Bored at work. Decided to make a thermal paste stencil

[bignazpwns]

Got bored at work so I decided to make a thermal paste stencil for Ryzen's.


Rev one to test the paste amount and coverage. Final version will look much different and be set up for ease of use. Just squeegee it on then lift the stencil and your done.



Now this is 100% pointless. But for doing test on thermal paste this makes the application amount the same every time as well as the application. So it removes the possibility of user error.

 

[Super Nade]

You should try standard recommended patterns such as an "X", dollop in the middle, "Z". What CAD program are you using? NX? Solidworks?

 

[bignazpwns]

You should try standard recommended patterns such as an "X", dollop in the middle, "Z". What CAD program are you using? NX? Solidworks?

I do. I just wanting something extremely consistent for when I check temps between brand x and brand y. Remove the user part of it where it's the same application and same amount every time.


Solidworks.

 

[Bustos]

How does it make sure the same amount is used each time? It's just a grid with holes. How are you measuring it? I may be missing something here.

 

[UltraTaco]

I think he applies it and scrapes off the rest, like filling in the gaps, then remove the grid and you have super ultra consistent coverage instead of "x" or round blob because it doesn't have to spread far away on it's own from original application spot.

Edir: I just auto realized, you will have to be ultra precise at tightening everything at the same time to have maximum consistency in results.

 

[Blaylock]

Right, the thickness of the stencil will regulate the amount applied once the excess is scrapped.

I have a couple concerns for this testing though. First, the design of the stencil could lead to air pockets under the heatsink creating an inconsistent contact surface. Also, Ryzen CPU's will auto regulate the clock/voltage based on temperature. You'll need to verify that the clocks and voltages are fixed and do not fluctuate once testing begins.

I'm interested to see the results. To my knowledge no one has tried to standardize the TIM application to any degree of accuracy. I love the idea and hope this works out.

 

[EarthDog]

What is the thickness of it/the TIM application?

 

[MaddMutt]

I may be WAYYYY - off base here

You do know that you can get internet movies to watch while at work. All you would then need is some and

Warning : I can not be held responsible if you get caught ^ doing something COMPLETELY irresponsible. If this happens then it is your

^ You have been warned by the Dog Pound

 

[UltraTaco]

Mr.Mutt came in with the most reasonable solution. I do agree though, watch some movies!

 

[bignazpwns]

i got a bunch of revisions im going to test. in a few shapes.

from a 30um thick plate all the way up to 100um in 5um increments. If need be i can go thicker. Also working on finalizing the design.


Main idea is get a version for Intel as well as AMD and threadripper. Send out a few to some of the guys who do the big shootouts of thermal paste and let them do there thing. Then upload the STL online so anyone can make one. But if its lower then 50um FDM printers might have issues with it. And FDM printers really cant get the surface finish needed for this. It needs to be flat on both sides. But air bubbles wont be an issue. When you are swiping you are pushing down so it packs them full.


Some times i play world of warcraft at work lol.

 

[EarthDog]

Before you send it off, I would work on seeing what the spread is like with that odd honeycomb pattern.. like press down on it with a thick piece of plexi and see if you get air bubbles...

 

[UltraTaco]

One way to migitate happiness of air bubbles is to intentionally start mounting tapered to one side and ar the end tighten it flat

 

[EarthDog]

One way to migitate happiness of air bubbles is to intentionally start mounting tapered to one side and ar the end tighten it flat

Which then pushes more to one side and out that side. The best way is equal pressure. Most AIO's and any coolers with four mount points tell you to tighten cross corner and never tighten fully... you go around incrementally.

 

[UltraTaco]

Ok, thanks!
I was just trying to sound smart. I stand correctioned.

 

[Blaylock]

If you look at all the recommended methods there is always a path for air to escape. X, I, II, , & the one that represents "5" on a 6 side die.

You'll want to emulate that concept somehow.

 

[bignazpwns]

Yeah I got some ground lab glass on the surface of an inspection station that's 100% completely flat. That's why I made a few patterns. I'll be putting a camera under it want watching how it spreads. I might even die some white paste multiple colors and see how it spreads and how much overlap I have when a waterblock is installed.

 

[freeagent]

Honestly, I think way to much thought is being given to this. I bought some Thermalright paste, and it came with a spreader and a picture of how to spread tim over your IHS. On all of my Intels, I just put a good dot in the middle and plant my cooler on it, and I give it a titty twister before I clamp it down. On my GTX 970 and 980 I just put a good dot in the middle of the core and planted the cooler back down. Works awesome. On AMD cpu;s it looks like cores sprouted chiplets, so probably a good idea to coat the IHS. Could also be all these new cpu's are based on such small and dense packages, they cant disperse heat as well as their older and bigger brothers can. So that heat could be built in, as in meant to run hot. To draw a quick picture, my 3770K runs hotter than my x5690, but puts out way less heat than my x5690. 3770K=22nm, x5690= 32nm.. Not to mention it seems AMD IHS is not the same as Intel IHS, meaning coolers that work great on Intel, may not work so good with AMD.. seems like AMD likes a flatter base, whereas Intel likes to be poked with some convexity. I could be way off here..

