What do you think about this simple watercooling idea?

[BonzaiDuck]

I was window(s) shopping fans today, and I started windows(s) shopping water-cooling apparatus (as I've done occasionally for a few years now).

My attention is currently on chipset heatsinks, and how I can eliminate the 60mm stock fan and actually get better cooling with more CFM. Those here that know me will understand immediately what I'm attempting to do. But that's not the point.

I was looking at these water-blocks. CPU water-blocks. And it occurred to me that water-cooling effectiveness might be enhanced if the base and interior of a CPU water-block looked something like this:

Copper Northbridge cooler at Sidewinder

We've always agreed that with smaller and smaller cores, processors are getting smaller -- but for the grace of the god Intel, the IHS caps are a constant size for each socket-standard. This in turn makes TEC cooling less effective, because TEC cooling depends on areal size of the item being cooled, and smaller and smaller TEC devices require more and more power -- at which point, there is a limitation with the heatsink and heat-removal requirement on the TEC's "hot" side.

But here, I would think you would extend the effectiveness of water by adding extending the third dimension with greater areal contact between copper spires and the water. Of course, you can only pump so many cc's of water per second based on the pump and the tubing, and 1/2" tubing seems to be the performance "high-side." Therefore, the waterblock becomes a miniature reservoir, but water is going in and out at the same rate, and it would seem that there would be more efficient transfer of heat.

I'm interested in the thoughts of certain people about this idea. AigoMorla? For instance?

 

[mpilchfamily]

The interior of many block actully looks like that or has a good number of micro chaneles all to increase the surfance area like that. Water blocks are as thin as they are to prevent flow restrictions. If you make a water block that tall or half as tall to acomidate more surface are then you sevearly effect your flow rate. Water would come to a near stand still inside the block. Your really want to keep that flow going to keep that heat moving out and away from the CPU.

Actually the flow rate of the water entering and leaving the block would not be the same. The flow would actully be slower coming out then going in just like it does in the res and every water block. Every time that water has to make a turn or hits one of those pins it will slow down. That block would add allot of friction that the water hass to fight threw.

 

[BonzaiDuck]

I thought there might be a hitch to it like that. If not, someone out of thousands of people would've designed it and put it into production. You're saying that there would be "backwaters" in the tall water-block design, which would just get warmer -- at least I think that's the implication of your remarks there.

 

[mpilchfamily]

In a block that size and set up like that you would have some dead space and turbulance. You would have a dirrect stream shooting straight down at the base of the blaock. Then the center would be swirlling around. Not much of the water in the center would be getting pusshed out so it would just continue to get hotter. When you open a water block you'll see there is a dirrect path the water follows from one inlet to the next. You'll also notice the size of the cutout in the block holds ruffly the same amount of water as a straight peice of tubing would running threw that block. When you take a flow of water from say a 3/8" tube and push it into a 1/2" tube the flow slows down. If you go from a 1/2" to a 2/8" the flow increases so long as the pump is able to sustain its flow with the added back pressure that is created. Some blocks will use smaller channels so there is a faster flow. but in general most water cooling pumps are not very strong and can't deal with allot of back pressure.

Anyway an alternative would be a semi-tall design like that. Basicly the block would be in a "U" shape with those columbs shooting up each branch of the "U". This would certainly add more surface area for the water to com in contact with. But at teh same time it ofrces you to bend the tubing over more to fit inside a case. As mentioned before more bends, and tighter bends cause flow restrictions. Especially in rubber tubing. The tighter the bend the smaller the diamiter of that are of tubing becomes.

 

[WoodButcher]

IIRC the d-tek v-2 pins were shortened from the v-1. ( I own both , just don't remember!) IMO the thought is to have the water as close to the core as possible and removed. There is a balance between flowrate and "thermal conductivity" (? forgive my ignorance!?) or mass, a point of "diminishing return", otherwise water blocks would be more like the tec waterchiller I built. My block because of the mass of copper heatsink will take longer to respond to the temp change but once cold, will maintain that temp better hopefully slowing the switching cycle. Aigo had a block made more resembling the standard water block that we're all waiting results on so I'm sure he has a very definitive idea in answer to your question.

 

[BonzaiDuck]

Originally posted by: WoodButcher
IIRC the d-tek v-2 pins were shortened from the v-1. ( I own both , just don't remember!) IMO the thought is to have the water as close to the core as possible and removed. There is a balance between flowrate and "thermal conductivity" (? forgive my ignorance!?) or mass, a point of "diminishing return", otherwise water blocks would be more like the tec waterchiller I built. My block because of the mass of copper heatsink will take longer to respond to the temp change but once cold, will maintain that temp better hopefully slowing the switching cycle. Aigo had a block made more resembling the standard water block that we're all waiting results on so I'm sure he has a very definitive idea in answer to your question.

No -- no ignorance there, Bro'.

Conductivity is the inverse of resistance. I THINK these are equivalent in that respect -- the higher the conductivity, the lower the resistance and vice-versa. You can measure the overall thermal resistance of a water-cooling loop or a heatpipe cooler, or compare the conductivity of different metals (copper versus silver, for instance, and I thought I once saw a Danger-Den (?) water-block with at least the base made from silver.)

Especially, your note about increasing mass tends to confirm what I'd observed -- even second-hand. I was comparing idle temperatures on the same winter day at the same room-ambient between my heatpipe cooler and a friend's Zalman CNPS-7700 "Monster-flower." Since the copper flower cooler had a higher thermal resistance than my heatpipe cooler, I kept wondering how his idle temperatures were significantly lower (significant, meaning they could not have been due to errors in measurement or variation in ambient, even if they weren't crucial to the overall performance.)

So even though the variation between idle and load could be less for a lighter heatpipe cooler, increased mass would reduce the the temperatures at which that range is effective. [But more mass means more stress on the PCB.]

It's intriguing that reducing temperatures, either through CPU design or cooling enhancements, may have some practical limit for the various technologies we apply. I think some white paper -- maybe it was posted at Tom's, maybe it was a post on a thread here -- noted that different parts of the CPU respond differently to lower temperatures. The metal parts display increased electrical conductivity at lower temperatures, but the silicon parts actually have better conductivity at higher temperatures. [Or something like that.]

And TEC applications, as with your chilled water cooler, means more power consumption.

I'm on an "island" in the California grid, where power is provided by a municipally owned utility, and our rates are lower. In about three years, that's going to change. It may be that my 420W of power-consumption under full load for this one PC may not have that much of a dent in our bill. Then I wonder what some exotic TEC application might mean, but just to keep the cores at room -temperature, it would run intermittantly -- not constantly at full-bore.


[ . . . . and I had to evict a tenant this year -- place was vacant for six months, and I'm still recovering . . . ]