11/12/2018 at 18:10 •
I'm building a robot arm similar to the SCARA architecture.
It's for plotting on a wall or chalkboard! It could also be used for pick and place on a flat surface.
Scale is ~ 2m radius, so it will have a large working volume (a very short large diameter cylinder)
The goal is 0.5 - 0.1mm positioning accuracy / repeat-ability and low cost.
Goal speed is ~ 1 rev / sec
This will be a significant challenge.
- Motor Selection - Aiming for Brushless DC for power density / low cost
- Driver Selection - Currently using Odrive, looking to implement with FOC (field oriented control) for low vibe / high torque at low speed.
- Encoder Selection - Investigating implementing 12/14 bit magnetic encoders w/odrive
- Arm Material / Cross Section - Prototyping in Aluminium, later considering Carbon Fiber Reinforced Polymer (CFRP)
- Bearings - Considering low cost IGUS slew bearings, may have too much lash, will also look into thin section roller bearings but may be cost prohibitive.
- Gear/belt reductions - starting with a 2 stage 1:11 (so total 1:121) cycloidal reduction, need to investigate machining tolerances with available equipment ( Tormach PCNC 1100 or a Haas TM-3 ), After speaking with Evil Mad Science peeps, also considering a differential belt drive (similar to how micometers work
- End effector - Starting with 1 or 2 simple microservos, may move to a bowden cable to keep mass away from end point. https://en.wikipedia.org/wiki/Bowden_cable
- Payload mass / load at end effector ( normal force and friction on writing instrument ) - High end constraint would be something like a gopro on the end.
- Continuous rotation (slip joints / wireless control) vs limited rotation (wire harness) - Need to investigate cost of making slip rings that can be clean enough to transmit control signals or power signals. Might make more sense to have multiple controllers connected wirelessly, and batteries in each stage.