I’m also experiencing something which I think might be related.
I did: power cycle, retract, extend, mount to frame, take up slack. at this point it would jog around. and move Z
However I ran some Gcode and had it on mm not inches so I hit STOP. (I could not see what else to do other than let the code complete.) When you hit stop it says machine must be power cycled. If you do that there appears to be no way to recover. It seems you have to take it off the frame and redo everything?
On repowering it and reloading the browser tab the red alarm button is there. Hitting that changes it to green/idle, and trying to jog returns these suspiciously extended belt length looking errors. I’ve now seen this about 5 times since I keep tripping up at various steps.
MSG:INFO: Caution: Unlocked]
[GC:G0 G54 G17 G20 G90 G94 M5 M9 T0 F0 S0]
[GC:G0 G54 G17 G20 G90 G94 M5 M9 T0 F0 S0]
[G54:0.000,-450.000,-0.020]
[G54:0.000,0.000,-0.020]
[MSG:ERR: Position error on Bottom Right axis exceeded 1mm, error is -1754.638mm]
[MSG:ERR: Position error on Top Right axis exceeded 1mm, error is -1749.787mm]
[MSG:ERR: Position error on Top Left axis exceeded 1mm, error is -1745.916mm]
[MSG:ERR: Position error on Bottom Left axis exceeded 1mm, error is -1764.477mm]
so somewhere in there it is not recording belt positions correctly. it would appear to be zeroing it in one place and then computing the error to be exactly the extended belt length
Yeah, the stop button is more of an emergency stop button at the moment. Someone pointed out last week that it wouldn’t always stop the machine which seemed like a hazard. I need to dig into making it more gentle.
This is without doing the retract → extend cycle, right?
Correct - since it was just running correctly and I hit stop. this is what it returns on power cycle and browser refresh. and then hitting red alarm/unlock and trying to jog.
does it not record the position across a power cycle? Seems like that would be a key functionality
because… despite all kinds of chaos in starting and stopping jobs and redoing things after missing some setting … I would think the belt lengths should only be screwed up if you force them past the encoders or force them when the encoders aren’t computing length (such as powered off) in other words the functional priority would be to assume the encoder and stored position is correct and carry on.
I agree, I think that we can make it more robust to things going wrong without needing a full reset. Certainly hitting the stop button shouldn’t require turning it off and back on again. The whole system is very brittle right now so if anything goes wrong it doesn’t recover well, but there is no reason that it can’t.
while there are pretty high gear ratios involved, there’s nothing preventing the
belt length from changing between power off and power on.
Also, unless the machine is idle at power off, there’s a chance that recording
the position doesn’t happen correctly.
With the earlier versions of the maslow, where the encoders recorded only
relative movement, this was a major risk. With the absolute encoders, you would
have to rotate the encoder magnet 1/2 turn to become ambiguous, but right now
the system doesn’t trust anything after a power cycle.
@bar, given the very high gear ratio, and high friction of the spools, I think
it’s reasonable to ask at startup if things have changed, and if not, assume
that the encoders have not rotated more than a 1/2 turn.