming into contact with aluminum. This solution is immobile and not made of gallium. The only "issue" I've read about is that the material has a thermal set requirement, a burn-in if you will. After installation, the die needs to reach a certain temperature for a length of time, causing it to adhere to the CPU and heatsink.

I've recently purchased both the putty and metal pad with the intent of trying it out in my daily driver.

 

I was looking for ICD but it's not available in my country, same for Kingpin KPx at the moment, although this might be available some time later.

 

I can really go on about my disappointments for Lenovo T480 but I'll just limit it to its thermal performance downfalls. The standard 3 cell battery makes the laptop so flat, no circulation happens under the laptop. 6 Cell battery raises the backside a bit and I automatically lose 10~15C.

Without external fans, with te i7 4GHZ core and NVS MX150 running at full, thermal load is obviously way too high for a tiny wimpy fan and thermal solution even while fan runs at 4700RPM continuously.

Now, there are 3 thermal pad locations, all of which I just keep the original pads for. It probably does more harm to be in thermal contact with the heatpipe regularly reaching 100C. As for the main surface to contact with the CPU+GPU are rather flat. The pump out usually happens uniformly outwards from the die showing that the contact is even.

Now, going back to thermal pastes. It seems that more rigid and hard the paste they are, quickly they pump out. For example Arctic MX-4 is rigid and performs really well but pumps out within a couple of months. Whereas Kryonaut is very sticky and less rigid and did not pump out in >6 months at the cost of performance.

Now with Hydronaut, I am noticing that the heat transfer between the die to heatsink must be slower. I guess the heat capacity of Hydronaut must be much higher than Kryonaut as a comparison. It seems to be really good at maintaining a temperature but heat spikes really do show.

 

It's the first time I've read from someone that Arctic MX-4 has better performance than the Kyronaut. Usually Arctic MX-4 is more a watery paste in my eyes which isn't the best fit for a laptop. But maybe it's really CPU depending. The Phobya NanoGrease Extreme had good laptop reports whereas on my 10nm CPU it's bad. Your laptop seems to run on a 14nm CPU.

 

Nanogrease does better on larger surface areas as there is more space for it to expand and contract. Higher heatsink mounting pressure also helps kepp it from pumping out.

That metal pad sounds interesting. Let us know how it goes and how it compares to liquid metal!

 

Have you ever got to try prolimatech pk3 or kingpin kpx?
They are a very thick thermal pastes which can not pump out

 

Kingpin kpx have changed. Buy old batch if you want the thick stuff. Now it's a much softer paste. Still worth a try.

 

So, remain only pk3

 

Speaking of which, I have just applied it on my GS75. Its super liquidy. Like very diluted toothpaste.

Temps are worse than TFX and mastergel maker

 

Have you found a suitable thermal paste for your gs?

 

They will always create the kingpin paste to be good for Ln2 and sub zero cooling (Fast and apply and remove). Remember the same is also for Kryonaut. And I won't use that paste on any of my computers.

http://forum.notebookreview.com/thr...2-owners-lounge.832848/page-253#post-11069652

 

Yes, I was using Lm on the cpu and carbonaut on the gpu. I'm just testing pastes at the moment.

 

Unfortunately here in Italy Kingpin Kpx, TFX and PK3 all are difficult to find and Kryonaut extreme is quite expensive per gram.

Personally I have good long term results with MX-4 even if the paste doesn't seem to be held in high esteem on this thread.
Obviously the correct application is paramount and if you put too much on the CPU/GPU die it will result in poor thermal conductivity (too thick) and if you put too little it will dry up fast and lose it's heat dispersion effectivenss. I think I apply it correctly most of the time, yet when I test it results can vary up to 4-10 °C from application to application. It's really a trial and error.
But When I apply it correctly I get very good thermals both on GPU and (relatively since the Intel contact die are so small) on CPU.
Also, form my 8 year experience using MX-4 on three laptips (2 of which were gaming) I can say MX-4 has a very good duration (deteriorates slowly over time IF you use max fan settings to take the heat away efficently) and temp stable over 1-2 years usually before I have to repaste. Aslo I have never seen it 'pump out' a lot, just the excess part, which is good since the thermal paste is supposed to fill only the microscopic irregularities in the contact between the die and the heatsink. If you can't see thermal paste on a die when you dismantle the heatsink, it's probably because you used too little to begin with. Just my 2 cents.

 

You can manually disable PWM on your fans, allowing the fan to spin up at 100%. 4700RPM is an arbitrary firmware fan speed. To do this, go to RW-Everything > Embedded Controller and write "40" into offset 0x2F.




EDIT: Just to clarify, this address only works on ThinkPads.

 

Hello,
I think the pump out effect also depends on a heatsink, in the sense:
-The space between the heatsink and the CPU, if it were small, would be enough to have a medium high temperature for the heat sink to expand thermally.
-As you said, keeping fans to the max causes the entire heatsink to be "colder" than at a lower fan speed which reduces expansion of the heatsink.
So, in short, pump out occurs with inefficient dissipation systems that depend on the speed of the fans. The strange thing is that with Laptop with steam chamber heatsink the problem seems to be less present.

 

You can get Thermalright TFX on Amazon Italy

 

Thank you I wasn't sure it was the 'good stuff' so I went with MX-4. Will try out TFX in the future for sure, for comparison.

 

Just to clarify 4700 RPM is the maximum fan speed, not some arbitrary firmware fan speed. I appreciate your input but I use tpfancontrol( https://sourceforge.net/projects/tp4xfancontrol/)/hdinfo(https://www.hwinfo.com/download/) controlling fan fron W10 and thinkfan for linux and these I find are easier and user friendly alternatives to rw-everythiing.

 

In which case you can input "64" into your fan speed, which will disengage the PWM signal. I still believe that a fan speed of "7" is a firmware maximum, not the mechanical 5volt fan maximum.

 

In case if you didn't know "64" yields 4700RPM. 7 yields 4000. I've been trying to tell you I know how to use the fan...

It seems that the general feedback at this forums is "the users who post here don't know anything, but I know everything because I have the internet. Let's try to get as condescending as possible."

 

Dude, chill. You've been pretty snarky with anyone here that's attempted to provide information you don't agree with. It's a buffet; take what you like and leave the rest. From undervolting to fan control, it's all an attempt to help you, not challenge your ego.

 

I don't consider them challenge to my ego. I consider them malicious distractions from the topic. I mean if I started posting at a thread regarding undervolting and input my "knowledge" in thermal paste and fan control, I am 100% not helping anyone.

By far within this thread, I find only the following relevant inputs:

1. Possibility of bent heatpads/pipes.
2. Recommended pastes and pads.

Ironically the two items above have the least number of posts. The following items I consider irrelevant and even malicious to the topic at hand:

1. Undervolting.
2. Fan control.
3. Explaination of "Oh, you need to do this to operate the fan/CPU properly." without a single consideration that maybe, just maybe the user knows what he's doing.

I do find useful information from the Panasonic community and Alienware within the forums but I can't help but being actively counter-malicious towards people in this thread who have so much to contribute but very little patience to read even the title of the thread.

 

Undervolting is a tried and true method. For a lot of less experienced users (of which I'm not including you), it happens to be something that's overlooked. You, on the other hand, have made it very clear that your use case is specific enough that it isn't an option. I didn't detect any malice or condescension between you and Che0063, on the latter party, that there was anything other than an attempt to help. That user simply was conveying what they believed to be true and useful. I believe it's worth pointing out that tone is lost in the written word, far easier than can be stressed.