I put coolabs on copper last night and took a pic… You have to rub it in first for it to perform right. Then put a little in the middle of the pill and mount strait down.
Your practically soldered then 
Best overall is MX-4, spread, and performance.
Yeah have to disagree… MX4 is probably one of the worst offenders of failing over time. I’ve used them all- even the diamond pastes. They go on- anyways already explained in this thread
I got some of that stuff comfy recommended… Silicone heat transfer compound. It’s metallic oxide and silicone oils- mine came open though they squeezed some out… Going to file a complaint but I still used some. I like the applications it’s used in though, maybe the best for flashlights
It seems legit at least. Says does not separate on the tube, but found there was oils in the top- but it was opened too so who knows. Using it on the threads between the sink and the pill
I’ve always just used Arctic Silver for CPUs. Sometimes boxed CPUs come with a little thermal paste, which I use when I build non-enthusiast machines where the people will probably never know or care as long as the machine doesn’t overheat.
Well, 4 months ago i opened my wife`s CPU and used AS5 on the die- after 4 moths it still shows exacly the same temp levels, so i could say it doesnt degrade as PK1-2-3, Gelid or other top CPU thermal intefaces
So, apart from the liquid metal it seems its the best choice for our flashlight use
The best standard is Arctic Silver 5 and it is not expensive either. Here is a great roundup of best thermal paste for CPU and GPU.
No it’s not the best.
It’s just commonly used and pretty cheap so a lot of people buy it.
Arctic silver is garbage, compared to even newer cheaper Arctic brand stuff.
It WAS the best one years ago when it had a huge following. It was THE brand to get. Like buying a voodoo video card.
Umm, okay, so everyone’s poopooing AS5… so what is the best (and isn’t obscenely expen$ive)?
Absent anything else, plain ol’ Fujik will work better’n nothing, and AS5 has always worked fine for me.
I don’t want anything that’s either going to run when it gets hot, nor dry out and require reapplication in, well, ever. So again, AS5 pretty much fit the bill for me all this time.
So what else is out there? Assume Al-on-Al or maybe Cu-on-Al.
Thermal Grizzly Kryonaut, MX-4, NT-H1, and IC-diamond are the ones I see most used without problems.
Gelid GC extreme also seems very good but I rarely hear of people buying them.
I use MX-4 because I can get a huge 20g tube for pretty cheap, unlike the others.
MX-4 also has the advantage of being less viscous (so it spreads very well without tons of pressure) and also can be cleaned with a simple moist paper towel or cloth, unlike the other thermal pastes that require cleaning with alcohol.
In terms of performance, nobody has really done a good scientific comparison of all pastes with almost no variables.
There are multiple large comparisons online, but the differences between all the best ones is within margin of error.
For example:
https://overclocking.guide/wp-content/uploads/2015/06/Tim_test_update1.png
http://media.bestofmicro.com/6/G/396808/original/02-Air-Cooling-High-Pressure.png
http://media.bestofmicro.com/6/I/396810/original/04-GPU-Cooling.png
Liquid metal is far better than regular thermal pastes, that is very clear, but between the best thermal pastes it’s too close to tell a clear difference (and even if you could, it’s so small that it would not matter)
So that is why there are other things to consider such as cost, ease of cleaning, viscosity, rather than which one is 0.1 degrees cooler in test X or Y.
The problem with these “thermal paste benchmarks” is that there are way too many variables in the tests.
Everything from non-uniform layer thickness (due to different viscosity), to inconsistent computer load, to different amounts of paste.
Even though some of the tests show numbers with a resolution of 1 degree or less, the margin of error is likely higher than that.
.
A proper test would consist of an insulating tube (maybe like 5cm long) filled with the thermal paste, so that the distance is is held constant and the longer distance allows for much clearer differences to be observed.
The tube would have a regulated wattage load on one end to produce the heat, and a high accuracy thermometer on the other end, both with the end of the tube capped with insulating material.
The 5cm of thermal paste would be the only material conducting the heat between the load and measuring device.
Both the load and the thermometer should have the exact same amount of area contacting the paste.
The room should also be temperature controlled, even though the delta temperature is what matters there is possibility of pastes changing properties depending on the ambient temperature.
If you apply ICD and MX4 in the same flashlight, do you end up with the same TIM thickness?
You mean so that both of the pastes are the same thickness? Yeah.
The thickness would be between having all ICD and all MX-4.
Unless if you tighten down your MCPCB a lot then it’s possible the ICD would become as thin as MX-4 simply due to pressure and the uneven surface of the materials sandwiching it.
The maker of one of the pastes, might’ve been AS5, says that best results are obtained after a few cycles of hot/cool/hot/cool/… to I guess let the thermal ick “settle in” somewhat. So probably just bulk heating through the samples might not be the best way to do it unless that’s also taken into acccount (ie, another variable).
Basically, since I’m not heating an LED to incandescence, as long as it gets decent results when, say, a small dollop is squished between the mcpcb and shelf, it’s fine by me. Again, with the caveats that it doesn’t run or dry-out for a reasonable life of the light.
True, temp-cycling all the pastes would allow the data to settle over time and see which pastes require burn-in time.
If done continuously over a few thousand cycles it could also give us data on the longevity of the paste in terms of temperature cycles (but not time lifespan, a long-term test would be required for that)
You’re right that for LEDs the marginal difference shown in the benchmarks becomes even more marginal, so upgrading from AS5 to something else will make pretty much no noticeable difference (except with liquid metal, that may be n improvement)
Nothing wrong with AS5. If you still have you tube from 2006 of it it will do fine.
Arctic also makes MX4 which is better now. But finish your 1st serving first!
I did not mean mixing pastes, just 2 separate applications with the same amount of force. As much force as bezel tightening or whatever else tightening method provides.
I learned so much from reading this post. Thanks guys, you really know your stuff.
With the same force a thicker viscosity paste will (almost always) result in a thicker layer.
But probably not a very significant difference in performance unless the pressure is weak.
You need to look at some good reviews (for example, this https://www.gamingfactors.com/best-thermal-paste/) on this topic and choose the most interesting for your studies. The only thing that should be noted is that metallic thermal pastes are the most popular because it conducts heat better than the other two types (ceramic and silicon). The main problem with metallic thermal pastes is that it’s also inherently electrically conductive. With smartphones there are another situation: when I wrote an essay on harmful effects of mobile phones on students, then in my research I used a standard silicon thermal interface. This is a good thermal paste for people who are beginners.