Building Our Own CPU Water Block?!?

[907rider]

Here’s an idea, for extremely small "fins" (really grooves) these 9 thou thick cutoff wheels for a Dremel...

 

9 thou thick cutoff disks

 

One thing I saw about 5 years ago in the machine shop I worked in at the time was this.... We cut a piece of 1/4" round stock about 4" long, then drilled directly into the center of it on the lathe, not very far, maybe 1"  We threaded this to fit a 6-32 screw. Then we took and mounted this 4" long cut off wheel mandrel we had just made into the chuck of the Bridgeport mill.  We affixed a 1/64" thick cut off wheel for a Dremel to our mandrel using the 6-32 screw in the end.  Next we mounted the piece we were cutting in a vise that held it perpendicular to the table. This meant that it was now in the correct orientation to have grooves cut into it by the cut off wheel.  We cranked the drive on the Bridgeport as high as shed go and let'r eat.  Obviously the RPM on the Bridgeport was wayyyyyy slower than a Dremel so things were painstakingly slow... It took quite a bit of finesse, but eventually we got a feel for the feed rate and such and were able to cut some fairly accurate  and thin grooves into the piece.

 

 

[akhilgeothom]

10 hours ago, AlexTheGreatish said:

How hard can making your own water block be? Actually not too hard if you have a CNC Router... Also really hard when you have a Linus...

 

CAD: https://lmg.gg/WaterBlock

 

Buy Single Flute Bit.
On Amazon: http://geni.us/nAxR

 

Buy CPU Water Block
On Amazon: http://geni.us/QQAVw
On Newegg: http://geni.us/9ADQ

 

 

Please link to what swift tech does?

[tjd10684]

59 minutes ago, 907rider said:

Here’s an idea, for extremely small "fins" (really grooves) these 9 thou thick cutoff wheels mount into a Dremel...

 

9 thou thick cutoff disks

 

One thing I saw about 5 years ago in the machine shop I worked in at the time was this.... We cut a piece of 1/4" round stock about 4" long, then drilled directly into the center of it on the lathe, not very far, maybe 1"  We threaded this to fit a 6-32 screw. Then we took and mounted this 4" long cut off wheel mandrel we had just made into the chuck of the Bridgeport mill.  We affixed a 1/64" thick cut off wheel for a Dremel to our mandrel using the 6-32 screw in the end.  Next we mounted the piece we were cutting in a vise that held it perpendicular to the table. This meant that it was now in the correct orientation to have grooves cut into by the cut off wheel.  We cranked the drive on the Bridgeport as high as shed go and let'r eat.  Obviously the RPM on the Bridgeport was wayyyyyy slower than a Dremel so things were painstakingly slow... It took quite a bit of finesse, but eventually we got a feel for the feed rate and such and were able to cut some fairly accurate  and thin grooves into the piece.

 

 

Your probably on to something rather than a Dremel you could use a slitting saw or a jewelers saw to do basically the same thing but a bit faster. You might have to get creative with the fixture but I see no reason it wouldn't work. Actually if you look closely at the EK supremacy fins that may actually be their method of manufacture.

 

Edit: I just looked up slitting saws on MMC and they have them in widths down to 0.006 cost is under 30.

 

.... I might just have a new project in the works.

Edited February 15, 2019 by tjd10684
added info

[Clovenedhoof]

Can I please be your technical adviser/machinist? 
In all seriousness, you guys are doing great, and I love these videos.  I am a jeweler by trade and run a small job machine shop that makes hard to make parts from exotic materials.  I did not read all the above comments, I'm sorry if anyone brought these up already, but I wanted to give a couple tips that might help you guys a lot!

1. Consider aluminum for the water block instead of copper.  It's MUCH easier to machine, it's not as conductive as copper, but it should probably be enough for your uses.  Maybe even just for prototyping. 

