June 19, 2023
Description
Turn your Kossel from a wobbly tower that rings like a bell into a rigid, no-nonsense, crisp-printing etc. etc. etc. thing of awesomeness with these snazzy-looking anchors for steel wire frame braces and increase either your quality or speed/acceleration by a lot (or meet in the middle someplace), hurray!
(2023 EDITOR'S NOTE: this description was written up back in 2018 when I thought 3000mm/s^2 was a lot of acceleration. D'aaaaw. It's too cute to update. Just be aware it also works for much higher print speeds. Also, everything linked on Thingi is also posted on my Printables profile.)
Really, this mod is more about improving quality than speed; in quantitative terms, I'd say my ghosting/ringing artifacts for perimeters printed at 40mm/s, 2500mm/s^2, and 10 jerk (inner/outer) are nowless visible than they were pre-mod at 30mm/s, 1200mm/s^2, and 5 jerk (outer) + 35 mm/s, 1500mm/s^2, and 10 jerk (inner), which translates into a seriously healthy speed boost for an actual improvement in visible quality.
But on the other hand, this mod will also set your printer up to withstand some pretty foolish print speeds, and let's be honest, which is more fun?
It's basically the same benefit you'd get from screwing big, thick panels of something stiff to all three sides of your printer. Or close, anyway, and waaaaaay less involved.
IMPORTANT DISCLAIMERS: this mod is fairly quick to print, remarkably easy to mount, has a dead cheap BOM, and offers huge and instant benefits, but it comes with the tradeoffs of a moderately reduced build envelope (UPDATE 10.24.18: unless you use these new spreader arms: https://www.thingiverse.com/thing:3173546 -- otherwise it's going to cut off three outside chunks of the build platform somewhere near the inward extent of the horizontal frame members) and the need to really verify your bowden/wiring clearances and look for new potential sources of wear -- plus you could theoretically destroy your frame corners if you don't take your time tensioning the setup evenly, although this mod is specifically designed to make it very difficult to apply uneven tension that would wrack your frame to death. Also, I would not recommend doing this mod without also having some compatible corner braces, like mine (https://www.thingiverse.com/thing:2474583), that will transfer twisting moments on the uprights into the horizontals, because otherwise I think this mod could eventually cam the uprights enough to split your corners.
But hey, you're a big girl/boy and you want WARP SPEED, right? Read on to the design section and I'll share a semi-secret that's holding you back if you've never monkeyed with the stock Anycubic firmware settings, too!
Anywise. Also see the design section for all my reasoning that led to this design.
You'll need 3 each of the adjustable and fixed anchors, 3 of the screw adjust mounts, 6 clamp washers, and 1 double eye join, plus:
Special tools:
After I put this on my printer I couldn't resist dialing up the speed and acceleration to stupid speeds and trying to print a Benchy, and I was pretty shocked when something came out in half an hour that actually LOOKED like a benchy. Is it great? Heck no! My long ol' bowden and direct drive extruder can't even approach keeping up with these kinds of speeds. On the other hand, the visible ringing artifacts wouldn't have been out of place on this machine at half the speed, if that, before I put this mod on it. So at the end of the day it is a vast improvement, but like with most such mods you push out one limitation and put the pedal down only to find the next limitation waiting :) In the case of my AnyCubic Kossel, it's definitely the extruder followed by the risk of the steppers skipping when doing psycho travel moves near the outside of the build envelope.
UPDATE LOG:
6.16.18: I beefed up the eye join with more meat in the middle and a thick cross-brace; the original V2 had no cross-bracing and could crack under the compression of the wire loop, allowing loss of tension.
10.24.18: Spreader arms posted to help fix most of the volume loss issue: https://www.thingiverse.com/thing:3173546
Printer:
Anycubic Linear Kossel XL
Rafts:
No
Supports:
Yes
Resolution:
.2 for the anchor pieces, .1 for the washers and eye join
Infill:
95% for the anchor pieces, 100% for the washers and eye join
Notes:
I printed in PLA with 3 perimeters + Cura's alternating bonus perimeter, 6 top/bottom layers, and ironing on the anchor pieces. I also used Cura's tree support for the screw adjust anchors to make sure the small lips on either side of the nut socket (gross) didn't droop. The eye join doesn't technically need much for support at 0.1 layers, but fiddling with tower/tree settings to get some bonus material around it will make it more likely to print successfully without getting knocked over or warping a lot.
Super tip: fitting M4 nuts into vertical 2020 slots and holding them in place with something so you can screw bolts into them is a pain in the butt -- UNLESS you stick a little rolled-up piece of scotch tape (sticky side out) into the slot first, and stick the nuts to that.
If you have corner braces on your machine, start by loosening up the bolts fixing them to the frame horizontals (you can leave the vertical-side bolts alone; this just ensures that if your frame wasn't perfectly trued up because of the corner braces, you won't be fighting that imperfection when you set up the wire braces).
