My biggest 3D printer yet! Analyzing the CerbrisReborn design

The is my first delta 3D printer and also the first one i’ve completely designed from scratch! After about two years in the making, i’ll call it “done” (for now).
In this video, I’ll go through the design and have a look at what worked out and what didn’t.

What’s up everyone, Tom here, and i’ve been getting quite a few questions recently on this big-ass delta printer behind me. So today, i want to shed some light on what it is, why it is and how it is, and most importantly, what mistakes i made along the way so that you hopefully don’t make the same ones.

So i’m calling this one the CerbrisReborn, not to be confused with Cerberus, and this one is a one-meter tall delta printer which was mostly a learning exercise for me, but did end up with a few nice tricks up its sleeve. But actually, it didn’t start out as a delta, my original intention with the parts used in here was to build a compact, high-precision machine, similar to what Rene Jurack is doing with the DICE. Unlike him, i actually gave up on that concept because i had backed that design into such a restricted corner that, essentially, i ran out of space in the frame, and because i really didn’t want to throw the original design parameters out of the window and just build another mediocre printer, i ended up scrapping the Cerbris design and started over to design the CerbrisReborn from scratch. This was my first hands-on experience with a delta printer, so i wanted to try out as many things as possible to learn as much as i could on them. So my rough parameters for this machine were these: Completely design it from scratch. Check, did that. Make it affordable. Also did that, though affordable is pretty relative. Make it overkill. Yeah, some parts, yeah, others, mmmh, maybe not. And make it modular. And i think i kinda pull that one off, too. Sorta.

So the design centers around these slides. These are actually using Openbuilds wheels, but not on Openbuilds extrusions. The real Openbuilds-compatible rails are sort of expensive here in Europe and if you’re simply rolling the wheels in the grooves of a standard aluminum profile, they’ll wear out in no time because they don’t have that flat surface to ride against. So what i did here instead was to use standard profiles, V-wheels and have them ride on the edges of the profile, which i’ve turned 45°. And these do run smoothly, though it looks like there is some wear after extended use, and the biggest downside to not having a rigid aluminum plate to mount the wheels to and instead relying on the plastic’s springiness for tensioning is that these really aren’t too stiff at all. They’re ok for a delta with a light toolhead, but nothing more than that. Maybe a brace on the backside of these would help to keep them tightly tensioned.

So the carriages have belt clamps and rigid tensioners built in, plus some guidance to keep the belt loop perfectly aligned. And obviously, they also provide mounts for the ball ends on the delta arms. Now, at the time, i just wasn’t comfortable using any sort of magnetic joint because i didn’t know what sort of force to expect from the delta arms yet, and i didn’t want to couplings to fail and be a headache. So i went with these cast iron and brass ball ends, way overkill, much too heavy and as it turns out, some of these have a lot of play to them. The full set of bell ends cost only about 10 bucks, so that was worth a try for me. They compatible to igus iguball ball ends, and that might just be what i’ll end up upgrading to. In the carriage, these are screwed directly into the plastic, not a good choice, i should have at least used this sort of brass insert, because the plastic threads just wear out much too quickly with this sort of an alternating load on them. On the other end, the ball ends screw into these aluminum arms. Now these actually worked out really well. Yes, they do look flimsy, but they only need to take up axial forces, and for that they are plenty strong. Plus, with them threaded on each end, i can adjust the arm length without having to glue anything in place.

On the other end of the arms is the effector, and this one has two purposes. One, to provide a quick mount for hotend and sensor assemblies, and two, to get the hotend as far up as possible to get the maximum printing height out of the machine. Which, by the way is 380mm due to the arms being so long, and they are only so long because the ball ends have a fairly restricted range of motion.

So for the quick change, the effector works out incredibly well, it’s just three screws and the electrical connections to swap in a different hotend. This effector was sized to have enough space for three E3D v6 hotend in a circular shape with a 12mm sensor in their center. I don’t think i’ll ever be using that, but, hey, i’m not going to limit me there. There is also a spot to mount a 80 or 70mm LED ring, which i had to take out to make space for the Deltaprintr Mini mount. What this carriage is lacking are hardpoints for things like part cooling fans or the PCB that attaches the wiring to the connectors. The next revision of the effector is simply going to have some M3 inserts all around the rim.

What’s driving the filament through the almost one meter long bowden tube is a fairly simple direct-drive extruder with an E3D HobbGoblin, threading in the bowden coupling was a bit fiddly, but it works at least well enough for the job. The endstops on the top of the printer are these simple 6mm wide switches, the carriages just bump into them, and while it sounds a bit rough, they work perfectly and all three of these together cost less than a tenth of what you’d pay for a single microswitch.

So the base of the CerbrisReborn houses this heated bed. Overkill here we go. This is a 230V, 650W silicone heater, for some reason they sent me two, so i can show you the second one, it’s stuck to a 6mm cast and milled aluminum plate. And i’ve got to tell you, this thing rocks! Not only does it heat up insanely fast, but it also gets insanely hot. I’ve actually had it go up to 200°C when i ran into a bug in Smoothieware, and it did just fine. Not that you’d actually use those kind of temperatures, but it’s nice knowing that i’ll easily do 150°C should i ever decide to give plain polycarbonate another try. It’s actually this bed and SSR i showed off in the solid state relay video and it’s been going strong ever since.

