Which o-ring for Kryos on lapped CPU?

[Kayin]

I lapped my i7-860 back when I was doing air cooler reviews as the HDT coolers I had to test were getting poorer temps than stock. As such, it's razor blade/engineering rule flat and makes an excellent impression with TIM subtitute against lapping glass. It's as flat as I could get it without decking it with a flycutter.

So, I find that the Kryos comes with a thick and thin o-ring, and it seems from Vapor's testing that the bowed baseplate works better on an unaltered IHS, but if it's as flat as I can get it wouldn't the stock o-ring actually function better for reducing air gaps and maximizing contact? Or are our temps less a functionality of contact patch and more an issue of contact pressure?

Anyone with a scientific opinion on such wanna weigh in?

[Church]

thick o-ring increases bow and mount pressure -> better contact -> better heat transfer. Imho same should hold true for lapped aswell. Isn't it simpler to just test with both?

[Kayin]

Any changes mean full disassembly of block, but honestly I shouldn't have too many problems with that either.

You have a point, I should just test it myself when I get it assembled (whenever that is.)

[NaeKuh]

pm vapor.

As he's would tell u best what u need to do, as it was his tests which he pulled off to beat all the other blocks.

So ask the source is what im saying.

Vapor is a cool guy too, if get pass the baton.

[theseeker]

You would be better off with the thick o-ring, just like churchy stated.

[[XC] Synthetickiller]

+1 for thick ring.

I think overall, the cooling performance is better with the thick ring. That's how I roll.

[Kayin]

Vapor explained the o-ring situation to me. While the thick O-ring was the correct choice, I wanted to know why. He gave a very clear, direct answer that will mean yes, I will tear apart my block and replace the o-ring.

I understood certain parts of block topography and why we want a perfect surface for mating, but I did not understand completely concepts like a flexible bow or IHS deformation, which are taken into account. He set me straight, and for that I'm thankful.

It doesn't matter to me which way we do things, so long as I know why I'm doing it. I simply wanted to know why the bowed would perform better. Maybe he should post the real answer here-I'm sure I'm not the only person who has wanted to know why.

[warriorpoet]

Vapor explained the o-ring situation to me. While the thick O-ring was the correct choice, I wanted to know why. He gave a very clear, direct answer that will mean yes, I will tear apart my block and replace the o-ring.

I understood certain parts of block topography and why we want a perfect surface for mating, but I did not understand completely concepts like a flexible bow or IHS deformation, which are taken into account. He set me straight, and for that I'm thankful.

It doesn't matter to me which way we do things, so long as I know why I'm doing it. I simply wanted to know why the bowed would perform better. Maybe he should post the real answer here-I'm sure I'm not the only person who has wanted to know why.

I second this motion. All in favor?

[NaeKuh]

I second this motion. All in favor?

LOL i got him on aim to see he can shine in.

I think eric has been busy lately.. and also he's kinda tired of having to do clean up on every last EK thread in this forum.

[Vapor]

Per request, here's what I said in PM.

Why a bow works, even with a perfectly flat IHS:
Because the IHS isn't very good at its job (spreading the thermal energy), having the least thermal resistance over the hottest parts of the CPU* can be (and is) a really big help with temperatures. TIM is a decent contributor to thermal resistance, so you minimize its bondline thickness over the hottest parts of the CPU (with an modern Intel CPU, that's on line drawn down the center). By lowering the contact area, you get significantly more pressure at the right spots without an increase in overall mounting force. Additionally, with most blocks these days (kryos blocks included), the base plate is larger than IHS and the base plate is much weaker, structurally, than the IHS+CPU. You actually can't mate two flat surfaces together and assume they'll both remain flat--the base plate of most CPU blocks will actually flex a little and become concave, really hurting contact (thickest bondline will be over the center, too). So a bow, in most cases, does two things that help contact: ensures the thinnest possible TIM joint is right over the center, and 'preloads' the shape of the base so that, in the worst case, it deforms to be flat.

* Intel buys into the hottest-in-the-center arrangement so much that they don't place the CPU die in the middle of the package, they offset it to one side slightly so the cores are aligned on the center and the cache and I/O is off to the side. On top of that, they actually bow their IHS slightly so that even with their stock coolers, the best best TIM joint is right over the hottest part of the CPU, the center.

As for which o-ring to choose:
Both the stock and thick o-ring provide a bow, just how much of a bow is the difference.
[...from first segment...]
With these things considered, I'd say the thick o-ring is better. But not by much, especially if you use a great TIM (like the included PK-1 or the better Shin-Etsus).

Is lapping a good idea?
I think, generally, lapping a block is a bad idea. Once it's flat, any deformation will make it concave, which is bad. As for a lapped IHS, it might get that last .1c if the block had a just-right bow, but that's kind of a big if.

I think two slightly convex surfaces is easily the safest, with the best either two slightly convex surfaces or a flat IHS with a just-right bowed block. And as the TIM improves, the overall gains from enhanced contact get smaller and smaller.