We have had a few reports of the maslow getting into a strange state (at least sometimes at the end of calibration) where all belts spool out slowly, like you were extending, but you aren’t pulling on the belts.
I asked the copilot ai to analyze the codebase for what could cause this problem, it identified something, but I’m not sure if it’s a symptom or a cause.
If you see this happen, please grab the logs, and your maslow.yaml file, and upload them to this thread with a description of what you were doing when this happened.
If it’s something that you can repeat, that is even better, because we can then have you run a new firmware to do more investigation.
I think this might have been what was happening to me. 4.1 machine. I was not able to download the logs. I think I might have had an old yaml file with the 1.12 firmware. When mine was slowly spooling out even beyond the end of the spool and then kinking up backwards it was only on one arm. I did take apart the machine and put that arm back accidentally in a different place and the problem seemed to follow the physical arm not the position. Today I took the arm apart to check the magnet but the magnet seemed solid. I also reloaded all the firmware and the new index and yaml file and the problem seems to have disappeared. I will play with it again tomorrow. Also possible that I had a loose or bad wire connection as I have unplugged and replugged all of the parts at this point. Not terribly useful but I can confirm slow movement in one arm.
just getting a report of it happening with one arm vs all 4 arms is useful.
it makes the problem harder, but it also means that we know there is something going on that’s not always going to be all 4 arms
One of the things that the AI speculated is if there are problems communicating with the encoder (including the encoder having low voltage), so your unplugging and replugging could have helped
May be a different issue, I uploaded on my other thread but just in case here is the serial from a run where the machine reported that it was in the middle of calibrating and was maybe successful but then started to draw in the belts in with no command untl they broke and it fell off the waste board trying to pull itself between the last two anchors. It was doing this movement after it had reported calibrating calculations.
update, going to tear the whole machine apart it looks like one of the belts flipped itself over in the gears? weird. There was no twist in the belt when I attached it.
I believe that Bar replied to you that the firmware detected you had a frame larger than the defined max, and instead of stopping, it redefined the frame to be no larger than the max.
as a result, it was trying to pull the belts, thinking that it needed to pull them in more than it really did. And the torque of the motor was enough to snap the belts.
@bar we should do some stress tests with a maslow between two solid anchors, and ramp the current up until the belts break, 4000 for the current limit appears to be too high
