rgb.sh

loowit eurorack enclosure

July 8 2025

About the design and construction of my adjustable-curve, flat-pack Eurorack case

Half-size rack powered on
Loowit in action.

Loowit is a custom Eurorack enclosure I designed.

Loowit offers the following features:

  1. Extensibility.

    Multiple rows can be stacked vertically to whatever height the user desires. Rows can also be joined horizontally to extend the total width of the rack.

  2. Adjustable curve.

    Each row can be curved inwards up to 22.5°. This allows the user to configure the shape of their rack to their choosing.

  3. Flat-pack construction.

    Nearly every component is flat sheet metal, making it compact and easy to ship.

  4. Limited unique parts.

    The core enclosure is comprised of only 5 unique pieces.

  5. Open-source.

    Loowit is open-source and available on GitHub.

Background

This project came about as the intersection of three different threads going on in my life: I had outgrown my Eurorack enclosure, I wanted to actually be able to use CAD software, and I wanted to try out online sheet metal fabrication.

This first thread is likely familiar to most Eurorack fans. I had reached a point at which my module collection had outgrown the amount of available space in my rack. In my case, this was a 6U 104HP (horizontal pitch, Eurorack’s unit of horizontal space) enclosure.

The second thread is one that had been pestering me for perhaps a decade or more. I’ve had multiple 3D printers, done a ton of different DIY projects and have basically spent my entire life at the computer. I even work in computer graphics for a living. And yet somehow, being able to use CAD software has always been the glaring hole in my metaphorical tool belt. This project seemed the ideal opportunity to dive into FreeCAD, with the goal to be able to quickly custom design 3D-printable parts after I was done.

The third thread was just something cool I’d been seeing a lot of in the last couple of years. I had learned of services like SendCutSend which make it easy to get designs manufactured at small scale. I’d seen excellent results from my roommate’s custom amplifiers, as well as from a myriad of projects at Open Sauce. I figured being able to send off custom CAD designs to a fabricator would be a really useful skill to have.

Design

The original source of inspiration for this project was the Teenage Engineering Computer-1. This is a flat-pack computer case, where the user actually bends the sheets into the final shape. As I love this design, and the idea of flat-pack in general, I decided to try to emulate it.

Most Eurorack cases are boring rectangles, following the form of the server rack standard from which Eurorack was derived. There are some generally quite expensive cases which place rows along a curve instead. The idea is that this is more ergonomic; in practice, I don’t know if this is the case, but I personally find these the most visually appealing in that high-tech sci-fi instrument panel kind of way.

Functionally, I wanted a case which could morph between both extremes. While there are some cases that offer limited curve adjustment, they most often seem to be limited to one or two hinges connecting flat sections with multiple rows. I instead wanted to find a way to put each row along a continuous curve.

On top of the adjustability, I designed the case with modularity in mind from the beginning. My thinking was this would mean I never need to upgrade to a whole new case ever again. Instead, I can extend the existing structure. This means Loowit adds extra modularity on top of the existing modular Eurorack standard, so I can call it a modular, modular synth enclosure if I aim to be obnoxious.

Lastly, I wanted to use off-the-shelf parts where possible. Being that Doepfer is the originator of the Eurorack standard, I chose their Bus Board V6 and PSU3. For rails, I chose the TipTop Z-Rails in 84HP. The length represents the original standard for Eurorack (and the standard from which it derived). I picked the Z-Rails over some slightly cheaper options on the market, just because they tapped their rail ends in metric sizes and not SAE.

Adjustable Curve

Drawings working through how rows would rotate relative to one another, and how to place their axes of rotation. The squares (or sometimes just their bottom corners or edges) represent rows viewed from the side. Dotted lines generally trace to the axis of rotation, usually notated with a circle. At bottom right, I am working through how the corners and rails impact the offset from the axis of rotation.

