Prototype Bendy 8 - Fabrication Focused, Modular & Expansive

[SnoWFLakE0s]

Hello SFFN,

I'm back with what is a drastic revision of my original custom case concept. Having gone through an initial "real" prototyping cycle, I am now pretty well informed in my tooling and manufacturing capabilities... and I also just wanted a much nicer product then what the first prototype was shaping up to be. With these lessons in mind, I went through a total renewed design from the ground up.

Some key design goals that stayed past the initial concept:

• Easy to maintain - modular
• Accommodate "standard" ITX components--no Flex PSUs (easier to source reliable hardware)
• Wider hardware support (up to smaller 2-fan GPUs, 220mm)

Some "new" goals and features of this rendition:

• Be very easy to manufacture--despite my use of relatively simple geometries in earlier designs, I found actual manufacture, especially for a hobby machinist at best, is rather difficult. With only a simple press brake at hand (a pretty beat-up one at that) I limited myself to linear parallel bends in part geometry, and also did not use any overlapping corners (which makes setting up tooling much, much easier). All in all, tolerances have been made looser to allow for parts
• Be cheaper. For one, the nesting layout for my earlier design required a larger, custom cut of steel to produce--I've been able to compact this down into a 2x2' sheet with stock to spare for extra replacement parts. With the right timing, you can buy a sheet of 20 ga steel for around 8 bucks--pretty economical to acquire. I've also made it a point to make a common failure point (like angles between parts) a separate piece that is very cheap and easy to add to the nest for extra potential replacement parts should you mess up some.
• Have modular I/O. I've had this long obsession with panel mount I/O connectors--Amphenol makes them in a form factor commonly used for A/V applications (with various names, but typically called type D panel mount connectors)
• Support notable amounts of 2.5" storage. Me personally, I have quite a few 2.5" drives that I use, so I really needed something that would allow me to stick some extra drives in my case. This rendition of the case can support up to 3 total 2.5" drives in a small bay.

Anyway, some renders...

 

[SnoWFLakE0s]

Steps in Production​

I prepared a 2x2' A1008 sheet for the laser cutter and got cutting... With more optimized geometry, cutting the entire thing only took 36 minutes from start to finish. I'm sure this is more obvious for the more experienced, but using fillets liberally across a laser cut design helps the tool head retain more velocity through passes, cutting down on machine time notably. This is a bit different from machining where fillets should be used very purposefully as they are an extra processing step, but for a "2D" operation like this, they are very useful especially on repeated geometry like the vent patterns shown below.

Here is a layout of all the parts after some basic deburring.

I really should have cut these so that all the engraving lines were on the correct side for bending, but it's nothing a scribe can't fix.

None of these intersect and are very easy to bend on a super basic press brake, in poor condition at that, minus the drive bay which was a little more hard to fit. I've also tested using a vice as a makeshift stamping tool--if you had some sort of V-block and a square bar, you can bend 20 gauge steel fairly nicely and cleanly. More on that later.

Here is a layout of the parts after all the bending. They look pretty damn good!

Some parts are not that great out of the brake, and need some retouching to get right. Particularly, with my tooling, the blade of the press brake was not quite linear, resulting in a notable variation of bend angle along flanges. You can help flatten this out with a vice and some parallels or other flat sided stock.

Clamp the un-flat flange between the two faces of the stock and use the vice and apply some good pressure to form the metal into one straight piece. You can also slightly change the placement of the flat stock, placing one corner against a flange corner, a piece behind the workpiece, to sort of bottom bend a flange to spec.

A nice resource that was rather quite helpful: https://www.komaspec.com/about-us/blog/guide-to-sheet-metal-bending/

Due to the geometry of my drive bay, I couldn't fully bend it to 90 degrees on the press brake--you can either opt to use a technique I forgot the reference link for, in which you bend a centerline between the two bends on the opposite side to gain clearance, or do what I did and only bend the metal about 60-70 degrees, then form it to spec on a vice. Makes me wonder what I can achieve with a makeshift press using some dies and a vice...

 

[SnoWFLakE0s]

Rivnuts​

If you've ever seen the most recommended post on this part of the forum, you'd probably see a thing or two about PEM self-clinching nuts. They're nice, but they're not exactly an easy install job, and require a bit of tooling to get right. Some quick research tells me that you need about 3500 lbs of force on each nut to install them properly on steel, and I did not have access to the kind of punch to install those to spec, so I opted to make use of "normal" rivet nut or rivnuts, which are far more abundant and easy to find. They rely on the buckling of a midsection on the threaded rivet to provide the backstop on the rivet and anchor it in place, which is pretty easy to get right and can easily be done at home.

I initially bought a pack of aluminum M3 rivnuts. This was a terrible decision--I will strongly urge anyone who thinks to use rivnuts to go for the steel variety. The aluminum threads strip far too easily, and are nigh impossible to install correctly. Even with steel nuts, the installation of fasteners were not exactly easy... I initially tried to use the tooling that typically is used for this application, where you have a bolt (M3 in this case) against a thrust bearing, which allows you to hold the bearing in place and use the threads to pull the rivet into place. I ended up with issues accurately and reliably getting the rivets to install, and the threads weren't too happy. The twisting action of the thread-pull method doesn't seem all that great.

