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

Fortunately, that was the only time I've had to move a clamp out of the way. I have some double sided tape and the classic superglue + painter's tape method for the trickier parts, though I try to avoid those workholding methods as they can be messy, at least in my experience.

...yeah, I try not to think about what was living in that wood too much
Shouldn't be hard to fill in those divots! That's a problem for future me.

 

Update 05: Spooky Scary Double Sided Parts

Now things are getting spicy! Each of these panels are close to being done, and I'm starting to get excited about the end products. Finishing them will be a massive milestone in this project. It'll mean that some of the largest and most geometrically complex parts of the build are done, and I'll get a much better idea of what this will actually look and feel like in real life. Staring at an imaginary 3D model of this build for a few dozen (hundred even) hours has completely thrown off my sense of scale

There's still plenty of cutting to do before that point, however, so nothing for it but to get back to the machine. Though most of the hours spent in this project, like any other CNC machining endeavor, are squarely in the planning stages of things.



This ^ chamfer is present on every one of the wood panels, so I just ran a single gcode file for all of them.



I started by cutting most of the material away into this stair-steppy looking feature. Pretty standard. You can see that I used some scrap pieces of wood to prevent the edge clamps from marring the parts, and they did a good job.



Then I used a 4mm two flute ball end mill to finish the big chamfer, making it look much nicer
I was worried that the feature might be offset from where it's supposed to be since I don't have the best method of probing in the world, but this is turning out really good!



Lastly, I shaved off what little extra material was left with my chamfer end mill. Programming this toolpath was really weird... I had to set the cutter as a ball end mill to start with and then change it after the fact. I suppose that makes sense as this one isn't really meant to do stuff like this as far as I know.

Despite that, the end product looks pretty nice! I only have to sand it a tiny bit to remove some of the lines and call it a day. In hindsight, using the chamfer end mill wasn't really necessary at all, but if it saved me 30s/panel of sanding/post processing, then I call it a win.

I went through this process for all of the panels, but they all went pretty much the same so there isn't much reason to show it.



This is where the lack of a good x/y probe really bit me... For the finishing passes on the I/O panel, I used a dainty little 2mm single flute end mill as it's the smallest square one that I've got. I wanted the filleted corners in the IO cutouts here to be as small as possible since the edges of the motherboard's I/O shield are fairly sharp by machining standards. Problem is, I didn't probe this amazingly well, and the end mill cut a little bit too far in the +X and the -Y, leaving this ugly step-looking thing everywhere:



The lighting here makes it look a lot worse than it actually is. It's only off by like 0.1-0.2mm, which for what I'm doing isn't terrible. Still, I had to spend quite a bit of quality time sanding these into oblivion on the parts that will be visible . The tight corners also made it difficult to get everything, meaning I have my work cut out for me in the post-processing department as I haven't gotten it all yet... At least i can do most of that while waiting for the other parts to be done.



Now to do the opposing side of the panel containing the cables. This one is a little bit trickier than the previous two. Still, it went well!



A little acrylic window will cover that pocket in the center, while the PSU cables will route through it. It's super extra and serves no practical uses at all, but I wanted something to show off the pretty cables

Last but not least, the front panel. Which is, of course, by far the hardest of all. Gotta save the best for last, eh? That reason being that it consists of two main parts: the panel (obviously) and a cover that makes it look nice and clean. Plus, it's simply the most complicated of the three. There's like 5 tool changes just for the panel, let alone the cover piece.



For probing in Z (and for the tool length setter on the right), I use this button in a 3D printed box. It doesn't seem like it would be terribly precise, but I've tested it and it can do +/- 0.02mm assuming there aren't any confounding factors. Not bad for some random Amazon push buttons! It's more than good enough for what I need.




Started off by making the cover first. This is the only stock I have left at this point (that hasn't been uh... lived in), so I'd better not **** this up!




I had to remove so much material for this. Those chips off to the side are like 2% of it, with the other 98% clogging my shop-vac filter. Speaking of which, I've been hogging so much material out of these walnut planks that the vacuum filter's clogging almost instantly, forcing me to clean/replace it. At least this part's simple. It just has this L shaped boss in the center and some cutouts for magnets so it stays put when it's in place.

