HELPR — Part 0: Research on a new 3D printer design
I’ve been gearing up to make some more cosplay elements and props recently, so I’ve been tying to get my old 3D printer working, and it’s… not going well. What I have now is a MakerGear Prusa Mendel built from a kit in December 2011. It has never worked terribly well, and the whole 5mm threaded frame and slightly-bent Z-screws have made it wobbly and imprecise. I mean, it’s been fine for building cases and prototypes where I don’t care what they look like, but it’s recently started jamming on every print (maybe thermal damage to the hot end). For something that’s supposed to be able to reliably print objects on a regular basis, it’s been an amazing disappointment machine.
My journey started out looking at the Prusa i3 as a possible model to build, reusing many of the parts from my Mendel. While I was thinking that, I came across this post on Hackaday about making your cheap 3D printer better. While I was there, what really caught my eye was a comment by Dr. Mark Rehorst dismissing most of the items in the article as band-aid to make up for the cheap un-sturdy construction of the consumer-grade printers. He argued that for not much more money, you could build a printer that was way more rigid, accurate, and reliable.
This is what I want. I don’t mind spending the time to build the printer myself, but I want something that I can build once and stop fucking around with. My hobby is not building 3D printers — it’s making cool cosplay parts and props, and the printer is just a tool.
Now, I promised myself that I wouldn’t buy another 3D printer until I could find an affordable SLA printer, and I briefly considered building my own, but again: I don’t want to fuck around with the printer. Turns out the Wanhao Duplicator 7 is an SLA printer that is finally good enough and cheap enough that I went ahead with the purchase. I mean, I’m probably going to have to fuck around with the printer a bit, because it’s cheap enough that it’s definitely not going to be a one-touch process, but hell, even the semi-professional Form 2 still requires a lot of fiddling about with post-processing, and for less than 1/10th the price, I’ll take it. The D7 should arrive from China soon, and get me good detailed prints for smaller items. However, I still need a bigger FDM/FFF printer to make larger cosplay parts, like the Starlord helmet I’ve been wanting to make.
Rather than regurgitate everything I’ve read, allow me to link to the series of articles that have greatly inspired my design thinking about this project:
- Build a 3D Printer Workhorse, Not an Amazing Disappointment Machine — lots of good tips for band-aiding my current cheap Prusa Mendel printer
- CoreXY cartesian motion platform by Ilan E. Moyer
- Hypercube 3D printer by Tech2C
- Hypercube Evolution by Scott Alford
- An Almost Reliable, High Precision, 3D Printer: Son of MegaMax (SOM) by Mark Rehorst
- Ultra MegaMax Dominator (UMMD) 3D Printer by Mark Rehorst
Since the UMMD seems to combine the best of both worlds from many of the printers above, let’s take a look at some of the design features. Photos here courtesy of Dr. Mark Rehorst, designer of the UMMD.
The frame is very rigid, made of 4040 T-slot aluminum extrusion. The motors and electronics are outside of the print area, to avoid heat damage, and it’s enclosed in polycarbonate, to trap heat when printing ABS. It’s also very tall. I don’t think mine needs to be that tall — 300mm seems like a very useful print height. But then again, if I’m designing this to be the one true 3D printer that I build once, and use forever, maybe a taller model would be wise.
The design uses a CoreXY platform, which means the print head can make all of the lateral moves, and the print bed only needs to be raised and lowered.
The Z-axis uses a unique belt-driven design, which seems excellent for this very tall design. For my 300mm tall envelope, perhaps this isn’t necessary, but it seems like it might be a simpler design than using two lead screws on a belt/pulley system? I’ll have to compare the two further.
And last, the print bed itself, leveled with a 3-point system, evenly heated, and milled perfectly flat. It’s the kind of print bed that you level once and never have to think about again — no auto-leveling or mesh-based compensation necessary.
3D Printer Design Goals
The most important part of this is rigidity, which leads to accuracy.
- If the frame is rigid, that means it won’t shift out of place, leading to repeatable actions.
- If the build plate is on a rigid linear positioner, and the bed itself is made of machined cast aluminum, I should only have to level the bed once and be done with it forever.
- All of the linear motion axes should be as rigid as possible
Smarter Weight Distribution
Right now, my Prusa Mendel lifts the full X axis with the Z axis. The Z axis should only have to lift the print bed and the (light) weight of the printed object. With the CoreXY platform, that means the weight of the X motor is eliminated from inertial calculations.
- A true 300×300×300mm plus margins print envelope. The UMMD is much taller, but I don’t currently see myself needing it. I suppose if I do, I can get some longer parts, and I’ll be good to go with an upgrade.
- Capable of 0.2mm Z resolution — 0.1mm resolution would be nice, but I feel like if I need smaller resolution, it will be a small item and probably get printed on the D7, so I don’t care that much about 0.1 layers. That said, I expect with the rigidity of this frame, that 0.1mm layers won’t really be a problem.
- Single Z-axis motor, so the platform never gets misaligned
- Evenly heated print bed
- Enclosed print envelope, so it can be heated for warp-free ABS printing
- Electronics outside of the heated enclosure
- Capable of printing PETG, flexible filaments, PLA, ABS, and their various heating and cooling needs. Probably optimize for PETG, but flexible enough to print the others.
- Graphical LCD, rotary controller, and SD card slot
- Solid extruder and hot end that’s not going to jam
- Dual extruders, so I can print break-away (or dissolve-away) support material
- 99.999% reliable first layer adhesion. A lot of this will be taken care of by proper alignment and a flat bed, but a PEI print surface also seems wise.
- Maybe future convertible to a laser cutter or light-duty CNC? Probably a terrible idea, but something to keep in mind.
I was thinking that my design would be a blend of Hypercube with the SoM, until I realized that Mark Rehorst had already designed the UMMD, which basically combines the two in the way I was thinking. However, the “Ultra MegaMax Dominator” is a way more aggressive name than I’d like to have in my life, so I came up with HELPR, the “Hypercube Evolution with Lots of Parts that are more Rigid.” Yes, it’s kind of a terrible backronym, but it’s also a bit of an homage to Venture Bros., and just sounds generally cute and helpful. Maybe I’ll come up with a better backronym later.
- After reading about the CoreXY method of positional control, and looking at the Hypercube-based printers, this really feels like the way to go. Both of the α and β motors that control the X and Y axis are stationary, which means I don’t have to move them, and also that the build platform only has to move in the Z direction. The CoreXY platform will only have to move the print head.
- 0.9º stepper motors. Again, they don’t cost that much more than 1.8º motors, and the increased resolution is worth it.
- Undecided on whether I should use lead screws or the belt-lifted Z-axis of the UMMD. I’m kind of leaning toward the belts at the moment, since it seems like a simpler mechanism overall.
- 30×30mm T-slot aluminum extrusion for the frame
- Probably eliminate as many 3D printed parts as possible. I love the RepRap project, but on this particular printer, I really don’t give a crap about it ever being able to replicate its own parts. I just want the parts to never warp or melt or snap.
- Use linear slides instead of steel rods with bearings. It doesn’t cost that much more, and everything ends up much more rigid. Rigid is good.
- Maybe upgrade to 24V power. More power overhead is good.
What do you think of the design goals so far? What am I forgetting? Leave a comment here, or hit me up @vixenlabs on Twitter.