Why the small waterblocks?

[TheWesson]

I was perusing this site's waterblock test results and something occurred to me.

The "best" waterblock shown achieved .13 C/W.

This is - I am guessing - the C/W when a perfect radiator is assumed. That is, the inlet temperature is kept constant at the listed temperature. Like attaching your watercooling system to the tap.

Call it .10 C/W if these results are moved from the world of die testing to the world of CPU testing.

So - that's good - but there would seem to be substantial room for improvement.

The waterblocks I've seen look kind of ... small. Why don't people make bigger waterblocks? You could have more surface inside the waterblock for heat transfer, and you could generate more turbulence while keeping impedance the same (or lower impedance while keeping good heat transfer.)

Is it the weight? 1 cc of water weighs one gram. Hence a limit of 400 grams of water would limit you to a cube 7.4 cm on a side. I don't think it's the weight.

Maybe for a CPU less than 100 W a thermal resistance of .10 or .15 is "good enough" ...?

Maybe the limiting factor on waterblock performance is the thermal resistance of the radiator+fan? Ehhh ... I kind of doubt that. At a guesstimate, the thermal resistance of a good radiator/fan looks like about .06 C/W.

Well anyhow. I guess I'm waiting for someone to tell me that yes they have a huge waterblock and it works really well, duh ...

Or do the test results listed come from a system with a radiator (and the waterblock thermal resistance is trivial) and I missed that?

or bigger waterblocks couldn't help much, you're just dealing with the thermal resistance of getting heat from the CPU into the waterblock in the first place?

or??

the wesson

[Graystar]

Apparently the heat simply doesn't spread very far from where the CPU core is touching the block.

[JFettig]

If you can suck up all the heat right away, then do it. thats what these are doing, they are sucking up every bit of heat before it spreads very far. If it has to spread, theres a bigger thermal gradient of temperature.

Jon

[eobard]

Bigger blocks may not provide more heat removal but one thing I will say for a large, open chamber block like the Swiftech MCW462, if your pump dies having all that water there gives you more time to shut down.

[Cathar]

Originally posted by eobard
Bigger blocks may not provide more heat removal but one thing I will say for a large, open chamber block like the Swiftech MCW462, if your pump dies having all that water there gives you more time to shut down.

This is why we have thermal shut-down features on the motherboard.

The converse disadvantage of having large open-chamber blocks is that when the tubing melts ('cos you didn't have thermal shutdown enabled while running Folding@Home when you were out) you now have quite a large quantity (30cc or so) of super-heated water that's going to spill out all over the board and other components, as opposed to the ~2cc of water that these smaller blocks hold.

The fear of pump-failure should not be a reason to not take proper precautions with your setup. It is an issue that people entrust their many thousand dollar setups to the cheapest pump they can find.

[rogerdugans]

Both designs (large and small) have good and bad points:
Small-
low weight- less "pull" on the motherboard
Simpler mounting often (see above point)
fit easier on the mobo/in the case

Large-
reserve capacity in case of pump failure
Likely to work better with pelts (and chillers, perhaps...)

Both can work very well and neither seem to have a tremendous advantage over the other, IMO, but most current work seems to be concentrating on the smaller designs.

[Diggrr]

Well, being the smart alec that I am, a 3" (7.6cm) cube of copper would cost $47 (before shipping/milling/fittings/lapping) and weigh over 8.5 lbs.....:D

I think waterblocks are too big still, and am waiting for some micro-blocks to hit the stores. I think it'll happen sooner or later, and cheaper to make is cheaper to buy.

[SK8]

The only "microblock" that is really good now is either rbx or white water , they are mostly targeted @ amd users b/c the size of die and it doesnt have ihs so it doens spred heat accross the block.

[Diggrr]

No no, I mean MICRO blocks. Ya know, so small that you can tell if a proc is unlocked without removing the block? One that's so small it needs additional tabs to hit the four pads.
One that's as big as my thumbprint would suit me fine.

