Direct-die WCing! Cheap and easy.

[drizzt5]

My god, you're brave.

You should try suggesting this to microsoft and see what they say. Although you should see if the idea is patented already.

This could make your career if it works.

You mean intel? Why would he send it to microsoft?

[[XC] riptide]

Need results.!

[Hondacity]

My god, you're brave.

You should try suggesting this to microsoft and see what they say. Although you should see if the idea is patented already.

This could make your career if it works.

intel engineers have been working on direct die cooling for some time, but they're not doing water, they're doing tec. they know how hot this cpu is. direct water is gonna be ummmmm we'll see the results...

[fallwind]

Sorry for the delays guys, I set up a temp loop to leak test the block and discovered a microscopic leak between the block and IHS. I don't know how it happened since I wasn't exactly skimpy with the epoxy. Anyways, it may have been for the best since I managed to remove the block fairly easily. The epoxy I used wasn't so great afterall.

I took this opportunity to pump some silicone into the IHS and hopefully seal the wafer from the water chamber. Maybe not necessary but it can't hurt. After that dried I cleaned up the mating surfaces and stuck it together with JB Weld this time. Gotta let that cure for 24 hours before another leak test.

You should consider narrowing the entrance of the water in order to accelerate it at the entrance of the "block" or just before the die. As more speed you get there, more performance you will have.

There's a 3/8" OD Swiftech plastic hose barb screwed to the underside of the inlet inside the block. You can sorta see it through the exit hole it in the last pic I posted. I trimmed the barb so it terminates about 1/4" away from the core. Trust me, the core will get scrubbed hard by the incoming water. I have a D5 with Detroit top powering the loop for plenty of pressure.

In a tradition WC setup, you have 4 layers of material the heat has to transfer through to reach the water (core>>solder>>IHS>>TIM>>waterblock>>water) Here it will be straight from the core to the water. I see no possible way that this won't be a HUGE improvement.

[SoulsCollective]

In a tradition WC setup, you have 4 layers of material the heat has to transfer through to reach the water (core>>solder>>IHS>>TIM>>waterblock>>water) Here it will be straight from the core to the water. I see no possible way that this won't be a HUGE improvement.

Surface area. That die is damn small and has no crenelations, fins, pits, grooves, anything to increase surface area. Although you are removing all those possible sources of inefficiency in thermal transfer, which should count for a pretty significant temp decrease, it's oversimplifying things to say that there is no possible way this could not work.

[fallwind]

I see what you are getting at. In a normal cooling setup, the heat is transferred to a larger area before being absorbed by the water. I looked it up and water has a thermal conductivity of 0.6, copper is 400. So copper is definitely better at transferring heat than water. But all that is meaningless when you have to use a TIM that has a thermal conductivity 3. http://en.wikipedia.org/wiki/Thermal_conductivity Maybe the ideal solution would be to solder the core directly to the waterblock? I don't know, maybe someone with a background in thermodynamics can pop in here and explain the science behind it.

[SoulsCollective]

I see what you are getting at. In a normal cooling setup, the heat is transferred to a larger area before being absorbed by the water. I looked it up and water has a thermal conductivity of 0.6, copper is 400. So copper is definitely better at transferring heat than water. But all that is meaningless when you have to use a TIM that has a thermal conductivity 3. http://en.wikipedia.org/wiki/Thermal_conductivity Maybe the ideal solution would be to solder the core directly to the waterblock? I don't know, maybe someone with a background in thermodynamics can pop in here and explain the science behind it.

It's not meaningless. The TIM, if applied properly, is only filling in the microscopically small pits and grooves on the surface of the waterblock and IHS - over 80%+ of the area, you should be getting direct metal-on-metal contact.

[hellcamino]

I see what you are getting at. In a normal cooling setup, the heat is transferred to a larger area before being absorbed by the water. I looked it up and water has a thermal conductivity of 0.6, copper is 400. So copper is definitely better at transferring heat than water. But all that is meaningless when you have to use a TIM that has a thermal conductivity 3. http://en.wikipedia.org/wiki/Thermal_conductivity Maybe the ideal solution would be to solder the core directly to the waterblock? I don't know, maybe someone with a background in thermodynamics can pop in here and explain the science behind it.

In a normal water cooled setup you would be transferring heat from die to solder to IHS to TIM to copper to water to copper (or aluminum) and finally to air.

