Go Back   Pro/Forums > ProCooling Technical Discussions > General Liquid/Water Cooling Discussion > Water Block Design / Construction
Password
Register FAQ Members List Calendar JavaChat Mark Forums Read

Water Block Design / Construction Building your own block? Need info on designing one? Heres where to do it

Reply
Thread Tools
Unread 07-20-2005, 03:11 PM   #1
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35
Default waterblock design

Attached is a picture of a .dwg of my latest design. So far, I think the biggest problem is maybe the horizontal channels are a bit too wide and the water won't flow around through the top and bottom. I basically have all the g-code written, and my head's about to drop into the keyboard. The measurements are in inches.

There is also space for the o-ring around the channel.
As for the top, I guess delrin would work best. Should I just screw it into the 4 corners?


Please give any suggestions. Currently cutting this into an aluminum test block to see how it looks.

EDIT: I forgot to mention that the top and bottom parts of the design were inspired by some of Jaydee's workings. Other than that, I'll post pictures of the aluminum once it actually finishes cutting. Also, I lost the link of the o-ring web site. Could someone please post it?

Thanks for reading.
Attached Images
File Type: gif waterblock2.gif (6.3 KB, 123 views)

Last edited by ljohn787; 07-20-2005 at 03:16 PM.
ljohn787 is offline   Reply With Quote
Unread 07-20-2005, 03:35 PM   #2
tex707
Cooling Savant
 
Join Date: Apr 2001
Location: Europe
Posts: 164
Default

www.oringswest.com

Do you have a 3D file of the block...like IGES or STEP?
tex707 is offline   Reply With Quote
Unread 07-20-2005, 04:09 PM   #3
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35
Default

Thanks for the URL. I actually don't have a 3D file, but all cuts are at the same depth of .500". I also realized that the fins in the middle are actually TOO small. I'm going to remove 2 of the 5 cuts so that 3 channels remain. The fins are paper thick and I guess I couldn't tell looking at it through ACad, lol.
ljohn787 is offline   Reply With Quote
Unread 07-20-2005, 04:46 PM   #4
maxSaleen
Cooling Savant
 
maxSaleen's Avatar
 
Join Date: Dec 2004
Location: Orlando, Florida
Posts: 383
Default

Let me make sure I'm seeing this correctly. You have a single inlet and a single outlet located on the left and right sides of the block. You have fins in the middle AND pins on the outside edges of the block? If so I would be curious as to why.

Quote:
Should I just screw it into the 4 corners?
If it worked for the old swiftech blocks I'd imagine it should work for yours. Derlin is a great choice for the top.
maxSaleen is offline   Reply With Quote
Unread 07-20-2005, 07:01 PM   #5
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35
Default

Quote:
Originally Posted by maxSaleen
Let me make sure I'm seeing this correctly. You have a single inlet and a single outlet located on the left and right sides of the block. You have fins in the middle AND pins on the outside edges of the block? If so I would be curious as to why.
Is there a problem with that? The water goes in through either end, doesn't matter, and most of it is supposed to be routed through the middle, but I tried to have a design that would create some turbulence as it went around the top and bottom and all 3 channels would meet up at the outlet.
ljohn787 is offline   Reply With Quote
Unread 07-21-2005, 10:55 AM   #6
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35
Default

I forgot to mention: I have PRO/Engineer Wildfire 1.0 on this machine. I can create the 3D model of my waterblock if I so desired, but I have no idea of how to then create g-code from that. I've been using ACE converter, which is a free tool that I can convert .dxf files into g-code, but it lacks feed rate commands, so I'm stuck making the whole program slow as hell so I don't snap a tool.

Also, the waterblock was a pretty big failure, thanks to me not having a 1/16" tool. Instead, a 3/32 tool was doing the work, but I replaced that with a 1/16" ballmill with 1 degree draft. The tops of the fins in the middle might be a little thin and sharp, but should actually be existent this time around.