Terminology notes: the fixed anchors are the ones with TWO holes for M4x12s, while the adjustable anchors have ONE hole for an M4x12, and the screw adjust mounts are the guys with the big slot for the M5x20's. The M5x20 bolts are the ones you'll use for applying final tension later.
Assembly Steps:
Now it's the really fun part: applying the tension!
FIRST, give the top of your frame a rock and a twist to see how much it can flex, just so you have a point of reference when you rigid it up.
NOTE that this design uses a single loop of wire strung all the way around the machine so that you can avoid all the hassle of trying to get six different pieces of wire the exact same length and tension -- as long as you don't lock down the clamp washers until the very end, and make your tension adjustments in small increments to each joint in turn, you should get perfectly even tension all the way around. You can test for even tensioning as you go by simply plucking the wires -- if they all generate the same tone, you know you're good.
Okay, last steps:
I genuinely don't know why nobody's thought of this one yet (or if they have, I couldn't find jack about it on Google).
The problem: if you've got a Kossel, you know it's not a rigid setup. You've got plastic corners (or possibly metal) to keep the uprights true, but they don't have much leverage, whereas the forces inflicted on the uprights are pushing way up high, where they have tons of leverage. Take the top frame and push it or twist it and watch what happens to your effector head -- it's going to shift all over the place. It's not a good recipe if you want to go fast, because going fast takes high acceleration rates, which means you're pushing your frame harder, which means every start, stop, and corner is like hitting your frame with a bat.
I figured this out pretty quick when I got my Kossel, and my first fix was designing some bolt-on corner reinforcements. They definitely helped with the ringing that came from the stock frame's lack of rigidity, but they only reduced it, and the reason is pretty obvious from an engineering perspective: they just don't have good mechanical advantage compared to the forces they're meant to counteract. I got especially worried about this when I took out my steppers to put in some dampers and discovered that two of the three corners had cracked across the motors' mounting faces. No idea when that happened, but heaven knows I'd been pushing my printer hard.
Anyway, from the inadequate corner braces, the next logical step is to look at stiffening up each side of the frame with longer braces, and the best-case for this would be braces that go all the way from corner to corner. Rigid braces would be ideal, but difficult in practice -- you've got to get the lengths and mounting angles perfect to avoid wracking the frame out of true, plus at those lengths you're not going to be able to count on good resistance to compression forces without having pretty fat chunks of CF tubing or similar, which is going to cost. The other idea (as done on printers like the Fisher) is to mount nice, stiff plates either around the corners or across the sides; I'd regard this as probably the ideal solution, but not the easiest one to implement at home without access to equipment for doing a really precise job of cutting the plates (and/or bending them) and getting all the holes in exactly the right spots. Plus, again, the plates will probably need to be made of something not-cheap even if you don't have to pay somebody else for the custom fabrication.
So I wondered: why not just go for purely tensile bracing, which would mean nothing more complicated than steel wires? They'd self-align, and if I used a single loop they'd also self-adjust for even tensioning around the structure. Plus, with all the anchors mounted to the uprights, I figured the wires ought to be able to take on a lot of tension without putting terrible strain on the corners. The main tradeoff, then, would be balancing how far to mount the wires from center on the uprights, because too far out would mean giving the anchors a big moment arm to translate the uprights' deviations from vertical into twisting forces on the uprights themselves (bad for the corners), but too far in would start cutting into the usable print volume by limiting how far side-to-side the printing arms could move. My first design iteration errs well to the "closer to the center" side of this, but if I really start to miss my lost printing volume, I'll probably start experimenting with how far I can spread the wires without endangering the plastic corners.
Initial testing has shown that these wire braces can improve frame rigidity a LOT for a minor build volume loss; time will tell how it translates into more (or less, hopefully) wear and tear on the rest of the printer from reduced cyclical wear on the frame.
Oh yeah, and that secret to warp speed? Go into your firmware and change your DEFAULT_MAX_FEEDRATE to something like {400, 400, 400, 400} and your DEFAULT_MAX_ACCELERATION to something like {9000,9000,9000,9000}. If you've never done this, you've probably been stuck all this time with a speed limit of 200mm/s and 3k mm/s^2 on all axes. Even if you don't push your speeds/accelerations up to these new limits, you might be surprised at the effect of raising them -- in my case, I think my extruder had been constrained somehow by the lower limits, because until I made these changes my prints always suffered from underextrusion above about 30mm/s at 0.2 layer height.
But once you've raised the limits, it's worth seeing what a 350mm/s travel move really looks like at 5500 or 6000 mm/s^2. (Answer: sorta scary, honestly)
Designed from scratch in SketchUp.
Category: 3D Printer Accessories
License:
Creative Commons — Attribution — Noncommercial
7