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

Stereolithography for water block tops?

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

MoreGooder

Member
Joined
Dec 13, 2003
Location
Saint Louis, MO USA
Hey friends,

I was thinking about doing a design for a water block using a traditional copper base but using Stereolithography for the more complex top portion. Has anyone tried this, or know more about the materials used?

My thought is this: If it is so difficult and expensive to machine small jet impengement tubes, why not creat them via stereolithography? There are several websites that offer free online quotation for your 3D model. I did a quick one just for grins and was auto-quoted for only $78.00. I'm sure that if I optimized it I could get it lower. And, perhaps just the jet impengement portion of the top could be manifactured this way, with the rest of it simply encased in poly.

Your thoughts please!
 
Cathar has thought of doing the same thing, the problem with that from what some others have mentioned is that slm isnt very strong, the structure isnt always the best or something like that.


Jon
 
I believe that resolution can also be an issue, with 0.1mm resolution typically being quoted. That can be a problem when you're talking about things designed down to 0.05mm resolutions.
 
Thanks for the quick responses!

Cathar, are resolutions down to .004 of an inch not good enough in general or for specific features? Wall thickness, hole diameter, hole location, concentricity..... There's more to it than just .004 standard tolerances I would think. And, they probably achieve better than .004 in general if they guarantee .004. So, water blocks are machined to the nearest .002 of an inch (sorry, metrically challenged here in the US) That's crazy tight specs for something that will be hand assembled. No wonder that machining them costs so much.

I'm just trying to think of less costly methods of production. Bear with me.

There are a variety of plastics available now for stereolithography. I once held a prototype Motorola Star Tak cell phone in my hands created with this process. You could feel the ridges from striations of the plastic material. But, that was 10 years ago now. Things have improved.

Imagine creating the complicated tube section using this process, but the rest using traditional CNC milling techniques.

Cather, the part I had auto-quoted was the last design I emailed to you some months ago. I just "unioned" all of the strips into one solid model.
 
Last edited:
Here's a link to a spec sheet for a machine used by stereolithography.com:

Viper pdf

And that's the low res one. There is a high res system now capable of .001 inch thick layers.

Edit:
And here's a spec sheet for water resistant resin they use:

Watershed
 
wow, thats pretty cool, wish i could completly understand the mechanical properties..
hm.. what about the the water absorption? Although its very small, what could happen to the state and strength of the material after long submergages in water? hm. ptb looks like the best :)
 
Stereo-Lith is an interesting option, and with newer materials it may eventually become a viable method for producing some WB prototypes such as the XXX. However, all of the stereo lith parts I've seen have been quite brittle and weak. They kind of remind one of dry angel-hair pasta, not something that I would consider very robust.

Luckily, we haven't reached the limits of conventional machining resolution. For any good machinist, +/-0.005" is a walk in the park, +/-0.001" isn't too hard to accomplish, and +/-0.0005" is still achievable. Unfortunately, with the tiny tools and such that must be used to manufacture the current waterblocks, machining costs to achieve these tolerances are very high due to the excessive time necessary to manufacture said waterblocks.

However, there are other methods for producing small, strong parts. One project I am currently working on involves the LiGA process (It's a German acronym but in English it stands for: Lithography, Electroplating, Casting). Anyone ever seen those videos of dust mites riding around on little machines? Yeah, those parts were made with the LiGA process. With this process, 5 micron (and finer) resolution is not unheard of. However, getting a single LiGA wafer made for you will cost at least $3000-$5000. A modified version of the LiGA process is a possible option for this type of prototyping, but not commonly available. What makes LiGA parts so enticing is the high strength possibilities (medium to high carbon steel type strengths). 2D is very simple to obtain with this process, true 3D parts aren't as simple,

Along the lines of the stereo lith idea would also fall SLS (Selective Laser Sintering) and metal printing. I've seen both processes, and they, too, offer great strength advantages over the plastic stereo lith parts with similar or finer resolution. However, these processes are not cheap, and metal choices may be limited.

Then we come to the question that really hurts. Is it worth it? For 99.9% of overclockers, the answer is no. The best possible waterblock we could hope to obtain by these exotic processes will allow us to run the CPU just slightly above water temperatures. Is that extra 5-10% worth the hundreds or thousands of dollars spent on building these? Only to a very few who want 'bragging rights.' There are some scientists working on this that will take heat exchangers to the absolute maximum efficiency no matter the cost, and they will be noted in history for that. I don't want to belittle the achievements of others, but it has been said many times that anyone can take exotic materials and expensive, exotic processes and create the absolute best performing product. It takes true genius to use average materials and inexpensive processes to create a product that performs 95% as well as the best but costs only 5% as much. That, my friends, is engineering in a nutshell.

It seems many are missing the forest for the trees. Many people are still caught up in the hype of waterblock design and are ignoring other issues that can yield much higher results than a new WB design. Really, look at this thread that is over 6 months old. Many of the obvious defecencies in current watercooling were identified there, but I have seen very little if any work done on most of them. It makes me believe two things: 1.) most people are satisfied with the current level of performance in watercooled systems and/or 2.) most people are not creative/ambitious enough to go out in a different direction on their own.

Let's make an analogy to put things in perspective. If we were talking about cars, we could say that the average computer user is driving a Geo Metro (OEM aircooling) the average overclocker is driving a Monte Carlo (High end aircooling) and the average watercooler is driving a Corvette (good watercooling). There are a couple of us like Cather, JFettig, and myself who are driving Formula 1 cars or LeMans GTP cars (completely custom waterblocks, but extremely expensive). We're all interested in driving at top speed. However, to make this analogy correct, we must assume that the Corvettes and faster cars are still using farm implement tires instead of racing tires and have no suspension. It's obvious that large gains in speed could be gained by tire and suspension development, yet no one has chosen to do that because engine horsepower (WB C/W) is currently more prestigous to work on. To further slow down progress, new regulations in F1 and LeMans limit the cars to 5 gallons fuel per week (wife, kids, jobs, etc...) To make everyone go faster, it would really help if Bridgestone and BF Goodrich (new guys in WC development) started making advances in tire technology (pumps, radiators, chillers...) that provided better traction (more efficiency). Anyone wanna help out?
 
Thank you NeoMoses. That reply was very insightful.

I became enamored with the Cascade (as many have) and have tried to come up with ways to help lower the manufacturing cost endured by Cathar. My efforts have been largely nonproductive unfortunately.

I read that entire thread from six months ago you linked to. Not much has happened since then.

Here's an update on PCM+ for those that might be interested from reading the other thread:

PCM+Update
 
Back