In the meantime, I'll be working on a new design that's faster to create (image attached). I am mostly unsure of whether there is enough turbulence in this block. The center is probably where it gets the hottest, so that's where most of the water will be flowing directly. The water will have to flow around the top and bottom sides, and I made some cuts to the center of some of the channel walls.


If you could, please post any feedback on the new design. It should also be a lot faster to make, but that's hardly an issue.
Attached Images
File Type: gif waterblock3.gif (6.1 KB, 38 views)
File Type: gif waterblock4.gif (11.8 KB, 41 views)

Last edited by ljohn787; 07-21-2005 at 02:05 PM.
ljohn787 is offline   Reply With Quote
Unread 07-21-2005, 03:22 PM   #7
tex707
Cooling Savant
 
Join Date: Apr 2001
Location: Europe
Posts: 164
Default

Why do you make those channels straight...I think that "waves" would do better (take a look at the attached screenshot).

Maybe I could help you with G-code creation.
Attached Images
File Type: jpg WB.jpg (90.7 KB, 91 views)
tex707 is offline   Reply With Quote
Unread 07-21-2005, 04:28 PM   #8
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35
Default

Yeah, I think I have to agree. I'm really new at this kind of stuff, so I'm just playing around with things that are fairly easy to make. My prime concern with this design was definitely turbulence. I think I might revise it again tomorrow and make it so that they are waves. Although the program I use can convert splines, I would rather not use them because it plots too many of the points and the CNC actually has trouble catching up, lol. Is the design you made with arcs or is it splines?
ljohn787 is offline   Reply With Quote
Unread 07-21-2005, 05:58 PM   #9
tex707
Cooling Savant
 
Join Date: Apr 2001
Location: Europe
Posts: 164
Default

Quote:
Originally Posted by ljohn787
Yeah, I think I have to agree. I'm really new at this kind of stuff, so I'm just playing around with things that are fairly easy to make. My prime concern with this design was definitely turbulence. I think I might revise it again tomorrow and make it so that they are waves. Although the program I use can convert splines, I would rather not use them because it plots too many of the points and the CNC actually has trouble catching up, lol. Is the design you made with arcs or is it splines?
No problem with CNC mill catchin' up...all you have to do is just use circular interpolation. However, you have to take care to define the geometry right...sometimes you have to use parts of a circle (arcs and NOT ellipses or splines) to get circular interpolation right later during G-code post-processing.

I use circular interpolation wherever possible.
tex707 is offline   Reply With Quote
Unread 07-21-2005, 08:48 PM   #10
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35
Default

Quote:
Originally Posted by tex707
No problem with CNC mill catchin' up...all you have to do is just use circular interpolation. However, you have to take care to define the geometry right...sometimes you have to use parts of a circle (arcs and NOT ellipses or splines) to get circular interpolation right later during G-code post-processing.

I use circular interpolation wherever possible.
Yeah, I've really been trying to use arcs over splines (less messy code, anyway). The thing that sucks about ACE Converter is that I can't give it two feed rates: one for -Z movements and one for X,Y movements. I either have to manually give all -Z movements slow feed rates and then give X,Y movements faster feed rates. Or, if the program is umanageably huge, I just make the whole damn thing slow, lol. Do you know of any converters that I can use? Given, I like everything else about ACE, especially since it's free.
ljohn787 is offline   Reply With Quote
Unread 07-21-2005, 09:05 PM   #11
maxSaleen
Cooling Savant
 
maxSaleen's Avatar
 
Join Date: Dec 2004
Location: Orlando, Florida
Posts: 383
Default

Quote:
Is there a problem with that? The water goes in through either end, doesn't matter, and most of it is supposed to be routed through the middle, but I tried to have a design that would create some turbulence as it went around the top and bottom and all 3 channels would meet up at the outlet.
Not necessarily a problem, just not an optimized design. The heat won't really spread out very far from the center of the CPU die. Thus the pins on the outside edges will be useless. By replacing the areas where you have pins, in your original design, with solid Cu, you will increase the velocity of the water over the die area. This will result in better temps. The second design will be too restrictive and will not create much turbulance. Take a look at the first block #rotor made or JD's lummpy channel, for examples of highly effective, easy to make blocks. I believe both can be made with as little as a drill press, though you obviously have access to CNC mahinery.
maxSaleen is offline   Reply With Quote
Unread 07-21-2005, 11:49 PM   #12
tex707
Cooling Savant
 