Not much more to see of this cover piece, so now onto the front panel!

Unfortunately, I completely forgot to take photos of this job . I think I simply got too caught up in trying to make it turn out good. Which, in fairness, seems to have worked out pretty well in that respect



The big L-shaped cutout in the center is a maintenance window so I can reach in the system and flip the lever on the drain valve for the watercooling loop, while the smaller cutouts on the bottom are front IO stuff/the drain port for the loop. Those smaller pockets on the sides are for some neodymium magnets so the cover stays in place. I'm hoping a lot of these quality of life features make this build at least somewhat functional, since I plan on daily driving this thing!

​

Rather predictably, the cover for this didn't fit perfectly right off the machine. I spent an hour or two sanding the L shaped boss down on its edges to make sure it fit the way I wanted. I had to sand the top and bottom for a while as well... thanks to tool deflection, parts that are supposed to fit together like this almost never do without some post processing. I do try to compensate for this with spring passes and a negative stock to leave, but it's difficult to predict how over/undersized the final product will end up. Thank the stars I invested in an orbital sander otherwise the sanding would drive me insane. Sanding (at least to me) gets boring incredibly fast. I can file all day though, not sure what's up with that

Since I made these parts from the same chunk of stock, the grain matches up amazingly well! I'm incredibly happy about that and I didn't even attempt to take it into account initially. I'm sure there are things I could have done to make it line up even better, but I'm perfectly content with 'pretty good'. Especially since it was entirely unplanned

Now that the main panels are basically done (at least on the machining side of things), I might as well finish off the rest of the wooden parts. Not much left now! Just small stuff.

 

Lovely. You may have planned way more magnet than you need.

 

Probably. Better too many than too few, eh?

 

That depends on if you want to be able to remove the panel... ever.

 

Update 06: Wrapping Up The Walnut Parts

Given that I've decided to separate this build's construction/manufacturing (those words make this operation sound way more competent than it actually is) by material, it would seem that I'm running out of walnut parts left to make. What a milestone! I'm tired of inhaling sawdust every time I enter the garage, but now I'm moving onto materials where surface finish matters a lot more and mistakes become more expensive. My track record in that respect isn't terribly promising, but I did start with the most forgiving material first for a reason. Boy am I glad I did.



Not all of the currently un-milled parts are visible here, but the rest of the parts are all just small fries like these covers. Starting with these instead of those massive, complex panels might have been a good idea huh?



Remember the first side panel that I killed with that poopy tool path? Might as well use its corpse for something. I simply faced down the thing into oblivion until there was only a 5mm layer of walnut on the bottom, which is where I'll cut the rest of the parts out. Well, almost.




Unfortunately, I lost some steps during the making of the this part, which threw the spacing of its mounting holes out of whack. Oh well, there's plenty of space left for a redo! It's a cover that goes right underneath the GPU. Its sole purpose is to hide ugly RGB cables, so it doesn't need to do much except look pretty.








Here's ^ the milling a bunch of cable covers. They came out pretty good! Some quick sanding took care of whatever machining marks or burrs were left. It's nice to do some easy parts for a change, lemme tell ya.

All that's left are the case feet! They are a bit thicker than 5mm unfortunately, so I can't just keep abusing this current piece of stock I've been using for them. Time to grab some random stuff I still have lying around.




Just like before, I had to core out the majority of this stock to get to the thickness I wanted. Unfortunately, I ran into a bit of an oopsie in the process



In making the second case foot, I forgot to face off the material above where the part would be cut out. Good thing I was watching, otherwise the 8mm end mill I was using might have been toast!
It plunged straight into the material and starting cutting away like nobody's business. Well, given that it sounded like a banshee being thrown into a wood chipper...I'd say it was definitely somebody's business.

On the bright side, everything ended up OK. I just put on a new pair of underwear and carried on.





The case feet also have magnets on them too, so to cut them out I used this rather creative workholding technique. It worked like a charm! A vise or something would be nice though.



Some small blemishes of as per usual, but I really like these parts. They're small, but they put just that little bit of extra flair into a build-which is something I've always found lacking with most off the shelf cases (like the one I'm using with my current PC).



