Scratch Build – In Progress Dimidium - A Build In Two Parts

Cant beat a good bit of CNC , glad you got all those issues sorted. I agree on the silicone for the Glass to Alu bond as well, if you used any hard glue/resin, I foresee bad bad things when the difference in expansion rates with heat take effect when the case warms up during gaming lol

 

I have to wonder how much thermal expansion actually happens to PC cases when they warm up. You do bring up a good point though!

 

It depends on the materials but some have a fairly big difference. I learned that the hard way on my previous case build, where the acrylic window bows outwards when it gets hot lol, as I rigidly mounted it.

 

Dang, that sucks. Thanks for the warning

 

Update 13: Righting some Old Wrongs

Some acrylic stock arrived to my door recently, so it's now time to get crackin' on the stragglers that I wasn't able to complete last time. Thanks to uh, you know, issues...





The first few ops went pretty well! The better finish also means I won't have insane amounts of polishing to do down the line.




Drilling and threading a billion holes! Unfortunately, I had to give these holes a final pass with an M3 tap after the fact as the milled threads were a hair too tight for my taste. I have to admit though, the reason for that is hilarious. I bought a new name brand (Techniks) 1/8" ER20 collet, and the runout on it is so much better than the previous one I was using that it's causing all my milled threads to end up way too tight. Now that the thread mill isn't wobbling like crazy, it's cutting a tighter radius than before even with the same exact cut settings.



Aand that's it for OP 1 on this part! Time to turn it over and cut some clearance pockets for the GPU riser...





Now that this infernal part is finally done, I can move on to the others! Just two more left.



Uh.... that's not supposed to happen! I have no idea why this happened, but the CNC randomly shut down and upon restarting it, I was greeted with this...thing. Static electricity? Cosmic rays? Gravitational spacetime warping? Whatever it was, it was pretty serious. The machine was out of commission for a day or so, refusing to respond to my commands. Each time I sent it something, it would output jibberish in the console, which was a bit terrifying! Fortunately, I eventually fixed the problem by reinstalling the controller's firmware.





A window for the cable channels! This should look very pretty if all goes to plan




This part is just a spacer for the pump. It should allow it to fit without bottoming out on the distro plate, assuming I got the thickness right!



Finished! I had already polished these by this point as I couldn't help myself. They definitely didn't look this nice right off the router. Now I just have to do the back of a lot of these parts.









Looking good! The probe is really paying for itself now. Naturally, I saved the best for last. The raid boss of the project returns!



Remember how I milled the 'wrong' side first? Well, Here's why. Since it's stepped down on one side, with one half being 15mm thick and another being 10mm, I have to make up the difference somehow-otherwise I can't attach this to the bed of the CNC accurately. If it's not flat, I'll be cutting at a slope, which is obviously bad.



Fortunately, this aluminum cutoff from one of the parts I made recently is just below 5mm thick, so I was able to use it as a pretty accurate and flat spacer to keep this part flat on the bed! Took some elbow grease to get it to that point, though.



Leaky superglue bottles...





God, there's a lot going on here. I'd better not have screwed something up!






Whew! This cut was too much for my machine unfortunately, but tuning the feed override helped with that.




Finally done! And the pump fits too! Thank the stars for that, I was worried about this as I've never milled a pump mount before. This also means I only have aluminum parts left in this build. What a milestone!

Anyways, that's it for now. See y'all next update!​

 

Im impressed how nice of a finish all these are for just a mist coolant, had always thought you really needed to smother acrylic in coolant to get a nice finish.

 

It's probably the high-dollar bits. I have my new setup's rpm at minimum and I'm still getting globby chips... With $2 carbide bits.
I'm curious about your feedrate and rpm now. The results are lovely.
I see you found McMasterCarr.

 

Me too, honestly! Most of the good finishes I'm getting are from my single flute Datron end mills. Gotta love the precision German engineering! I'm too cheap to get the polished ones, so they're just the bog standard sub-$20 ones that are really tiny and take forever to do anything. Not a big deal for one-offs though!

Glad you like the results! I think I was going at 20k RPM at a 0.04mm chipload? Something like that. So the feedrate hovers around 1000mm/min, though that is a little much for my machine with the chunkier end mills. And yeah, I can't get Chinesium end mills to make good finishes either

McMaster is expensive, but I keep coming back anyways. Their selection is too hard to pass up!