You removed the first 4 from the equation, I think this is one of the best experiments I have ever seen! With the flow you are planning for this I think it will work well and am looking forward to seeing it's conclusion.

[rge]

I see what you are getting at. In a normal cooling setup, the heat is transferred to a larger area before being absorbed by the water. I looked it up and water has a thermal conductivity of 0.6, copper is 400. So copper is definitely better at transferring heat than water. But all that is meaningless when you have to use a TIM that has a thermal conductivity 3. http://en.wikipedia.org/wiki/Thermal_conductivity Maybe the ideal solution would be to solder the core directly to the waterblock? I don't know, maybe someone with a background in thermodynamics can pop in here and explain the science behind it.

The surface area matters and is one negative, but still I am very interested to see what you get....too many variables to predict.

The best way to get good temps would be if intel designed an IHS that was a water block complete with milled water channels that was attached via solder tim to die. Or, if someone could get ahold of solder tim and attach one directly to die...but intel has years of refining the attachment process.

Normal water cooling = die (with very local hot spots at conductance of ~120 W/M*K) >> STIM (87 W/M*K) >> copper IHS (which increases surface area many times as the ideal thickness of it spreads the hot spots in addition to increase size at cond. of 400 W/M*K) then TIM (2-4 W/M*K) then copper with milled channel to increase surface area to water (.58 W/M*K) then to rad

You will be die with local hot spots...very small surface area, smaller than the die as hot spots will be cooled directly by water with thermal cond of .58 W/M*K.

Too many variables to guess, but one of the 2 things I am interested in seeing tried.

The other you said, which I would expect to be the most effective would be...die (120 W/M*K) >> solder tim (87 W/M*K) >> IHS water block (400 W/M*K which bottom part both spreads the hot spots from die increasing the surface area many more times than just increase in size, and is milled on inside to increase the surface area for transfer of heat to water (.58).

[moiraesfate]

Oops, sorry. Yeah, it should be Intel. I was at work when I wrote that and doing a bunch of other things too lol.

[Mech0z]

If you didnt lap your cpu and then instead cut channels in the surface, wouldnt that work?

[Sadasius]

In a normal water cooled setup you would be transferring heat from die to solder to IHS to TIM to copper to water to copper (or aluminum) and finally to air.

You removed the first 4 from the equation, I think this is one of the best experiments I have ever seen! With the flow you are planning for this I think it will work well and am looking forward to seeing it's conclusion.

What is important is the thermal dynamic pump. As long as heat is being transfered quickly enough and out of the system to the surrounding air. Eliminating steps does not always equate to a better performance in the thermal pump process. It is an interesting experiment nonetheless. I am anxiously awaiting the results.

[rge]

If you didnt lap your cpu and then instead cut channels in the surface, wouldnt that work?

Yes, but then you would have to attach/weld a bottomless waterblock to the IHS, so the IHS forms the bottom, assuming the channels could be cut deep enough to be effective, not to mention be pretty difficult, but would be interesting if someone would do that.

[paul_]

need to see these results. I'll buy one of these babies off you for my Xeon, it's dies get hot. May have to alter it to work properly with dual dies tho...

[Sparky]

Make sure to switch that pump on first and get water flowing before kicking the PC power on. With an IHS and waterblock a second or so delay of the pump turning on isn't an issue, but with nothing but water I'd think you'd need the water moving right away.

[Mech0z]

Yes, but then you would have to attach/weld a bottomless waterblock to the IHS, so the IHS forms the bottom, assuming the channels could be cut deep enough to be effective, not to mention be pretty difficult, but would be interesting if someone would do that.

Isnt that what this is about? Else I have misunderstood something, I see water touching the IHS, did I see wrong?

[Movieman]

Ok, got my popcorn and waiting to see the results..

[Nightstar]

Surface area. That die is damn small and has no crenelations, fins, pits, grooves, anything to increase surface area. Although you are removing all those possible sources of inefficiency in thermal transfer, which should count for a pretty significant temp decrease, it's oversimplifying things to say that there is no possible way this could not work.

That same damn small die is just as small when in contact with a copper water block. There is no way to increase the die size post manufacture. Adding material to the top of the die will only make it transfer heat less efficiently.

The only positive thing the block lends to the equation is a greater thermal mass than an equivalent volume of water. Flow and pressure will be the limiting factors in the capabilities of hot-side heat exchange in this direct die design.

