• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Designing a Heatsink

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Originally posted by: CycloWizard
You could do one just for novelty... Instead of a pump, build an Archimedes Screw to elevate the water, then a waterfall instead of a radiator, for example. It would be totally novel, as I doubt anyone considered watercooling computers prior to the advent of the modern pump. 😛
Click to expand...

well its not entirely new. there was this one cooler a while back that was an open design that rained water down into its reservoir for cooling. i forget its name, but it had a funny one. did used a regular pump though.
 
CNC machining is where it's at for something like this. Basically you input a design into a computer and it drills whatever shape you decide. You might be able to find a "machine shop" to give you a discount or even do it for free if you play the kid working on a school project angle.

As for desigining a heatsink... One thing that would be really neat is to use a fractal pattern -- something easy like the H-fractal would work quite well. You'd need to make it 3-d though. The cool think about a fractal is that it is space filling, so it will give you a maximum surface area. Fractals are neat and are all over (e.g. look at your circulatory system. Pipe size and length from artery-capillary decreases by a constant ratio at every branch, and each pipe branches out the same. This provides maximum blood flow and maximum reach 🙂. Apply the same idea to a heatsink and you're in business.
 
Just that machining fractal surfaces is a bit more difficult than cutting straight lines thru metal. Yes, the advantage of extra surface is there (theoretically at least), but the machining could be difficult.
Anyway, it's just a school project, so it can be tried
 
Water will not work for Phase Change since it takes 100° C (212°F) to change it's phase. Basicly you are trying to boil water with your CPU and that is not a good idea. What is needed here if you want to go in to phase change is a degree in chemistry. My best sugestinon for you is to go to Wal-Mart and buy a can of R134a ($5) and use that as you Phase Change material.


When it comes to manufacturing anyone who is taking a basic machine shop coarse at your local high school can do the cutting shaping and assembly work for you. The $econd option is to open up the good old fasioned yellow pages and look up metal working or CNC Milling.
 
When it comes to manufacturing anyone who is taking a basic machine shop coarse at your local high school
Click to expand...
That would be me 🙂. Next term I'm taking it.

About the R134A, thanks fo the model number. Sounds pretty scary though
The exposure concentration was 4000 ppm (0.4% v/v) and was
scheduled to last for 30 minutes with a 5 minute postexposure evaluation
period as was accomplished in the Halon 1301 portion of the study.
Approximately 4.5 minutes into the exposure, the subject lost
consciousness and both pulse and blood pressure dropped to zero
Click to expand...
....erm.. Died from a small concentration? I should definately be careful with this stuff (I haven't read the article through yet though).

0roo0roo: Evaporative cooling eh? You'd need to fill this up once in a while, but if this was done in a closed environment .... think of the wonders it would do..

As for fractals, I feel it may be expensive. Best thing to do is to NOT use laser-sharp cutting tools (except for the base), but rather something more rugged so that more surface area can be found.

Right now I'm flipping through sone engineering journals regarding heatsink design. IEEExplore has a lot of them - but I can only access them from school as it requires a registration fee.

Currently my idea is to create a heatsink, something similar to a heatpipe cooling solution, except with a pump attached that recirculates the cool liquid (or R134A or something) to the base, and the hot liquid would naturally evaporate to the top of the heatsink. I haven't really thought this through yet, but at this point I'm in the information gathering stage.

-The Pentium Guy
 
Originally posted by: The Pentium Guy
When it comes to manufacturing anyone who is taking a basic machine shop coarse at your local high school
Click to expand...
That would be me 🙂. Next term I'm taking it.

About the R134A, thanks fo the model number. Sounds pretty scary though
The exposure concentration was 4000 ppm (0.4% v/v) and was
scheduled to last for 30 minutes with a 5 minute postexposure evaluation
period as was accomplished in the Halon 1301 portion of the study.
Approximately 4.5 minutes into the exposure, the subject lost
consciousness and both pulse and blood pressure dropped to zero
Click to expand...
....erm.. Died from a small concentration? I should definately be careful with this stuff (I haven't read the article through yet though).

0roo0roo: Evaporative cooling eh? You'd need to fill this up once in a while, but if this was done in a closed environment .... think of the wonders it would do..