 

Pray MSC or Grainger don't buy them if you think they are expensive now. Msc bought Enco and promptly started selling their stock with a 3X markup.
Edit: I stand corrected. They are selling bits Enco sold for $2 for 22.50.

 

Jesus... Allow me to be cocky for a second, but I always KNEW industrial equipment was often ludicrously overpriced! I've been eyeballing the used market for anything mildly expensive I might pick up in the future so I can keep my kidneys intact. I just HAD to choose expensive hobbies...

 

Update 14: Odds and Ends

Sorry for the small hiatus! Would y'all believe me if I said it was because I was working on the mod? I made lots of progress, so I'll be sure to tell that story in the coming weeks. As of last update, the largest parts of this build are done being milled, so now I'll be tackling the itty bitty parts. There's a lot of them, so the sooner I start, the better.

Space in this build is hella tight, so I need to make a cutout in one of the structural parts so that the GPU will even fit in this thing. Shouldn't be too hard (famous last words).




This toolpath I'm using is kinda old... I've since updated my strats when roughing in aluminum. Well, it seems to be cutting fine anywa-



...oops.

Welp, my piddly attempt at workholding failed in combination with my hyper aggressive DOC and feedrate... It was just an el cheapo chinesium bit so I didn't lose out on much, but this still sucks. It's the only single flute end mill I had left with a reach longer than 12mm or so, which I'm gonna miss. The longer reach end mills only have niche uses but when they are useful they're absolute lifesavers.

'Oh well. I can salvage this part at least. This clearance cutout really doesn't need to look pretty anyways.'

- I chant softly under my covers as I cry myself to sleep
​

Much better! Really goes to show how important it is to 1. have good workholding and 2. make sure your CAM-fu is halfway decent. I got lucky though, normally mistakes like that mean your part is irreparably marred/gouged.



Now moving on to some other bits! There's a bit of leftover aluminum just chilling out on the bed of the machine, so I might as well turn it into some useful parts. I'm starting with an extraordinarily complicated part, the power button!





...that's all she wrote, folks.

Next part is a really tiny cable comb.





Wonderful! This guy will be sandwiched into one of the acrylic distribution plates. It'll organize all the non-power cables going between each half of the system, so RGB, USB 2.0, fan controls, etc.
The power button on the right will just be glued to a key switch I have lying around. Should work pretty well I think!

The next bit I'm cutting are some cable combs for the GPU PCI-E power. I'm hoping these carbide drills from Datron hold up, because I don't feel like waiting for replacements in the mail and I have a ton of holes to drill. They're surprisngly cheap though, at only $5 a pop for most of them. This is just the beginning of the mountain of holes I need to drill though...







Not bad, not bad! I need to make another one though.




I gotta make this specific cable comb 4 times. Not that this took forever or nothing, but knowing I have 3 more of these does make this feel pretty tedious...




...especially when the end mill I'm using breaks.

Ugh this SUCKS. Two end mill breakages in one update!? And this time, it wasn't just some cheap chinesium bit that ate it. This was my 3mm single flute carbide cutter from Datron, which also happened to be my favorite end mill too! The finishes this guy was getting were unparalleled... I think I was running too much coolant here, which eventually soaked into this painter's tape holding the part onto the machine bed and caused it to weaken -> pop off under all the cutting forces. Then the end mill dove straight into the loose part causing it to try to wrap around the end mill, which... well I think you can infer the rest.

I don't really have an alternative solution to this, and I have this end mill show up in almost every G-code program this machine will see in the next few weeks/months. In other words, I'm just gonna have to wait for a replacement to show up in the mail. These aren't horribly expensive at least, at around $15 a pop. This is going to be agonizing...

