Incorrect dimensions on printed objects

[bot]

I highly doubt that a finely tuned Taz printer has "the exact same issues" as a stock Rostock max. It's been a long time since you ran a stock rostock, hasn't it michael? Mine is still mostly stock: arms, carriages, etc etc. I can say with absolute certainty that even my other delta is far far better (but not even close to perfect) with holding tolerances than the rostock is.

[mhackney]

For the types of features the OP is discussing here, it - and the Taz - have the same issues. True, I have highly modified my Rostock. My friend has a stock V2 with the older arms (not the new balljoint arms) and it can print the parts for my fly reel with more than adequate accuracy and precision. Take the "gcode challenge" for a couple of slicers and see what you find.

[Polygonhell]

My MAX is pretty close in tollerances to the Cartesian printers I have here. Like Michael I have an early V1, built before there was a manual, and until very recently outside the hotend (which changes every couple of weeks) and additional drivers, it was entirely stock, nasty 1st gen arms (you had to sand them to fit) and metal on metal U-Joints included. I put Trick laser arms on it a few months ago, which certainly does remove some of the slop in the effector.
I printed a mechanical part yesterday that was 90x60mm and it was within 0.1mm in both axis, the holes were undersized, but that's caused by a different issue that all printers share.

I'm not totally sold on the Delta design in general (subject of a different post), but there are some things I like about it.
Layer alignment tends to be better than most cheap cartesian printers, though this is largely because they tend to use cheap screws for Z, and insufficient rigidity in the linear motion systems used on the Z axis.

Outside of complicated calibration, what the Delta lacks is a good way to resolve asymmetric inaccuracies in X/Y, these are usually indicative of inaccuracies in the build.
The tendency with cartesian printers is for people to tweak steps/mm for each axis independently to get what appear to be accurate dimensions, but usually they are correcting for build inaccuracies, not as they tend to believe differences in the timing belts. There is no equivalent quick fix on a Delta because any change effects all of the axis.

Thingiverse is full of models with errors and no tolerances, people will tweak their printers to make them work, or fiddle with the model after the fact.

[bot]

My main gripe is the difference in thin wall thickness depending on where on the buildplate it is situated. I find this is due to the direction the path is laid in. The arms seem to follow different paths when they move in opposite directions. (ie clockwise vs counter clockwise, when drawing a hollow rectangle).

I'm not saying the rostock max isn't a perfectly capable printer... but we are all lying to ourselves if we don't realize the dimensional tolerances it actually can or can't hold. I find variations in thin wall thickness of .3mm or more, and this isn't over large distances. This doesn't happen (to nearly the same extent, if at all) on my other delta, and I've seen countless reports of it online with the max.

[mhackney]

What other deltas do you have bot? I have several mini kossels with Traxxis joint arms and either linear rails or carriages and a K250 with mag ball arms and carriages. I print my reel parts on all 5 of these machines (along with the Max) and I mix parts from them with no problems. But I've spent a lot of time calibrating and minimizing motion issues. I don't print a lot of thin wall rectangular parts so my experience may be more positive.

[bot]

My only other delta is a custom frame with berry bot parts. I would never mix prints from it and the rostock max... it would be too plainly obvious which was from which printer. Prints that reach near the extremities of the build plate on the rostock max can be off by as much as 1mm or more. My other delta, while not perfect in this regard either, has much less variation and a much larger usable accurate build envelope.