[paul_]

Isnt that what this is about? Else I have misunderstood something, I see water touching the IHS, did I see wrong?

Not quite - the IHS has a hole in it with a retrofitted nozzle shooting water DIRECTLY ONTO THE CPU DIE
Though the former does sound like a very economical, smart wayto do things, if you could make a grooving system or something similar from the IHS, or maybe make a new IHS that is thicker with a swiftech-style pin matrix.

[paul_]

Surface area. That die is damn small and has no crenelations, fins, pits, grooves, anything to increase surface area. Although you are removing all those possible sources of inefficiency in thermal transfer, which should count for a pretty significant temp decrease, it's oversimplifying things to say that there is no possible way this could not work.

Except that it has worked - there is a thread floating around here that I saw a few days ago where someone in fact had already done this - his temps, if my memory serves me correctly, were mid-30s full load, over 4GHz. It just comes down to how much water is moving over the die - Over time, the surface area of water that is exposed to the die over, say, a minute, is astronomically higher than the surface area of the TIM/solder/IHS, which has a much higher thermal conductivity, but a static surface area.

[Mech0z]

Not quite - the IHS has a hole in it with a retrofitted nozzle shooting water DIRECTLY ONTO THE CPU DIE
Though the former does sound like a very economical, smart wayto do things, if you could make a grooving system or something similar from the IHS, or maybe make a new IHS that is thicker with a swiftech-style pin matrix.

Tbh I dont think just spraying water onto the die is going to get good temps tbh :/ channels will be needed, but I might be wrong, but I think it would be better to make channels in the IHS.

[hellcamino]

The arguing over the validity of this idea in this thread cracks me up, I just hope the OP doesn't decide not to share their results due to the arguing from people who clearly have no idea what they are talking about! To the nay sayers: go come up with something innovative.

If direct die cooling was a bad idea than no one would ever remove the IHS to cool it, all that achieves is the removal of a couple layers of material to transfer heat through.

[Mech0z]

The arguing over the validity of this idea in this thread cracks me up, I just hope the OP doesn't decide not to share their results due to the arguing from people who clearly have no idea what they are talking about! To the nay sayers: go come up with something innovative.

If direct die cooling was a bad idea than no one would ever remove the IHS to cool it, all that achieves is the removal of a couple layers of material to transfer heat through.

Yes you get rid of some, but you get less area to transfer the heat, and a flat plate is not good at transferring heat compared to a rilled plate, but as I said might be wrong.

[rge]

That same damn small die is just as small when in contact with a copper water block. There is no way to increase the die size post manufacture. Adding material to the top of the die will only make it transfer heat less efficiently.

The only positive thing the block lends to the equation is a greater thermal mass than an equivalent volume of water. Flow and pressure will be the limiting factors in the capabilities of hot-side heat exchange in this direct die design.

If water's thermal conductance was 80 W/M*K then I would completely agree. But water has thermal conductance of .6 or 130x less than solder tim and 667x less than copper.

The IHS increases the surface area from size of hot spots on die which are a fraction of even the die size and spreads the heat along entire size of the IHS using high thermal conductance of 130-667x greater than water. Then with the dramatically increased surface area with more copper and milled channels you suffer the piss poor transfer rate of heat from metal to water...once you have enough surface area to do so.

That is very different than having a very small surface area trying to transfer same heat using .6 W/M*K conductance.

The same is true for air. Air sucks as coolant, very low thermal conductance. To make up for this need massive surface area. So air blocks need tons of thin fins to have enough surface area to use air as coolant. If adding more material only hinders cooling, try removing the fins on true and just blow air on base of copper block.

To use water as a coolant, logically it would seem better to first dramatically increase surface area using high thermal conductance 100's of times more efficient than water before getting to piss poor water....need massive surface area to make up for piss poor heat conductance.

Nevertheless it will be interesting to see what happens.

[Petra]

To quote Stew from four years ago...

Don't really need a massive flow rate to keep something cool. The problem, as always, is to do with convectional efficiency.

Water does hold a heck of a lot of heat per unit volume. 1 litre per minute (~0.3gpm) will rise in temperature by just 1.0C per 70W of heat. Could happily cool a 140W super-overclocked heat-monster CPU with just 1LPM, provided you can get the convectional rate up high enough.