Christ, that layer of sawdust and wood chips down there in the shop-vac is a solid 20cm thick or more! No wonder the filter kept getting clogged so quickly eh? No way I'm emptying this thing inside the garage, otherwise I'll be breathing in this stuff until the heat death of the universe.

Well... that's it! For the walnut parts anyways. What a rollercoaster this has been, and I'm barely 1/3 the way through the machining stage of this project. I still have a mountain of post-processing left to do...

Still, I couldn't be happier with the results! Actually, I could. But I'm just glad to hit this milestone either way. Next up are all the acrylic components that make up all the distro plates in this build! Those are going to be a little tougher to manage than the walnut, as surface finishes are going to be significantly more important (unless I want to spend days/weeks sanding+polishing ) and mistakes can't as easily be buffed out after the fact. The machine seems to be running much more reliably than when I started, so I'm not too worried about that. I think the biggest danger here is my tiny pea brain commanding my machine to do dumb stuff, since I'm not the best CAM programmer on the planet. Yet ​

 

Milling 1" down to 5mm? That's some expensive and increasingly rare mulch you have there!
Somebody get this guy a band saw!
The parts are looking great.

 

Thanks

I was determined to use that failed part for something useful, dammit!

 

Update 07: Distro Time!

I've been into PCs and PC building for almost a decade at this point, and as soon as I figured out what a distribution plate was, I NEEDED one. Issue was, until recently they've mostly been relegated to crazy custom builds designed by modders or to people with deep pockets (or both), neither of which I considered an even touchable prospect-until now! Clearly the solution was to just do both of those things as I have

Either way, this build is getting positively inundated with distribution plates, and nobody can stop me

To do this, I bought several large sheets of cast acrylic, and I'll cut the parts I need out of them. Might as well use all the space available to me with the CNC router!



Since I already milled the top half of this distribution plate, I'll start off with the bottom half here ^. This one plumbs directly into the GPU since I didn't have space to add even right angle fittings out of the stock terminal on the waterblock. And because I think it's cool




As per usual, I faced down this 1/2 in stock to a much-cleaner 10mm in thickness. Unfortunately, the facing cutter and the toolpath I used for this op kinda sucked, so the finish is hot steaming garbage. If your eyes bleed while looking at the rest of these photos, just know that mine were almost certainly bleeding way worse

Well, I can always sand and polish crappy finishes away. Is that a godawful strategy? Yep! Oh well.



Drilling and thread milling these M3 holes is stressful. I'm using a puny 2.5mm carbide drill from Datron here, but that's a chunky boi compared to what I'm using for thread milling these holes:


That's a 4mm end mill on the left for reference.

With a ~2mm cutting head and a ~1mm neck diameter, this tool gives me the heebie jeebies big time. These thread mills are not cheap by the way. I would have used M4 holes like everyone else for these distros, but space is at an absolute premium in this build, and in many areas, there is legitimately no room for M4 screw clearances.



Whew! That went much better than expected. I did do some tests before starting this part to get the settings right, but using such a tiny tool is always scary. Always.




And it's done! Mostly. I still have a good amount of work to do on this part, involving some cutting ops on different sides, but I couldn't help but do a test fit. Naturally, that's a rather small part to be cutting out of such a huge slab of acrylic, so time to get onto the real show: the distro plates that go in the center of the build!





I'm starting off each of these with drilling and thread milling all the holes. Unfortunately, I ran into a major issue which I've previously eluded to in my last interim that nearly scrapped this part... It was the issue with the X axis drives randomly triggering their OCP protection because their power supply was in the process of slowly dying. So frustrating...




Did you know that in acrylic, you can drill with thread mills??? Me neither

Somehow the little guy survived cutting these gouges in this part! I was not happy about this predicament, but at least I didn't end up with a gouged part and a broken tool. Now if I had standards, I would have just stopped here and tried again. However, I figured that these gouges are so small that I probably won't even notice them at the end of the day, so I just soldiered on. Begrudgingly.