As for fractals, I feel it may be expensive. Best thing to do is to NOT use laser-sharp cutting tools (except for the base), but rather something more rugged so that more surface area can be found.

Right now I'm flipping through sone engineering journals regarding heatsink design. IEEExplore has a lot of them - but I can only access them from school as it requires a registration fee.

Currently my idea is to create a heatsink, something similar to a heatpipe cooling solution, except with a pump attached that recirculates the cool liquid (or R134A or something) to the base, and the hot liquid would naturally evaporate to the top of the heatsink. I haven't really thought this through yet, but at this point I'm in the information gathering stage.

-The Pentium Guy
Click to expand...

R134a is not a dangerous material at all. It has been used in Automotive Aircoditioning systems for the last decade.


http://www.airliquide.com/en/business/products/gases/gasdata/index.asp?GasID=141
http://wiki.ehow.com/Fix-Cars-Air-Conditioning
 
I was thinking you could do a heat pipe design using Butane or R134A as the transfer material. If you want to get extreme (and destroy the Ozone Layer) R12 and R22 are much colder.

R12 and R22 are the primary reasons why we have a Hole in the Ozone Layer.
 
I've been thinking of some CRAZY ideas. Just for the hell of it.

Say you have a peltier cooler. A peltier cooler is basically a thingy where a current runs through this pad and it's cold on 1 side, HOT on another side. I could design a heatsink around one, using a mini-sterling engine to harness that energy to power the fan (maybe). Or, just heatpipe it to a sterling engine and use that to generate the electricty for the peltier and the fan. Now also, becuase a peltier makes one side of the pad FREEZING, there'll be some condensation. That condensation could be doubled as water for a watercooling-type thing, dissipating MORE heat out.

Crazy idea. But whatever, I gotta look into it 😛.

-The Pentium Guy

 
Oh hey, I got FEMLAB from my school now. I'm dead confused though, but I'll figure it out. They've also got MatLab available too, but I'm not sure if I'll use it or not.
 
Personally i wouldnt bother with matlab, use a dedicated fluid dynamics package if possible (your school may well have one), i have used a couple, but not for several years and the names escape me .... i know ANSYS has one built into it (although its not great)
ahhh i think its coming back to me.. flotran i think it was thats a good (and reasonably simple cfd (computational fluid dynamics) package)

the other thing i have come across that might be helpful is there is a synthetic oil that has only recently been invented with most of the properties of water, but does not conduct electricity so would be ideal for heatsinks, you could even have the fluid in direct contact with the core.

the other thing is a website, www.matweb.com, which lists a LOT of different materials and thier properties. have a look through, there are defintely better materials than aluminium out there 🙂


oh - one last thing. i did once do a similar thing to what your doing (only for a different purpose, so design requirements were a bit different) and found that even using a very good milling machine, very thin fins are near impossible.

given the time and budget constraints i had i just had to design around this, but bear in mind that if you want lots of thin fins, you are looking at a zalman-esque contruction (lots of thin pieces of metal stuck together) rather than a 'traditional' heatsink.


hope that helps 😀
 
Awesome first post 😀. Welcome to Anandtech btw (either that, or if this was the Off Topic forum I'd ask "Who were you before you were banned?" 😛)

Thanks a lot, I was primarily focusing on Aluminum, but you've opened up some new possibilities for me. Tons of different materials out there...I looked into FloTran, and it seems out school doesn't have it. It's fine, Femlab's got some good fluid dynamics tools in there too...our school even even has the license for the heat transfer module...

Thanks for all the tips.
-The Pentium Guy
 
Here's a crazy idea, make your own mold and cast the heatsink. You could make the mold, melt your old lawnchair, design a press to really jam the metal in there, and machine it to your design specs once cooled. It would definitly get you an A+ for serious effort.

Or, like SES mentioned, try a stacked plate design using aluminum fins, copper tubes and a silver base (one ounce of silver is 10$ I think). Friction fit the fins, press in the copper tubes then machine the base it to a 0-2 surface finish, and if you are using a heatspreader-equipped cpu, machine the heatspreader to the same finish, then wring em together without thermal paste. It would take an enormous amount of work and time but be exceptional for heat transfer.
 
Originally posted by: Howard
ThePentiumGuy, you can't really tell if a chatroom is dead just by staying in it for 20 seconds.
Click to expand...

