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

Best Block Design

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

country_3030

Member
Joined
Oct 13, 2002
Location
Paris, Texas
What is the best type of watter block design. I am not talking about copper, alumin, etc. but what design spiral, micro channels, etc.
 

The Overclocker

Member
Joined
Aug 21, 2001
Location
Milton Keynes, UK
when choosing a waterblock, you need to look at what makes a good heat transfer in water, this must be done by looking at the qualities (and problems) of water,

water has a huge thermal capacity, this means it can absorb lots of heat withough increasing in temprature, however it has lots of surface tention, this makes it hard to transfer heat to the water, turbulance is the ideal way of transfering heat as the water may take longer to move around due to some flow resistance, but that dosnt matter because the water can store alot of heat, this also reduces surface tention so heat can be transfered better.

to chose the best waterblock, simply look at one which will create lots of turbulance, bear in mind in general a waterblock with lots of surface area will work better then one with not much due to the added turbulance from the obstructions.
 

Cathar

Senior Member
Joined
Jun 8, 2002
Location
Melbourne, Australia
The two best design methodologies that you can currently use are:

micro-channel
jet-impingement

Almost no commercial blocks on the market make use of these concepts.

My micro-channel jet-impingement hybrid design (called the "White Water") is achieving CPU temperatures up to 6C better than any other commercial water-block on the market which has been backed up by numerous customer experiences, and is achieving CPU temperatures that match up with some direct-die cooling efforts that have been recently demonstrated here and elsewhere.

Yes, you want maximum turbulence, and using both of those concepts at once allows for some pretty dramatic improvements.
 

JFettig

Hey! I showered! Senior
Joined
Jan 5, 2002
Location
MN
High surface area, high velocity, low restriction.
high velocity also means high turblience..

cathar, how exactly does this jet thing work?

-edit- heres what i mean, i dont see were it can be used on other waterblocks becides yours, it would be hard to do it on most other ones, and mostly pointless for some... if you understand waht i mean..
 
OP
country_3030

country_3030

Member
Joined
Oct 13, 2002
Location
Paris, Texas
I am wanting to build my own water block i have acess to a machine shop. I was just wandering what type of design would be best. I am leaning to a maze type block with micro channels in the bottom of the groves. If i can acheve such precsion.
 

The Overclocker

Member
Joined
Aug 21, 2001
Location
Milton Keynes, UK
i dint think there is one set answer for what is the best base thickness, i am sure if cpu power outpust keep increasing (which i am sure they will) then you will need to increase the base thickness to spread the heat to be cooled perperly again, at the moment a thin base is desirable as not much heat is being emmited from the cpu, in a few years time the ideal base thickness may be much more due to the inefficantcy of a thin base when removing that much heat.
 

Cathar

Senior Member
Joined
Jun 8, 2002
Location
Melbourne, Australia
The base thickness you need is tied to the internal design of the waterblock.

Block performance does not degrade as CPU heat is increased. Block performance is linear (or near so) with increasing heat loads The only thing you need to be careful of is the water boiling, which isn't going to happen until you have very low flow rates in a bad design and a ~2000W heat load. So no worries there.

So, as stated above, the base thickness relies solely on the design. The thermal gradient radiates out from the CPU core in a rough dome shape. A very thick base spreads the heat over a larger area but is limited by the thermal conductivity of copper ultimately.

When using micro-channels and jet-impingement it's possible to get substantially higher rates of thermal convection that greatly surpass the thermal conductivity of copper, and therefore a thicker copper base becomes a liability as it starts to act like an insulator compared to the convection process. In this instance you want a very thin copper base.

So, the short of it is that "it depends". If you're doing an open pool design, a base thickness of ~10mm is good, if you're doing a channelled design then 3-4mm seems to be a good value, if you're doing micro-channels then <2mm seems to be good, if you're getting more advanced then you can push down well below 1mm.
 

Voodoo Rufus

Powder Junkie Moderator
Joined
Sep 20, 2001
Look at Swiftech. They've got super thick blocks. Thicker is better methinks for distributing heat across the copper. I'd like to know the performance differences between thin vs. thick.
 

nikhsub1

Unoriginal Macho Moderator
Joined
Oct 12, 2001
Location
Los Angeles
Voodoo Rufus said:
Look at Swiftech. They've got super thick blocks. Thicker is better methinks for distributing heat across the copper. I'd like to know the performance differences between thin vs. thick.
Reread the post above yours by Cathar, it seems you didn't.
 

obliv

Member
Joined
Jul 14, 2002
swiftechs new block is substantially thinner than the rest of their old designs and out performs them. the turbulance in that block is sick.
 

Cathar

Senior Member
Joined
Jun 8, 2002
Location
Melbourne, Australia
Voodoo Rufus said:
Look at Swiftech. They've got super thick blocks. Thicker is better methinks for distributing heat across the copper. I'd like to know the performance differences between thin vs. thick.

Again, it depends on the internal design. Using a thin bases for open pool designs like the Swiftech MCW-462U sucks. It needs the thicker base to spread the heat out.

Performance differences? If done properly?

My 0.75mm thick based block is outperforming Swiftech style thick-based blocks by up to 6C for AthlonXP's at ~1900MHz/~2.00v.