Join Date: Apr 2001
Location: Europe
Posts: 164
Default

Quote:
Originally Posted by ljohn787
Yeah, I've really been trying to use arcs over splines (less messy code, anyway). The thing that sucks about ACE Converter is that I can't give it two feed rates: one for -Z movements and one for X,Y movements. I either have to manually give all -Z movements slow feed rates and then give X,Y movements faster feed rates. Or, if the program is umanageably huge, I just make the whole damn thing slow, lol. Do you know of any converters that I can use? Given, I like everything else about ACE, especially since it's free.


I’m not quite sure what you mean…why should you change federates for Z-axis so frequently? This work is to be done by so-called prismatic machining, which means that the tool enters the stock in a certain path and federate up to the predefined cutting depth and than does it’s job (X and Y axis movement) while constantly staying on the same depth. Usually the approach and plunge federates could be the same as machining federates. What could change is the federate on a sharp turn along the toolpath in a XY plane, though.

I don’t know about the converters, I use a complete CATIA package.
tex707 is offline   Reply With Quote
Unread 07-22-2005, 11:12 AM   #13
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35
Default

As I said, I'm totally new to this. I don't plan on being an engineer.. I hate math and science. I have a basic understanding of physics, but hardly anything thermal related.

Correct me if I'm wrong, but won't the heat dissipate fairly evenly throughout the copper block? Using this logic, I was trying to have as much surface area as possible for the water to cool down. The machine actually finished cutting the aluminum test block, so I will post pictures of that later. I'm also modifying the new design to replace the straight fins with wavy ones in the middle.

As for feed rates and machining in general, I'm fairly clueless. I've been working here for around 2 months now and am trying to learn how to properly cut into aluminum without breaking the tool. Just as a side note, we do injection molding here and were cutting the cavity part of the mold for a "Revlon" display. I came this morning and one of the machinists told me that 3 tools broke. Once the biggest tool breaks, the rest are doomed.

I don't even know where I was going with that, but thought it might be interesting, lol. Are you basically saying that I can plunge the Z as fast as I move it across the X and Y axes? I recall my father telling me to plunge slowly, and then move about laterally at a normal feed rate.

I greatly appreciate you guys taking your time out to help a newb like me out. As I said in my first post, I tried to implement Jaydee's lumpy channel design for the top and bottom parts of the first design. However, if the heat really doesn't spread out from the center of the die, then I will definitely just crap both designs and create something with only a center channel. It'd be cool if there was an image showing the heat spectrum throughout a block.
ljohn787 is offline   Reply With Quote
Unread 07-22-2005, 02:49 PM   #14
maxSaleen
Cooling Savant
 
maxSaleen's Avatar
 
Join Date: Dec 2004
Location: Orlando, Florida
Posts: 383
Default

The heat will not spread out from the die very far. Don't forget that the die is not the whole CPU. It is located in the center of the heatspreader of current P4s/A64s and is about 2cm on each side. The most important area to cool is directly over the die. Having extra channels around the sides of the block will just reduce the velocity of the water over the die area.
maxSaleen is offline   Reply With Quote
Unread 07-22-2005, 04:14 PM   #15
tex707
Cooling Savant
 
Join Date: Apr 2001
Location: Europe
Posts: 164
Default

Quote:
Originally Posted by ljohn787
<snip>
I don't even know where I was going with that, but thought it might be interesting, lol. Are you basically saying that I can plunge the Z as fast as I move it across the X and Y axes? I recall my father telling me to plunge slowly, and then move about laterally at a normal feed rate.
<snip>