Looks pretty good, right? I thought so too, but unfortunately... This part ended up being scrapped anyways. Issue was, one of the roughing toolpaths (for that half circle-shaped pocket in the center) had a Z offset in the wrong spot, so it cut about 1.5-2mm deeper than it was supposed to. Meanwhile, I was watching the machine cut this like a hawk, ready to stop it if the OCP issue from before reared its ugly head again (which it did, several times). Only for this infernal piece of overpriced plastic become scrap anyways

...I had to take a few days off the project after this to calm down. And to figure out why the drives were giving me lip. At this point, I hadn't discovered the fact that it was their power supply yet.

Once I dried my tears and dusted off the machine, I got right back to work. Not much else to do but to keep pushing forward.




Much better! Well, mostly. The parts I was able to make successfully came out quite nicely, with one caveat: the finish, as you can likely see, is hot stinky poo poo. Especially in the big pockets in the center. Eeesh. My machine was making quite a racket during the roughing (and even finishing) passes. That's mega disappointing on the part of this machine honestly... I thought building it out of the largest steel plates and aluminum extrusions I could afford would let it at least avoid chatter in such a forgiving material like acrylic. I'm not expecting this to be able to fly through hardened steel at supersonic feed rates or anything, but this really isn't what I expected. Doesn't help that my CAM skills are still severely lacking, too.



...yeah about that, turns out my selection of materials for the machine frame wasn't the problem, but more on that later

Anyways, that's about as much information as I'm willing to cram into a single post, so I'll leave the rest for the next few updates.
​

 

My rig started messing up like that. It was caused from an overheating motor controller.
-Bad finish/loud: The material is getting too hot or the bit is old. The recommended feeds and speeds depend heavily on the ambient temperature too. I'm running about 7-9 IPM at 4200 rpm. I found really high rpm throws coolant off the bit instantly too. Spraying it with (more) water on the finishing pass may help a bit too. I've gotten good finishes with old bits by flat out swamping the cut.

 

Thanks for the advice! I cheaped out on some of the tooling early on and now I'm finding out I would have just saved money in the long run getting the good stuff straight away. Buy nice or buy twice, as they say. That facing op that looked crappy was the work of a $12 facing cutter from Amazon. Does well in wood, tragic in anything else. Didn't seem to matter how much coolant I threw at it. In hindsight, I'm not sure why I thought it would do a good job

My brother was doing a surveying internship this summer. That meant 50 hrs/week of standing in the desert sun. He would come back from that, step into the garage, and tell me straight up that it was way worse in there than anything he experienced at work! If I could cut the acrylic somewhere cooler, I would.

Fortunately, I did figure out how to get much better finishes pretty soon after this. I stepped up my CAM game immensely.

 

Update 08: Distro Time, Part II

While that large slab of acrylic is stuck in the machine, I might as well finish off the parts I'm currently cutting from it. Less part setups = more accurate = less time investment.





This part came out a lot better than the others! I changed my cutting strategies a little, and that seems to have produced better results. Especially in these O-ring channels, they look immaculate.



No vise? No problem! I covered this small distro from before in tape to keep the surface from getting scratched, and then used a combination of dowel pins, edge clamps, and superglue/painter's tape to secure it. It's certainly not optimal, but it worked beautifully.




I also cut water channels on the bottom so liquid can get in and out of the GPU waterblock:





My GPU waterblock from alphacool came with this little logo plate on it, and I knew I wanted it in the final build. I don't think it could have turned out much better than this! It fits perfectly. I just need to polish these parts and it will be good enough to be in the final build.





Almost all of these parts are double sided because I hate myself. Without a proper probe, this is a huge pain. I just put a dead endmill dowel pin in the spindle and use a feeler gauge to locate the parts. It works OK, but it's sloooow. The dowel pins in the fixture plate make lining these parts up at the correct angle a breeze, at least.





Thread milling! Side note: It's a small miracle that watercooling fittings thread onto these ports. Most threads are set to a 60 degree angle, which is also what this thread mill cuts. However, G1/4 ports (along with a handful of other British pipe threads) are dumb and have 55 degree angles. Seriously, why do people do this? Technically I should be using a whitworth style thread mill or just tap these ports, but I'm cheap and I'd much rather just use the tools I already have. It's not optimal, but it works great!