The bed is sitting on a scrap 3mm aluminum plate, this is some sort of a construction site color sample, doesn’t really matter, but i liked the upcycled aspect of it, plus, it’s red, and on its underside, it’s home to the electronics. Now, i wanted an ATX supply in here because i’d be able to run the control board and a Raspberry Pi from the always-on 5V standby rail and then easily turn on the rest of the power supply with the PS_ON signal. Since then, i’ve become less and less of a fan of ATX supplies for 3D printer applications because, unless you buy like a 850W unit, which the my Mendel90 uses, you often end up running into weird power issues that you just don’t want on top of the challenges you have anyways when trying to get a new printer running well. Plus, i had to take this one out of its case to fit, which means there’s exposed 230V AC in this compartment. I know where not to touch it, but i wouldn’t be comfortable with anyone else using this printer. The frame is connected to protective earth, so i should be fine on top as long as i don’t reach into the bottom compartment.

So the brains are a Smoothieboard 5XC, and i couldn’t be happier with it. There’s no way i would have used an Atmega-based 8bit board for a Delta, and now that Smoothieware has implemented mesh autoleveling for the Delta kinematics, i can’t think of any reason not to use any other platform for a Delta. Except, maybe price, sure, but i do love how flexibly Smoothie can be configured once you wrap your head around how the setup works.

So all the motors in here are 0.9° or 400 steps per revolution, which do have a lower top speed ceiling, but give the printer more accuracy and precision without having to rely on microstepping. This works well and i’m making 0.9° motors my new default for new machines.

So for wiring, i’m entirely using Cat5 ethernet cable. It’s super cheap, gives me color-coded twisted pairs, ideal for motor, heater and thermistor wires to reduce electromagnetic interference, and for routing it, i actually just jammed and clipped it into the slots of the aluminum profiles. These clips aren’t ideal and routinely fall out, a better way to hold the wires in place and make the profiles more appealing would be to use these aluminum covers, they sit flush with the profile’s surface and hold in very tightly. Another thing i haven’t solved particularly well is strain relief, on the effector side i’m using this proto PCB with the wires soldered on and zip-tied to add stiffness, on top i’m mostly relying on the bowden tube to keep things in place. What i want to do is try and establish a standard at least for my hotends, so that i can use a single type of PCB that has wire connectors on one side and proper connectors for the hotend and sensor on the other end to make them swappable and universally compatible between most of my printers. At least on this one, i’ve made wiring plans and prepared the machine for a dual-extruder setup, so adding that intermediary PCB for one or two extruders should be pretty straightforward.

Now one more thing i want to talk about is the frame and its stiffness. So when i originally designed it, the CerbrisReborn only had the 20mm aluminum profiles connecting the bottom and top triangles, and that was one of the wobbliest printers i’ve ever seen. It did print, but you could easily see the effector swinging around b y more than a centimeter after an abrupt move. So first i tried adding polystyrene sheets to two sides, the plan was to use a clear one on the front as a door. Unfortunately, polystyrene is a) not that stiff, and b) super friggin brittle. So within a week, both white panels had a corner broken, and at that point i switched them for these aluminum plates. These are four and five millimeters thick and give the printer frame an enormous amount of stiffness, plus, i can still attach what’s left of the polystyrene sheets between them and eventually enclose the build volume that way. I just have to figure something out for the corners because i can’t actually attach anything to the aluminum profiles as the wheels ride on those corners.

So overall, i’m already very happy with how the CerbrisReborn turned out and i think it’s about time for it to replace my Mendel90. Let me know if you’d enjoy a design breakdown like this for the old workhorse of mine before i send it to a better place. And to leave out that euphemism, yes, i am planning on gutting this printer and, maybe, rebuilding it? I don’t know yet, but it’s in need of a lot of care and love before i’d call it representable and i’m not sure if i’m ready to commit to that. We’ll see.

Well, anyways, that’s it on the CerbrisReborn, if you’ve got any questions left on it, leave me a comment below and i’ll try and answer them. If you enjoyed this video and want me to make more in its style, like and share it, and you want to support the general thing i’m doing here, consider subscribing, using the Amazon affiliate links from the video description or directly throwing me a dollar or two over on Patreon. And that’s it for today, thanks for watching, and i’ll see you in the next one.


Learn more about the DICE

Extruder used on the CerbrisReborn

This video is licensed as Creative Commons Attribution Share-alike thanks to my supporters on Patreon!

Support me by shopping on eBay or Amazon through my affiliate links! Anything you buy through these gives me a small bonus to run the channel – at no extra cost to you!

Music is Jahzzar – Lift Off, licensed CC-BY-SA


You can support me without spending a single penny!


This video is licensed as Creative Commons Attribution Share-alike thanks to my supporters on Patreon!