Then I attempted to use a drill press with a dowel pin to use as an improvised hydraulic press, but that resulted in what you see below...

All too often I came across this defect, in which the pressure applied was off the axis of the nut, causing it to buckle to the sides. From what I could see, it was hard to get uniform axial loading on the rivet, and often the rivet would buckle off axis like seen here. I eventually figured out how to install these nicely, and reliably:

A washer, a dowel pin partly turned down to M3 internal size ~2.8mm, and a 123 block all in a milling vice does the trick. You have to align the dowel pin through the nut, with the hole of the washer aligned over a hole in the 123 block. This solves a couple of different problems:

• The dowel pin provides an internal "guide" for the rivet to ride along as it is crushed.
• Using a vice prevents the pressure from backing out as you install.
• You can go nice and slow, monitoring the rivet as you go--much easier to get it right.

Once aligned, this setup gives very reliable, clean results by simply crushing the nuts into place. I just used a dowel pin that was lying around (and it was hardened steel--the lathe wasn't all that happy) but you could probably just grind down a regular M3 socket head screw to get the exact same results.

This is about as far as you want to go. You can give it a little more, but avoid over-crushing the nuts. The result is a very nicely aligned, secure installation of threaded fasteners fixed to your sheet metal. You can install some as test pieces to see how far you want to go, I opted to go as far as the seam lines.

With all fasteners installed, it's time for a test assembly...

We're still missing the plywood front panel, but this is pretty dang good. I'm not getting extreme blowout anywhere like I used in my older design and I've got nice clean angles everywhere! Very happy with the results. Looking like a nice case.

 

[SnoWFLakE0s]

Test Assembly​

I got some of my components actually in the box. Great results, here are some images.

Please kindly ignore that the side panel wasn't screwed on properly

Now all that's left is finishing (painting) and the front panel.

 

[Arboreal]

This looks a good design and a great project.
Having had a 4.5L case with a Flex ATX PSU, I have avoided those PSUs since then.
Your design looks similar to the old Metalfish S3 Plus, which has now become the Metalfish T40.

I have thought for a while that this design layout is a good compromise between keeping the size small and having an easily available quiet power supply.

The front panel has added a dash of Densium style to add some variation in finish.

I look forward to seeing the front panel added and how the modular IO works out. The male/female D connection combo may be a bit bulky, but I am happy to be proved otherwise!

Keep up the good work

 

[SnoWFLakE0s]

This looks a good design and a great project.

Thanks. Definitely Densium-inspired, but that Metalfish case definitely feels very similar to what I'm building. Not a huge fan of the design language there though, so I'm pretty happy with what it's coming to. Going completely dual-chamber with the inner wall might be a great idea, although that would mean I'd have to design a bit more around fitting everything. Building in this has proved to be very easy due to every panel of the box being just a screw-in job.

Your design looks similar to the old Metalfish S3 Plus, which has now become the Metalfish T40.

I'm looking at the T40 now and looks like they actually changed the design to be more similar with what I did. In all honesty the drive bay was a bit of an afterthought after I arranged the layout, so I'm not 100% sold on the way it looks, but it works and helps bear some load of the motherboard tray. This still comes in smaller than the T40/S3 Plus in what I deemed more critical dimensions so I'm pretty happy about that.

I look forward to seeing the front panel added and how the modular IO works out. The male/female D connection combo may be a bit bulky, but I am happy to be proved otherwise!

I plan to release the plans for this case, and have parameterized it such that it should be very easy to configure expansion in pretty much any direction, especially if one would prefer having a top exhaust fan. Even in its current configuration, with an open front panel instead of a solid variant a standard 120mm fan can easily be mounted, so I have also been designing a ThinkStation-inspired plastic front panel to add more front panel variety.

I'll probably be posting something about the I/O modules, but the nice thing about these is that they come in pretty much any configuration... from audio to USB to DisplayPort and so on. Practically hot-swappable, I just became so infatuated with this style of connector for no reason a while back and bought a whole load... so I better use 'em!

 

[Arboreal]

I'd love to do a custom build but I have no design or fabrication skills.

The T40 is OK and has allowed me to downsize my LAN box by nearly half.
It will be going out tonight and seems pretty well cooled with the AXP120-67 and the metal mesh sides. They look to be a real improvement on the fairly basic vented acrylic sided they originally used.
I only have a GTX 1650 in there currently, with a short RTX 2060 in mind to.up the horsepower.

Also, the handle. I'm a sucker for small cases withbl handles! My first one was a 22L mATX rig with a mag alloy Toughbook collapsible handle modded on to it.

Will you be painting or powder coating it?

 

[SnoWFLakE0s]

I'd love to do a custom build but I have no design or fabrication skills.

