Austin Yuming Conversation Minutes

Owned by Yu-Ming He
Jul 10, 2024
2 min read

How they did calculations:

  1. Calculated the loaded torque o n the joint at maximum possible moment, plus inertial forces with an assumed acceleration then added a safety faction when picking out the joint torque

  1. Assumed this specced motors off nominal torque or higher operation torque for maxon motor as they provide curves and data for how long the motor can be engaged past nominal limits but before stall

  1. Updated axis 2 3 4 joint calculations for shortened arm

2022 Arm Actuator Selection Options

  1. Matlab calculations are over estimated, purpose was to size the motors for the prototype

  1. Based off 5kg load, not 10kg

  1. 22 torque for axis 3 is not including 3:1 belt drive, which would bump torque to 66

  1. Reason why they have a lower safety faction and close torque values:

  1. Pressured to design the arm pretty fast, because team switched to BLDCs and new voltage architecture (which requires new arm in 1 year)

  1. Arm is most expensive part, so cutting it closer to required limits had a lot of benefits (smaller form factors so less weight, less cost, etc)

  1. We did all our calcs when arm is at max moment (joints in straight line) we planned to integrate some software checks to prevent stall, and reconfigure the arm to a more advantageous position if required

  1. Prepared to lose points on some of the heavier items in the comp if necessary bc not a lot of mission points and time are associated with moving max 5kg load at all times, so if we found we were close we would ignore as a last resort

  1. When arm got wrapped the plan was to always build this arm, stress test the controls and IK and improve the axis later

  1. But by the time arm was built we lost FW team

  1. Took time to rebuild resources, esp BLDC motor control not beginner friendly

  1. He thinks axis redesign for any of the 6 axis would be worth doing AFTER following is done:

  1. Intelligent Testing and control (simulating competition tasks, logging motor performance data, see how torque spikes with diff activities, seeing if we r losing position control due to bad encoder placement, or just seeing what and when arm struggles)

  1. Through testing might identify one set of joints or one joint fails the most often and use to redesign

  1. Hard to redesign without knowing what the performance pitfalls are

  1. Even though torque are too close to joint outputs, question is:

a.      How often are relevant is the max loading condition to the arm bottom line performance

b.      If you can hit 60-80% of arm mission with current design, but science is only 15%, make sense to leave arm and focus on science for example

10. For weight reduction can be done with more data

  1. Structural components like (strain wave gearbox, bracket, or big CNC metal link for arm) have been designed for published loads

  1. But need to know how good that guess was before shaving weight off