2. Make sure you are using an easy to machine copper.  You are getting weird chips, it's hard to tell on a video, but you might do better with a copper with a higher amount of Tellurium like C14500 Tellurium Copper

3. Consider a silver base plate soldered to a copper mid-plate.  Silver is more conductive than copper.  And it's easier to machine if you are machining through the metal, than trying to machine a pocket. plus you can water jet cut it, or laser cut it.  (If you need someone to do the soldering, I will gladly do that)

4. When tapping, especially a larger thread like you guys are doing, consider using a drill press to hold your tap.  You want to chuck a live center in the drill press, then mount the piece you want to tap in a vice.  Then use the drill press to apply stead pressure and hold it straight, while turning the tap with the tap handle. 

Let me know if I can help you guys.

-Chris

[RejZoR]

You don't need to grease up gasket on direct pressure sealings. Greasing of gaskets is only required on screwdown sealings, because the twisting pressure warps the gasket, stretching it and potentially destroying the seal. Greasing it allows the two surfaces to glide across the gasket on final pressure moments without warping or stretching it.

 

Greasing the gasket only brings contaminants into the water cooling loop...

[ahuckphin]

Scrapyard Wars 2 throwback vibes 

[Velcade]

8 hours ago, Ezio Auditore said:

Would a gasket maker work instead of an o-ring?

Yes it would.

[XeSquid]

So... I want a LMG CPU Block, just for the heck of it.

[XeSquid]

1 minute ago, VegetableStu said:

the fin pattern would be dickbutt ._.

It'd be a reason for me to get a case with a window! For the first time in ever x)

[Woyta]

Yep. That video hurt my eyes. Really Linus missed DIY watercooling era? I am playing with watercooling since Athlon XP.

 

I recomend to check this old site of czech creator Erty. It shows lot of history. http://erty.sweb.cz

 

First blocks were just cooper pipes soldered to cooper plate. Thist method is still great for VRM cooler.  I used to have later revisions of these blocks.

Or second solution were blocks  with few drilled holes that were crossing inside.

 

Then we get acces to maching tools. Lathe were most common machine, many of us get acces. So design looks like christmass tree. For example this is waterblock from czech creator Erty.

 

With more and more milling machines design starts to change. That is time before CNC become standart feature. So usualy  there was only S shape inside.

http://erty.sweb.cz/gpu/DSCN2302a.jpg

Or few ribs. Still OK for high flow chipset block.

 

There is still few guys who are still making  block. Check http://www.aquacool.cz an look for discontinued models. They are showing progress too.

I got his 4 ram stick cooler.

 

 

There is lot of desigsn, I did some too. There are in atachement. Destiny should provide lot of performence but 15 years ago cost of manufacturing were to high for two students.

[Kroon]

On 2/14/2019 at 11:22 PM, tjd10684 said:

 

Its very rarely a good idea to flame polish acrylic. Melting the surface introduces stress into the material that WILL cause cracking eventually. You would be much better off polishing with Novus polish with a hard felt cloth. If you must flame polish then you need to anneal the part to even out the induced stresses. 

I use flame polish almost every day and never heard of anything cracking, not even horror stories.  Unless you are using extremely thin acrylic there is no risk what so ever of cracking. 

[Thermosman]

On 2/14/2019 at 3:06 PM, AlexTheGreatish said:

How hard can making your own water block be? Actually not too hard if you have a CNC Router... Also really hard when you have a Linus...

 

CAD: https://lmg.gg/WaterBlock

 

Buy Single Flute Bit.
On Amazon: http://geni.us/nAxR

 

Buy CPU Water Block
On Amazon: http://geni.us/QQAVw
On Newegg: http://geni.us/9ADQ

 

 