Yes, that's what I'm saying. Of course, that could depend on a material, but sometimes the approach "macro" (the approach and plunge toolpath) can be more important than the feedrate. Take a look at the attached screenshot...that is the example of the 2mm OD endmill plunging into the material to make an O-ring groove...there's no way it would break this way.
Attached Images
File Type: jpg Approach.jpg (79.5 KB, 54 views)
tex707 is offline   Reply With Quote
Unread 07-23-2005, 12:34 AM   #16
JFettig
Cooling Savant
 
Join Date: Feb 2003
Location: Willmar MN/Fargo ND
Posts: 504
Default

I would not recommend ramping small cutters at all(mostly in metal) but ramping bigger ones that can take the side load and the load from plunging can handle it. I have ramped my 1/2" and other end mills larger than 1/4 quite a bit, but when it comes to ramping something like a 1/8" end mill, I burned it up good.
You should not plunge at the same speed you feed at, its actually quite a bit less.


On my mill I plunge 1/16" end mills at 1ipm in copper, 1/8" end mills between 1 and 3ipm, and because my machine cant handle much more on bigger cutters(1/4" I sometimes get away with 5ipm) but depending on the cutter, you can plunge quite a bit faster with say a 1/2" 3fl in aluminum. I have actualy seen/heard(yeah it was noisy) a 1/2" endmill plunge at 40ipm in aluminum, it managed to do it on a Haas, but not likely on much else(nor with a diff end mill).


Jon
JFettig is offline   Reply With Quote
Unread 07-23-2005, 02:41 AM   #17
tex707
Cooling Savant
 
Join Date: Apr 2001
Location: Europe
Posts: 164
Default

It was not my intention to mislead anyone. Considering the fact that we are talking soft materials here (aluminum, copper, Delrin) plunge feedrate that equals machining one, with high spindle RPM (7k-10K) and pressure cooling, should do no harm...ramping approach either, especially if the machining feedrate is not set close to the inner edge of the "envelope".

Interesting fact is that some CAM programs even have no option for different plunge and machining feedrates.
tex707 is offline   Reply With Quote
Unread 07-25-2005, 09:47 AM   #18
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35
Default

Alright, thanks a lot guys. Loads of useful information here. Sorry about not posting the pictures, had a pretty stressful weekend. And Guild Wars...
In terms of design, I guess I'll get rid of the outer channels. I also have no idea of how to ramp, so I'll just cut a little bit at a time and keep at about 20 IPM. Usually, I plunge at 1 IPM for a small tool, sometimes even less. Since I'm using the 3/32" endmill, I have to be extra careful. I always cut under coolant, so heat and small chips shouldn't be a problem.
ljohn787 is offline   Reply With Quote
Unread 07-27-2005, 03:49 PM   #19
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35
Default

Pretty much decided to go with Tex's "wavy channel" design. Already cut my first design, now in the process of cutting the second. The other thing that sucks about the converter I use is that it writes some of the weirdest g-code ever, and I'm not sure the time spent optimizing the code would be worth it in the end. However, the first cut looked good. It's been cutting for almost an hour now.

Some questions:
1) Can I get orings from a hardware store? Delrin?
2) Best method of securing barbs into the Delrin?

Thanks.
ljohn787 is offline   Reply With Quote
Unread 07-27-2005, 04:54 PM   #20
tex707
Cooling Savant
 
Join Date: Apr 2001
Location: Europe
Posts: 164
Default

Quote:
Originally Posted by ljohn787
Pretty much decided to go with Tex's "wavy channel" design. Already cut my first design, now in the process of cutting the second. The other thing that sucks about the converter I use is that it writes some of the weirdest g-code ever, and I'm not sure the time spent optimizing the code would be worth it in the end. However, the first cut looked good. It's been cutting for almost an hour now.

Some questions:
1) Can I get orings from a hardware store? Delrin?
2) Best method of securing barbs into the Delrin?

Thanks.
I don't know where you can purchase Delrin, but some Teflon tape is sufficient for securing the barbs.

