• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

what makes a waterblock good?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

HoratioJackson

Registered
Joined
May 10, 2003
what are the seperate components to a water block? thickness? in and out holes? material? i am very new to water cooling.

also, what is the best block to get and y?

thanks
hj
 

Cathar

Senior Member
Joined
Jun 8, 2002
Location
Melbourne, Australia
Mark Larson said:
Lots of surface area, low flow restriction and a good clamping pressure.

Agreed on good clamping pressure.

Lots of surface area is a false idol, as is low flow restriction.

In fact, I would have to say that the low flow restriction approach is the one that is most directly holding back water-block advances.

Blocks that effectively make use of mini-channels and/or jet impingement, coupled with very thin base-plates have the best hope of offering any real advances over the standard maze designs.
 

Since87

Member
Joined
Jul 30, 2002
Location
Indiana
Mark Larson said:

Interesting... how come?

Because in White Water, the majority of the heat is convected (is that a word?) from a 'small' amount of surface area just above the die. On the other hand, I'd say that White Water's surface area to volume ratio in that region is very high.

Perhaps if you had said, "lots of EFFECTIVE surface area"...
 

Cathar

Senior Member
Joined
Jun 8, 2002
Location
Melbourne, Australia
Mark Larson said:

Interesting... how come?

Only so much surface area can ever really be used effectively.

Beyond a certain point, the extra surface area is coming at a distance from the heat where it's well beyond doing anything useful. A waterblock can have heaps of surface area, but very little where you want it, while another can have 1/10th the surface area, but because it's placed correctly, will offer better cooling performance.

The further heat has to move though the block to get to somewhere cool, the greater the temperature difference between the hot thing (the CPU) and the cooling (water hitting the surface). If heat has to move more than about 5mm to be cooled, then it's going too far. Now if all the useful surface area is concentrated within 5mm of the heat source, this means that
there's not a lot of surface area you can use, so it's important to use it well.
 

Mark Larson

Disabled
Joined
Feb 26, 2003
Location
Assembled in Malaysia
:eek: - i thought it would be understood that i meant effective surface area, since surface area farther from the core will do nothing for efficiency. i see where you're coming from though.
 

Cathar

Senior Member
Joined
Jun 8, 2002
Location
Melbourne, Australia
Mark Larson said:
:eek: - i thought it would be understood that i meant effective surface area, since surface area farther from the core will do nothing for efficiency. i see where you're coming from though.

Yes - but when explaining it to someone who isn't familiar with waterblock theory, then we need to be a little more specific, because they might run off thinking that some waterblock that looks like a SLK-800 heatsink wrapped in a shell will be the best thing on the planet.
 

strokeside

Member
Joined
Jan 10, 2002
Location
Dublin, Ireland
how thick would you recommend for the base, 3mm? 2mm?
So if the intel Heatspreader is 3cmx3cm (approx), then your effective surface area is 4cmx4cm, centralised about the middle of the core? And any other surface area outside this is pretty ineffective/useless?

Am doing up my first Cu WB, so want to get the design right.
 

Intrepid

Member
Joined
Jun 30, 2002
Location
Boise, ID, USA
hmm. just wondering. does anyone here have a degree or has studied thermodynamics and hydrodynamics in collage or graduate school to back up one or more of these sides? just wondering, there are diferent sides i would like to see what people studying related fields say on te issue.
 

gone_fishin

BandSaw King
Joined
Feb 11, 2002
Location
U.P. Michigan
The die underneath is much smaller than the heatspreader and an aircooled heatsink is what the heatspreader was designed for. The heat is intended to conduct further throughout a standard aircooled heatsink than a waterblock needs to and this is what the heatspreader aids in doing among other things (protecting the core).
 

Cathar

Senior Member
Joined
Jun 8, 2002
Location
Melbourne, Australia
The other main reason for the heat-spreader is to act as additional thermal mass to give time for the overheating safety mechanisms to kick in should the P4 CPU ever be turned on without a heatsink attached. Without the heat-spreader sections of the CPU could fry itself before the in-CPU thermal sensors register what's going on (like on the bare-die Athlons).

The heatspreader is there more as a protection and "idiot-proof" device. It actually hinders the effective cooling of the P4 CPU to a certain degree. The heat-spreader is 1.27mm thick (0.05") and if applying highly focussed cooling, it really isn't required to focus on cooling much more than a 16x16mm area in the middle of the P4.