This part will go in front of the motherboard so it's gotta turn out good, as it will be one of the most visible parts in this PC. Also, I came so close to this edge clamp while facing this thing




Finish here is much better, but still not great. Well, on the top anyways. On the bright side, the finish in that cutout in the middle looks amazing. It's a bit hard to show on camera, but I won't need to do any post processing in here at all. And I can easily sand the top to make it look nice, too.



Alright, time out, time out. My CNC just had a stroke and the X axis (the long axis the gantry rides on) has been binding and stalling. Methinks it's a misalignment in the linear rails from my rather haphazard initial construction of this machine. If the rails aren't parallel within a very tight tolerance, this can happen. Time to fix this!




...don't judge. Despite looks, this setup did work, I promise.



During the alignment process, I had to rest the gantry on the rail blocks without any sort of support. It was incredibly sketchy, as a strong gust of wind or an accidental bump coulda sent this whole thing crashing down. Fortunately it didn't, and I got the rails lined up within 0.03-0.05mm across their entire 1m length. Took forever, but...





...it was so worth it.

Turns out, the gantry wasn't properly bolted onto those rail blocks this whole dang time. It was literally just resting on them, able to jingle around in place like a giant maraca or something-which explains a lot of the chatter issues I was running into before. Fixing that means my finishes just got waaaay better overnight! Oh, and I guess there's no more binding either.




I had to do this part twice due to a misalignment, unfortunately. It's one half of the distro plate that has the fill port on it for the loop.




Who cares about that when the finished product looks this good, though? Well, at least compared to what I was getting before!

Alright, next update will continue the last bit of these acrylic parts. I'm much more confident in this machine now, lemme tell ya. Which is very much needed, because next time I'm going to be cutting the most complicated part in this entire saga by a landslide. It's basically the raid boss of this project. Just look at how many operations there are for this monster:



Stay tuned!
​

 

That is MUCH better! ...wait, You didn't have the gantry bolted on?

 

Wow...that was a lot of update to digest...
It is great that you discovered the issue with the gantry...by accident
Maybe worth to check on all the bolts again with a torque wrench some fine day...

And don't worry. I source my dowel pins exactly the same way like you do

 

Yeeeeeeeah... it was super subtle, in a very hard to reach spot. They were countersunk screws and they were too long by like half a mm, so they weren't actually securing anything.

I would have used thread locker on all of them, but I didn't think to when I was building this thing initially. In hindsight, I didn't think to do a ton of super important things initially. Live and learn, I guess.

 

Update 08: Conquering the Raid Boss (Mostly)

I left off yammering on about this huge distro plate that was gonna be insane to make, and now I guess I actually gotta make it now. This is one of those parts that I really really do NOT want to mess up, because the stock was expensive and this whole milling operation is going to take forever. Know what takes even longer than forever? Two forevers!

This part is complicated for quite a few reasons. First of all, it will have space for a D5 pump, which require the plate to be >15mm or so in terms of thickness. It also has ports on it for a temperature sensor and the drain port. Not to mention it's got a small reservoir in it, its thickness varies because of space constraints, there are some ports for the CPU waterblock, there's the function generator, the zero-backlash hypocycloidal speed reducer, etc.

Small tangent incoming:

Ideally, a small ish build like this would use a DDC pump for space constraints. I'm repping one in my current build, but tbh it doesn't give me the warm fuzzies. It makes an annoying whine if it runs at anything above 20% speed, and more frustratingly, it has issues starting up when I turn the PC on. If there's any crud in the loop, it'll get on the impeller bearing and then you have to tear the pump apart and clean it for it to work properly again. My current build sucks, so the loop gets dirty almost instantly no matter how hard I try to clean it and the cycle starts all over again (galvanic corrosion... ). My brother has a D5 pump in his system and he just sets it at 100% speed and forgets about it. Totally silent and has 0 startup issues. So despite space being tight, I'm squeezing one in this build dammit!​

*Ahem* Anyways, I think it's high time I showed y'all why this is complicated instead of just yapping about it.



Here's ^ the part in the context of the model. But before I jump into it, I'm gonna have to finish the stuff that's still on the bed of the machine. Sorry, I promise this update isn't clickbait!