This was definitely sort of the target. This is easy enough to bend on the simplest of press brakes in a machine shop, and although I have free access to a metal laser cutter, it would be fairly cheap to buy these parts from a fabricator like SCS and bend them yourself (or just have them bent, but that would defeat the purpose of this design). It took me 5 minutes to figure out how to use the janky press brake, you can also get good bends with a vice and a wooden jig.

Also, the handle. I'm a sucker for small cases withbl handles! My first one was a 22L mATX rig with a mag alloy Toughbook collapsible handle modded on to it.

It would be fairly simple to add a handle onto this--it's still made of nice strong real steel and can take a beating.

Will you be painting or powder coating it?

My local powdercoat shop gave me a pretty absurd quote for these, so I am going to be painting it myself. Again, as I have access to a nice shop, I can do some really good prep work (sandblast, clean, prime, paint).

I am also entertaining a slight modification that would expand the case along its width, as currently it is basically flush with the sides of the SFX PSU. This change would allow the PSU to sit pretty flush on the bottom panel, although I am debating the efficacy of such a design. It would give about ~10mm total extra clearance for either CPU or GPU (probably better spent on the CPU) while cutting down the height, which would also warrant a change to the layout of the storage drives. If anyone any suggestions for improvements, I would gladly try to implement them (although my version will likely just stay as is).

 

[SnoWFLakE0s]

Finishing - Paint​

I went ahead and (finally) painted the parts.

I began by sandblasting and acetone-cleaning all the parts.



Next is to prime the parts. I applied a pretty decent amount and let dry for about 2 hours while I went and did some other things, so I guess I didn't take a photo.

Finally, I applied my top coat. I tried my best, but the lighting in the painting room isn't exactly spectacular so I ever so slightly underpainted some spots.




I'll probably leave these out to hopefully cure nicely for a day or two, then touch up the less covered spots then matte clearcoat it. I've been fairly thorough so hopefully this is a nice, durable finish.

 

[SnoWFLakE0s]

Finishing - Cont.​

Applied a light extra dusting of paint, finishing my can, and some matte clearcoat today. The clearcoat really works wonders! Makes it feel far more finished.

I'll work on the front panel soon...

 

[Sicaris]

Thanks for the detailed write up on rivnuts. I've never heard of them before but just from your post I feel confident I could install them with some simple tools.

Also, beautiful case. I look forward to seeing the final results!

 

[Jexkin]

Instead of an SSD mount, would it be possible to create a mount for a 120mm or 92mm radiator for a small, custom water cooling?

 

[SnoWFLakE0s]

Thanks for the detailed write up on rivnuts. I've never heard of them before but just from your post I feel confident I could install them with some simple tools.

Also, beautiful case. I look forward to seeing the final results!

They're definitely more common in M4 and bigger sizes, but are much easier to source than PEM nuts (and are cheaper). Even on Amazon, you can get a 200+ piece pack for less than 10 bucks. Barring the extra clearance required to use these, they're a very easy solution to work with.

Thanks! I should be done with everything soon.

 

[SnoWFLakE0s]

Instead of an SSD mount, would it be possible to create a mount for a 120mm or 92mm radiator for a small, custom water cooling?

Most certainly. 92mm would be a easy mount, but a 120mm one might need some careful management of the clearances around the front (especially with the front panel elements). I will release the designs for this case, so you could definitely experiment with hardware choices. This is a more air-cooling focused design though, so do keep that in mind.

 

[SnoWFLakE0s]

Milling, Milling, Milling!​

It's finally time to start milling and I'm here with some results.

Front Panel​

I finally got a time slot on the CNC mill so I went ahead and got my front panel machined.




And for the final assembly... here's what it (tentatively) looks like. I'm just seeing now that there still are some slight tolerance problems with the slots for the I/O, so I've got some sanding to do.

GPU Retention Block​

As a quickie on the manual mill I also milled a small piece of aluminum to use as my GPU retention block. Lovely part! You could definitely print this instead, though.

Almost finishing...​

I also had to modify the SSD tray slightly to allow for proper spacing for mounting hardware.

Some quick printed spacers & drive tray reassembly.




Some reassembly with a dash of some cable management to tidy things up, and using some plastic push rivets to secure the motherboard tray to the drive tray.




We're almost at the end!

 

[SnoWFLakE0s]

Nominally Complete..?​

I got around to final assembly after some sanding and final touches. Here are some beauty shots:





As you can see, the birch ply is still unfinished, and I ended up having issues with having the depth of the power button rear cutout be so small... but I'll figure it out soon.

The front panel solution is as pictured below:



I technically am only running 2 x USB 3.1 Gen 1 (5Gbps) USB-A at the moment--the other I/O module is currently not set up for proper use. I will probably rig up a USB-C to USB 2.0 header custom cable to get my extra IO port running, as my board only supports that much IO. But overall this has been a very interesting and educational project--I do have enough material to do a whole extra 2nd case, so I may or may not fabricate an extra for the purposes of selling it if anyone is interested enough.

Thanks for following along. I will release designs for this project once I am confident that the models are good enough.