Cooling block design change?

[OldChap]

Whilst this may seem that I am having a blonde or senior moment, I would invite the thermally aware among you to pick holes in the following idea.

Currently, cooling blocks are far larger than skt 1155 ihs and I can see the advantages of this when considering barbs etc.

A number of Ivy Bridge processor matings to such blocks have shown me that the ihs is far from flat (the corners and edges seem high)

Ivy Bridge seems to have a more concentrated heat dissipation path through ihs

Would there be any advantage/disadvantage in building a block that was stepped so that whilst the overall size could remain the same the contact area could be sized to fit within the periphery of the ihs thereby obviating the need to lap the corners and edges from the processor ihs. the step would not need to be huge, perhaps a few thou to overcome ihs edge unevenness.

If the die size is 37.5mm x 37.5mm then I am proposing the contact area be 30mm x 30mm approx.

I feel that whilst this might reduce the total block to cpu contact area it would have a better chance of being a closer fit to the ihs and spring pressure would be more concentrated also resulting in thinner layers of tim so better heat transfer.

Your thoughts on this are invited

[Martinm210]

Stepping is an alternative to bowing and it does help over a flat block but bowing is better. I think there were a couple of blocks that came with stepped bases back in the days where the fuzion v1 ruled which was a couple of years ago. Bowing just ensures the maximum pressure is focused on the center better where stepping can still have some issues with pressure along the perimeter of the step. Bowing works kind of like camber in a bridge too in that it anticipates the bending of the block under load a bit as well. When mounting with 50+lbs the blocks do bend a bit.

Stepping is also getting difficult nowdays with the extremely thin bases. Just not much meat left on the bone to play with.

It is something to consider, but I think bowing is a bit better. If anything, I would like to see more blocks come with bow "tuning" options such as various oring thicknesses or shims so you can better tune the block toward the platform needs. I have seen both done but most current gen blocks have only one fixed bow amount.

[AVulgarPicture]

Wasn't one of the Enzo blocks stepped?

[eXa]

Fuzion doesnt come with a step, but bow.

[Martinm210]

I know, I was talking about timeframe when I remember seeing it.
It was the OCZ hydroflow that had the stepped base:


I think it was Niksub1 that started the stepping idea with his block testing back in the days if I recall.

[AVulgarPicture]

the OCZ. thats the one i was picturing in my head.

[nikhsub1]

Whilst this may seem that I am having a blonde or senior moment, I would invite the thermally aware among you to pick holes in the following idea.

Currently, cooling blocks are far larger than skt 1155 ihs and I can see the advantages of this when considering barbs etc.

A number of Ivy Bridge processor matings to such blocks have shown me that the ihs is far from flat (the corners and edges seem high)

Ivy Bridge seems to have a more concentrated heat dissipation path through ihs

Would there be any advantage/disadvantage in building a block that was stepped so that whilst the overall size could remain the same the contact area could be sized to fit within the periphery of the ihs thereby obviating the need to lap the corners and edges from the processor ihs. the step would not need to be huge, perhaps a few thou to overcome ihs edge unevenness.

If the die size is 37.5mm x 37.5mm then I am proposing the contact area be 30mm x 30mm approx.

I feel that whilst this might reduce the total block to cpu contact area it would have a better chance of being a closer fit to the ihs and spring pressure would be more concentrated also resulting in thinner layers of tim so better heat transfer.

Your thoughts on this are invited

Yeah, you are 5 years late http://www.xtremesystems.org/forums/...nd!&highlight=

Anyway, the latest is that Ivy Bridge IHS's are put on with a crappy TIM and not soldered. I think the STORM on a delidded Ivy Bridge would be best cooling available. Don't dare put a bowed block on a bare die.

[NKrader]

Don't dare put a bowed block on a bare die.

just tighten it down enough. :p

[AVulgarPicture]

and crack the cores lol.

[Church]

some might try to work out process of removing IHS, applying some liquid metal or IX & reinstalling back. I predict very high chance of fail though.
EDIT
on second thought careful partial removal of bow via lapping of modern block + very low mount pressure and LM or IX kind of TIM (to solve worse contact because of less pressure) might be better choice imho then reinstalling IHS or using prehistoric blocks.

[AVulgarPicture]

Almost certain fail.

[Kayin]

I'm not certain calling a Storm a prehistoric block is proper. it was designed for bare dies, something no block we have now is designed for-so if you took a delidded Ivy, and put a block from now on it, A. you'd most likely kill the CPU because of the bow and B. it wouldn't perform as well. A purpose-built bare die block should beat it every time.

[OldChap]

some might try to work out process of removing IHS, applying some liquid metal or IX & reinstalling back. I predict very high chance of fail though.
EDIT
on second thought careful partial removal of bow via lapping of modern block + very low mount pressure and LM or IX kind of TIM (to solve worse contact because of less pressure) might be better choice imho then reinstalling IHS or using prehistoric blocks.

Pure Indium seems to have all the heat transfer qualities but not the required lower heat to flow it. Indium/silver lower but not so great transfer...????

