Vapor Chamber Water Block Question

[prava]

I don't see the point at all. A vapor chamber does nothing with the heat: it only moves it somewhere else. In this regard, its quite effective in gpu cooling because there is not enough space to implement a functional heatpipe-based solution: you don't have space for enough fins because the pipes themselves take a ton of space.

On cpu's a vapor chamber would be useless unless you didn't have any space on top of your cpu. So, with a LC system you would probably lose performance rather than gain it, as you are adding stuff between the water and the processor, and ideally you would put your processor into the water itself

[subtec]

I don't see the point at all. A vapor chamber does nothing with the heat: it only moves it somewhere else. In this regard, its quite effective in gpu cooling because there is not enough space to implement a functional heatpipe-based solution: you don't have space for enough fins because the pipes themselves take a ton of space.

On cpu's a vapor chamber would be useless unless you didn't have any space on top of your cpu. So, with a LC system you would probably lose performance rather than gain it, as you are adding stuff between the water and the processor, and ideally you would put your processor into the water itself

VCs serve to increase the surface area available for heat transfer to occur. Take a look at this pic from the first part of the linked article. Without the VC, only a portion of the heatsink is actually useful in transferring heat to the surrounding medium (air, in this case). It stands to reason that it could allow a significantly larger water block = more surface area = better thermal transfer = lower temps.

Though, as the article shows, the size required for a VC to effectively increase performance might not be practical for a CPU cooling application. Or it might. It seems like an idea worth investigating further, at any rate.

[NaeKuh]

once again.. if the waterblock was limited and not the transfer, it would work.

But we are at a transfer point in waterblocks.

Vapor chambers would prove more surface area on cooling head.
But Bowing Proves it otherwise.

So we went with BOWing instead of Increasing the cooling head area.

[BeepBeep2]

once again.. if the waterblock was limited and not the transfer, it would work.

But we are at a transfer point in waterblocks.

Vapor chambers would prove more surface area on cooling head.
But Bowing Proves it otherwise.

So we went with BOWing instead of Increasing the cooling head area.

Don't know how many times I've said this, but a lot of CPU IHS'es are not flat, they are also convex. The bowing idea works well with flat surfaces, the bases are thin and flex under mounting pressure. Take a flat base and mount it with considerable pressure and you end up with a concave base, more contact on the outside than in.

I've had three CPU's to test this on, unfortunately all AMD CPU's, but nonetheless. I owned a 965 Black Edition, (now sold to a happy buyer) and I was not happy with the surface area of the block on the CPU. Only about 40% of the IHS had paste on it indicating contact with the block. I ended up lapping this CPU to 1000 grit, flat, and temperatures both idle and load dropped by 4c. I had perfect contact after that throughout the whole IHS. Its called a heatspreader for a reason.

I also have a 565 Black Edition and 1100T Black Edition. My 565 BE at stock, runs almost the same idle temp as my 965 BE, my 965 BE idled at 28c after being lapped, 32 before, my 565 BE idles at 27c. All of the contact I get on my 565 BE is right at the center, (maybe 50% of the CPU, I can still read the date code) even though I use about twice as much paste as I feel comfortable with as well as a much harder mount than is recommended. The 1100T is the same way, and I'm idling at temps with it higher than some guys on air, as well as have load temps that are pretty ty tbh.

The bow is there so the block equalizes flat under pressure, I'm sure you know that when you have a base this thin the copper flexes quite easily.

Your idea makes no sense, considering there is inherently heat on the parts of the IHS that are not even contacted with the block. You would be better off aiming for a FLAT mount (hint, not flat base, flat mounted) and more pressure to equalize whatever pressure you had in the middle to begin with over the whole surface area.

[Vapor]

once again.. if the waterblock was limited and not the transfer, it would work.

But we are at a transfer point in waterblocks.

Vapor chambers would prove more surface area on cooling head.
But Bowing Proves it otherwise.

So we went with BOWing instead of Increasing the cooling head area.

The bow is there to improve contact where it's most important. Bows are implemented because we use thermal paste that is, typically, a couple magnitudes of performance away from being near-perfect. Bows are one way to 'solve' the transfer from the IHS into the copper. Using a near-perfect TIM is another way (CLP, CLU, IX).

The water block is limited by the abilities/properties of copper and surface area *where it matters.* (1)

A VC would increase the size of the "where it matters zone" and allow manufacturers, without shrinking their tooling, to increase useful surface area. A VC would be potentially useful for improving performance of the block, regardless of TIM.

(Without a bow, near-perfect TIMs still get near-perfect performance when normal pastes don't even come close.)

That said, it seems like VCs, or at least the ones examined in the write-up subtec provided, are not ideal for WC--we keep temperatures of the IHS and base of our blocks way too cool for them to be efficient. Maybe one with a lower internal pressure or a different coolant (not the right word, but replacing water in the VC) would have better efficacy at lower temperatures.

A bow would still be ideal where the VC meets the IHS.

[Boogerlad]

Creating a vacuum inside of the vc using water will lower the boiling point. This could raise efficacy. I had this idea in a thread of mine a while ago, but no tools to actually make haha.

[NaeKuh]

Eric we showed a long time ago, its not cooling area that matters but location.

This is why we kept telling people the location under the IHS directly is the important part, and all the excess around will give you almost no benifit.

The transfer at which heat is moved from die -> ihs -> waterblock is already at the limiting factor.
Adding a IHS -> VC -> Waterblock would only add a extra thermal layer.

If the waterblock was limited at picking up heat, then id be all over your statement saying yea.. it works.. but when working with how much heat the IHS will allow to be transfered from die system, i dont see how adding a VC would improve anything.

Were at the mercy of the IHS in short.
Without fixing the IHS so its more compatable with a VC, i dont see any gain in this.

Creating a vacuum inside of the vc using water will lower the boiling point. This could raise efficacy. I had this idea in a thread of mine a while ago, but no tools to actually make haha.

yeah but once again.. the moment it got to the block, that is the limiting factor.
If the block could do as you guys are saying, then we would see better spreads off the top blocks.
But we dont.. because were at the mercy of what the IHS will allow off the DIE, and the location of such.

As i said, were adding another layer on a part which is the handicap.
Without fixing the handicap, i dont see how this technology would work at all.

[bamtan2]

2012 bump

We're at the mercy of the IHS in short.

so now that ivy bridge is tiny, hot, and ready, who will be the first to pop the IHS and increase the surface area?

[thegoatman]

Umm, isn't vapor buildup between coolant and the object being cooled a BIG problem in conventional liquid cooling applications, I.E. Thermonuclear reactors and automobiles?

[bamtan2]

Umm, isn't vapor buildup between coolant and the object being cooled a BIG problem in conventional liquid cooling applications, I.E. Thermonuclear reactors and automobiles?

what the hell? I think you stumbled into the wrong thread.

[wez]

2012 bump
so now that ivy bridge is tiny, hot, and ready, who will be the first to pop the IHS and increase the surface area?

How would you increase the size of the core?

[zeroibis]

Unfortunately all the possible gains are lost due to the amount of exchanges the heat would need to go though. However it could be possibly to reduce that in the future with better materials or design.