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

Cathar, my waterblock idea needs insight...

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

fafnir

Member
Joined
Jul 9, 2003
Location
moved
Cathar, JFettig, Thorilan, and Fushy,... or anyone else who's got some experience with this stuff....

i need some insight on a waterblock idea i've got....

since we're almost at the material limits of heat conduction, would a waterblock with a heat pipe/column defeat the 4.5celcius limit of a waterblock?

i know such a design would be extremely limited by the performance the the heat pipe used, but since the current best heat conduction rates for a heatpipe are approaching diamonds, it shouldn't be too much of a problem right?

e.g. use: heat_p_01.jpg

(two center pieces)

or

heat_c_03.jpg


from...

http://www.npowertek.com/product/index_heat_pipe.htm

as in use the center of this cooler...

http://www.overclockers.com/articles754/
http://www.overclockers.com/articles754/ttic2.jpg


and machine a baseplate with a hole just big enough for the heatpipe/column to fit through just like the air cooler listed above,... and then water the heat pipe/column....

would that work?

and if it would, how well?


thanks for the time everyone
 
Last edited:
as in take this:

ttic2.jpg


ttic3.jpg


mill off all the fins or something... baseplate with a hole and then solder shut, and then water....
 
ive seen a few heatpipes on older systems but not on newer ones. i hink the reason would be that the total capacity for a heat pipe is very low due to the lack of cooling liquid so eventualy it will balance its convection higher than a normal WC setup. BUT THATS A CONVENTIONAL DESIGN .
you may come up with something that might work but it wouldnt be easy i think .
if you have very good building skills then it might be feasable but as a heat pipe is really a non compression version of a phase change system it might end up just turning into a custom phase change block.
 
No a heatpipe wont do anything to help cool the CPU on a waterblock. The cooling of the CPU gets done right over the core, cooling a bigger area of the block makes no difference.
 
yes, nikhsub1, you were supposed to be lumped together in the first sentence too but i just really couldn't remember how to spell the name...

so cooling a larger area or more surface area won't help even if the heat *does* conduct further and more efficently? the cascade block IS what the pinnicle of copper waterblocks might be (unless Cathar goes crazy with the fluid dynamics again or something) but its still being limited with the material used right? e.g. the silver version was a bit better...

so besides doing a reliable direct to die we've exhasted ALL possibilities? can't be right, right?



and thorilan, you're good too,... you've come up with some great ideas too and you've gotta have the best, so don't be ashamed... i know you wouldn't be if you were talking to a cute japanese woman or something,.. so lol... and "welcome"
 
fafnir said:
yes, nikhsub1, you were supposed to be lumped together in the first sentence too but i just really couldn't remember how to spell the name...

so cooling a larger area or more surface area won't help even if the heat *does* conduct further and more efficently? the cascade block IS what the pinnicle of copper waterblocks might be (unless Cathar goes crazy with the fluid dynamics again or something) but its still being limited with the material used right? e.g. the silver version was a bit better...

so besides doing a reliable direct to die we've exhasted ALL possibilities? can't be right, right?
With the use of metals (copper, silver) and not getting into exotic materials such as synthetic diamond and the like, we are VERY close to cooling limits with conventional machining. With the silver cascade, it is getting to the point where .5C means A LOT in terms of relative performance, but for most people, .5C is meaningless. What I am trying to get out of my clouded head is that it will take a substantially greater amount of resources and money to even get a little better performance than the silver cascade and most won't foot the bill for the performance gain.

Now, about the heatpipe idea, heatpipes are great when they are BIG and used for passive cooling, they can do the job OK when it is large enough. For waterblocks, cooling surrounding areas of the core makes no difference in core temps.
 
fafnir said:
so besides doing a reliable direct to die we've exhasted ALL possibilities? can't be right, right?

Right. We are nearing the pinnacle of waterblock efficiency, and we've passed the point of diminishing returns. Is this as good as it gets? NO. We just need to look elsewhere in the system for improvements.

1. Thermal Paste? No. The TIR is about as good as we can get it right now, and I don't think many people here are materials engineers capable of developing new thermal pastes, although I've heard rumors of new ones in the near future.

