Welcome! Log In Create A New Profile

Advanced

DiffBot - Differential Screw-Driven XY Gantry

Posted by Apsu 
Re: DiffBot - Differential Screw-Driven XY Gantry
September 29, 2019 01:03PM
Quote
JoergS5
This a very interesing idea.

If you suffer from backlash in the gears, you can try a friction drive (with 90 degree pivoted), or the idea of [www.youtube.com]

Looking forward your further development!

Thanks! I'm really trying to stay away from flexible materials entirely, that's one of the goals here (as well as simplicity and low part count/high temp materials), so I'm not looking at 90deg belt drives and similar.

Friction drives are possible, as well as some more complex gear tooth profiles, but I don't actually think the gear backlash part is much of a real concern. Here's my reasoning:

---

Any tooth-tooth backlash doesn't translate directly to linear motion in X or Y, because only the nuts move the axis. That means any backlash in the gears is divided into the screw/nut resolution, which is backlash/360 * lead.

For example, 1º of backlash would be: 1/360 rev * 8mm/rev = 0.022mm linear backlash, or 22 micron The worst backlash possible on my 20T (tooth) gears is 360/20 = 18º, and that's like, missing a whole tooth. So practically speaking, let's say it's about half, or 10º for simplicity. That would be 0.2mm linear (200micron), which is certainly not great, but I just want to illustrate that even with extremely poor meshing, the gear -> screw -> nut relationship makes it far better than it seems at a glance.

In comparison, 20T belt driven XY is often quoted at the calculated 25micron resolution that gets you, but as we all know when it comes to belt systems, there's often variations in tension, alignment, concentricity of idlers/pulleys/shafts, and a variety of other factors that reduce the resolution. So it's not like my design is wildly worse in comparison, even in a worst-case tooth mesh scenario.

Of course, there's considerations and concerns in mine as well, such as linear guide alignment, stiffness, straightness, deflection tolerance, screw straightness, etc. But those are present in a belted design as well, in terms of the mechanical gantry.

---

Now all *that* said, my videos had some gear mesh issues and also some linear rail/frame squareness issues. I saw and have improved them since then; I replaced the gear sets and adjusted spacing and can see/feel no discernible tooth-tooth backlash, and I replaced one of the linear rails as the old one was relatively rough and loose, and I printed an X carriage to further connect and stiffen the X screw/X rail portion. I'll have new video soon, but the behavior seems to have improved a lot.

Sorry for the lengthy response, thanks for your thoughts!

Edited 1 time(s). Last edit at 09/29/2019 02:08PM by Apsu.
Re: DiffBot - Differential Screw-Driven XY Gantry
September 29, 2019 03:05PM
I offered the ideas only because in your second video [youtu.be] at second 18, when you change from X motion to Y motion, there is a jerk in the gears which may produce X direction errors. I thought this is due to backlash, but this may have another reason.

I agree that using flexible material will create other problems.
Re: DiffBot - Differential Screw-Driven XY Gantry
September 29, 2019 07:32PM
Quote
JoergS5
I offered the ideas only because in your second video [youtu.be] at second 18, when you change from X motion to Y motion, there is a jerk in the gears which may produce X direction errors. I thought this is due to backlash, but this may have another reason.

I agree that using flexible material will create other problems.

Yeah, I saw it as well. Some of it was definitely backlash between the gear teeth. I have since changed them out for a tighter spacing, and that effect is gone. However, there is some twisting in the rail clamps underneath the gearboxes that I'm investigating. There's no longer a jerk as backlash is taken up, but there are some small inconsistencies on direction changes and differential transfer because the X rail is rotating a little in its clamps. I have some ideas to minimize or fix it, so we'll see what I can come up with.
Sorry, only registered users may post in this forum.

Click here to login