This is where it all goes a little pear-shaped through. Silicon CPU dies are pretty small things. Achieving average convectional rates, via jet impingement, much above 50000W/m²K on a flat surface (cpu surfaces are flat of course) is very hard to do unless you're using pumps significantly stronger than what most people use for watercooling their computer.

Now at 50000W/m²K, with a 0.0001m² CPU (100mm²), results in a C/W of 0.20. Since we're talking direct-die here, this is pretty much what the cooling effect would be against our hypothetical 100mm² CPU. There's no metal conduction, and there's no thermal paste interface.

Let's compare that with some of the upper-end waterblocks.

The thermal paste barrier, per 100mm², with AS5 has been fairly accurately ascertained through a number of different methods over at Procooling, to have a C/W somewhere in the order 0.065 for a good mount (most estimates are coming in around 0.06-0.07, so we'll pick the middle average for now).

Now waterblocks, when looking at the convectional rate, can be stated in one of two ways. We can look at the actual convectional rate per unit of surface area that the water touches the metal, which is an incredibly complex way to model the waterblock's cooling effect, albeit the most accurate way to do so. Alternately we can utilise what is called the "effective convectional rate", which sums up the net effect of the additional surface area and all the little variances in cooling effect over that area, including the conductional costs of the metal in the block, and compacts it all down into what would be the "equivalent" convectional rate as if just acting on the area of what's being cooled (i.e. the CPU area)

By utilising the simpler "effective" cooling effect method, for a 100mm² CPU die, a block like the Storm/G4 is estimated to approach the 110000W/m²K mark, and the Storm/G5 approaches the 125000W/m²K mark, when matched with the uppish range end of water-cooling pumps that people use, but these values also include the conductional cost of the metal path as well.

At 110000W/m²K, the effective C/W for a 100mm² CPU is ~0.091C/W. For the thermal paste interface, the cost was 0.065, and these two are added together to arrive at an estimate of the total C/W for our 100mm² CPU, and it works out to ~0.156. At 125000W/m²K, it works out to around a 0.145 C/W.

Now in our direct die example, the C/W of a jet impingement device on a flat surface is 0.2, or substantially worse. In order to beat the waterblocks above we would need to achieve an average convectional effect of at least 70000W/m²K over the direct-die impingement of the CPU surface area.

This is where it gets a little difficult to construct such a device, and have it work with ordinary pumps. An jet impingement array, while good for larger areas, results in localised "dead-zones" where the jet washes meet. When impinging on a conductive plate of metal, this isn't so bad 'cos the metal just conducts the heat to where it's cooler, but when dealing with the surface of a CPU die we cannot afford this - the CPU will get MUCH hotter in these regions - which pretty much forces the single jet model for jet impingement cooling of a CPU die.

The other concern is the "wall" region of a jet impingement effect. The "wall" of a jet impingement effect occurs after about r=2.5d, where d is the diameter of the jet orifice. Once you get beyond the "wall" diameter, the convection efficiency drops off fairlu quickly. This places some important to consider restrictions on the size of the jet orifice. Ideally, in fact, the jet orifice should be tuned on a per-cpu-die basis.

For a 10x10mm CPU die, the diagonal area is ~14.14mm, which means that our jet orifice should at least be 14.14/5 = 2.83mm (7/64") in diameter, and the jet of course positioned centrally above the CPU die.

Using a variety of calculations which I won't get into here, the correspondent jet velocity for a jet of this size to achieve a net cooling effect > 65000W/m²K, is 11m/s.

Using yet further pressure drop calculations (which I also won't get into here), it all works out to requiring a pump that can deliver 4.7LPM (~1.25gpm) against a pressure resistance of around 9mH2O, and that's just for the waterblock alone. All up, we'd be talking about needing to use at least something like a US-Spec (60Hz) Iwaki MD-30RZ pump in order to deliver a cooling effect that would begin to outclass what top-end waterblocks can achieve.

If the die-size is larger then the problem becomes even harder to solve for the direct-die effect to achieve something better than what a good closed waterblock can achieve.

So the end answer is, yes, it can work IF you give it a strong enough pump (and by strong - we're talking REALLY strong), and you're blasting the bejeezus out of your small and fragile die of CPU silicon directly.

When coupled with all the other risks associated with direct-die cooling, my personal opinion on the matter is that it's simply not worth it.

Original thread: http://hardforum.com/showthread.php?t=895401

Old information, yes... but still useful/informative. Either way, direct die cooling always makes for a fun little project.