Which would be this guy ^, for context. It has the fill port on it, so it is pretty important-but it's also relatively simple which is a nice break from the chaos that is to come.





Started off with drilling these M3 clearance holes and then I moved onto facing the part. The finish is great again, so I'm glad that last part wasn't just a fluke or something! Still, I had to claw this kind of performance from my machine's cold, dead rails, so I would be lying if I said my expectations weren't high.





I completely forgot to take a picture of this part when I cut it out of the stock here. Just ignore those o-ring channels on the back, OK? I totally didn't cut them weeks later and then hastily take a photo after the fact or anything. Nope!

Ok, now it's time for the main course!





Even that crappy facing cutter is getting much better finishes now. It's still not good, but I changed up my cutting strategy anyways, so this doesn't matter. This next part involves cutting away a ton of material, so I'm just gonna get that started and do something else while the machine does its thing. It'll take a while.




Christ, what did I come back to?? Did like my cat take control of the CNC while I was gone or something?!

Seems like the dust shoe popped off its fixtures and got absolutely choke-slammed by the machine while I was gone... I just hope the part is ok. Sucks that I have to make yet another dust shoe though. That marks the second time I've had one die on me



Wheew! Looks great! All of that was planned, naturally.








The 3mm single flute datron end mill I was using for this is awesome. It leaves spectacular finishes compared to many of my other end mills. It also looks like the pump cutout/ O-ring channel is turning out well, which I'm super pleased with as I was dreading this initially.




Would it be a proper distribution plate without copious amounts of tapped holes? I'm super glad I invested in thread mills now, lemme tell ya. Hand tapping all of these would suuuuuuck.






Some fine detailing work with a ball end mill cleaned up these stair-steppy surfaces quite nicely. Not much left now, just have to cut this part out of the stock!



And this is why you check your tool stickout/clearance ahead of time




Ladies and gentlemen,



We got em. Kinda. What, you thought this was just a one-sided part?

I'll be doing the other side later, though. In hindsight, I should have done it first because of the stepped surface of this panel. It's going to be mighty tricky to lay this part flat on the bed of the machine now. Plus I'm planning on knocking out Op 2 on these acrylic parts all at once. Some of them need to be redone, so that's coming at a later time. But if I'm being perfectly honest, I'm mostly just dreading Op 2 and I don't know if I have the stones to subject myself to more of this thing.

Well with that, I think I'm ready to start the aluminum parts! Might seem weird to say, but I'm rather excited to be moving away from wood and plastic-for now. Sure, aluminum is less forgiving in general, but most of the parts made of the softer materials in this build are egregiously complex, as many of you can tell. A good portion of the aluminum parts are just rectangles with some holes in them. That sounds nice. ​

 

Well done
Needs some polishing and you are good to go...

What D5 pump (brand) does your brother use, that is silent at full speed?
I tried several so far and they are my standard choice too, but none of them spins higher than 35% - for noise reasons.

I made a distro for an O11 front in the early days of that case, but it was a complete fail, because the integrated D5 just wouldn't go silent...

 

Thanks and indeed! The polishing steps shouldn't suck too bad.

He has one from EKWB circa 2017 or something. It's in one of those integrated pump/reservoir combo units, which I'm guessing helps with noise since it has a built in vibration dampening ring...thingy. But yeah, he just runs it at 100% all the time and it's practically inaudible. If my new one for this build is even a fraction as nice as that, I'll be happy!

 

He can consider himself lucky then.
I have EKWB's D5 with their pump tops in 2 builds as well and they are noisy AF at 100%...
I am ok with 35% though, so I keep them and carry them on from build to build...so far

 

Parts look great!
Getting something in the spindle like the part coming free is always fun. You dodged a brown pants moment there.
It wasn't complicated until you cut the wrong side first. You have the offset numbers in the model. Use that to make a clamping spacer.
D5s were originally popular because you could run them on lower speeds and make them silent. They were loud at full speed, though. Your brother probably won the lottery and got one that was balanced absolutely perfect. Your best bet to isolate vibration is to use a FAT O-ring on both sides of the mounting bracket.