HeatSprings Sign Up Here

My guess is that its difficult to get the CPU heatsink to be 100% level. A slight angle can still be quite good with paste, this isnt that forgiving.

 

how many layers are you using? can you stack it?

 

@HTWingNut, do you have any mods on your CPU heatsink to increase pressure?

Are you running it on your Clevo W230ST? With the M17x R2 we can do the c-clip mod to add more pressure but stacking like pinoy_92 mentioned should help as well.

 

Yes w230st. I will try stacking later.

Beamed from my G2 Tricorder

 

Installing in my R2 now.... results very soon

Sent from my SM-N9005 using Tapatalk

 

(pant)(pant)(pant)(pant)(pant)

 

After disassembly and installation of heatspring I cannot get my machine to power up. No post, no lights or fan spin, just plain dead. What is going on ?

Sent from my SM-N9005 using Tapatalk

 

Do you have a spare cpu you can use? Same thing happened to me last month. I had to replace the cpu.

Sent from my DROID RAZR using Tapatalk

 

Stupid flingin did not knew he have to connect the keyboard to get power .....

Sent from my SM-N9005 using Tapatalk

 

Cpu 39-50 idle so far

Sent from my SM-N9005 using Tapatalk

 

2 cores heats up much more than others 90 vs 57 'c

EDIT

Ok so after adjusting the heatsink I was able to run 24 fid on 4 cores, but because of pressure not being equal 2 cores operated at approx 71 + 'C and other 2 at 95+, which tells me HeatSpring is really really AWSOME, but you have to get the heatsink pressure right, also tried 2 layers and temps gotten worse. Screenshots from TS later

It will be hard to get optimal pressure in a laptop, but this heatspring thing will be a BEAST in desktops !

peace !
Sent from my SM-N9005 using Tapatalk

 

Awesome feedback flingin. Are you running a stock heatsink or with the c-clip mod?

 

did you try to adjust the screws while running flingin to try to level the pressure?

 

No c clip mods. And yes I tried to adjust heatsink while cpu was on load but this is very dangerous because cpu temperature would spike up by 20'c sometimes reaching 100. I also do not have to remind you that if you put to much pressure on one cpu corner you will end up cracking CPUs DIE, which I did yesterday. But is is a very small chip and my cpu works just fine anyways.

This has super potential. If i would somehow manage to get pressure even, I could run cpu @24 on all cores and the highest temps would be in regions of 75+ !

Testing Rig



These are temps 53% into TS bench at 24@ All cores TDP/TDC 105/90, you see the coolest core was at 78"C, which is pretty damn good, and without any temp spikes - I see potential Here



These are temps when i installed it for the first time and there was major preassure contact issue on one side of CPU, shame i do not know where exactly LOL



And this is is screen when i stacked 2 Heatsprings on top of each other, i ran only 32M test



I was thinking that if i would attach the Magnesium cover on it would help a bit, just because magnesium thingy rests on CPU heatpipe and it could help adjusting the preassure levels ( or not )

We could also put something between CPU Heatsink/Magnesium Cover to increase/adjust the preassure...

 

Thanks flingin! That is awesome info! I do feel bad because I feel my heatsink is very uneven on the surface. So without a way to easily level it out I'll need to do something else. Thanks!

 

Have you tried polishing your heatsink flingin? Just to be sure you have removed microscopic bumps?

 

Too much of a hassle. Also my heatsink looks perfect and I also tried 2 heatsinks as I bought another one just in case 6 months ago

Sent from my SM-N9005 using Tapatalk

 

Ahh ok.
Polishing is very easy BTW.
If you use brass shine or crome polish on a pice of paper it will shine like a mirror in no time
Dont rub hard anywhere even though that pitting is a lot of work with polish, just rub around the copper surface for a little while in cirles without adding a lot of pressure.

 

While it is easy to get mirror finish, it is not that simple with achieving technically ideally flat surface, using just paper and your own hands. Believe me, the end result might be even worse.

We are talking here about 0.1mm gaps. If not smaller.