[Graystar]

Originally posted by Diggrr
No no, I mean MICRO blocks. Ya know, so small that you can tell if a proc is unlocked without removing the block? One that's so small it needs additional tabs to hit the four pads.
One that's as big as my thumbprint would suit me fine.
You mean like this?

http://home.nyc.rr.com/graystar/WB4.jpg

[Yuriman]

Whatblock is a realy cool block. I wouldnt mind having one for my next project.

[Graystar]

Originally posted by Yuriman
Whatblock is a realy cool block. I wouldnt mind having one for my next project.
I've got some copper coming tomorrow...I'll make you one if you'd like to test it out and tell us what temps you get. 1/2" or 3/8"?

[Diggrr]

That is exactly what I'm talking about. Just couldn't remember who made it.

[SK8]

i wouldnt expect that to be a performance king with that much surface area and it looks VERRY restrictive

[TheWesson]

Thanks, it's nice to see all this interest in my question.

That micro block blows my mind. It actually works well?

So ... basically you're saying that longer fins would be totally useless because the fins would be virtually the same temp as the water for most of their length?

You don't want a lot of copper between the water and the CPU if you can avoid it, because the moving water is better at removing heat that the highly conductive (yet immobile) copper .. ?

Immersing the die in moving water would be ideal, then, if that were possible (and that has been done, yes?)

So anyhow then the sole purpose of a bigger block (visavis C/W efficiency and demands on the pump) would, possibly, be the ability to pump a lot of very turbulent water through very fast without a lot of impedance (that is, without a mega-pump)?

I'd still like to see what would happen if you pumped water through a modified SLK-900 (or whatever aircooler you like) instead of air.

It would perform much like a decent waterblock, do you think, then? You'd have to be sure to direct the flow of water toward the core, I suppose ...

I doubt it would perform worse than the same cooler in air, anyhow.

the wesson

[Yuriman]

Im going to compare it to a homemade cascade. I want to see if it is close at all.

[9mmCensor]

The new InnovaCool rev. 4 is pretty small and cool looking too.
http://www.innovatek.de/contentServ/3.0/www.innovatek.de/data/media/744/rev4_boardSA1.jpg
www.innovatek.de

[Graystar]

Small is "in". hehe :D
Originally posted by SK8
i wouldnt expect that to be a performance king with that much surface area and it looks VERRY restrictive
I can't tell how well it performs because my board is notorious for bad temperatures. For example, I can make the temperature drop by 10 C just by placing a cheap low power 80mm fan in just the right place. I do know that it performs better than the high-end ducted Dynatron heatsink that I was using.

The block is somewhat restrictive. When I hookup my Pondmaster 250 (a Danner Mag model 2 with 3/4 barb on input) it slows the flow from 250 GPH down to about 60 GPH. This is also due to the 3/8” tubing and 3/8" hose. The next version will have 1/2" barbs, and possibly a .0625” chamber height instead of .043” (1.64mm vs. 1.09mm.) With those changes, the cross sectional area of the chamber will double. So I hope test will show at least 120 GPH, because I don't want a drop in water velocity through block. The pump's P-Q says I won't get it, but these relationships are rarely so simple.

[rogerdugans]

That is a GREAT looking block, man!

lol- I never thought I'd see a water block smaller than the barbs the hoses connect to :D

I hope testing results get posted....? :D

[Diggrr]

Originally posted by TheWesson
I'd still like to see what would happen if you pumped water through a modified SLK-900 (or whatever aircooler you like) instead of air.

It would perform much like a decent waterblock, do you think, then? You'd have to be sure to direct the flow of water toward the core, I suppose ...

I doubt it would perform worse than the same cooler in air, anyhow.

It's not that easy. You still need to find the right ratios of waterflow, surface area, velocity etc... Bigger and more isn't necessarily better. To make an hsf like that into a watercooler would work, but it wouldn't beat a waterblock by any means because the waterflow and velocity characteristics would still suck. If you had a ginormous pump and 1" tubing to make the velocity against all that copper high enough though.....

[Yuriman]

What about a dannermag3? Thats pretty big.