Strain Wave Gear Robotic Arm

The empty space within the FlexSpline is where the magnetic encoder will be placed for final joint angles. This design is still a little larger than I hope the final design is. I may look into optimizing some of the space where the motor connects to the planetary gear, possible make I direct connection to the input of the gear rather than the yellow piece that is currently between the two.

• Added Grease to the Planetary Gear for the first time, this seems to fix all of its friction problems
• Made the mount to connect the motor to the assembly and did the first proper test
  • Runs at low Current
  • It seems to be loud, I will see how enclosing the final Gearbox changes this
  • The Gearbox is extremely strong, I will be properly testing the strength and backlash soon

• Slimmed the walls of the FlexSpline
• Redesigned the outer Wave Generator profile

Both the Wave Generator and the FlexSpline seem to work very well now. I am still struggling with the Planetary Gear box as there is too much internal friction. It seems to be the lowest efficiency part. I am holding back from adding grease until I build out the body of the gearbox to hold the motor properly

The roof of the FlexSpline where the output is still flexes when in motion. I suspect when I attach the output bearing and lid the FlexSpline will become much less flexible and will require redesign.

If the problems with the friction in the Wave Generator come back, I may switch the outer casing to TPU. 

The New Design takes a different approach the the issues from before:

Results of First Test

• Wave Generator has too much friction making it quite stiff
• Embedded 5:1 Planetary Gear does not run smooth, could be fixed with lubricant
• FlexSpline is not Flexible enough, I will be thinning out the walls in the next version
• Wave Generator Slips deeper into the FlexSpline, I plan to add a ridge in the FlexSpline to fix this
• PLA threads in the poles that run down the center do not work at all. Not sure how to secure these yet, possible threaded inserts or one long bolt down the center.
• Other than that this version feels much better than the last. 

5:1 Planetary Embedded in the Wave Generator. The green/yellow rods act as a rigid pass through giving access for an encoder in the unused space in the center of the FlexSpline Cup

Found This Git Repository https://github.com/abssimon/harmonic-drive-tooth-profile-generator

Used it to Generate This Result:

Flex Spline teeth skip under load. same results occur no matter the tooth profile, although I am using default tooth profiles from Fusion 360 addins, not sure if this will be an issue. Either way I believe the problem lies in the material of the FlexSpline (TPU)

I am using TPU in the first place to avoid the large flexure area needed if stiffer materials are used. TPU is super flexable, but maybe so flexable that the teeth are flexing as they are forced into the Circular Spline

I will be creating the next version. this time with a PLA FlexSpline to test more rigid materials. If this is successful the final design will not use PLA as it is too brittle for my liking.