You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
While flashing moteus firmwares on Upkie Zero, we observed that flashes would repeatedly hang at the "verifying" stage on the right leg (bus 2, a.k.a. JC2), but not on the left leg (bus 1, a.k.a. JC1).
Investigation details
Reducing the bus to "2=4" (one servo, the two others powered off although I didn't disconnect their CAN cables): no effect
Powering off bus 1 and using --pi3hat-cfg "2=4,5,6": no effect
Swapping terminators (checking their resistances at the same time): no effect
Disconnecting JC1 and connecting the JC2 connector to JC1 on the pi3hat: worked
Try again with --pi3hat-cfg "1=1,2,3;2=4" (this time 5 and 6 properly disconnected, both CAN and power, not just power): high failure rate
Try again using a CAN cable not previously part of the bus, with --pi3hat-cfg "1=1,2,3;2=5": high failure rate
Try again with servo 1 on JC2: high failure rate
Checked the resistance between CAN_H and GND (:warning: black cable of multimeter touches GND, not the other way round) on all pi3hat JC connectors: 2.2 MΩ on all of them
Same tests on JC3 rather than JC2: low failure rate, a.k.a. it works™
Decision
After investigation, we found that the issue does not happen when connecting the right leg to JC3. Since JC1-JC2 and JC3-JC4 are "siblings" on the pi3hat, the bus layout (JC1, JC3) is at any rate more efficient than (JC1, JC2), so we should switch the right leg to JC3.
The text was updated successfully, but these errors were encountered:
While flashing moteus firmwares on Upkie Zero, we observed that flashes would repeatedly hang at the "verifying" stage on the right leg (bus 2, a.k.a. JC2), but not on the left leg (bus 1, a.k.a. JC1).
Investigation details
--pi3hat-cfg "1=1,2,3;2=4"
(this time 5 and 6 properly disconnected, both CAN and power, not just power): high failure rate--pi3hat-cfg "1=1,2,3;2=5"
: high failure rateDecision
After investigation, we found that the issue does not happen when connecting the right leg to JC3. Since JC1-JC2 and JC3-JC4 are "siblings" on the pi3hat, the bus layout (JC1, JC3) is at any rate more efficient than (JC1, JC2), so we should switch the right leg to JC3.
The text was updated successfully, but these errors were encountered: