TEC1-12740 direct die thermal performance at various voltages (Skylake 6700K 4.8GHz)

cypres · Nov 24, 2015

Quote:

Originally Posted by LiamG6

you should look into de8auers delid die mate, its a tool for safe and foolproof delidding/relidding, looks fantastic and shouldn't be that expensive, can be used for Ivy/Haswell/Skylake and most likely future mainstream chips.

my method to delid is to clamp CPU and IHS in a clamp, one side on substrate, the other side on IHS, apply a slight amount of tension, then get a hair dryer and heat the IHS, increase the clamp tension a bit then keep going with that method until it "pops" ie glue will release the IHS and it will pop over a mm or two and then just peal the IHS off. less likely to damage anything than with razor/hammer/or plain vice method, but not as good as de8auers tool.

you really need to delid and lap to get a better result

the fact that your cold side temp only rises 9*c at load is interesting, even if actual temp is wrong the delta between load and idle should be right, so say you might be at an actual 0-10*c cold side temp. or -10*c to 0*c at best

What is your CPU core temp with CPU in idle?

I'll look into it for sure, I've seen the tool but I'm just kinda nervous since this is my second skylake

Core temp at idle stays at zero, not really sure how far under that it goes

LiamG6 · Nov 24, 2015

I think between the stock TIM on the die, the concave IHS, possibly a bad waterblock mount on CPU and not enough torque on TEC you're probably losing 30*c with loaded CPU temps. Maybe more.

cypres · Nov 24, 2015

Quote:

Originally Posted by LiamG6

I think between the stock TIM on the die, the concave IHS, possibly a bad waterblock mount on CPU and not enough torque on TEC you're probably losing 30*c with loaded CPU temps. Maybe more.

Yeah I agree. I'll lap the die probably tomorrow or Friday when my new TIM comes in and give that a shot first. Yesterday I ended up removing much of the foam that could have been obstructing the block - it used to be a compression type fit but now there is a gap so the only thing the block touches is the cpu, and the edges of the block are sealed with mounting putty. Now that I look back on it I don't think my foam was a problem at all, it was probably always the concavity of the IHS and stock TIM causing my issues.

cypres · Nov 25, 2015

Quote:

Originally Posted by LiamG6

I think between the stock TIM on the die, the concave IHS, possibly a bad waterblock mount on CPU and not enough torque on TEC you're probably losing 30*c with loaded CPU temps. Maybe more.

Whelp, good news is the cpu has been delided, IHS lapped and polished and TIM replaced (trying Gelid Extreme for now), bad news is that I might have killed another mobo. I bent a couple of pins when I was goofing with it, fixed a few and could get it to post but PCIe bus isn't right, first x16 slot not recognized and second x8 will boot into Windows but acts flaky. Instead of messing with it and risking my cpu I think I decided to just replace the board to be safe.

Problems actually started before the delid so that's untested as of now. I used the razor method and verified that I didn't get any nicks but idk if the shorted pins could have killed anything. We'll see, maybe I'll get another $400 keychain out of it

LiamG6 · Nov 26, 2015

ahhhhhhhh. careful man, money doesn't grow on trees, if you are really careful you can bend those pins back into place. did you have dielectric grease in the socket on the pins? if you didn;t have dielectric grease in it could have corroded socket already. if you did have grease in, then it is very easy to bend pins without noticing before you clamp CPU in, I was extremely careful when I applied to the grease to my socket and I still bent a pin, luckily I found it before I clamped CPU and bent it back into position before it got totally stuffed.

you're a brave man doing the razor method, I wouldn't risk it, but I guess with skylake and no FIVR capacitors on the substrate and a thinner substrate the other methods might be more likely to damage it. I still think vice with slight pressure and hair dryer is the safest option aside from that delid tool.

I hope you can get it all going again.

at the very least you need to run with Coolaboratory Liquid Ultra on the die, it makes a huge huge difference, gelid extreme is fine on the IHS, but CLU will show a big benefit over gelid on the die.

cypres · Nov 26, 2015

Quote:

Originally Posted by LiamG6

ahhhhhhhh. careful man, money doesn't grow on trees, if you are really careful you can bend those pins back into place. did you have dielectric grease in the socket on the pins? if you didn;t have dielectric grease in it could have corroded socket already. if you did have grease in, then it is very easy to bend pins without noticing before you clamp CPU in, I was extremely careful when I applied to the grease to my socket and I still bent a pin, luckily I found it before I clamped CPU and bent it back into position before it got totally stuffed.

you're a brave man doing the razor method, I wouldn't risk it, but I guess with skylake and no FIVR capacitors on the substrate and a thinner substrate the other methods might be more likely to damage it. I still think vice with slight pressure and hair dryer is the safest option aside from that delid tool.

I hope you can get it all going again.

at the very least you need to run with Coolaboratory Liquid Ultra on the die, it makes a huge huge difference, gelid extreme is fine on the IHS, but CLU will show a big benefit over gelid on the die.

Yeah, it really sucks

I was using grease, just a light coat on the pins and that gap in the socket as well as on the pads of the cpu and covering the SMD components. No signs of corrosion at all this time. I did pull the board and had it under my bench magnifier and got most of them straightened out but there is one that is twisted in an odd way and looks like it's shorting on the piin beside it when it compresses. It was just out of frustration that I ordered the new board I think, I was going cross eyed from staring at those pins

I practiced on the old one first which is why I just went for it, its actually pretty easy once you get the razor into the goop. The new one was a bit harder because I had some trace silicone on the PCB part that I couldn't get off and I had to push the razor through that also otherwise the angle would have been off. But it looked pretty good. Also on the old one I practiced the vice method and wasn't happy with how the edge of the PCB was looking after a few hits with a wood block (could tell it was stressed), and I also tried some solvents in the name of science but I didn't have anything that would dissolve it. It's almost certainly silicone based it seems.