.....
Oh I know what you're talking about. I was just neffing there. Besides, the forums are more active than that chatroom. But let's save the neffing for OT k? (I'd rather keep this thread "Clean" of ATOT material).
 
Say you have a peltier cooler. A peltier cooler is basically a thingy where a current runs through this pad and it's cold on 1 side, HOT on another side. I could design a heatsink around one, using a mini-sterling engine to harness that energy to power the fan (maybe). Or, just heatpipe it to a sterling engine and use that to generate the electricty for the peltier and the fan. Now also, becuase a peltier makes one side of the pad FREEZING, there'll be some condensation. That condensation could be doubled as water for a watercooling-type thing, dissipating MORE heat out.
Click to expand...
I've been doing research on this for the past month. This is looking to be very expensive.
1) I'll need at least a 230W peltier cooler for my purposes
2) A modern CPU consumes 60W. That means 230W+60W of heat released from the top side of the peltier
3) This requires exotic cooling (watercooling perhaps)
4) The peltier needs to run on a seperate power supply as it's essentially an inefficient money hog. Electricity bill will skyrocket 😛.

Or, like SES mentioned, try a stacked plate design using aluminum fins, copper tubes and a silver base (one ounce of silver is 10$ I think). Friction fit the fins, press in the copper tubes then machine the base it to a 0-2 surface finish, and if you are using a heatspreader-equipped cpu, machine the heatspreader to the same finish, then wring em together without thermal paste. It would take an enormous amount of work and time but be exceptional for heat transfer.
Click to expand...
Would it really? I'm seeking to acheive "excellent" results, that's my primary goal (performance), next to cost and quietness....so I'm looking for alternative methods that are witty but get the job done - well that is.

Looks like I have a ton of research on my hands. My goal is to have this built by the end of december...so I can test it over christmas break and type up my paper in the latter half of January (because it's due in the beginning of February).

I'm attempting to contact a WPI professor as my interviewee so I can get a direction for my project. Think I should look in the Thermodynamics dept? My research seminar advisor told me to take a peek at the Computer Science dept though.

Any general ideas, books to read, comments, direction would be extremely welcome. I really have to thank all of you who participated in this thread - all the information you posted sincerely helped me a lot.

Thanks,
-The Pentium Guy
 
I know one of my old roommates did something just like this for a project in an electrical engineering thermo class. I'll see if he can send me his report - might save you a lot of time. I recall that the conclusions he reached was that for the peltier to be financially viable, it would have to continuously power the fan for 150 years (or something similarly ridiculous). Of course, I think he was also considering the case where the thermoelectric device isn't used as an active part of the cooling process, but rather to harness the output energy as a power source, so it's slightly different than what you're trying to do.
 
Originally posted by: The Pentium Guy

I'm attempting to contact a WPI professor as my interviewee so I can get a direction for my project. Think I should look in the Thermodynamics dept? My research seminar advisor told me to take a peek at the Computer Science dept though.
Click to expand...

I go to WPI. But, I'm Biochem...so probably of no use to you. 🙂

 
Originally posted by: Calin
Originally posted by: The Pentium Guy
I was thinking about Phase with water. Phase is: Liquid at room temp, Gas at cpu temp. Right? Adding a solute to water (anything that increases its molality) would cause it's boiling point to go up. Fine tuning it by creating a low-pressure environment (not the case, the cpu cooler itself) would cause the water to evaporate even faster.

Not sure. This sounds VERY interesting. Thanks for the idea!
Click to expand...

:thumbsup:
Yes, something like that. Play with the pressure so you could use some strange liquid. Or you might choose a different agent and a different pressure.
You might think at using ethilic alcohol (ethanol) as its boiling point is at some 76 Celsius (I think). Find different liquids, research each other (volume as gas, heat taken by evaporation, extra heat transported from extra heating (if any). You'll find some liquids working better than others.
Also, you could use a presurised liquid (instead as - with water - depresurised one). See how much could you pressurise the heat pipes, and how much it will help at keeping the CPU cool

Oh,you might want a special core block that is able to withstand the pressures/depressurisation. Not sure if normal water blocks will survive
Click to expand...

I just read somewhere that the XBox360 have a similar cooling system - using depressurised water for a phase change
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top