Sucks to end this on a low, but I can't do much about it I'm afraid. I promise next time will be a little better!​

 

I quit using tape on metals for just that reason. I think the force of cutting is too much for it, although in some cases the part would heat up enough to melt the tape's glue.
I've been using caulk/window glazing to stick my billet to MDF. When you are done you soak the mdf in water to free the part, because destroying the mdf is about the only way to free the part.
Also, You are leaving material all around the part when you can get right on the edge of the billet with tape. It reduces the amount of material you need, but it REALLY increases the life of your bits.
(Sorry if I'm repeating. I seem to tell people this stuff a lot. I'm betting I've said it here too. )

 

Thanks for the advice! I knew I wasn't being super efficient with the material I had at hand, but I wanted to guarantee that I wouldn't be cutting air where I should have been cutting metal. Plus maybe when I stop being a bonehead I can actually use my end mills to their fullest extent and actually worry about their lifespan. I am getting a lot better though!

 

Update 15: Spring Cleaning

The replacement end mill has arrived, so it's back to the grind! Well, kinda. While I was waiting for it to arrive, I decided to take care of some operations that were festering in the back of my mind, just waiting for me to finally get to. These are mainly things like little double-sided ops that require me to put parts back onto the bed of the machine for the 2nd time, a prospect I lowkey dread quite a bit. The more times I have to put a piece of metal on the machine, the higher the chance it'll get scrapped for whatever reason...


Starting off by drilling the holes for some M3 standoffs in the base plate of the build. These are for mounting the PCIE riser cable for the GPU to the case. I did the half that's facing the machine bed already (not visible for obvious reasons), so this is a wee bit stressful.

Thread milling the M3 holes, with the tiny 2.5mm thread mill from before. This was the first time I've thread milled metal on this machine, and it was surprisingly painless!

Perfect fit!

Now time to move onto Op 2 for some other parts. I'm cutting chamfers on the TG window mounting...bracket...things. Not sure what to call them. I'll elect to show more than tell for this portion of the update






This process looks simple, but it was a pain in the butt. I elected to just do these cutting moves manually with the DRO and some typed G-code commands rather than write entire programs for what was just some back and forth straight lines . That was actually pretty simple, but the main issue I ran into with these pieces was workholding (seems to be a recurring theme here, huh ).

Fortunately, I never had any of the parts slip or anything, but I had to be incredibly careful not to crash into any of these edge clamps, which also had to be put at just enough pressure to keep the piece from moving while not smashing the double-sided tape underneath and causing it to sag, making the chamfer inconsistent in its width. Still, they came out well, I think.



*sigh* You again.

Yep, this is the same part as earlier. It failed a test fit with my motherboard as my chipset cover was just a half millimeter or so too tall for this to fit without some clearance. Not the end of the world, though!




Alright, it works much better now. I probably should have taken a picture of me test fitting the motherboard to this, but I don't have 3 arms



SO,

I made an oopsie woopsie. A big one.

See the highlighted part in the image above? That's my motherboard's sodimm.2 slot, which if you're not familiar, is basically just an expansion board for some high-end Asus motherboards that adds things like extra fan headers, M.2 slots, etc. Well, in my original design for Dimidium, I measured it totally wrong, and what you're seeing is the corrected version of it. Which happens to cut directly into the largest, most expensive, most labor intensive to make, most visible, most... everything part in this system. Wonderful. Using this board isn't optional, I'm afraid. I won't have any kind of storage in this build that isn't m.2 storage, which is only available with this board.

However, all hope is not lost! Taking the heat sinks off this board massively decreases its footprint to the point where it should fit. Well, after some light modifications to this distro plate anyways.


Wish me luck. Hopefully this is the last time the raid boss of this project sees the end of an end mill.




Pants? Brown. Armpits? Sweaty. Success? We got em boys

I can confirm via test fit that the sodimm.2 board will fit now! I will have to forego the stock heatsinks, but that's way better than the alternative.

Okay! There's just one thing left before I go back to crackin at the aluminum parts I still have left. The wooden panels. I need to make a few edits to them to accommodate the anchor wires properly.







I had to do this four times since there are, well, four of these panels, but I don't think I need to repeat myself! The process was more or less identical for all of them. Wood is super forgiving, so this was also very quick.

Well, that ends the log for today! A much needed one for sure, but a little too stressful for my taste. My lack of experience in CAD design and CNC machining really shows in logs like this, but digging myself out of these self-made holes is immensely satisfying.

