Belts are tight before starting calibration.
1 Like
At least 20 minutes. Once it works you donât need to do it again unless you change the frame.
The endless Fitness Too Low, Restarting Fitness should run for 20 mins? I let it go for 30 mins over the weekend and it never changed.
To be clear - I understood that if you get a Fitness To Low, Restarting Fitness error, that your calibration has failed. The machine keeps pumping this message out over and over.
Are you telling me that these Restarting Fitness errors are not a problem and that I should ignore them and let the system keep running for another 20 mins and eventually they will resolve?
one other thing I note from the log, it looks like you are significantly out of
square with the top anchors being to the left around 600mm
this could be that your bottom right anchor is down from where it should be,
rotating the baseline
David Lang
Just measured - every corner of my frame is equidistant from every corner of my 4x8 spoil board except my bottom right anchor, which is 3/4" (19mm) too close. That doesnât seem significant - itâs certainly not 600mm out of square.
Measure the diagonals, should be close to equal. If they are your setup is square.
I never tried running the config with a calibration grid of 300300 Grid size. smallest I tried was 1500800 Grid size 7.
Initially took around an hour before it finished.
Yes, it will keep trying over and over and hopefully eventually find a solution, although usually it wonât take more than at most 10 or 15 tries
Maslow
Thanks everyone! Itâs great to know that the community is so helpful and communicative. And, thanks, Bar. I know what comes with running a one-man company!
So, I got it to calibrate! And have a few new questions.
I set the grid area to 30.5â x 48â (out of my total 88 x 121â frame area). It wouldnât let me do a 3x3 or 5x5, and it went into a fitness loop with 9x9. But 7x7 worked.
The belts unwound a couple times at the start of the calibration, but didnât get tangled and wound themselves back up. I didnât apply force to simulate it being on a vertical / wall-style frame.
The retraction issue with the one belt went away. Perhaps and friction in that armâs belt guide got rubbed out.
I donât know what belts went to which anchor last attempt, so perhaps I unknowingly did a 90 degree reorientation.
Another change from my first attempts was getting the robot off the foam (like what Bar uses in his home garage) and directly onto the ground. I assumed being on the foam would be a more accurate representation of the jobs, but perhaps not??
Some clarifying questions:
What is the current consensus / best forum thread for determining the ideal concrete floor frame size? Using Barâs Github work area visualizer I was bummed to see my best area is not that big. What is the maximum size frame possible?
Is there a thread on best feeds / speeds / bits for substrates? A starting reference chart would be amazing, but Iâm guessing thatâs also on the to-do list.
Added here are some photos of my setup and config settings used.
1 Like
So even though itâs saying DONâT USE THESE CALIBRATION NUMBERS and Restarting Fitness, I just just let it keep running like that until something different happens?
Also - congrats @caddad - nice to know this works for someone.
caddad wrote:
Another change from my first attempts was getting the robot off the foam (like
what Bar uses in his home garage) and directly onto the ground. I assumed
being on the foam would be a more accurate representation of the jobs, but
perhaps not??
you do want to calibrate at the height that you will be working (and make sure
your z offsets match)
What is the current consensus / best forum thread for determining the ideal
concrete floor frame size?
in general, the bigger the better. you have 14.5 ft belts (but you donât want to
unspool them all the way) you need to be able to reach from the anchor to the
far corner of your workpiece.
Using Bar¢s Github work area visualizer I was bummed to see my best area is
not that big. What is the maximum size frame possible?
you can go almost to 14âx16â before you run out of belt with a 4x8 sheet of
plywood.
my calculator shows the belt length required so you can play around with it.
David Lang
Please remind me does the calibration calibrations run on Maslow PCB or on our computer?
If on computer the age /performance of the computer may take longer to calculate ?
Dano
1 Like
I note you have orientation set to vertical but you a running on the floor, should be horizontal
1 Like
Calculations are on the computer and speed of computer makes a huge difference.
1 Like
Maslow-serial-8.log (72.5 KB)
20+ mins of restarting fitness⌠no change
1 Like
boslaw wrote:
So even though itâs saying DONâT USE THESE CALIBRATION NUMBERS and Restarting Fitness, I just just let it keep running like that until something different happens?
usually, each loop through give a better estimate of the frame, so the fitness
improves. Itâs not clear why this isnât happening with you.
usually a small calibration size and small number of points gives a high fitness
estimate (it doesnât know enough to know how wrong itâs calculations are)
If you can try to measure the distances between the anchors (including diagonal
distances) we can plug them into manual calculations for the anchor positions
and you can enter them in directly instead of continuing to fight the automated
calibration (which is just trying to compute the anchor locations)
Once you get a set of values that work, save a copy of that maslow.yaml file and
you can always get back to that.
David Lang
2 Likes
Dano wrote:
Please remind me does the calibration calibrations run on Maslow PCB or on our computer?
If on computer the age /performance of the computer may take longer to calculate ?
the caclulations are done in your browser, so the computer/browser performance
can make a big difference in how long it takes to compute
David Lang
1 Like
That message has been in the firmware since before it could automatically try again if it failed to find a solution. You are absolutely right that it needs to be updated.
Does anyone have a suggestion for what it should say to communicate that it failed to find a solution, but it will automatically try again?