CPU block design

marttisiitsman · 03-14-2006, 08:03 AM

Few months ago I posted a design that could be done with lathe. Eventually I dropped that idea and thought a new one.

Here it is:

Base thickness is 3mm. Channels width is 2mm and the waves are 1mm thick.
How would it theoretically perform?
Any improvements considering that the main goal is low restriction and simple construction?

Or is something like "R-Type" the better way to go?

JFettig · 03-14-2006, 09:35 AM

Looks ok, but unless your making it on a cnc mill, theres much simpler designs

That block would likely perform good, but not any better than any blocks on the market.

Do you have capabilities of getting this made?

Jon

marttisiitsman · 03-14-2006, 11:57 AM

Quote:

Originally Posted by JFettig

Looks ok, but unless your making it on a cnc mill, theres much simpler designs

That block would likely perform good, but not any better than any blocks on the market.

Do you have capabilities of getting this made?

Jon

Some guy from my local hardware forum would make it for an modest sum.
But if I could get a really good and simple design, I'll prefer making the block myself to cut costs and for the joy of DIY.

PS: Sometimes my sentences sound funny etc. So don't hesitate to correct me, acctually I have to practise, I have to do an English exam after high school if I wan`t to get to some good college.

phide · 03-14-2006, 03:19 PM

Waved fin designs are really quite capable. EK blocks make a very similar block called the Wave Water (some sort of spin on the White Water, I suppose). Although I don't have any measurements of that block, it seems like you're on track for this one performing very comparably to the Wave Water. Joe over at Overclockers.com reviewed it not too long ago - you might want to check that out.

Your English seems to be quite good. I wouldn't fret the exam, so long as you take your time.

wildfrogman1 · 03-14-2006, 06:50 PM

I think this is what you are looking for. It would be alot cheaper than a one off custom build. Interesting block design

http://ekwaterblocks.com/shop/images/CPU5.jpg

marttisiitsman · 03-15-2006, 12:51 AM

I have a question about those o-rings. I see all sort of GPU coolers which cool the whole card, where to they get those huge o-rings? In what stores those o-rings are sold? I work in a hardware store (something like HomeDepot or something

) and we only sell up to 20mm (4/5") ID and 38mm (1 1/2") ID ones. I used the latter as the measurment (<?) for the design. It's restricting me of making a bigger block. I had in mind that I would make a HDD block aswell.

And and GPU.

2Shane · 03-15-2006, 11:31 AM

Go for simple and round.

A round bit of copper bar, round O ring, and keep it really simple and easy to make.

Lathe and drill press - and a hand tap.. = easy.

davidzo · 03-15-2006, 06:13 PM

I like your waterblock. It should perform well.

But it is nothing new. The zigzagdesign is quite old.
I did most of my blocks in this design because it is well suited for my simple 3-axis CNC-mill with small drill diameters.

From left to right:
UCD from 2003: (Universal Cooling Device - you can mount it on CPU,GPU or the Chipset)
UCD-V from 2004: with optimized structure for a target price of ~20$
UCD2 from 2005: zigzagchannels 1,2mm width (now added two channels to the final version and only 1mm wide channels). This cooler is somewhat restrictive but creates a lot of turbulences on each bent, less restrictive than micropin or jetdesign but with nearly identical performance due to high turbulences and lots of dissipation area.

There are o-rings in every lenght and thickness, you just have to search for an industrial manufacturer. In Hamburg (my Hometown) I know a Company which manufacturs and has on Stock at least 10.000 Different O-rings, so that should be enough to find the right one.
For prototypes i use O-ring string which you can buy at such O-ring manufacturing companies for prototypes. You can either buy some vulcanizing machine to get the ends together or use a special glue which costs a little fortune for a few gramms.
When i did my first cooler with a dremel, i used UV-string and melted it together over the flame of a candle. After cutting the rest off it, it gives a decent and very realiable UV-active sealing but you need a few tries before you getthe right lenght (can't predict its exact lenght because of the melting and pressing process).

I hope this could help you.

marttisiitsman · 03-15-2006, 11:47 PM

Quote:

Originally Posted by davidzo

I like those.

How does the middle one perform compared to the others?
At the moment I am waiting for the material. And then I'll go to the search of o-rings.

marttisiitsman · 03-16-2006, 09:40 AM

I have ran to a problem. It seems that the width of the channel isn't same on the whole lenght (hopefully I said it clearly

) and I can't find a way to draw them correct.
Any ideas?

By the way, I made a another one.

JFettig · 03-16-2006, 03:05 PM

looks pretty good, the round o-ring groove isnt necessary, nor round stock.

With that one you just did, see if you can gradient it into the pins rather than having the tiny channel into the end.(only a gradient about half way)

Jon

phide · 03-16-2006, 03:40 PM

Quote:

Originally Posted by marttisiitsman

I have ran to a problem. It seems that the width of the channel isn't same on the whole lenght (hopefully I said it clearly

) and I can't find a way to draw them correct.
Any ideas?