Some of these have already been said but I will reiterate regardless

• You can easily flame polish acrylic to give it a clear finish
• Any fins are better than no fins, you could get a 1/16, 1/32, to cut slots which would be beneficial regardless of how many fins you can fit
• Alternatively you could use an engraving tool with a steep tip angle which could cut grooves, increasing the surface area significantly and hence increasing heat transfer
• Always face off any mating surfaces between parts to ensure that they are flat and go together correctly to avoid leaks
• While you didn't have leaking problems, you may want to add more crush distance to the o-ring by making the groove not as deep to avoid leaking
• Some have mentioned using a slitting saw to cut grooves, however this would be challenging on a router because you would have to mount your part sideways, and with such thin slitting saws you would need to ensure that they are very concentric to prevent breakage
• If you modify a heatsink with micro fins already machined as others have already suggested, I would also suggest to use a sander in the direction of the fins, as trying to machine them would likely cause major damage
• When tapping, use a center, such as that that would go in a drill press, with constant downward pressure to keep it straight. I was not able to determine if you were using a taper or a bottoming tap, but either would cause issues, as the taper tap would not form the full thread in the bottom of the hole, resulting in you not being able to fit the fitting in. Using a bottoming tap would also not be ideal as it would be very hard to center if you have not already started the thread. The proper way to do this would to use a taper tap, but ensure that you have the full thread form through the entire hole, not only until the tip is flush with the bottom of the part. Chamfering your holes quickly, even with a hand drill would make it much easier to get the tap to go in straight even if you don't have a live center
• Other materials than copper such as aluminum have similar thermal conductivity to copper and are much easier to fabricate, as mentioned above, This would make it much easier to machine grooves
• If using any type of grooves, design with the criteria in mind to maximize flow directly through and over the grooves to optimize the use of the surface area, such as restricting the vertical area above the fins
• Minimize restrictions to flow through channels, and ensure that the copper is thick enough that it will not bend. Especially with this design, the center that was much thinner was likely not flat leading to worse temperatures. If you have problems you could face the bottom of the waterblock after assembly to ensure that it will be almost flat regardless of bending in the material, just leave extra thickness on the bottom.
• O-ring grooves and o-rings in general can be a pain, especially when your design isn't a simple as a small circle. Other solutions include various epoxy or silicone sealants that will do the same.
• In some cases it may be easier to put the o-ring groove in the acrylic block.
• Always test your waterblock with pressure while not over hardware first.

[Woyta]

Also turbulent flow removes more heat than laminar flow. But it is causing restriction.

 

Acrylic is tricky it depends what type they are using. After some bed expirence with cracking, bending on water parts, i have switched to delrin/POM. For me it is way better material. Less risky.

[PlayerX]

As I had some spare time and I found this quite an interesting topic, i tried to give it a shot.

I did some thermal simulation with a heat source of 140W input in this circle (30mm in diameter). For water I used a water-flow of 0.025kg/s at a input-temperature of 30°C.

On the left my design approach, on the right the LTT design approach. The temperature scale should be the same for both variants.

 

First the bottom of the cooler:

 

next the top of the copper plate:

 

then the water-flow:

 

3D-water-flow with surface temps:

 

Section with water-flow:

 

Section with water temperatures:

 

As I saw in this short simulation, the water-flow in the LTT design leaves a hotspot, which can also be also observed on the bottom (peak temperatures should be near 78°).

 

For my design I also moved the grove for the o-ring to the acrylic plate.

 

The fins can be machined with the following tool (I designed them with the use of this tool-dimensions: d1 63mm; t 1.6mm):


Was a fun thing to do. Hope you find it also interesting

[Emanbaird]

lmao i'd love it if LMG released their own line of manufactured hardware. The closest i think we've gotten is fans 

[Woyta]

PlayerX: If I send you  IGS files of Destiny waterblock (that one with jet) would you bee so kind and run simulations on it?

I would like to see how we designed it 15 years ago.

 

I asked coleaguges from simulation department but they are doing only crash simulations.

[PlayerX]

9 hours ago, Woyta said:

PlayerX: If I send you  IGS files of Destiny waterblock (that one with jet) would you bee so kind and run simulations on it?

I would like to see how we designed it 15 years ago.

 

I asked coleaguges from simulation department but they are doing only crash simulations.

Yeah, sure. Just send me a private message.