concentric non-captive shafts for Z+A axes?

[WayOutWest]

Upshot: The idea here is to have no motors (i.e. heavy weight) on the part of the machine that moves in the Z-axis.


Has anybody built a PnP using two concentric non-captive lead screws for the Z+A axes? Here's how it works:

You've probably seen non-captive stepper motors like this one before. The inside of the motor shaft looks like the black plastic nut on the liteplacer's top plate.

8702dd9a7c46c9059b67c482ac6b967d.jpg (36.61 KiB) Viewed 5540 times

Now instead of a solid screw, use a pair of nested screws, like the green and pink parts of this video. Imagine that the yellow part is the motor housing. The threads on the pink part should run the entire length of the shaft, but the green outer shaft is still shorter than the pink inner shaft.

xxx.png (192.03 KiB) Viewed 5540 times

Now add a second non-captive motor whose interior teeth contact the pink inner shaft instead of the green outer shaft. Both motors' positions are fixed (i.e. bolted to the X-Y head). Mount the pickup head on the tip of the pink inner shaft (which should be hollow).

If the outer-shaft motor holds a fixed position while the inner-shaft motor rotates, you'll get A-axis-only rotation like you see in the video (the outer shaft moves along the Z-axis, but that doesn't matter -- it is shorter than the inner shaft).

If the inner-shaft motor holds a fixed position and the outer-shaft motor rotates, you'll get linear motion in the Z-axis and also rotation on the A-axis. To get Z-axis only motion, do the same thing but instead of holding the inner-shaft motor fixed rotate it just enough to create a counterbalancing A-axis rotation (the two motors must be synchronized, stepping together, or else you'll get "rotational vibration" when you want only linear Z-movement).

So basically you choose the mix of Z-translation and A-rotation by the "interference" between two co-axial motors.

Since no motors need to move in the Z direction the head would be feather-light, allowing extremely fast movement and probably all-plastic (i.e. even lighter) shafts since they carry almost no load -- just the component, shaft, and needle-tip.

There's probably a way to do something similar with the Z+X axis motors and again with the X+Y motors by using multiple "interfering" timing belts -- ultimately resulting in a machine where none of the motors move at all (like a Kossel/Delta machine), but the price paid is a very complicated rat's nest of belts. I'm not sure it's a win there. But for the Z+A axes I think the machine gets simpler rather than more complicated.

[Picky]

It's a nice design but I guess one will need a bunch of encoders all around this thing to make it useful. I've never seen anything like this implemented.

If you want to go all out, google Anorad Cyclone. That Z-Theta actuator blows minds with speed and precision yet a major overkill for a garden variety pick and place application.

[WayOutWest]

I guess one will need a bunch of encoders all around this thing to make it useful.

Why?

I mean encoders are always great if you have the budget for it. But if you don't, I can't see why this would need encoders to achieve the same accuracy as a conventional stepper+limitswitch design.

Of course you still need limit switches and a homing sequence on startup.

[Picky]

I guess one will need a bunch of encoders all around this thing to make it useful.

Why?

I mean encoders are always great if you have the budget for it. But if you don't, I can't see why this would need encoders to achieve the same accuracy as a conventional stepper+limitswitch design.

Of course you still need limit switches and a homing sequence on startup.

Theoretically speaking, you are correct saying that ideal actuator wont need encoders, but in real world...

Just watching the animation gave me the impression that axial backlash will be inevitable and difficult to counter. Even in Z-axis (vertical) application, the weight of the center shaft may not be enough to always eliminate tooth play. These non-captive "linear steppers" usually employ Acme leadscrews with ample of play and gobbles of grease. Now, multiply that by two for this type of actuator and remove all kinds of linear slides which would stabilize the Z... It will be impossible to tell the actual position of the center shaft without some kind of encoder due to axial backlash.

Radial play would be another (bigger) issue to solve and I don't even know how. Ever tried to wiggle the shaft on those non-captive motors? It will give at least 3 degrees in any direction and maybe more if you force it.

[JuKu]

For Z, the play doesn't matter. The variation in tapes, trays and paste viscosity (which affects the placement z) are bigger, and you need a spring loaded tip anyway. For A, arrange the software so, that the specified position is always approached from the same direction. (Standard LitePlacer software: emable A slack correction.)

[bobc]

There's probably a way to do something similar with the Z+X axis motors and again with the X+Y motors by using multiple "interfering" timing belts -- ultimately resulting in a machine where none of the motors move at all (like a Kossel/Delta machine), but the price paid is a very complicated rat's nest of belts. I'm not sure it's a win there.

There are some 3d printers using that technique (look for h-bot, core-xy), but doesn't seem to have caught on. I also saw some experimental designs for an extruder with a remote drive motor, but again not really popular.

[Picky]

For Z, the play doesn't matter. The variation in tapes, trays and paste viscosity (which affects the placement z) are bigger, and you need a spring loaded tip anyway. For A, arrange the software so, that the specified position is always approached from the same direction. (Standard LitePlacer software: emable A slack correction.)

Agree that axial play in Z does not matter much for pick and place. Matters for dispensing a whole lot more, but that's a different story.

Radial play will be the most troublesome to detect and counteract in such arrangement. Random tilt of the core shaft is unavoidable and unpredictable. Never know where it points at and even small deviation of 1 degree may result in large displacement of the long business end with the nozzle on it. I'm guessing one could add a couple of long sleeve bearings to reduce the play.

Here's a coaxial Z-Theta actuator I like better: https://www.youtube.com/watch?v=TueUrVKbzho

[WayOutWest]

Here's a coaxial Z-Theta actuator I like better: https://www.youtube.com/watch?v=TueUrVKbzho

Kirill, thank you so much for posting this! This is by far the closest thing I've come across to what I have in mind. And it is much more elegant than my approach.

The video does not show many details (despite all the text!). After digging around their website for a while I figured out how it works. This image from their catalog is the best explanation. Basically the shaft is spiral-threaded along one section and the other section has straight (i.e. parallel to axis) grooves. The nut on the second section has ball bearings that sit in the straight groove; this allows it to "force" the rotational position of the shaft. When the motors turn together you get A-rotation with no Z-motion. When the straight-groove motor holds position while the other one rotates you get Z-motion with no A-rotation. Very nice.

bssp.png (100.52 KiB) Viewed 5489 times

[WayOutWest]

Radial play will be the most troublesome to detect and counteract in such arrangement. Random tilt of the core shaft is unavoidable and unpredictable.

Yes, but a steel spring like the one on the liteplacer will at least force the tilt to be consistent -- always the same amount at a given rotation. So you can compensate for it, and that compensation can be calculated automatically (at, say, 30-degree intervals) using the up-camera. I'm doing this and the recalibration process takes less than a minute, although I do need to do it every time I change the nozzle. Thereza was the first one on this forum to do it, he calls it "needle wobble".

[EdwardsKathleen]

If the inner-shaft motor holds a fixed position and the outer-shaft motor rotates, you'll get linear motion in the Z-axis and also rotation on the A-axis. To get Z-axis only motion, do the same thing but instead of holding the inner-shaft motor fixed rotate it just enough to create a counterbalancing A-axis rotation (the two motors must be synchronized, stepping together, or else you'll get "rotational vibration" when you want only linear Z-movement).