These diagrams show me working through how the rows would rotate relative to one another. There are two important decisions I made here:

  1. Rows would share their axes of rotation, as opposed to each having their own. This meant that the rotation would be one fluid motion, because the rows would be moving along the same arc. I still believe this was the right decision.

  2. The axis of rotation would be in line with the front of the enclosure. This would keep the gap between rows pretty close over the entire range of motion, and be more visually appealing. I now think this was the wrong decision, as it made the front frame joint an arcing slot rather than a simple hole. This made it much more difficult to connect rows and to adjust the angle than a simple hole would have been.

Frame

Iterations on the design of the frame leading to the X frame.

Here I show a number of my iterations on the frame design. My goal was to come up with a design that would be strong enough to support holding up multiple overhung rows. While doing so, I also wanted the design to avoid adding too much extra weight, limit the number of unique parts and pack well for laser cutting to not waste material. I eventually settled on the X frame, which used two of the same bar stacked in opposite orientations.

These diagrams really showcase my influence from Bowellism, one of my favorite types of architecture. This is “inside-out” design, where the internal mechanism of a building are placed on the outside to be shown-off rather than hidden. I was able to apply a a similar approach here by placing the frame and power supply on the outside of the structure, and relying on an external frame to hold the entire structure up.

Enclosure

Design experiments with user-bendable sheet metal for joinery.

I really wanted get the bent sheet metal thing working. I thought through a lot of possible designs, and felt like I was closing in on a design.

Then, I looked at the cost of getting perforations added to the part. The amount of extra laser time needed meant this would be significantly more expensive than I expected, so I gave up.

Of all of this work, the one design component that remained was tab and slot. This is where joins are made between and protuding tab entering a like-sized slot. I love this design because it can form a relatively strong joint along 2-and-a-half axes (or whatever the proper machining term is…), with no bends or fasteners.

Materials & Dimensions

I really had no idea how to chose materials and thicknesses, so I leaned pretty heavily into copying. Pretty much every Eurorack enclosure and PC case out there is either steel or aluminum. Even long after abandoning the Computer-1 for the basis of my design, I still stuck with aluminum for its strength-to-weight ratio; with the possibility of multiple overhung rows, I wanted each to add as little weight as possible.

You can roughly model the X frame as a lever arm, where a weight is placed on an arm over a fulcrum. The front joints of the X frame are in compression when the above rows are bent inwards, acting as the fulcrum, while the rear joints are in tension. While very early designs considered only a single joint per side, having this second joint offset from the fulcrum would make for a significantly stronger hinge.

And because I am clueless in this area, I then ran to ChatGPT to figure out what the math would be. I gave ChatGPT a couple of extreme examples of heavy weights on long arms in order to home in on the proper type and thickness of aluminum.

Don’t worry though: I actually ended up taking material properties from other sources and doing the math myself. I really just needed to know what I was supposed to look at. I know that I oversized the aluminum by quite a bit, but it’s better safe than sorry when the cost and weight difference is so small.

Likewise for bolts, I looked at the required shear strength and then significantly oversized them. I also opted for serrated nuts and flange bolts, to get a better grip on the frame and avoid any possible slippage. Fun fact: the strength of bolted joints is from the friction between joined materials, not the shear strength of the bolt.

Enclosure

CAD

Prototyping

First CAD draft
The very first prototype in CAD. Here you can clearly see that I don't understand how bend radii limitations work.

I started in FreeCAD barely understanding anything about how it or the materials I would be working with would work. I was targeting the thinnest 5052 aluminum available from SendCutSend for the enclosure, as it was pretty similar in thickness to what the Computer-1 uses and easy to bend.

Teenage Engineering inspired CAD draft
Prototype trying to emulate Teenage Engineering's flat-pack design language, without bends.

This was the final state of the Computer-1-esque design. As shown, the case would rely on L-brackets to hold together. I didn’t like amount of extra hardware this would require.

It was also around this stage I found out that SendCutSend couldn’t powder coat their thinnest gauge of aluminum, but this didn’t have any serious impact as I’d already given up on the user-bendable design.