 

[WhitehawkEQ]

Honestly, I think way to much thought is being given to this. I bought some Thermalright paste, and it came with a spreader and a picture of how to spread tim over your IHS. On all of my Intels, I just put a good dot in the middle and plant my cooler on it, and I give it a titty twister before I clamp it down. On my GTX 970 and 980 I just put a good dot in the middle of the core and planted the cooler back down. Works awesome. On AMD cpu;s it looks like cores sprouted chiplets, so probably a good idea to coat the IHS. Could also be all these new cpu's are based on such small and dense packages, they cant disperse heat as well as their older and bigger brothers can. So that heat could be built in, as in meant to run hot. To draw a quick picture, my 3770K runs hotter than my x5690, but puts out way less heat than my x5690. 3770K=22nm, x5690= 32nm.. Not to mention it seems AMD IHS is not the same as Intel IHS, meaning coolers that work great on Intel, may not work so good with AMD.. seems like AMD likes a flatter base, whereas Intel likes to be poked with some convexity. I could be way off here..

Hey keep it G rated JK

 

[bignazpwns]

Honestly, I think way to much thought is being given to this. I bought some Thermalright paste, and it came with a spreader and a picture of how to spread tim over your IHS. On all of my Intels, I just put a good dot in the middle and plant my cooler on it, and I give it a titty twister before I clamp it down. On my GTX 970 and 980 I just put a good dot in the middle of the core and planted the cooler back down. Works awesome. On AMD cpu;s it looks like cores sprouted chiplets, so probably a good idea to coat the IHS. Could also be all these new cpu's are based on such small and dense packages, they cant disperse heat as well as their older and bigger brothers can. So that heat could be built in, as in meant to run hot. To draw a quick picture, my 3770K runs hotter than my x5690, but puts out way less heat than my x5690. 3770K=22nm, x5690= 32nm.. Not to mention it seems AMD IHS is not the same as Intel IHS, meaning coolers that work great on Intel, may not work so good with AMD.. seems like AMD likes a flatter base, whereas Intel likes to be poked with some convexity. I could be way off here..

It's the only way you can say how much better paste x is then paste y. It's not about just putting it on the cpu. It's about fine control over the amount and distribution of the paste over a given surface. Put it on. The paste is in the same spot and have a thickness that's accurate down to +/- 78nm. Yes the accuracy of the z on a sla printer is amazing.

I can look at one site and it has NT-H1 2c cooler then kryonaut. Then check another review and the Kryonaut wins by 2c. So that's a 4c swing thats not small.

So if you eliminate the human error. The numbers are what the thermal paste provides. Is this for someone who wants to put paste on there CPU? Can be. But it's designed to be used for a consistent paste aplacation that's repeatable for testing purposes.


And we're also not talking about compairing x CPU temp to y CPU temp. Your completely missed the point. It's about paste. That's it. A test with the same system.


Also a side note.

Tomorrow we will be looking at Cooler master mastergel nano under extremely high magnafcation and taking some measurements of the diamonds. Because? Why not. I'll check out some NT-H1 and take some pictures under high magnafcation.

 

[ShrimpBrime]

You're going to need a stable testing environment.
As BTU is added to the room, the room temp can raise thus creating a warmer running system assuming say 10 pastes are running a burn in period before testing (as some pastes require). 4 degree Fahrenheit is a 2 Celsius raise in temps. My thermostat for the house has a 4 degree swing, pumps to 2 over desired temp and drops to 2 under before pumping again in example. This type of ambient action is difficult to regulate.
So if these tests are done in a small room, I'd want to see ambient temps before, during (multiple times if needed), and after.

These kinds of tests will be within a margin of error I'd have no doubt. So if 8 of ten pastes are with 2 degree Celsius of each other, I would be in no shock simply because the testing environment will likely fluctuate that much.
I have found one type of cooling that is actually very very stable when it comes to temperatures. It has nothing to do with atmosphere actually. No cpu BTU is dissipated with end results, fans.

If we saw a swing of 4-6 degree on a stable testing environment all things accounted for, that would be pretty awesome results. One of my biggest issues with TIM application and one being (THAT MUCH) better than another really does not justify the testing facility done with these thermal compounds. In short, I could make any thermal paste look better than another without an actual stable constant testing environment.

Super interested in the applicator design though. I really like it. Hope this works well!!

Subbed.