I've experienced this in a number of drawing and drafting programs. I couldn't tell you what the 'fix' is (each channel must be horizontally offset by some factor), but I can tell you that it doesn't matter with these tolerances. I don't believe there's any chance of these channels intersecting, and I can say with some degree of certainty that variances in width at different points along the wavy fins are actually a good thing.

marttisiitsman · 03-17-2006, 04:50 AM

Quote:

Originally Posted by phide

I've experienced this in a number of drawing and drafting programs. I couldn't tell you what the 'fix' is (each channel must be horizontally offset by some factor), but I can tell you that it doesn't matter with these tolerances. I don't believe there's any chance of these channels intersecting, and I can say with some degree of certainty that variances in width at different points along the wavy fins are actually a good thing.

Ok, but I wan't that they are the same width because it's better to mill them when the channels are the same width.

How small is the headspreader on Athlon 64?

phide · 03-20-2006, 02:51 PM

The channels will all be the same width. The fins will not be, but this isn't terribly important.

davidzo · 03-23-2006, 10:11 AM

Quote:

Originally Posted by marttisiitsman

I like those.

How does the middle one perform compared to the others?
At the moment I am waiting for the material. And then I'll go to the search of o-rings.

The middle on is nearly on par with the first one due to its higher turbulences, thinner baseplate and higher velocity but it also has more pressuredrop. It is milled in a third the time of the first one. The Cooler to the right was never tested, i milled a new one with 8 channels instead of six. It should perform much better than the other two, testing is in progress, on a real platform (A64 3700+ @2,7Ghz) it is only 2° better than the middle one with similar pressure drop. It is on par with a nexXxosHP at least, maybe even better than that.

marttisiitsman · 03-26-2006, 01:33 AM

How bad is this? Compared to a center inlet maze design?

I think I'm going to search for inspiration from the "Post pictures of your homebuilt blocks here" thread.

I don't need high perfomance, main objective is to cool my Athlon XP2000+ better than WhisperRock IV and at the same time quieter.
I have a big aluminium radiator (could hold two 120mm fans) which might be able to cool it passively (<spelling?) and I'll buy an L20 because it's cheap and I have made some kits with it.. I would prefer a 12V pump, but they are tad too expensive. Or are there any low price quality 12V pumps?
I also have an copper radiator which is a bit bigger than 120mm fan. If I could get copper from somewhere I'd use the copper radiator, it's smaller and it would be easier to fit it into my PC case.

I attached pictures of the radiators and some designs.

davidzo · 03-26-2006, 04:20 PM

as long as you can't make the fins smaller or use a center inlet just stay with the fins, from the designs in your last post this one should be the best:

But performancewise i would prefere something like this but its much more complicated to mill:

I would make the waves a little longer, so that the outer edge of the inner bent of the channel is on line with the outer edge of the next bend to the other side (that the flow can't get straight through and is forced to take the bents).

The copper Rad uses a more efficient Honeycomb pattern as the aluminium rad (looks like a kind of heatercore or is it not?). I would go for it even if the alurad would be slightly better performancewise, because of the noise of only one fan. If you only plan to cool your CPU the rad will have the smallest temperature in your loop, i would wonder if both rads show more than 2° difference.

XyBeRWaReZ · 03-28-2006, 05:48 PM

Is there a guide for making channels and waterflow for waterblocks anywhere?

What is the best design, regardless of how complicated it is to mill?

Also I've been reading and I'm not sure if creating turbulence is better than having the water go through as quickly as possible.

Sorry, I'm new to designing waterblocks and what not so please try to to pelt me too much.

03-14-2006, 08:03 AM	#1
marttisiitsman Cooling Neophyte Join Date: Oct 2005 Location: Estonia Posts: 18	CPU block design Few months ago I posted a design that could be done with lathe. Eventually I dropped that idea and thought a new one. Here it is: Base thickness is 3mm. Channels width is 2mm and the waves are 1mm thick. How would it theoretically perform? Any improvements considering that the main goal is low restriction and simple construction? Or is something like "R-Type" the better way to go?

03-14-2006, 09:35 AM	#2
JFettig Cooling Savant Join Date: Feb 2003 Location: Willmar MN/Fargo ND Posts: 504	Re: CPU block design Looks ok, but unless your making it on a cnc mill, theres much simpler designs That block would likely perform good, but not any better than any blocks on the market. Do you have capabilities of getting this made? Jon

03-14-2006, 03:19 PM	#4
phide Cooling Neophyte Join Date: Feb 2006 Location: Los Angeles, CA Posts: 41	Re: CPU block design Waved fin designs are really quite capable. EK blocks make a very similar block called the Wave Water (some sort of spin on the White Water, I suppose). Although I don't have any measurements of that block, it seems like you're on track for this one performing very comparably to the Wave Water. Joe over at Overclockers.com reviewed it not too long ago - you might want to check that out. Your English seems to be quite good. I wouldn't fret the exam, so long as you take your time.

