Belts

[orcinus]

Eccentricity of the pulley/shaft is a good point.

However, that is going to always produce the same error in the same position, so it would result in deviations in, for example, slanted walls, but not straight vertical ones (since the deviation caused by the offset from the shaft axis center would always be the same for the same X or Y coordinates). Oblique surfaces would end up with regular undulations, though.

It would be interesting to see a comparison of direct effects on print quality (if any) for various belt systems.

[orcinus]

I just saw the post by Orcinus and something caught my eye in the marketing brochure: who uses a 10 tooth pulley with an MXL belt?

No idea, that really is a (more than a bit) tiny.
Probably an extreme edge case to illustrate the differences in geometry.

Note, however, that the text under the backlash diagrams says 20 teeth!

[Cre8ivdsgn]

Now why would marketing show a 10 tooth profile photo and provide chart notations based on a 20 tooth profile?

The error will only be in the same position in direct drive applications. It would have odd periods in rack and pinion applications (Dirktheengineer's buildlog showed something that could be attributed to this sort of thing - at least in part).

The only problem with running through tests of belt types is that it is hard to get belts that yield the same center distance between pulleys. The Gates sample I got is a good bit longer than any other MXL I have. And if I cut the belt and make the test stage have a straddle clamp over the belt ends for the carriage, do I run the risk of introducing another source of error that would invalidate the testing?

I think if you are going to do testing, it has to be with endless belts.

[r691175002]

Accuracy is not a primary requirement of 3d printers since the plastic extrusion process will make it hard to hit any exact dimensions anyways.

What you do need is flawless repeatability so that subsequent layers are always positioned exactly where the previous layer was.

I would avoid timing belts that have space between the pulleys and the teeth, its just asking for trouble. Even if you tighten the belts properly, they will still wander from one end of the tooth to the other over enough direction reversals.

The specification for the original trapezoidal pitches leaves a lot of room for the belts to move back and forth but I suspect that what you actually end up with depends a lot on who you buy from.

There are tooth profiles specifically designed for linear positioning (AT5) that have teeth on the belts that are larger than the corresponding grooves. GT2 also claims to have zero backlash.


MXL backlash clearly exists, it is documented everywhere and there are many alternative tooth profiles designed for accurate positioning. The real question is whether scrounging a few extra microns of repeatability out of the belts will make a difference when the rest of the machine uses 50$ in parts for the frame and linear motion.

[robotdigg]

Really good topic. I learn a lot.

Has any of the ORDbotters tried switching from MXL belts to GT2?
Or perhaps some other belt system?

Are there any visible benefits? Less backlash / higher precision?
Is there any sense in considering this?