Bracket CAD draft
First frame design in CAD, prior to the X frame.

My early designs of the bracket trended towards vertical bars to limit the amount of material used. I switched away from this design not only because of the number of unique parts, but also because it didn’t like how the load could be distributed unevenly across the bars when rows were angled.

Final Drafts

Full enclosure CAD draft
Later CAD draft featuring the X frame, but keeping the inter-row panel covers I would later remove.

After more iteration, I came up with this structure using the X frame. The idea here was that all rows would be open to each other internally, and rely on an outside inter-row cover to keep the structure enclosed. My inspiration was an arthropod exoskeleton, such as that of a millipede.

Ultimately, I didn’t feel the assembly would be particularly easy to put together, and I felt the visual design was too organic and asymmetric to fit the Bowellist style I wanted. This also kept rows from being truly independent modules of their own.

The handle/stand combination design didn’t proceed, either, but just because the offset between front and back arms of the X frame required the handle to either have a spacer or a bend. I would like to revisit the design down the road.

I did keep the new frame “dog bones”, as they served two purposes: they keep the side panels tensioned into place, and they provide a nice mounting point for the external support. This is particularly critical given that I abandoned the aforementioned stand.

Final enclosure CAD draft
Final draft of the enclosure.

This is, more-or-less, the final design of the core enclosure.

Bolt holes were added to the back panel for mounting the Doepfer Bus Board V6 and a to-be-design PSU enclosure. Additionally, I added a cutout on the top side for power cables. While these additions worked out well enough, I think I want to put more thought into how I’d make the mounting points more modular to support different types of bus boards and power supplies.

As shown, this is five unique, custom parts, plus the TipTop Z-Rails and Doepfer Bus Board V6. Screws are not displayed.

Assembly

Components

I ordered the case from SendCutSend in the following materials, thicknesses and finishes:

I probably would have opted to powder coat the frame if it were an option. However, tumbling gave an excellent finish for the dog bone brackets.

Structural fasteners came from McMaster-Carr. For the frame, I opted for serrated nuts to increase hold and flange bolts to apply pressure over a wider area without the need for a separate washer.

Single row

Parts for single row
Single row of the enclosure, in flat-pack form (excluding rails and bolts).

After many weeks of back-and-forth on my design, it was a relief to finally send out a single-row order to SendCutSend.

The moment I sat the flat-pack parts in a pile I knew I had made the right decision in taking the design in this direction. I know it doesn’t matter for a one-off personal project, but there is something so cool to me about this. And really, that’s the most important part of a personal project: satisfaction.

Single row panel test fit
Test of the enclosure tab-and-slot fit.

Having no real experience in design physical things with tolerances, I was curious how well the test fit would go. As it turns out, SendCutSend is crazy precise.

The initial test fit went well. I had toleranced all parts for the maximum thickness SendCutSend provided in their specs: 0.005”. One or two tabs took a small amount of force to join, but for the most part the joints were actually a little too loose. I didn’t end up changing this in the next order, but I’d reevaluate my tab-and-slot design and tolerances in the future.

Single row test fit minus rails
Test assembly of the whole structure, prior to installing the rails and modules.

The bolts, however, were crazy loose. I had found some numbers online for metric hole sizes, which turned out to have far too much clearance. After consultation with a fastener engineer (i.e. my roommate), I was informed that bolts should pretty much always come in just under their nominal diameter.

Take this with a grain of salt, but many of the online bolt holes dimensions seem to be intended for much larger and more complicated structures. In these cases, larger holes make assembly easier and give room for dimensional inaccuracies from manufacturing or temperature change.

For small, high-precision designs made on a laser cutter, bolt holes can have very tight tolerances. As such, I shrank all bolt holes to be at or near their nominal sizes.

Single row test fit with modules
Fully-assembled row.

Assembly was a little finnicky, but not terrible. I expect that there is an order of operations that would make this process somewhat easier, however I have not bothered to experiment with this.