2. Waterblock? pretty well optimized.

3. Water Heat dissipation (AKA Radiator/bong/etc...)? Bingo. I know that we have many gifted problem solvers here, but some of you just need a little direction. Here's where I see the next big advances in PC watercooling. I would like to see more people working in this area as well. For about a month I've been working on a water-chilling device that can drop water temps 5-8C below ambient temps. Know what this does to CPU temps?

Here's the Design Criteria I've set forth for myself:
  1. Inexpensive. Entire waterchiller should cost less than $100 US.
  2. Energy efficient. Power consumption needs to be less than 50 watts.
  3. Quiet. (of course.)
  4. Must yield at least 3C temperature drop at the CPU.
  5. Auto-regulating. Cannot create condensation inside your PC.

I've got the design and initial prototype done (man it's ugly right now, I miss my CNC :() and it's working as expected.

In order to not stifle the creative process, I won't reveal my methods for a while. Anyone else care to take a stab at this? I'd like to see some other solutions to this problem.

C'mon outside of the box. It's more fun thinking out here.:D :burn:
 
For the removal of heat from the system I think the rad/bong is also the biggest obstacle. Ever heard of squirrel cage blowers? :D I bet if you stick one of them on a radiator it will perform just fine.
 
yes i agree that in general the common heatercores we use are not the most effective things for dissapating heat, but yet we can't afford a direct exposed to air bong because the drawbacks of an open air system...

and we also can't use big things either, mostly because of flow restrictions... and dealing with something loud isn't great either...

so this leaves us stuck both ways, and thats not cool. serious.
 
L337 M33P said:
For the removal of heat from the system I think the rad/bong is also the biggest obstacle. Ever heard of squirrel cage blowers? :D I bet if you stick one of them on a radiator it will perform just fine.
Yes but the BEST you can do with a non-chilled radiator cooled H20 system is get the water temps to ambient. With my dual HC's in parallel I am almost there. Chilling like Neo said, is going to be the way to go, but, for me, condensation is NOT an option.

<--Edited for correctness-->
 
Last edited:
Nikhsub, you and I have the same opinion: Condensation is not an option. Many other watercoolers share this same concern.

Here's what I'm taking advantage of: relative humidity inside houses ranges from 20% (very dry) to about 60%. If you take a look @ a psychrometric chart, you'll see that even at 60% RH, dew point is much lower than air temp. This is the added cooling ability that I'm going after. Obviously this isn't going to help the poor souls who live in houston with no A/C where it's 100F with 99% relative humidity, but it will help many watercoolers.

Here's the trick. How do you measure humidity and regulate the temperature of your water accordingly? We all know there are many methods of chilling water. We also know that it's possible to measure dry bulb and wet bulb temps.... do you see where I'm leading?
 
NeoMoses said:
Nikhsub, you and I have the same opinion: Condensation is not an option. Many other watercoolers share this same concern...
Here's the trick. How do you measure humidity and regulate the temperature of your water accordingly? We all know there are many methods of chilling water. We also know that it's possible to measure dry bulb and wet bulb temps.... do you see where I'm leading?
Yes. In an ideal 'chiller world', the unit would be able to measure temps, humidity, barometric pressure etc... and then calculate the lowest possible water temp before condensation would occur and then to be safe, add 1C ontop of that :D
 
(qoute)
Yes but the BEST you can do with a non-chilled H20 system is get the water temps to ambient
(end-qoute)

I was under the impression that efficient bong-coolers could get s low as 5 degrees UNDER ambient. however, its been along time
since i first read the tutorial on the bongs, and i read it here :)

oh yeah, the above ideas are GREAT im not slamming ANYTHING ok?
however, isnt it true that ANYTIME you drop below ambient you CAN get condensation? i know that it may not happen inside the pc because he was talking about dropping the water temps below ambient by like 2-5 celsius, so maybe the die/rad automatically negate that?
MAYBE the MATERIALS and /or thickness of radiators could be looked at. also, perhaps someone more gifted then me
could research a new DESIGN of radiators, say maybe along the lines of insects book lungs, or sea creatures gills, something weird and esoteric that may yield results.
feel free to flame me, im going out on a limb here.
:)
 