I suggest you to put the water inlet in the very center of the block.

Could you send your G-code as an attachment, I would really like to run a backplot of it.
tex707 is offline   Reply With Quote
Unread 07-27-2005, 09:22 PM   #21
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35

Here are the pictures. As for putting the intlet on the center of the block, well, that's revision 3.

Tex, I will send you a link to the g-codes in the morning (don't have any of that stuff on my home computer).

Basically, the waterblock IS NOT that size. I'm going to shave it down tomorrow, and maybe even use that as my block.

EDIT: On second thought, I'm not going to use that as my block. I forgot to account for the copper in the heatercore and the resulting corrision will just be a headache in the end.
Attached Images
File Type: jpg Aluminum-010.jpg (85.0 KB, 42 views)
File Type: jpg Aluminum-011.jpg (64.7 KB, 34 views)
File Type: jpg Aluminum-013.jpg (39.8 KB, 46 views)
File Type: jpg Aluminum.jpg (93.5 KB, 56 views)

Last edited by ljohn787; 07-28-2005 at 08:35 AM.
ljohn787 is offline   Reply With Quote
Unread 07-28-2005, 11:45 AM   #22
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35
Default

I've got a block of copper and am going to cut into that. I'll leave an update on the status of the block, and I'll try to take some pictures along the way.

EDIT: Here are some pictures of the CNC machine. The first picture is finding Z0. The second one was cleaning the surfaces of the copper. The tool was actually working at around 1000RPM when I took that picture.
Attached Images
File Type: jpg copper1.jpg (35.8 KB, 63 views)
File Type: jpg copper2.jpg (58.3 KB, 58 views)

Last edited by ljohn787; 07-28-2005 at 01:03 PM.
ljohn787 is offline   Reply With Quote
Unread 07-28-2005, 03:20 PM   #23
JFettig
Cooling Savant
 
Join Date: Feb 2003
Location: Willmar MN/Fargo ND
Posts: 504
Default

Looking good, if your open to suggestions, hopefully before you mill it, leave fins in the center. itll increase performance decently.

Jon
JFettig is offline   Reply With Quote
Unread 07-28-2005, 03:59 PM   #24
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35
Default

Oh, I made that cut in the hopes it would increase turbulence. I kind of messed this block up anyway, so I guess I'll try that out tomorrow.
Attached Images
File Type: jpg copper3.jpg (24.1 KB, 34 views)
File Type: jpg copper4.jpg (15.1 KB, 32 views)
File Type: jpg copper5.jpg (34.8 KB, 32 views)
File Type: jpg copper6.jpg (55.8 KB, 37 views)
File Type: jpg cnc.jpg (38.0 KB, 45 views)
ljohn787 is offline   Reply With Quote
Unread 07-29-2005, 10:41 AM   #25
ljohn787
Cooling Neophyte
 
ljohn787's Avatar
 
Join Date: Jul 2005
Location: New Jersey
Posts: 35
Default

Alright.. the block I just made is a bit of a mess. I'm in the process of cutting a new one at painfully slow speeds, but I've got time.

As JFetting suggested, I removed that center hole. The mounting holes proved to be the biggest problem thus far, as I broke 4 tools on that and 1 tool on the o-ring channel (but that was my fault). The CNC is cutting the 4 mounting holes now, but I'm going in at maybe .1IPM. The part I as most worried about came out well: the channels. They're fairly clean, and the tool survived. I think this next block might be a bit better, so I will keep you guys posted on that.
Attached Images
File Type: jpg copper7.jpg (21.3 KB, 87 views)
ljohn787 is offline   Reply With Quote
Reply


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

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


All times are GMT -5. The time now is 12:13 PM.


Powered by vBulletin® Version 3.7.4
Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.
(C) 2005 ProCooling.com
If we in some way offend you, insult you or your people, screw your mom, beat up your dad, or poop on your porch... we're sorry... we were probably really drunk...
Oh and dont steal our content bitches! Don't give us a reason to pee in your open car window this summer...