Full Enclosure

5 rows of flat packed enclosure
Five rows of enclosure in flat-pack, as shipped from SendCutSend.

With minor adjustments to hole sizes made, I sent off the order for the rest of the rack.

It was crazy to see the entire structure laid out like this prior to assembly. This was really the moment where I felt like I’d “learned” CAD and how to design basic metal enclosures, and that I’d opened up a whole new category of approaches for future projects.

Full assembly test fit minus rails
Full enclosure assembled, minus rails.

Joining rows was not super easy, especially without a fixed mount of the bottom row. Luckily, it came together okay.

This highlights the importance of designing for assembly.

Full assembly with rails and modules
Full enclosure assembled with rails and modules.

I had to wait to assemble the rest of the structure, as my Z-Rails were on back order for a number of weeks. The moment they arrived, I bolted them in and added all of my modules.

Assembly curve test
Test of the curve mechanism on the full enclosure.

Finally seeing the thing I made in CAD fully assembled, with joints working as expected, was shocking. The feeling was much the same as when your code compiles and runs on the first try.

This was the point at which I realized that I didn’t really design in an easy way to move the structure. It is incredibly awkward to work with in this form, as it is quite large and I have to be careful not to bend the 5052 frame.

Material Choice

Bent 5052 bracket
With some effort, I was able to bend this extra bracket made of 0.125" 5052 aluminum.

On the fabrication front, my biggest regret was in my material choice for the frame. While I did the basic lever calculations with the chosen thickness (.125”) of 5052 and everything looked plenty strong, it turned out to not be as stiff as I would have liked.

To all of you mechanical types out there, this is where you should realize that us software types won’t be coming for your jobs any time soon.

Basically, stiffness and strength are very different things. I have zero concerns about the case falling apart on me, but this lack of stiffness means that there the structure can shake quite a bit. It’s particularly noticeable when standing upright with mounts only on the bottom row: you can shear the rack left and right a fair bit. It’s likely not helped by the loose tab-and-slot fits nor 5052 panels, either.

I ordered extra dog bone brackets in 6061, stainless steel and mild steel for use in mounting the enclosure and to give me a hands on comparison between materials. None of these could be bent by hand, no matter how hard I tried.

In the future, I’d probably switch to 6061 aluminum. It’s even stronger than 5052 and incredibly stiff, while remaining lightweight and similar in price.

Horizontal Stabilizer

Cross-bar CAD
CAD for the stabilizer.

In order to reduce shaking without ordering a whole new frame, I opted to design and build some horizontal stabilizers.

In a moment of irritation at breaking the second-gen mount, I gave in on not using SendCutSend’s bending services. I quickly threw together a cross brace, meant to add lateral stability, even forgetting to put mounting holes in the bar for attaching the case to a frame. Oh well.

This definitely reduced the amount of skew possible in the structure, but I don’t consider it a core part of the design nor something I’d want to keep going forward. I hope that a switch to 6061 for the frame, and potentially tightening of the tab-and-slot fit, is enough to mitigate the shaking.

Mounting

I really did not put enough thought into how the structure would mount to a stand or frame. As such, I abandoned my flat-pack-only approach, and went for what would get me the fastest result. This meant heading to my 3D printer.

CAD

V1

First mount CAD
CAD for the first version of the mount.

This first design shows the rough structure I wanted, but was too short to clear the PSU enclosure when mounted on the back. The cylindrical body did not print cleanly on its side, either, as it needed supports.

V2

Second mount CAD
CAD for the second version of the mount.

I thought this design was surely the one. Unlike the metal structure, for which I actually did simple math to validate strength, these 3D prints were completely guess-and-check. Of course, this was not the one.

The problem here is that the thinner top section kept breaking on me. It may be in part due to the switch from PLA to PETG, but it also seems related to the amount of flex available. The first version had the slot go all the way through, meaning that pinching the holes together could bend the entire sides inward. This design features a shorter gap on the back, matching the shape of the longer tip of the X frame, where only the thinner region around the hole could flex.

