I’ve been trying to get the calibration working, but it has never worked for me.
The only time calibration worked was when using a square frame using version 0.88. (newer versions would also fail on the square frame).
Now I’m using anchor points on a floor (rectangular, about 3m x 2m). Also tried using anchor points on another, bigger floor as well.
I’ve updated to v1.11.
My anchor points bottoms are 15mm from the floor, and wasteboard is 15mm thick, so I did not adjust the Z values in the config file. TRZ, TLZ, BRZ, BLZ are default values, and motors are at the correct position.
Maslow z is all the way down.
Also tried with different values for the calibration grid (different grid sizes, width, height).
This is a log file for when I let the calibration run for 1 hour. Maslow-serial (10).log (176.1 KB)
What is strange that it assumes the frame is square by saying:
[MSG:INFO: Frame size successfully adjusted to: 3043.217 x 3043.217]
But when looking in the code, it can only say it’s a square frame. There is no code for rectangular frames (if it is needed for the calculations, I have no idea)
when it first starts, it makes a guess on the frame size, assuming that it’s
square, and as it’s running the calibration calculations, it moves where it
thinks the anchors are to be a better fit for the measurements that it’s taken.
what is the grid size and x/y dimensions of your calibration area? if you try to
calibrate in an area that’s too big for your frame (getting ‘outside the green’
in the frame calculators) or the belt isn’t tight or catches on something during
some of the measurements, you get bad readings for that point, and that messes
up the calibration logic.
Oh okay, thanks, that makes sense about the square frame size calculation.
Grid size was 9x9 and x/y dimensions 700x700. I’ve also tried lower grid sizes (3x3) and 900x900, 300x300, …
With a 3m x 2m frame, the green limit would be approx. 1400x900
It’s hard to work out what is going on without more information.
Where the belts terminate on the post anchors, are they free to rotate?
Can you measure the distances between anchors manually and post here
BR → TR, TR → TL, BL → BR, BR → TR and the diagonals BR → TL, TR → BL.
A copy of your maslow.yaml file and the results you get when you press Test from the setup menu.
Are you seeing the machine take a grid of points like we would expect it to?
I pasted the measurements from the log file into the calibrations simulator and two things stood out to me:
The points aren’t really in a grid at all which isn’t normal
and
The simulator was able to pretty quickly solve for values for the anchor points from that set of measurements so I’m not sure why it didn’t give an answer
The belts are freely rotating on the post anchors (washers between belt termination points).
Here are the measurements:
BR → TR = 3818mm
TR → TL = 2023mm
BL → BR = 2000mm
BL → TL = 3820mm
DIAGONALS:
BR → TL = 4312mm
TR → BL = 4320mm
My frame is not perfect, does this have a big impact on the calculations?
This is the maslow.yaml file (named maslow.yaml on the maslow itself): maslow_09_09_2025.yaml (6.7 KB)
When pressing test, all test passed.
[MSG:INFO: Caution: Unlocked]
Test
Index.html Version: 1.10
[MSG:INFO: Firmware Version: 1.10]
[MSG:INFO: I2C Timeout: ]
[MSG:INFO: 10]
[MSG:INFO: All tests passed on Top Left]
[MSG:INFO: All tests passed on Top Right]
[MSG:INFO: All tests passed on Bottom Left]
[MSG:INFO: All tests passed on Bottom Right]
It does start with measuring a couple of waypoints. (waypoint 0 to 5)
I made a video about when it’s measuring the waypoints. Then it stops and tries to calculate the anchorpoints. (The small black dot on the wasteboard is the middle)
It doesn’t look like the points in the calibration simulation match the points the machine is doing.
If I put in the same calibration data (from the last log file posted here), and press on Parse Calibration Data, it shows something different, When pressing Converge, it is not able to find a good fitness. (I could be using the tool wrong)
I’ve added the numbers to the maslow.yaml and restarted the maslow.
Then did a retract, extend, apply tension cycle. But it failed after applying tension, so it was not able to jog.
I’ve tried retracting with less than 1300, and one belt would not retract.
I’ve opened up everything, and added some grease to the spools. (already opened it up before and added grease).
Now it can retract with a retraction current of 800.
Then I started the calibration, and it worked!
I then changed the grid size to 9x9 and 1000x1000, and calibration succeeded as well.
I have no idea if the solution was by adding grease to the spools, or by starting calibration with the altered maslow file.
Anyway, it is working now. I will do a test cut to see how precise it is.
One other thing that I noticed in the video is that you have things set up 90 degrees from the usually setup where the power cord is towards the long side of the wood.
I will recalibrate it with the power cord at the correct side. It will be easier for the dust collection as well. Thanks for noticing!