Good advice. I think I’ll go for a 2’x2′ or 2’x3′ addressable surface, since it sounds like a long y won’t be too problematic. I’ll keep Z to maybe 4-5″. (I know I’d like to carve a 3.5″ thick board, so that gives me a lower bound to aim for.) That’ll probably be a useful balance for me, and leave open the option of playing with aluminum.
Thanks for the welcome! Is this the mount set I should be building from? http://www.thingiverse.com/thing:1245848/#files
Any particular reason you did the mount in ABS? I just started my first roll of PLA (based on the MPCNC build advice), and I’ve been delighted with how much easier it is to get working well. If ABS is important for strength, I still have plenty to play with.
Before I continue, is that a Kossel delta in your picture? I’m looking for my next 3D printer and the Kossel XL is on my short list.
Why, yes, thank you for noticing! (blush) It’s a Kossel 250 from an Ultibots kit. It’s an aesthetically gorgeous design, just a pleasure to watch moving around, and the Ultibots kit was really nice to put together. Plus Brad provided really top-shelf support during the build.
However, I wouldn’t actually recommend the delta geometry. Other than the aesthetics, the main advantage of delta is no moving mass — no motors ride on carriages. For a pick and place machine, this is a huge win: low moving mass means a blazingly fast robot. For a 3D printer, it’s totally irrelevant, since the speed bottleneck is extrusion rate and cooling behavior of the printed part.
If that were the whole story, the aesthetics would still win. However, leveling this thing is a dog. On a Cartesian printer, you level it in physical space by turning a couple screws (or so I’ve heard), and you’re done for eternity. On a delta, leveling involves solving some crazy set of trigonometric equations; if any one parameter is off, the z=0 logical plane isn’t even a plane in physical space. I bought FSRs (force-sensitive resistors) with my kit, which lets the firmware probe the bed and then compute the parameters, which sounded great in theory. In practice, the algorithm is very slow — it tweaks one parameter at a time and hill-climbs, re-probing the bed between every adjustment, so “leveling the bed” takes an hour. (In principle, you could probably write a smarter algorithm that approaches the correct solution much more quickly using a multivariable derivative of the geometry, but that’s not what comes with the firmware.) The first few times it seemed to converge pretty well, but then, for reasons I don’t understand, it no longer does, and I can’t level the bed closer than 0.2mm. That’s enough that, on my bigger parts, the skirts don’t smush well enough to stay stuck together. The conclusion of this sad tale is that, as lovely as this printer is to watch, I’d love to be able to level the bed by turning a few screws. My next 3D printer will be Cartesian. (My next 3D printer will probably be a hot-end stuck to my new MPCNC, of course. :v)