If its not bending, its breaking, or so they say?

V3

Third mount CAD
CAD for the third version of the mount.

By the third iteration, I had given up on any type of fancy design. I increased the thickness of both the frame connection and the base, and this ended up making it strong enough to work.

Assembly

Test Fit

3D printed mount
Test print of the mount.

I have to say that 3D printing has really come a long way since I was in college. While both printers may have been Prusa’s, this more modern setup I picked up at a flea market is much easier.

Up to this point, this is probably the most complex thing I’ve ever designed in CAD and then printed.

Mount testing
Test of the mount supporting the full enclosure.

The first version of the mount at least validated the rough direction I was headed. The idea would be that multiple of these mounts would be used to attach the enclosure to an L-shaped wood frame.

For the purposes of getting something together quickly, this would be scrap 2x4s. I’ll revisit the external frame someday, but dimensional lumber has really never failed me.

Mount versions
The three mount versions, with failures.

At right is the first version. This didn’t fail on me in use, but I did break it on purpose to see its limits. I couldn’t break it by pinching, only by pulling the arms. Besides being too short to clear my PSU design, it seemed otherwise okay.

I had multiple of the second version, shown center, fail on me in use. The one in front is a real-world failure, while the back was broken on purpose by pinching. It was quite easy to snap.

The third version is shown at left. Why did I not sort this left-to-right?

Power Supply

While I was initially going to wait some time before designing the PSU enclosure, I ended up deciding that I wanted to get everything done ahead of Open Sauce. My roommate had been accepted to demo, so I decided to tag along.

CAD

PSU CAD with top hidden
CAD model of the PSU enclosure, with top hidden to show inside.

This follows the same design principle as the core enclosure: multiple tab-and-slot joints all held together by stiff, off-the-shelf metal bars in tension (here, the 50mm standoffs).

It was here that I realized that I had messed up in how I had placed the PSU mounting holes in the original enclosure: as they were so far towards the top and bottom edges, the PSU enclosure would cover the power cable access hole. This forced me to narrow the structure and place the tensioning standoffs externally to be able to clear the opening. Surprisingly, I ended up really liking how this looked.

I regret not just designing this alongside the rest of the assembly, though, as I think it would have netted a better overall design. That said, I’m still happy enough with it.

Full assembly with PSU attached
CAD model of the entire assembly, with PSU and mounts.

This shows the V2 mounts and the PSU enclosure. With the PSU attached, you can really see the Bowellist design come to life. And it only gets better with the power cables winding across the back.

Assembly

PSU wiring in-progress
Wiring up the PSU.

Assembly of the power supply was only marginally more complicated than the main structure. The only issue ended up being the wiring: I wanted to use 10 AWG cables everywhere to minimize resistances over the longer-than-normal runs between the PSU and the farther away rows.

As it turns out, 10 AWG wire is thick. It’s thick to the point that you actually need to design around it. Not only do the wires occupy a ton of space, but they have massive bend radii. Add on the crimp terminals, and it was basically impossible for me to use 10 AWG wire inside the PSU.

I worked around the issue by splicing crimp terminals Doepfer’s provided 18 AWG wires, and using the 10 AWG wires externally. I’d probably size up the PSU enclosure along the long axis and perpendicular to the PCB to compensate if I make a V2.

PSU attached to back of case
PSU fully assembled and wired up on mounted enclosure.

Once wired up, I made sure to test that every bus board was outputting the right +12V, -12V and +5V, and that all the grounds matched. As everything seemed good, I threw in one module at a time for a full test just in case the system decide it wanted to fry something.

Nothing went wrong, though. Everything worked as intended, so I mounted half of the assembly in preparation for Open Sauce.

Version 2

I think it’s unlikely I iterate on this design further, as the current one works great and I don’t need more than 504 HP of room in my system.

