I’m interested in building the MPCNC, but I’m in doubt if my printer could print the parts correctly since it is not the most accurate printer. What are the tolerances on the printed parts? Will the whole system wobble if some walls are (for example) 0.2 mm wider than the stl-file? My main concern is that the bearings will not be aligned correctly and therefore will be too loose around the pipes…
A 20 x 20 x 20 (mm) calibration cube prints at 20.05 x 19.96 x 20.01 (xyz) . A 40 x 40 x 40 (mm) calibration cube prints at 39.90 x 39.86 x 39.90 (xyz).
Secondly I am in doubt about the dimensions I should go for. Is the standard size in the “Cut Calculator” (300 x 300 x 100 mm) a good place to start for a rigid machine? The usecases would be as a router and/or lasercutter.
You can also measure diagonals to make sure it is square. The numbers are not that bad but it does seem like some easy improvements can be made. Look for some of the 100mm test parts, they are quick and much more accurate. Change a few settings print another and you should be fine.
Almost all diagonals are something around 27.4 or 27.8, with the lower 27.4 being caused by the rounded corners produced by the round nozzle (using 0.6mm nozzle). So I think it is pretty squared.
What settings should I dial in some more? The only thing I can think of is the extrusion width and the feedrate. If i raise one (or both) then the 40mm cube might get more accurate, but the 20mm cube is way off.
What exactly do you mean by 100mm test parts? A 100mm cube?
So I printed the proposed tests and dialed in the E-steps of all axis as to get exactly 100mm (after calibrating the E-steps, they are all within 100 +/- 0.05 mm). But then, when reprinting the 20mm calibration cube, all dimensions were a little off: 20 +/- 0,1 mm)
I suspect this might be due to a little over-extrusion, but I’m not willing to sacrifice strength of the part over accuracy.
Would this suffice to print all the necessary parts of the MPCNC and get a solid and accurate machine?
Before any tweaks on the step per mm, you should first get your extrusion right. Otherwise you’ll end up with inconsistencies and a lot of problems.
So first of all, you need to work on that. The steps per mm are not supposed to vary from the original settings, provided you are using the right belts, pulleys and motors, at least for X, Y and Z axis. I know it is sometimes tempting, but you should not touch them.
So check your extrusion, check your temps and your speeds. All of these can impact your prints quality and dimensional accuracy.
You do not need to over extrude to get stronger parts, this is a bad fix. You need to get the correct nozzle temperature, part cooling fan speeds, extruder steps per mm and actual filament diameter, and correct XY feeds speeds. I know it is tedious, but this is the only way that will work all the time.
Please post us some pictures of your machine and of the prints you’ve made, otherwise it is almost impossible to tell you precisely what might be wrong.
I’m not at home right now but will post some pictures tomorrow. I have no problem with consistancy; straight walls are straight, no blobs, no zits, no wobble… The temperature, extrusion rate, printing speeds etc. are pretty well calibrated as well. I have had this printer for about 3 years now, it is the “Velleman Vertex K8400”. The only thing is the dimensions of the 100mm test-parts were off by about 0.7mm. And i got the feeling they were dependent on the length of the part.
The only thing I don’t really account for is that the filament is 1.74 instead of 1.75, therefore I’m overextruding just a little.
My main concern is that my printer might not be accurate enough to print the parts for the MPCNC, what is required? How much tolerance is there? Will my printer be able to produce a solid MPCNC?
Yes, you are within a very safe range to print the parts.
We are all just super dorks around here and love to help and try to really dial things in. Accuracy and firmware options have been improving so much we can’t help but push them to the limits. I would stop even worrying about a 20mm cube. Lets say you had a 0.2mm backlash in the belt system for some reason, a 5mm cube would be 5mm+/-0.2mm (8%), the exact same error over 100mm +/-0.2mm (0.4%). So why I ask for big tests is to compare… If your 20mm cube has a 0.2 error and your 100mm test has a 0.2mm error it is a hardware problem, if your 100mm test has a 1mm error it is firmware.
Again, sounds like you are well within the limits. Make sure it is as square and accurate as possible, good to go.