Intrepid6546 said:
hmm. just wondering. does anyone here have a degree or has studied thermodynamics and hydrodynamics in collage or graduate school to back up one or more of these sides? just wondering, there are diferent sides i would like to see what people studying related fields say on te issue.

You're right though. I could have it all wrong. What I'm saying does seem to have some veracity in a physical creation though, being the White Water block design, which embodies the concepts I'm talking about above.
 

four4875

Member
Joined
Jan 6, 2003
Location
I can see walmart, 44906
i would think that direct-die cooling with alot of flow across the die would be best. caus there is no worry of thermal loss between the block and die, and the heat dont have to travel through all of the fins and stuff, then get restriction between the fins and water.
 

Cathar

Senior Member
Joined
Jun 8, 2002
Location
Melbourne, Australia
four4875 said:
i would think that direct-die cooling with alot of flow across the die would be best. caus there is no worry of thermal loss between the block and die, and the heat dont have to travel through all of the fins and stuff, then get restriction between the fins and water.

This is a little more involved than it seems.

The CPU die offers a very small amount of surface area to cool. Using regular pumps, it becomes somewhat difficult to apply water-cooling to such a small area in an effective fashion.

When talking about water-convection thermal efficiency, this is described in units of W/m^2K. Unless this can be gotten much higher than around 80000, which is quite hard (if not near impossible) to do given regular pumping pressures on a very small flat surface, it is often better to simply accept the thermal disjunction between the CPU and the waterblock, wear the price of conducting the heat to a larger effective surface area and apply effective water-cooling to that larger area.

ie. we're at a stage where the best non-direct die waterblocks are as good as, if not better, than about the best you can do with direct-die techniques, given pumping pressures that are used by most.

The other problem with direct-die cooling is that CPU's are not rated for constant submerged operation, and CPU death is not an uncommon occurance after shortish periods of time (within a few weeks/months).
 

Koooler King

Member
Joined
Nov 21, 2001
Location
North America
Re: what makes a waterblock good? Some parameters to consider ...

HoratioJackson said:
what are the seperate components to a water block? thickness? in and out holes? material? i am very new to water cooling.

also, what is the best block to get and y?

thanks
hj

My 2 cents ...

First, a good waterblock is constructed from highly conductive materials ... silver is best, but given it's expense and rareity, copper, which is second at about 97% of silver as a thermal conductor, is a pretty darn good comprimise. Avoid aluminum, it is about 60% of copper for conductivity and difficult to work with.

Second, you need a good balance of surface area and conductivity to the surface, balanced by coolant flow. A copper base thickness about 1/8" is a good starting point. Fins of copper less than about 0.020" thick will start to be less effective beyond 1/8" tall.

Next, you need to understand the flow rate of your fluid ... is it turbulant or laminer (e.g. Reynolds number), that is, is surface area more important than flow velocity? Sometimes extra fins help, sometimes they hurt ... that is where flow comes in ... balance flow vs. surface vs. turbulation. Your pump and fluid flow will be imprortant here. I would consider 0.25 gpm low flow, 2.0 gpm high flow, roughly speaking for cpu cooling.

Also, the working fluid ... for common applications, pure water, is best for heat capacity ... distilled water is better to keep your lines and coldblock clean. Don't bother with additives (e.g. water wetter) as the small gains that might be had in surface tension and air bubble reduction are not worth the hazards and hassles.

Finally, you need to understand the footprint of your heat source and the 3D conductivity of your waterblock design to conduct heat. The newer high clock cpu, despite the chip case size, the heat source can be very small, e.g. 10mm x 10mm, even smaller ... in that case, bigger is not always better ... you need to concentrate the cooling flow where it will do the most good, right in the middle of the chip.

Bottom line ... high tech engineering analysis will get you close, as will the try and test method ... yet there are no "secrets" beyond that ... both work, but neither method guarantees results the first time without testing and retesting ... beyond the waterblock, it all depends on your system dynamics, e.g.; pump, lines, radiator, and fan!

I hope I have not scared you away, it's a fun science to play with ... perhaps with 1000 iterations, you'll be the next "Edison"?!

Good luck,

Koooler King
 

Koooler King

Member
Joined
Nov 21, 2001
Location
North America
Intrepid said:
erm... correction copper is harder to work with than aluminum.

Thanks for heads up ... I meant aluminum is more difficult in that you can't readily solder it, etc. unlike copper. However, as you correctly pointed out, copper can be a "bear" to machine.