Didnt get a chance to measure what mine were. Still curious if we have a manufacturing defect. Did you get yours running?
Well, kind of… I started another topic about new issue…
I’m getting successful calibration but things go out of whack when machine moves to corners.
Can you share the link? I have a few other issues as well
I am having a really similar issue, I recently installed a new belt (because one of them caught in the gear) so I thought I would run calibration again (I successfully ran calibration previously). After it completes the first few waypoints it gets into a never ending fitness loop. Do I need to delete the old calibration files and start over?
Maslow-serial(1).log (34.0 KB)
1 Like
https://forums.maslowcnc.com/t/calibration-wont-come-to-en-end/24132/25?u=ross_o_connor
It should delete it after you calibrate. Do you have something binding?
The belts are not stuck if that is what you are asking, otherwise I’m not sure
I’m able to get the data here to find a solution if I throw out the first measurement which makes me think that something funny is going on with that first measurement (which is pretty common).
I think that we need to come up with a better system for taking that first measurement which is more reliable. I have some ideas but I will make a separate post for them.
Are you in horizontal mode? Are you giving it a little tug at the beginning of the calibration process like I do here?
1 Like
I am in horizontal mode and yes I give it a little tug to get it tight. I just watched the video and I am giving it the tug on the same side as the power cord. (I have the dust collection port on the other side opposite side from the cords)
Thanks
1 Like
It’s definitely that first measurement that’s throwing things off. Does anything seem strange with that first measurement?
1 Like
when it tightens the fisrt two belts the maslow moves about 4 inches before it tightens the other two belts on the first measurement.
I will video the start when I try it next.
1 Like
@bar can we try having it try tighening all belts twid3 so that if it moves
during the first pass, the 2nd pass around will make sure the belts are tight
anyway?
David Lang
I’m not quite sure I understand 
Bar wrote:
I’m not quite sure I understand
tighten tl, tr, br, bl, tl, tr, br, bl
this way if tightening the tr makes the tl belt slack (by moving the sled) the
2nd round will make sure that the tl belt doesn’t have slack in it.
David Lang
1 Like
Basically like go around in a circle and tighten each one? We currently toggle bl, br back and forth until neither one changes
that seems a reasonable approach for a vertical machine, but for a horizontal
machine, tighening the bl belt could pull the sled to where the tl belt is slack
(and tightening the br belt may not pull it tight if it would need to stretch
the bl and tr belts to pull the tl belt tight)
let me know if I need to make a diagram of this.
think of cases where the router is not centered (or at least, not centered after
the first belt pull), so you can have a case where two belts are ‘tight’ but the
belts are not in a straight line. if the next belt you tighten would stretch
those two (the outside of the angle), everything is good, but if the next belt
you tighten is on the ihe angle, it can move the sled and make both belts loose.
the best answer may not be to go around a circle, it may be something else
say tl, br, tr, bl and repeat until they are stable
I don’t think it’s likely to need to do more than two cycles through all 4 arms,
but the fact that it helps a horizontal calibration to pull the sled away from
the two tl anchors says that there the potential for movement that we are not
currently accounting for.
for debugging purposes, report each measurement as the belt becomes tight. That
way if we are seeing something odd/inconsistant, we will have the data
David Lang
I don’t think this latest batch is working at all, if any. I’ve been attempting to calibrate for 2 months with no success. Most of the time the machine just cycles through the same errors until it disconnects itself from the network.
Juan Salgado wrote:
I don’t think this latest batch is working at all, if any. I’ve been
attempting to calibrate for 2 months with no success. Most of the time the
machine just cycles through the same errors until it disconnects itself from
the network.
This is a long thread, so I have lost context, but if the problem you are having
is a very low fitness (<0.2) this is a known pathalogical condition, we need to
figure out what’s going wrong with the first measurements so that you can get a
more sane fitness to start with.
This is why we are focusing on the very first measurement. If that is wrong,
everything is going to have serious trouble because it’s based on that.
currently, when it’s doing the first measurement, it is only tighening the
bottom belts (a bug when in horizontal mode that you can work around by pulling
the sled away from the top two anchors when it’s doing this first measurement,
you are simulating gravity pulling down), Based on this thread, Bar is working
on a patch to avoid this problem.
so if you start again, retract, etxtend, apply tensions, and then start
calibration with you simulating gravity by pulling the top two belts tight, does
it do any better?
David Lang
David Lang wrote:
so if you start again, retract, etxtend, apply tensions, and then start
calibration with you simulating gravity by pulling the top two belts tight,
does it do any better?
One other thing to try is to set the workpiece and wasteboard thickness values
(currently only in the settings tab) or set the Z offsets to account for them.
We have seen some reports of this helping, which is why the workpiece and
wasteboard thinkness variables were added (I think in the last release,
otherwise they will be in the next one and you have to change the z offsets)
please let us know if this makes a big difference for you?
David Lang
Hey, yes, I followed the video that’s linked in all these posts asking about calibration assistance. Kind of mind blowing something so crucial on the fundamental function of a product wasn’t flushed out enough rather have the end user be the beta testers.