Creating a Heatsink, got questions.

Hey guys i have plans to design a heatsink, which my friend will make out of aluminum.
Im just wondering, what you guys think are little additivies or must haves on heatsinks.
I will be using fins but what would be better, thinner and lots of them, or thick ones with less?

I can be more specific. Basically, what do you think a heatsink should look and be like and act like.

Obviously its gonna have to displace heat well

Zoltan

for fins , thiker ones will heatup and hold the heat , making your CPU run hotter

the thinner will diapate the heat fast

thinner and more of them. you're trying to maxamize the surface area. but they still need to be able to stand up straight, so not too thin

dead_man311 said:

for fins , thiker ones will heatup and hold the heat , making your CPU run hotter

the thinner will diapate the heat fast

Click to expand...

Ok i wanted to go with alot of thinner ones.
SHould the base be solid and thick so it can hold more heat and then the thin fins will disapate it faster?

Zoltan

wow this sounds like quite the project, firstly, how will this guy make this heatsink? I have some ideas and drawings I have made for one and before I throw them out, i want to see what kind of seriousness we have here.


Jon

how thick ? hmm , you dont want it too tick make ir really heavy. dont want it to thin or it will heat up fast

say how does 3/8" to 1/2" sound

Zoltan,
What kind of experience/machinery does your friend have?

I have ( fairly) extensive experience as a machinist, and I think it would be far easir to build an effective waterblock than a comparable air cooled heatsink.
If you follow the path of Zalman, Gigabyte and Swiftec...making a fabricated sink from many parts, you run right up against the same problem that they have, namely an interrupt in the heatflow path.
Without some pretty sophisticated assembly/soldering procedures you'll have a sink that looks the dog's ******** and performs poorly.

Good luck though.

i might suggest trying to imitate the shape/form of well-performing heatsinks today - you probably wont' match their performance (but if you do be sure to patent your design!!).

if you aren't sure about how big/small/thick/thin things should be, then just cut the block and err on teh big/thick side of things, and if the performance stinks, then you can chop off the excess and see the improvement(s)..

i know that we're all about creation here but why would you want to take that kind of risk. I mean I guess worst case scenerio the HS would perform like poo doo, and you'll notice the temps and power down, but it just seems like a ton of effort. Cool idea tho. I wish you luck if you do try it.

I think the biggest danger in making an aircooled waterblock yourself is in mounting/remounting it - not in crazy temps... temps shouldn't be a problem as long as you use some common sense in the design of the fins/pins of the HS...

Lol, it cant be worse than some of the amd stock sinks ive seen. I happen to have gotten a good one that is solid copper coated in nickel.

Wow good luck on this project. Hopefully it turns out to be better than anything on the market right now so I don't have to pay an arm and lef for Thermalright sinks.

Well what im trying to do is make a custom heatsink for my friends card. I have already modded mine just sliced up a cooler master and an amd heatsink (stock) you can see it in card modding. its pretty sweet.

But for my friend, he has a geforce 2 mx 64meg and im going to customize the heatsink so it goes from the gpu to the ram sinks so everything is bridged thus more cooling capacity with fins all the way through. Does this sound like it would work. On my card i bridged the two ram sinks together for more cooling capacity and in theory i believe this is better, so what if everything was connected, or am i wasting my time.

My friend is not that experienced. Highschool courses, and alot of his spare time, he is quite good and i just want to have a fun side project. I have also worked in a machine shop on cnc lathe, mill, drill and pin machines. Obviously im not using these kinds of machines but i just wanted to have a fun project, see what happens


Thanks for the support at least.

Zoltan

Make sure you maximize the surface area of the fins... that's the key.

It's going one hard project trying to get the thin fins.

This'll be difficult but an interesting project nonetheless.

The point is to maximize surface area of the heat conductive material (aluminium, copper) so you have the maximum amount of cool air in contact with the heatsink at any one time.

The best design that I can think of at the moment is actually a base with many thin long pins pointing upward and a large powerful fan to top it off. Sort of like the pins on the bottom of a CPU, but much longer and more dense.

Good luck with it! Consider the pinned design!

The top of the gpu and the ram are different height, so you will have to take very exact measurements. Keep that in mind.

neat idea!! Like they've said so far maximum serfice area. keeep them thin as possible + Make sure they are far enpough apart to alow decent air flow. Think of how you want to rought the airflow in relation to where your going to put the fins. Thin fins are the way to go.

Wizdon said:

This'll be difficult but an interesting project nonetheless.

The point is to maximize surface area of the heat conductive material (aluminium, copper) so you have the maximum amount of cool air in contact with the heatsink at any one time.

The best design that I can think of at the moment is actually a base with many thin long pins pointing upward and a large powerful fan to top it off. Sort of like the pins on the bottom of a CPU, but much longer and more dense.

Good luck with it! Consider the pinned design!

Click to expand...

I've been wanting to make or even BUY a heatsink with that design... not only would it look sick with copper, but it would perform sick too!

I would work on varying thicknesses.The heat transfer is not even, the center core area soaks heat and the block transfers it through out it's surface..If it was me, I would try a base that starts at some standard thickness and tapers it's thickness out towards the edges, where of course the thinnest fins should be placed.

The thickness adjustment must be done at the fin side of the sink
so the contact area can be made perfectly flat reguadless of the actual thickness,and you would have to make your fins taller on the outer edges to maintain there even height with the center fin area. (Ultra thin outer fins might help as well)

I have always wondered why a heat sink maintains a constant thickness because it would make sense that thinning the block towards the outer areas would improve heat transfer and disipation. And i think help DRAW the heat away from the core rather then having the soaking effect that takes place with a single thickness.

Any way that's my 2 cents, and who knows it may not be worth that....

I wonder what effect air turbulence has on the cooling effort. i know you want to restrict the flow as little as possible when using a traditional axial fan on a heatsink, but say you were to design a heatsink where backpressure doesnt even have to be considered? I somehow think that coolermaster has gone about their design wrong with the aero series. you have this fan that is designed to generate higher static pressure than a traditional fan, and they put it on a pretty typcial heatsink design.
I would think a design like the swify would be a fantastic design assuming the use of a centrifigal blower fan.
when air hits the pins on the swifty, it creates more turbulence than if it were to hit a straight and even spaced fin.
If you were to design a all copper heatsink using threaded, sandblashed pins instead of aluminum turned pins, you could maximize the surface area heat can be exchanged by quite a lot.

also, fan placement is of concern, as modern heatsinks are not that efficiant about their design this way. in a pc case, air is blown horizontally "down" onto the heatsink. it blows at the base. this does not really help the air in escaping form the heatsink once it has collected heat.

it is helped some by the rotation of the newer ones, pointing their fins verticly instead of horizontal. in a heatsink that follows a threaded pin method, why not turn the fan 90º blowing up. that way, the air can escape easily from the heatsink verticaly, and you can make use of natural convection to transport the heat too.

There is of course a lot of math involed in this. machine shop skills are good but I would make sure a sound design is thought first.

if you want to try somthing similar, you can just machine down a 1/2" copper base plate and solder some copper fins to it. you might want to make a jig for this since you will want to hot air solder the peice together in one go. I did this once to try and improve performence on my r 8500. it works. im sure retail solutions perform better, but its not to say its impossible.

if you have an idea, and a dream to do something, damnit, do it. its not like there are people hovering over your shoulder ready to smack you for it.

although the offerings in the current market are impressive right now, they are anything buy brilliant or inovative. with a few notable exceptions ^^;;