03-14-2006, 06:50 PM	#5
wildfrogman1 Cooling Neophyte Join Date: Mar 2006 Location: Usa Posts: 6	Re: CPU block design I think this is what you are looking for. It would be alot cheaper than a one off custom build. Interesting block design http://ekwaterblocks.com/shop/images/CPU5.jpg

03-15-2006, 12:51 AM	#6
marttisiitsman Cooling Neophyte Join Date: Oct 2005 Location: Estonia Posts: 18	Re: CPU block design I have a question about those o-rings. I see all sort of GPU coolers which cool the whole card, where to they get those huge o-rings? In what stores those o-rings are sold? I work in a hardware store (something like HomeDepot or something) and we only sell up to 20mm (4/5") ID and 38mm (1 1/2") ID ones. I used the latter as the measurment (<?) for the design. It's restricting me of making a bigger block. I had in mind that I would make a HDD block aswell. And and GPU.

03-15-2006, 11:31 AM	#7
2Shane Cooling Neophyte Join Date: Mar 2006 Location: Antarctic Posts: 3	Re: CPU block design Go for simple and round. A round bit of copper bar, round O ring, and keep it really simple and easy to make. Lathe and drill press - and a hand tap.. = easy.

03-15-2006, 06:13 PM	#8
davidzo Cooling Savant Join Date: Mar 2003 Location: Hamburg Posts: 141	Re: CPU block design I like your waterblock. It should perform well. But it is nothing new. The zigzagdesign is quite old. I did most of my blocks in this design because it is well suited for my simple 3-axis CNC-mill with small drill diameters. From left to right: UCD from 2003: (Universal Cooling Device - you can mount it on CPU,GPU or the Chipset) UCD-V from 2004: with optimized structure for a target price of ~20$ UCD2 from 2005: zigzagchannels 1,2mm width (now added two channels to the final version and only 1mm wide channels). This cooler is somewhat restrictive but creates a lot of turbulences on each bent, less restrictive than micropin or jetdesign but with nearly identical performance due to high turbulences and lots of dissipation area. There are o-rings in every lenght and thickness, you just have to search for an industrial manufacturer. In Hamburg (my Hometown) I know a Company which manufacturs and has on Stock at least 10.000 Different O-rings, so that should be enough to find the right one. For prototypes i use O-ring string which you can buy at such O-ring manufacturing companies for prototypes. You can either buy some vulcanizing machine to get the ends together or use a special glue which costs a little fortune for a few gramms. When i did my first cooler with a dremel, i used UV-string and melted it together over the flame of a candle. After cutting the rest off it, it gives a decent and very realiable UV-active sealing but you need a few tries before you getthe right lenght (can't predict its exact lenght because of the melting and pressing process). I hope this could help you.

03-16-2006, 03:05 PM	#11
JFettig Cooling Savant Join Date: Feb 2003 Location: Willmar MN/Fargo ND Posts: 504	Re: CPU block design looks pretty good, the round o-ring groove isnt necessary, nor round stock. With that one you just did, see if you can gradient it into the pins rather than having the tiny channel into the end.(only a gradient about half way) Jon

03-20-2006, 02:51 PM	#14
phide Cooling Neophyte Join Date: Feb 2006 Location: Los Angeles, CA Posts: 41	Re: CPU block design The channels will all be the same width. The fins will not be, but this isn't terribly important.

03-26-2006, 04:20 PM	#17
davidzo Cooling Savant Join Date: Mar 2003 Location: Hamburg Posts: 141	Re: CPU block design as long as you can't make the fins smaller or use a center inlet just stay with the fins, from the designs in your last post this one should be the best: But performancewise i would prefere something like this but its much more complicated to mill: I would make the waves a little longer, so that the outer edge of the inner bent of the channel is on line with the outer edge of the next bend to the other side (that the flow can't get straight through and is forced to take the bents). The copper Rad uses a more efficient Honeycomb pattern as the aluminium rad (looks like a kind of heatercore or is it not?). I would go for it even if the alurad would be slightly better performancewise, because of the noise of only one fan. If you only plan to cool your CPU the rad will have the smallest temperature in your loop, i would wonder if both rads show more than 2° difference.

03-28-2006, 05:48 PM	#18
XyBeRWaReZ Cooling Neophyte Join Date: Mar 2006 Location: In the MIddle of Nowhere! Posts: 68	Re: CPU block design Is there a guide for making channels and waterflow for waterblocks anywhere? What is the best design, regardless of how complicated it is to mill? Also I've been reading and I'm not sure if creating turbulence is better than having the water go through as quickly as possible. Sorry, I'm new to designing waterblocks and what not so please try to to pelt me too much.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)