tr7 wrote:
Okay, that’s a sensible approach. Now, when your belts weren’t extending
enough, was this prior to your calibration? Mine aren’t extending far enough
anymore, after calibration.
can you please post your maslow.yaml file so we can see what it has in it post
calibration, and also please measure anchor distances manually (all 6 distances)
so that we can validate that the calibration results are plasable.
That should let us trouble shoot this a bit better.
David Lang
Sure, here is everything. I also included a photo of the machine, just in case there’s some obvious error (like mixed up arms). The top right belt is the one that’s not reaching atm.
This was pre-calibration. Don’t mess with yours unless you’re going to do another. My numbers stayed higher than the actual frame size on their own after calibration.
Ah okay. Looks like my situation is a bit more of an anomaly. Yeah I won’t mess with anything at this point.
Yep. I had misinterpreted and was thinking you still weren’t out of the calibration wilderness quite yet when I originally responded.
No worries - its hard to keep track in all these long threads.
Hmmmm, not sure what is going on, but the way I read it, your frame was measured by the Maslow as roughly 115.1 X 81.25 inches (2924.3 mm x 2066.5 mm) which totally explains why it didn’t extend the belts enough.
My suggestion would be to start over. Save the currently maslow.yaml file somewhere so you can go back and troubleshoot (the file is the only thing that saves the calibration numbers), but looks like it is just incorrect for whatever reason. Just put the measured frame values back in th setup screen and recalibrate.
Now, I am a complete newbie, so you may want to wait for someone who knows more than me, but I am impatient. 
I agree with @Aggiehouse. I’m not totally sure what’s going on, but as long as you make a backup of your Maslow.yaml file you always have the option to reset things to the way they are now. I would try running calibration again and see if you don’t get better results
Having just finished calibration on the spoil board and then doing a litlle reading here, I am curious if the 3/4" height difference will cause issues. Should I re-calibrate? Maybe shim the belts with 3/4" blocks? The latter maybe is better if I expect to cut pieces of different thicknesses?
hmm, manual calibrations show (assuming that I converted inches to mm correctly)
Maslow_blX: 0
Maslow_blY: 0
Maslow_brX: 3022
Maslow_brY: 0
Maslow_tlX: 4.318338170839752
Maslow_tlY: 2339.996015371702
Maslow_trX: 3029.3168431502318
Maslow_trY: 2336.988545929636
your maslow.yaml says:
Top left corner of frame. nominally 0, height
Maslow_tlX: -27.6
Maslow_tlY: 2064.9
Maslow_trX: 2924.3
Maslow_trY: 2066.5
Maslow_blX: 0
Maslow_blY: 0
Maslow_brX: 2953.2
Maslow_brY: 0
those are very large differences, something doesn’t seem right.
the Z height does matter, the maslow.yaml you are using says:
Maslow_tlZ: 100
Maslow_trZ: 56
Maslow_blZ: 34
Maslow_brZ: 78
dewaine wrote:
Having just finished calibration on the spoil board and then doing a litlle
reading here, I am curious if the 3/4" height difference will cause issues.
Should I re-calibrate? Maybe shim the belts with 3/4" blocks? The latter
maybe is better if I expect to cut pieces of different thicknesses?
It could make a difference, but I don’t think much (the belts will be a little
tighter), if this is an issue, or you have a much thicker workpiece you plan to
use, you can change the Maslow_??Z values to account for the extra height.
David Lang
Yes that would definitely explain things! I have no idea how that happened. But I will update the numbers and try again. Well if you’re a newbie, you’re an astute one alright.
So changing the frame numbers, indeed released more belt. Whew! And having finished another calibration, it did result in a slightly higher fitness score of .596 (previously .56). Definitely a step in the right direction. I did notice afterwards that one of my belts had the teeth facing the opposite way to the other 3 (I should’ve straightened out the single twist before anchoring) and also that the belts were rubbing against the stepper motor cables slightly at some points during the calibration (I could have brought the belts to clear the those cables better). Would any of this factor into effecting the calibration process/ fitness score? Should another calibration not be done from this point on? I’m curious as to what could be causing a fitness score on the lower end of the scale. The frame seems to be adequate and the frame numbers are jiving.
maslow(1).yaml (5.3 KB)
You mean the cable was just twisted? It didn’t go onto the spool like that? If it was just twisted in the open air, thats not a problem.
Not sure why you think your calibration number is low, I haven’t seen a graph of what others are getting, but .5 and up seems really good. Mine calibrated at .47, and I took that as good enough, but will try with concrete anchors later. I would say start cutting and see if it is good enough to you.
Congrats!
No, the cable was just in an open air twist. Okay, so that’s good.
Just from having read some threads where the cutoff fitness level was something like .45 or so or and seeing how some were reaching over 1.0, I formed the impression that a higher fitness score translated to greater accuracy. But yes, I haven’t seen any real benchmarks to compare to. Yeah, I think you’re probably right - time to try some test cuts. Thanks!
You are definitely right, the higher the better.
@dlang told me
fitness = 1/average error
where average error is looking at each point and measuring the difference
between the belt lengths that it thinks it should have at that point with the
calculated frame size and the belt length measurements that it got at that
point.
So, .5 gets you 2mm accuracy. Best thing to do is make some cuts and see if that is good enough for your projects. If you are cutting out large shapes chances are it is good. High tolerance fits might need more work.
Also, I am hopeful that in a few months time the software will be even better at calibration and will result in better accuracy without me having to do anything but just recalibrate (I’m kind of lazy that way).
Hope that helps
That’s definitely helpful. David had confirmed that a higher number is better, but seeing his explanation sure put things in greater perspective now. So this should be a good tolerance to start with and when I get to a point where I might need finer work (perhaps certain fonts types for a nameplate) then maybe fitness can be finetuned. Or like you said, it might just get taken care of with future updates. That’s CEO level thinking.