Spade said:
(qoute)
Yes but the BEST you can do with a non-chilled H20 system is get the water temps to ambient
(end-qoute)

I was under the impression that efficient bong-coolers could get s low as 5 degrees UNDER ambient. however, its been along time
since i first read the tutorial on the bongs, and i read it here :)
Yes you are correct! I was not even thinking of evaporative cooling, my mind was strictly on rads :p
 
I have to agree on the water-block being optimised thing. There are many factors now that I had to face with the Cascade block, and far more than I faced (well - considered) with the White Water. Pumping effort is a major consideration. Greater pumpage equals higher performance, but more power draw and more heat added into the water. What temperature differential that exists due to the Cascade's block design is more one of that limited by pumping potential. Waterblocks, in my mind, are about getting as close to a direct die cooling scenario (minimal material between the CPU and the water) and then cooling that properly. For metal conduction (copper/silver, etc) the metal that exists between the CPU and the water is contributing around a 0.02C/W level of thermal resistance for a ~100mm^2 CPU sized heat source. i.e. anywhere from 1.5-2.0C for a hot CPU.

A heat-pipe tries to solve the "metal barrier" problem too, but with only 0.02C/W to work with, we're talking about very small gains to be made, if they can be made at all.

NeoMoses is onto the right track. The various ways to move forwards from here for water-cooling is:

1) Improving waterblocks - we're almost done here - if there's another 0.02C/W to gain (on a testbed like BillA's) I'd be surprised.

2) Improving pumps - the pumps we use are typically around 10% efficient - that is for the motive pumping power they provide, they typically consume up to 15x the electrical power which usually manifests itself as heat in the water. An Eheim 1250 really is only provided around 2.2W of pumping power, but consumes 28W to do so. There exists to the ability (in theory) to triple the in-system flow rates over an Eheim 1250, and still be consuming around 20W of power, and all the resultant flow-performance benefits that go with that. Still, we're talking about maybe a ~2C gain here for a hot CPU.

3) Improving the TIM joint. This is the biggest barrier today, but the hardest to fix. Around 0.08-0.10C/W is attributable to the thermal paste juction, or around twice that of the thermal resistance of the waterblock itself. The issue here is servicability. It is possible to improve the joint by soldering or permanently fixing the CPU to the waterblock, but this then negates the ability to upgrade or change the CPU independently of the cooling apparatus. For easy to apply and remove thermal pastes we are at the point of diminishing returns too. Maybe a ~4C gain here at best with some very smart people working on the issue.

4) Improving the water temperature. As NeoMoses describes, for a non-condensing scenario we need a cheap, compact and efficient water-chiller that can track the humidity and keep the water temperature at say 0.5C above the dew point. This is achievable today with fairly cheap and commonplace components, but not efficiently. Depending on the relative humidity of the air, and the efficiency of one's existing radiator setup, we could be talking about anywhere from a 5-20C lowering of the water temperatures.
 
I dont know if this is the greatest idea or whether it has been done or not, but what about using peltiers on the radiator!?!? Wouldn't that cool the water quite a bit?!? I may be way off with this idea, but i thought i would share it.
 
Neomoses, have you thought about regulating a peltier directly on the CPU? Or have you decided that a waterchiller would be more effective? I'm an amateur electronics enthusiast, also. I'd be more than happy to help with a circuit to do what you've mentioned. Just shoot me a PM.
 
the one real obstical isnt even our water technology . its cpu technoligy. right now we are using ceramic and composits for cpus but its basicaly an identical problem because what limits a cpu is its ability to transmit pulses without electron tunneling ( we have .05micron process cpus already but once they go to .03 we start to get electron tunneling so the transistors arnt efficient . while this isnt directly related to heat , its the same idea as our NON exotic materials are reaching peak and we are at point of diminished returns for materials.

if we make diamond blocks more affordable and diamond cpus then we would have another step in the journey
 
Back