With a mini project designing a cycloidal drive using bearings, I wanted to put it to use in a larger project. I decided to design a 4-DoF cobot that would be driven by smaller versions of these cycloidal drives. I planned to use NEMA 17 motors for the joints and 3D print the majority of structural components.
I started out by conducting calculations on torque, on the required torque on each joint, based on the motor weight and the estimated weight of structural components. This provided me with weight and length goals for each joint and ensured the joints could be driven with minimal backdrive and that enough torque could be generated to lift its weight and a payload. I added a 2x safety factor for the initial calculations since mass of the linkage and cycloidal drive were largely unkown and was based on estimations.
I first started by designing the cycloidal drive used for the joints. I also created a rough CAD design to verify geometry and obtain a form factor for the joints. This joint form factor provided dimensions required for the cycloidal drive.
19:1 reduction ratio
Overall diameter = 70 mm, including wall thickness
Roller radius = 29 mm
Roller pin radius = 2.4 mm
Eccentricity = 1.2 mm
Width = 40 mm
These parameters were once again verified using the Desmos program used in Cycloidal Gear Design (WIP) (https://www.desmos.com/calculator/eqcjxzrkjv)
Rough CAD for Geometry Verification
Motion Analysis Result
I utilized the same process from my previous project to design the cycloidal drive. Several changes to my cycloidal gear design were made.
Instead of bearings, bushings were used on the output disk
I relied on a more robust system where the output disks were stationary in relation to the joint it was sitting on
The arm has roller pins essentially embedded into the structure, which increases ease of assembly
Using these new changes, I built and ran a motion analysis on the new cycloidal drive. It successfully showed the arm rotating.
Section View
Then, I created a more thorough CAD model, which includes the housing of the cycloidal drive and the motors. Again, I ensured bearing fit and conducted weight-saving on the linkages. The linkages were at the correct distance between the joints. I obtained a better weight for the linkages to change values within the torque calculations and verify that the reduction ratio is still valid. Since I was able to better estimate the weight between linkages and the weight of the cycloidal drive, I decided to reduce the safety factor to 1.5. As seen below, I placed each linkage into the Bambu slicer with Bambu PLA (for now) and 25% infill to obtain a rough weight estimate.
V1 Rough CAD (July 25)
Linkage slice results in Bambu
Static load testing between J2 and L2
Cycloidal drive demo
The cycloidal drive is the most important component within the entire project. To test its accuracy to hold the expected torque, I decided to manufacture only 1 of the cycloidal drive and conduct testing to determine the accuracy of its torque capabilities and accuracy.