That said, if I made a V2 I’d definitely do the following:

  1. Move the axis of rotation forward to align with the front frame joint.

    This would change the front mount from an arcing slot to a hole. While this would introduce a small gap between rows when curved, I believe it would make the structure easier to adjust.

  2. Switch the frame to 6061 aluminum.

    I think this should address most of my concerns with stiffness, and avoid the need for the horizontal stabilizer.

  3. Size the PSU enclosure up and add a cover to the output side.

    I’d like the PSU to comfortable fit 10 AWG cables for 3 rows, by increasing the length of the longest dimension and the height available above the PCB. It would also be a good idea to add some kind of cover to the side the low-voltage wires come out of.

  4. Tighten up tab-and-slot.

    This should help a bit with rigidity. Dogbone fillets offer a tighter fit than the rounded rectangle I used.

  5. Reduce powder coat tolerances.

    The fit was so good that it trended towards being too loose rather than too tight. Being less conservative with the powder coat tolerance would likely help without causing problems.

  6. Switch to generic mounting holes.

    I chose Doepfer mounts as being the Eurorack standard’s creator they are the closest thing to a standard for PSU and bus board mounting holes. I’d prefer to be vendor agnostic, though, by making the enclosure’s mounting holes generic. Each unique part could have its own mounting adapter.

  7. Make the cable cutout symmetric, and add a cover.

    I think it’d look better to have either a single central hole, or one at top and bottom. I’d also like the cutout to have some type of insert, whether it be a rubber grommet or 3D-printable plate.

I would also want to spend some time experimenting with the following:

  1. Ease of adjusting the curve.

    Right now, its quite difficult to adjust the curve, particularly in the middle of the rack. I’d like to play with designs where the front joint doesn’t lock and the rear uses large thumb screws. I’m also curious if replacing the arcing slot of the back arms of the X frame with multiple holes (e.g. at 0°, 12.25° and 22.5°) would be easier to use or not.

  2. Reducing unique part count.

    I’ve been thinking on how I could merge the top/bottom panel and rear panel into a single part. While I don’t think there’s a way to do it without some tradeoffs, I’d like to go through the full design process with it.

  3. Sliding tab design.

    There may be a way to increase the rigidity of the enclosure by switching some of the tab-and-slot construction for sliding tabs which interlock with one another.

Cost

I’m sure the first question many people will ask about this is “how much did this run you?”. Here’s your answer:

Total Cost: $2373.17

Relative Cost (at 504 HP): 4.71 $/HP

Honestly, I’m surprised the cost per HP came out as low as it did, given that I did not do a good job of minimizing costs. I suspect that this could be built again for a couple hundred dollars cheaper.

Breakdown

I’ve broken down all of my expenses for this project here:

Note that this doesn’t include the power cables, terminals, solder, etc. that I already had.

Learnings

If I were doing this all over again from scratch, here’s what I’d have wished I’d known from the start.

FreeCAD

  1. Make a file for a variable set, to manage all constants like “frame minimum width”.
  2. Use expressions everywhere for dimensions, based on these constants.
  3. Have each part live in its own file, and use SubShapeBinder to reference other parts.
  4. Reference external geometry in your sketches to avoid needing to constantly rewrite the same expressions.
  5. Rely on the fillet tool to make rounded corners rather than manually making every one by hand.
  6. You can reference assemblies in a sketch. Use this to avoid needing to reference and duplicate each piece of external geometry, because you otherwise couldn’t align multiple parts as wanted.
  7. Datum planes are your friend.
  8. Try out master sketches.
  9. Center your sketches.
  10. The Sheet Metal and Fastener workbenches are must-have add-ons.

SendCutSend / Fabrication

  1. I would use a tolerance below the maximum for powder coating for future tab-and-slot designs.
  2. Very small and very thin parts can’t be powder coated.
  3. Tumbling gets an excellent finish.
  4. 5052 is quite soft.

General

  1. Bolts should always be less than their nominal diameter. For laser cut projects, just design around the nominal.
  2. Lever calculations are not that hard, and doing the math is worth it.