2021-09-20 Meeting Minutes - Arm

Test 1 Results (Cytron control, monitoring current readout on power supply, gripper not installed an max 4.5kg weight was used instead of 5kg)

• First time testing, so data collection was a little bit questionable. Did not monitor true voltage into motors, so motor data is unreliable.

No Load Results

No Gripper, but 4.5kg attached to end of a6 results

• Keep in mind that adding the gripper holds the payload at a greater moment arm wrt a5, AND the gripper itself adds some weight

Test 2 Results (Proper Testing)

No Load Results

Max Load Tests

• Axis 2 belt fully died. Ran out of time trying to fix it and couldn't test A3. Ordered new belts, they should arrive tomorrow
• Need someone to finish testing this week (+ replace belt on A2). Use the kevlar reinforced belt. We likely won't be able to lift 5kg on A2 still, but just go as high as you can without skipping and we can run the calcs to see how they compare with the lower mass
• Also need to test A3 this week
• Austin cannot come to waterloo on the weekend, so someone needs to work with Ari and test out loads with PID - PM Austin Tailon Huang for how to test/what configs

General Observations

• Experimentally, A1 sees much lower loads than initially anticipated. As long as we meet our current torque requirements, we should be good
• Axis 2's current loading requirements are potentially sufficient - need to replace belt and confirm but experimental no load torque is very similar to calculated torque
• Need to test A3 properly - initial low loading scenarios may indicate that our current load spec for A3 is sufficiently high
• Axis 4,5,6 see higher loads than initially anticipated. Probably due to higher dynamic torque loads, and high torque losses due to friction in gearbox for the differential. 

Motors

• Obviously, we need to select BLDCs
• Will be selecting 24VDC BLDCs and scrapping 48VDC motors due to availability
• Integrate RLS or netzer encoders into joints, not broadcom AEATs

BACKDRIVING

• A1 is backdriveable but will not experience too many loads to backdrive it when driving around
• A2 and A3 both passively backdrive
• A4 has some backdriving (not much) under 5kg load
• A5/A6 occasionally had minor backdriving under 5kg load (only witnessed this once, but other times throughout testing it did not backdriving)
• Opinion: backdriving is only a concern on joints that passively backdrive when the arm is unloaded
  • Don't want to constantly consume power to hold arm position when unloaded
  • When arm is loaded, we will be drawing power/current anyways and will not carry high loads under extended periods of time - thus we can probably pass over fixing backdriving on A1, A4, A5, A6
• Worm gears or breaks for A2/A3? Or just choose a transmission option with high enough internal friction to counter natural mass moment (risky, easy to get wrong)

Axis 1

• Current belt tensioner loosened over time - and is somewhat difficult to readjust/tension on the fly
  
  • Solutions:
    • Apply a lead screw tightener
    • Use static tensioners (like the bearings idea on A3)
    • Switch to gearing (see kaydon bearing)
• Turntable bearing is not sufficent. Entire robot base tilts when larger moment loads are applied to A2
  • Solutions:
    • Spec a proper crossed roller bearing with high moment, axial and radial load tolerances Turntable Bearing Options (Kaydon, iGUS, chinese bearings, mcmaster)
    • Currently chasing down SKF on if we can spec a turntable bearing from Kaydon through our sponsorship
      
      
      • These bearings come with no teeth, or teeth on inner/outer race depending on config. If these are available, we can remove the belt and simply use the kaydon bearing as a gear/bearing at the same time - saving weight and components.
      •  Rated for various loads, weigh 200g-500g
• Better cable management solutions should be integrated for this pairing

Axis 2

• Obviously belt drive does not work due to slipping
• Single stage reduction option availability seems initially slim (based on Electromate comms, not much external research has been completed)
• Other than that, A2's linkage and frame construction seems OK
• Simplify construction and reduce number of parts for next revision
• Better cable management solutions should be integrated for this pairing

Axis 3

• Comprehensive load testing on A3 has not been completed, but from initial observations it works well. 
• Reduce weight if possible / simplify number of components for next design
• Should reduce the transmission ratio on the belt. A3 is currently fine, but if anything is learned from A2 large belt ratios can occasionally be unreliable
• Closer to chris annin
• Better cable management solutions should be integrated for this pairing

Axis 4

• Experimental loads are higher than initally calculated, however they are still within a similar range to A5/A6
  • We can likely still spec the same motor + gearbox for all three of these axis
• Design can likely be simplified, implement off-axis encoder to reduce weird components
• I think we should continue to have the A4 motor stick out on the opposite side of the link (counter act the moment)
• Better cable management solutions should be integrated for this pairing

Axis 5/6

• Differential is very rigid/stable
• Experienced loads are higher, internal tooth meshing friction may be high but also there may just be significant dynamic torque applied
• We are still within our expected calculated range (with SF), current requirements with SF are likely sufficent for upcoming design
• Reduce bulk for upcoming design
• Improve cable management
• Better cable management solutions should be integrated for this pairing