The next log, by the way, is gonna be funky. It's gonna be weird. I'll be making some 90 degree fittings out of brass to connect some of the distro plates in the build directly with each other, and it's so far removed from what I've been doing that it necessitated some...creative solutions to problems I encountered along the way. See ya then!​

 

Oh, boy. You get to learn about brass temper and part warping too. Don't go making any -long- brass parts. Keep them fairly cubic. I found out the hard way that brass will warp toward the direction it is milled.

Looking great so far. Every build I do I add even more space under the motherboard. -Currently using 1 inch risers and wishing I had designed for 1.5".

 

I did not know that! Fortunately, the brass parts just so happen to be quite cubic in nature. It's the 360 alloy so machining it shouldn't be too bad, I don't think. And yeah, I'm finding out the hard way that the space beneath/around the motherboard is extremely underrated!

 

1 more thing: Brass chips are the nastiest. It can look like harmless dust, but is razor-sharp and loves to get under your skin.

 

Update 16: A Process to make MacGyver Proud

Yeah I learned this the hard way.

Many of you may watch the Bit-tech YouTube channel and seen project Aquaceras. What a cool project, huh? That was one of the main inspirations for Dimidium. It was so cool that I decided long ago to shamelessly steal an idea from said project: connecting distribution plates together without fittings! To that end I'm making a set of 4 brass connection... fittings? Blocks? Unfortunately, I'm not experienced with machining brass, like at all. I don't even own a cnc vise (or a manual milling vise for that matter) either.

So uh... experienced machinists, look away.



BEHOLD!

Is it terrible? Awful? Dreadful? Hacky? Janky? Yep! But also, it somehow manages to have a death grip on the brass stock here, and that's all I need for what I'm doing now. It's simply some edge clamps, a chunk of HDPE scrap, and my 25-50-75 blocks. The edge clamps have a 45-degree surface on them which is what pushes the 25-50-75 block into the stock and clamps it in place. I can't believe this s*** worked... I was confident that this would work!







The proof is in the pudding! The finished surface here is unbelievably smooth to the touch. I think it's probably the best I can expect from this machine. I probably should have used the dust shoe, because holy mackerel are there a lot of chips on the garage floor. I was also getting pelted with them relentlessly as I was taking photos of this process too. They're clearly pretty sharp but no injuries or anything. The cuts sounded good and the chips look decent so I'm happy.




Perfect! The drilling and thread milling of these M3 (as always for this project) threaded holes was pretty painless. I have to do a lot more though, so I'm not out of the woods yet.





Cutting this slot for the watercooling fluid was suuuuper sketchy. I had it do a trochoidal toolpath (volumill in SolidCAM) to rough out this material, and it made some serious CNC dance moves that I wasn't even sure this machine was capable of making. It worked, but I won't be surprised in the least if something goes wrong in future attempts. To that end, I've slowed down the cutting parameters a tad. Chip evacuation was also not optimal, so I basically had my hand on the coolant line the entire time, making sure it was always pointing to where it should.



It worked though!

...now to do this 7 more times. This was just one side of one of 4 parts. I think it's time for a small highlight reel of the rest of the parts!







This is great progress, but I'm still not done though. Notice how the features on the faces of these parts are a little far to one side.



Much better.








Adding these chamfers was totally worth it



This was really funny. I used a bit of extra coolant occasionally, and it dries all sticky. Guess where the 25-50-75 blocks were sitting on the machine bed



Done! All that's left is to polish and then nickel plate these bad boys. The only real blemish on these parts is that one of the sides on one of these parts was cut at a slight angle, probably because the part moved in the junk pile masquerading as a cnc vise a little. Oh well, it won't affect functionality and I may be able to sand it flat.

That's it for now. Not many parts to cut on the CNC router left! I'm starting to see the light at the end of the tunnel ​

 

Been absent from the forums for a while so I only just now found this build log. What an amazing piece of engineering and art! I also love your detailed stories about your CNC, I can find a lot of myself in there, though you seem to have made quick progress judging by your later results, very impressive!

 

Lovely parts. Have you felt the strange burning sensation in your fingertips yet?
Alex isn't the only person that hates buying fittings. Sadly the link is dead, but there was an old log on X-Zone called "Superfast" where the builder used zero fittings. The cpu and multiple gpu blocks were connected by an aluminum manifold.
Also, I can't judge your clamping. I've done far sketchier. If it worked, it's good.