Project Debrief
I assembled all the parts from my quadruped actuator design section for testing. Everything was printed in PETG (higher melting temperature at ~ 75°C ~ 80°C vs PLA which is ~ 55°C ~ 60°C).
Key parts
Stator + Rotor
- 15m AWG 26 & AWG16 wire
- 1x heat shrink
- 3x bullet connectors (male & female)
- 22x N45 magnets
Planetary Gear System
- 1x planetary carrier outer bearing (45mm ID, 55mm OD)
- 1x planetary carrier inner bearing (15mm ID, 21mm OD)
- 1x stator anchor bearing (35mm ID, 44mm OD)
- 3x steel dowel pins
- 8x 10mm M2 screws
- 3x 5mm M3 screws
Assembly
Front
Back
Interior
Challenges
1. Motor winding
Issue #1: Inconsistent copper wire length per tooth due to difficulty manually wrapping under tight spacing
In my 1st prototype, I used 6 parallel strands to wrap around each stator tooth; however, the very tight spacing meant sometimes the wire wasn’t fully wrapped around a tooth and therefore some had 5 instead of 6 turns.
- Potential effects
- Improper spin from commutation timing error - as I was using an ESC, the combination of uneven wire count and the PETG stator and rotor may have been messing up its readings of the back-EMF zero-crossings, which delays the ESC’s commutation and could cause jagged motion in the rotor.
- Torque ripple from unequal MMF per tooth - Ampere’s law states MMF is directly proportional to N, number of wire turns; therefore with some teeth having one less turn, they may have created stronger and weaker magnetic field regions that create uneven torque output.
- Uneven heating - as some loops in each phase were bulkier, there’s extra heating in the copper wire which could cause some teeth in this PETG sample to soften more than others.
- Solution: Swapped AWG26 with AWG16 wire. This meant less copper filling and a lower current rating, but for purposes of testing my actuator CAD design, I thought having an even torque output was more important.
Prototype 1: 6x AWG26 strands twisted into 1 phase
Prototype 2: 1 AWG16 strand for each phase
Issue #2: Enamel causing difficulties soldering wire
The enamel on the wires were also very hard to scrape off, which made it very difficult to expose enough wire for soldering to create the Y joint.
- Circuit completion check #1: After measuring the connectivity between individual phases at their ESC connectors and the Y joint, I confirmed that they all had similarly low resistance levels. This was the same when I measured the resistivity between each of the 3 phases, meaning the circuit was complete.
2. Testing motor spin
Issue #1: Lack of stator and rotor centering mechanism
- Potential effects
- Torque ripple caused by unbalanced magnetic pull from inconsistent air gap - the coils with a smaller air gap have higher flux density, which may induce extra torque that causes inconsistent movement.
- Improper spin from commutation timing error - as mentioned before, the zero-crossings may shift away from the coil’s true position, hence the back-EMF values may be corrupted and trigger the ESC to commutate at the wrong time.
- Solution: test with stator fitted onto back housing cover, then upgrade stator to magnetic steel material & rotor to steel (in progress rn!)
Before: testing with no mechanism to center and anchor stator relative to rotor.
After: testing on back housing cover
Oops - 1st back-housing prototype was too tight. Had to pry it out.
🚧 (Troubleshooting in progress) Issue #2: Rapid jittering, but no spinning
Issue #3: Keeping magnets in place
- I used JB weld to secure the magnets to the rotor during testing and held them in place with masking tape. However there was still weak attraction between the magnets which caused some to be somewhat slanted, instead of fully fitted inside their slots.