We will wait for another users to publish their resaults, maybe somebody gets lucky with pressure distribution

And then we can do more tests

Pretty sure it will be a killer performance in desktops, especially in F@H systems



Sent from my SM-N9005 using Tapatalk

 

Hehe i know that leveling it would be near impossible with paper and polish

It was more removing the 0.1 mm gaps causing slightly more contact and even removing 0,3 tops if there are any since this cooling stuff is not very forgiving.

I think i will try coming week when i get the chance.

 

Good luck, have Batman !

 

To be honest, with my bad heatsink mount, I don't dare try the heatspring anymore I'll stick with the MetalPad which is coping ok for me atm, although max performance did degrade noticeably. But I'm keeping the heatspring on hand to have when I'll get a different machine I want to push.

 

does the heatspring require a burn-in process in which the tim itself melts? when stacking the heatsprings, does it require a thin copper sheet in between?

 

No copper shims in between needed but it can be tested this way. Good idea there. I think it requires the cpu to heat up a bit in order to sit in between DIE and heatsink well. Then it sticks to the copper and it can stay there.

Sent from my SM-N9005 using Tapatalk

 

initial tests tell me temp problems with core 0 and 2, especially core 0 easily hitting 100 deg in about 3 sec using TS bench. (25X and 90 / 90 ) and about 6 sec using 70 /60 24X

Lowest core 3 hovers in the 60 area for now, but havent run it very long for now.
Also done GPU2 to check my two GPU's compared.

Will try a bit longer to see if temps improve but dont expect it to be that much better than before after "burn in" since it hits 100 deg that fast with such low power.

I also see that decreasing the power increases the temp on the other cores.
Probably due to the temp throthling in the first core affects the other cores reducing all load???

 

Yes that clearly shows the contact is not perfect. Or far from being even correct.

Sent from my SM-N9005 using Tapatalk

 

Reducing force on the screws fastening varying screws does not change the values.

giving it a bit more burn in to test.

If i go to apx 60/55 and 24 i can keep it to right above 80 deg and difference of apx 12 from max to min core temp.
Temp difference way lower when power decreased.

 

12 is a bit too much
5 would be ideal

Sent from my SM-N9005 using Tapatalk

 

I know. But it might be uneven in advance since i have same two cores higher with paste too. At least 6-7 deg. Alway core 0 and 2 goes higher. Guess its the same for you?

 

Yes same for me. One side of the DIE is clearly not under pressure

It would be nice if we could say where the cores are situated and the way they are numbered/aligned, so we could point the side that is compromising heat transfer.

I will look up Intel schematics, maybe on one of them I can find each core numbered as 1 2 3 4.

What I was thinking is

Core 1 is on top left
Core 2 bottom left
Core 3 top right
Core 4 bottom right

But it also may be

Core 1 top left
Core 2 top right
Core 3 bottom left
Core 4 bottom right


Jesus.....

Sent from my SM-N9005 using Tapatalk

 

Probably one of those since others would be across from corner to corner.
I have bottomed the screws but no clip mod though.

 

I've not been part of this thread or the heatspring thing, but I thought it an interesting thread to read.

If you start adding shims, and other things that might require thermal paste as an interface (to get rid of the small air gaps) then aren't you defeating the purpose of the potential high heat transfer rates of the heat spring. Heat flow is only going to be as fast as your worst interface material, and the more material you use between the CPU & heatsink then the worse the heat transfer anyway right?

EDIT: Actually, I think I misread your intention, you were referring to jamming something above the heatsink (not on the CPU side) and between the outside cover of the laptop to increase overall pressure on the chip? I suppose you could test that by adding some finger pressure to various parts of the heatsink while it's running to see how that affects temperatures.

EDIT 2: Just had a look on their website to see what it was all about ( http://www.indium.com/thermal-interface-materials/heat-spring/#products). From one of their graphs entitled Bulk Thermal Resistance it seems to indicate that heat-spring thermal conductivity is greatly influenced by psi, whereas thermal paste is not (according to that graph). It also shows that at least 40 psi is required to match the performance of the pastes they tested. They were comparing 2mm of thermal paste vs 3mm of heatspring though, but maybe heatspring is 1.5 times thicker than paste after a typical CPU application (I don't know). 80 psi on the graph looks to be where heatspring shines, but no idea where laptop cooling systems sit in terms of how much psi they can deliver. Given some peoples results here, it does seem that the pressure is likely to be less than 40psi, given the sometimes worse performance than paste seen by users in this thread. Also, I think it shows why there is a lot of temperature difference going on between the cores. Because the thermal resistance of heatspring is so pressure dependant, it highlights any uneven pressure you guys are having in your heatsinks. Thermal paste is able to overcome those inaccuracies, as seen by that graph that shows that pressure does not greatly affect thermal resistance of pastes. Looks to me like there's not much margin for error for getting heatspring to work in a laptop, a bit of a knife edge, and that's assuming that laptop cooling systems can offer enough overall pressure full stop.