I might put in an order for liquid ultra, but the thing is I already resealed my IHS (I was concerned more about condensation getting inside if I let it float). What I used to seal it should be much easier to remove the IHS but it's quite firmly attached to the PCB bit so I'd have to worry about either leaving the old stuff on and not resealing it or adding a bit more silicone and risking having too much gunk pushing back on the IHS. If my temps aren't much better with the Gelid I'll figure something out I suppose

LiamG6 · Nov 26, 2015

the socket really needs to be packed full of grease, and I mean full, otherwise moisture will still form in there and corrode the pins or cause shorts. like this http://www.overclock.net/t/1526237/lightbox/post/23184457/id/2259909

this is the delid method I'm talking about,

no wooden blocks or hammers and no pressure on the substrate really, just slight pressure, and heat, worked extremely easily for me, in about 2 minutes I was done with almost zero chance of damage.

I actually lapped the bottom lip of my IHS so it sat lower on the die so that I could get a thinner layer of CLU and also have room to glue the IHS back with some liquid electrical tap, worked like a charm. Don;t go to thin or you might crack your die., just a mild lap, I lapped the bottom lip until I could see an even ring of copper, that was perhaps slightly too much but it is working fine.

Elloquin · Nov 26, 2015

Just put a big blob of grease in the socket then use a hairdryer to heat it up and it will flow into the socket.

cypres · Nov 27, 2015

Quote:

Originally Posted by LiamG6

the socket really needs to be packed full of grease, and I mean full, otherwise moisture will still form in there and corrode the pins or cause shorts. like this http://www.overclock.net/t/1526237/lightbox/post/23184457/id/2259909

this is the delid method I'm talking about,

no wooden blocks or hammers and no pressure on the substrate really, just slight pressure, and heat, worked extremely easily for me, in about 2 minutes I was done with almost zero chance of damage.

I actually lapped the bottom lip of my IHS so it sat lower on the die so that I could get a thinner layer of CLU and also have room to glue the IHS back with some liquid electrical tap, worked like a charm. Don;t go to thin or you might crack your die., just a mild lap, I lapped the bottom lip until I could see an even ring of copper, that was perhaps slightly too much but it is working fine.

One thing though is you need to be a little careful about using too much grease. I suspect that contributed to the failure of my last Skylake was just packing the socket until it squeezed out the edges when you clamped the chip down. I could see on the chip itself where the corners were bent up slightly, probably from a combination of lots of grease and clamping down the block too tightly.

And I really like that delid method.. if the tool was more available/less expensive I'd probably go that route, but it looks like I got lucky this time. I've heard a lot of people have bricked their processors doing it so yeah, probably not worth the risk, but I get impatient

Oh and I did order 2 tubes of Coolaboratory Liquid Ultra and the cleaning kit, and also great advice on sanding the lip of the IHS. I went ahead and did that earlier and also polished inside the IHS a bit by hand with jeweler's rouge. The nickel layer is still there but it's a mirror finish now. The top of the IHS is down to the copper and polished as well. Will probably use the Liquid Ultra on both the die and IHS to cold plate and switch both sides of the TEC to Gelid Extreme.
Quote:

Originally Posted by Elloquin

Just put a big blob of grease in the socket then use a hairdryer to heat it up and it will flow into the socket.

That's an awesome idea actually, never thought of that. I'll try that when I get my parts in

LiamG6 · Nov 27, 2015

hairdryer to melt the grease is my lesson of the day, thanks ello, don't have to freak myself out when smearing it on the pins anymore

yeah the vice/hairdryer method is so good, I don't know why the razor or vice and hammer method became so popular when they both have risks.

nothing more satisfying than a lapped and polished IHS and cold plate. between that and the CLU you should see 10-20*c less core temps. and yes CLU on DIE>IHS and IHS>Cold plate is what I have, and I'm using Thermal Grizzly Kryonaut on each side of TEC which is very similar to Gelid GC extreme in performance and also being a right pain to apply

cypres · Nov 27, 2015

Quote:

Originally Posted by LiamG6

hairdryer to melt the grease is my lesson of the day, thanks ello, don't have to freak myself out when smearing it on the pins anymore

yeah the vice/hairdryer method is so good, I don't know why the razor or vice and hammer method became so popular when they both have risks.

nothing more satisfying than a lapped and polished IHS and cold plate. between that and the CLU you should see 10-20*c less core temps. and yes CLU on DIE>IHS and IHS>Cold plate is what I have, and I'm using Thermal Grizzly Kryonaut on each side of TEC which is very similar to Gelid GC extreme in performance and also being a right pain to apply

Just finished switching the TEC to Gelid, a right pain to say the least. At least it spreads a little easier on the ceramic than metal but still... ugh...

Looks like I should have the new board and CLU early next week. starting a new job though so I might not have an update until next weekend

TEC1-12740 direct die thermal performance at various voltages (Skylake 6700K 4.8GHz)

cypres

LiamG6

cypres

cypres

LiamG6

cypres

LiamG6

Elloquin

cypres

LiamG6

cypres

Top Contributors this Month

Recommended Communities