There could be some mileage in only lapping the edges of the cpu but that would require building a special rig too.

nikhsub1; Yeah I'm old and slow but now and again I get these brain storms.... It seems I am not the only one

[nikhsub1]

I'm not certain calling a Storm a prehistoric block is proper. it was designed for bare dies, something no block we have now is designed for-so if you took a delidded Ivy, and put a block from now on it, A. you'd most likely kill the CPU because of the bow and B. it wouldn't perform as well. A purpose-built bare die block should beat it every time.

Yay! Someone with a brain that knows how to use it instead of following the heard like sheep. Good to see you Kayin.

nikhsub1; Yeah I'm old and slow but now and again I get these brain storms.... It seems I am not the only one

Nope, you are not the only one. Who is old

[Church]

Hmm, you guys wouching for Storm like blocks .. please correct me where my line of thought gone wrong and in what exactly way i'm not using brain and follow heard:
- Most probably it might be well fit for lid-less CPUs because of less mount pressure, that has lesser chance of chipping core (from my memory though in that era there were cases of chipping with careless mounts even with stock coolers designed with similar mounting properties);
- Modern blocks with modified mounting sys for less pressure should do same AND have better cooling internal structure.

I'm too lazy to search for old tests that might have happen to have also at least distantly modern blocks aswell in it, but my guess that with storm cooling might be at least 5C worse, which is a lot for LC. That's how i think in totally brainless way.

[nikhsub1]

Hmm, you guys wouching for Storm like blocks .. please correct me where my line of thought gone wrong and in what exactly way i'm not using brain and follow heard:
- Most probably it might be well fit for lid-less CPUs because of less mount pressure, that has lesser chance of chipping core (from my memory though in that era there were cases of chipping with careless mounts even with stock coolers designed with similar mounting properties);
- Modern blocks with modified mounting sys for less pressure should do same AND have better cooling internal structure.

I'm too lazy to search for old tests that might have happen to have also at least distantly modern blocks aswell in it, but my guess that with storm cooling might be at least 5C worse, which is a lot for LC. That's how i think in totally brainless way.

The Storm was essentially the last best block designed for bare die cooling... once the Swiftech Apogee came out, everyone move to making blocks based on cooling IHSes - and for good reason, they were soldered on and were here to stay. It was discovered that bowing a base would aid in cooling performance - there have been a gazillion discussions on this so I won't go into it now. Bottom line is, modern blocks are designed with IHS in focus so they won't theoretically work well on a bare die. Of course, this would all need to be tested - and my days of testing are done. There have been countless fights over IHS and non IHS based cooling and how to test in the past and I don't wish to dig that up. My only point here is that I still think a Storm (or a G5) would cool a delidded cpu better than any current block on the market.

*edit - to directly respond to your question: Mount pressure with an IHS can essentially be infinite - modern blocks with bows accentuate the pressue exerted on the IHS, you would not want uneven block/die mating that a bow would cause on a bare die. You would want an absolutely flat base like the storm has for a bare die -bare dies can still take some good mount pressure so long as the base is aligned with the die and not skewed as that would crack the die.

You are correct about the Storm being 5C worse in tests - WITH IHS CPU's! The storm was never designed to cool with an IHS in place - My bet is that without an IHS the storm would wipe the floor with modern blocks.

[Kayin]

It's good to see you again too. My health isn't the greatest, but I'm still sticking around-and I have a few tricks up my sleeve yet. Anyone successfully delidded a SB? Mine runs wayyy too hot for my liking. Currently working on a setup to liquid cool it, but it's idling in the 60s C. I'm pretty sure if I did the razor blade test, it would have daylight showing.

I've got an idea kicking around here somewhere, just need to get some time to work on it. Finally got access to machining tools, we'll see what comes of it.

[OldChap]

Think I might cut up one of the jet plates for the HF and just drill a nice hole over the cores for concentrated and higher flow. I can get this done before my 3770K arrives and for the cost of a plate .... nothing to lose.

If I have to lap it I will, once I see that I have some maximums while in warranty

[CrazyNutz]

I know, I was talking about timeframe when I remember seeing it.
It was the OCZ hydroflow that had the stepped base:


I think it was Niksub1 that started the stepping idea with his block testing back in the days if I recall.

LOL. I have this block (it was a freebee) on a 2500k, It has mediocre performance at best.

[Waterlogged]

I agree with nik about the Storm being the last great direct die mount block but, I can't help but think the D-Tek FUzion could be a good option for delidded IB CPU's, as long as they haven't been previously bowed as it offers some interesting tweaking options.

[nikhsub1]

I agree with nik about the Storm being the last great direct die mount block but, I can't help but think the D-Tek FUzion could be a good option for delidded IB CPU's, as long as they haven't been previously bowed as it offers some interesting tweaking options.

Yes it would but Fuzions are bowed by design... It would need to be lapped.

[Waterlogged]

Yes it would but Fuzions are bowed by design... It would need to be lapped.

, you are of course correct. I checked a brand new V2 last night before posting but apparently didn't have enough light when I did, just re-checked and saw that it is indeed bowed ever so slightly. Still, if I wind up getting an IB down the road, I think I'll take the risk of popping and lapping.

[nikhsub1]

Still, if I wind up getting an IB down the road, I think I'll take the risk of popping and lapping.

Of course you would, you are one of the smart ones

[PatRaceTin]

many people got IVY overheat!!!

[AVulgarPicture]

has anyone actually posted a delidded watercooled Ivy yet?