Interesting thread though, it's great to push the boundaries, and I hope it works out for some you guys after some more testing & fine tuning.

 

if the heatspring tim melts, i suggest LIGHTLY scratching the surface of the heatsink. i dont have a heatspring tim, but i tried it with the coollab metalpad few days ago and so far results are great.

 

I dont think its about 3mm or 2 mm either for paste or this heat spring.
It would me nm if anything

Like i mentioned before, its not very forgiving.
But i hope that burning it in for a while with almost 100 deg could deform it slightly and contact for the other side of the die would increase.
Its a hope, but not what i belive.

Think i have to check how to increase pressure and more important get it more in correct position between die and heatsink. You are right, the paste is WAY more forgiving.
Ill do more tests on that next week.
Might have to check if it is possible to schim the top of the springs on the screws to increase pressure and adjustment possibility or change the springs on the screws but still have springs to be sure to apply pressure more evenly and not crash the die.

Also this is physics. The formulas are quite clear.
40 PSI on what surface? Entire pad? That is not much.

40 psi = 2,58 bar= 2.58 kg per square cm if the 40 PSI is per square cm. If it is for a different area, its a quite fast recalculation.

Press your thumb with your thumb and you will see that 2.58 kg is not very much. We just have to clearify the pressure area where the 40 PSI apply to.

I think its more of an uneven heatsink/CPU mount giving us grief, but not sure how to fix it.

 

It's PSI or per square inch because it needs to be across the whole surface, minimum, not just one spot. 40lbs per square inch, or 2.58 kg per square CM is pretty significant for the small screws and components that were designed *NOT* to crack the CPU or GPU die. These aren't like desktop chips that have heat spreaders on them, the die can crack easily if too much force is applied.

 

Ah, you're right, I looked it up, 1 mil (which is the unit they used) is 1/1000th of an inch, so 3 mil is 0.076mm.

With the psi that you're questioning, I believe them to literally just mean exactly what they say with that, ie 40 pounds per square inch. So we can work out how much pressure 40psi is in terms of pressure on your CPU die.

My CPU, a 2630QM = 216 square mm
1 square inch = 645.16 square mm
Heatspring requires at least 40psi so: 40*(216/645.16) = 40*(0.3348) = 13lb (or 6kg) of weight required sitting on my heatsink to equate to 40psi.

The best pressure seemed to be about 80psi, so that would mean I'd need 12kg of weight pushing on my heatsink.

I don't know how realistic these forces are for laptop cooling systems? Anyone know?


EDIT: tried to find some standards for heatsink mounting pressure. I found this link here:
http://www.overclockers.com/heatsink-mounting-pressure-vs-performance/
They state in the "Pressure" section of that link when describing the heatsink designed for the Intel ® Core ™2 Extreme Processor QX9775, that: "The attach mechanism for the heatsink developed to support the processor should create a static preload on the package between 18 lbf and 70 lbf throughout the life
of the product."

So, that 18 lbf and 70lbf above is 8kg and 31kg of mounting force for that particular desktop processor. So those are the kind of ranges of force that desktop cooling solutions might apply to their chips. I'm not really sure what that tells us for our laptop cooling systems? I suppose if anything, it tells us that we need about half the mounting pressure of desktop heatsinks to make heat-spring work (when considering the 6-12kg range that I calculated earlier in my post for my particular laptop CPU) . And again, what does that tell us? Maybe not a lot!

 

To put it like this. In force each screw could carry a persons weight. There are 4 taking 2-4 kilo each. Easy. The pitch of the threads i can probably calculate for you but its a gear with ratio. Circle of screw/pitch- friction = force. Add bigger diameter of screwdriver and we can find apx what force on screwdricer is needed to get 2-4 kg. Not very mutch.
problem is as mentioned how much force the die can take. 100 % flat it can probably take some force. The more off you are the more force it gets and the die cracks.

 

I think I agree with you that the screws can create that kind of load that is required (6-12kg), especially when shared between the 4 screws. Yes, the difficult part is modding the mounting solution to create those pressures, and then making sure it's not too much pressure for the chip. Maybe that's the hardest part: knowing how much pressure is too much when you're screwing down the heatsink (after you've modified it for increased pressure). I've heard of the c-clip mod before, but don't know any others.

(Yep, the more unlevel your heatsink when you apply that force, then the more localised the force is on certain areas of the chip & it's those areas of high pressure on those localised areas that can kill it - because the load is not being spread and shared over the whole surface area of the chip).

 

The corner force can be calculated to but it boils down to way smaller area ( corner ) taking many times the force applied on the screws since there are angles and distance from screws into concideration.

 

Yes, I think I know what you mean, it's a very unpredictable scenario.

 

Don't you guys just hate how real life screws up the perfectness of formulae and numbers?

 

I just figured this now......

Take this under consideration. With the force already generated by the heatsink, I chipped my DIE already on both corners. This are bottom left and right corners of the CPU looking from the top on it,

So my investigation tells me that when all screws are screwed all in, then screws 1 and 3 on the heatsink are not doing its job and screws 2 and 4 are pushing just about enough.

This might be because of thermal pad that is on the voltage regulators ( too thick ) but I already tested 1 mm pad instead of 1.5mm that I had fitted normally, with no change.

We need more pressure where the screws 1 and 3 are :d

There is your answer

God I'm good
Sent from my SM-N9005 using Tapatalk

 

I guess if that's the case, then the challenge is just going ahead and doing that. You're a brave soul though, what with your already chipped CPU! Good luck, if you don't push you don't know where the boundaries are! (It's just a pain in the when you find those boundaries & you end up on the wrong side!)

 

Thanks.

Well the cpu still runs @25 on all cores just like it would be new. Like I said, a very little chips

Sent from my SM-N9005 using Tapatalk

 

Second gen mobile maximum allowable normal die load is 15lbs or 100PSI, whichever is lower.

So for 4C max would be ~45PSI and the smaller 2C would be ~65PSI

 

Ah, that's really interesting, and quite useful to know. So, effectively that's going to mean it will always be 15lbs really for our CPUs, which is 6.8kg. That's not particularly good news, because we worked out heat-spring is requiring 6-12kg of pressure, with 6kg theoretically equalling the performance of paste, and 12kg being the ideal for best performance. The 6.8kg that you talk of being the maximum allowable load is not far away from the 6kg minimum, so might be difficult to see increases in cooling performance on these chips without breaking the chips.

Where did you find out it was 15lb? I'd be interested to know the sources, can you link them?

(Yes, the smaller the die size of the chip, then the better these heat springs will work - heatsink able to apply a greater psi at any given force, eg at 6kg)

 

No problem.

http://www.intel.com/content/dam/ww...-guides/2nd-gen-core-mobile-thermal-guide.pdf

 

That's brilliant Dufus, thanks! Do you know if it's the same for Ivy Bridge & Haswell? Plus rep to you!

 

Unfortunately publicly available mobile data is lacking. You might have to approach Intel and request it. Maybe you could get the case temperature specs too, that would be interesting.

 

Ah, I just tried searching for matching documents for 3rd & 4th generation, and I couldn't find a similar pdf document for those generations. I found a load of datasheets for both, but searching for "die load" and "die pressure" didn't yield any matches within those documents, and a quick scan through the document didn't reveal anything either. Strange that they seemed to have provided that info for Sandy Bridge, but not for Ivy Bridge & Haswell. Might be safe to assume the max loadings are similar, as I'd be surprised if laptop manufacturers re-designed their mounting systems from one generation to the next.

Might be interesting to approach Intel for the info, but I might let someone else do that who has more invested in getting heat-spring to work. I'm a casual contributor to this heat-spring thread, not installing it on mine, but I would like to see some people be sucessful in getting heat-spring to work. Anything that can move laptop cooling forward is something I support!