Testing needed on Pending patches

Software Maslow Firmware
1

I have worked with the github AI to create a bunch of improvements, there are a few that need testing/review that I have not been able to do.

code review and suggestions are always helpful, and I don’t have a machine setup at at home that I can test that the hardware does what it’s supposed to, I can only check that it loads, doesn’t blow up, and the UI up to the point it needs to actually move something.

So if anyone is able to look at the patches, or test the results and report, it would be very useful.

4 Likes
2

PRs needing testing this week:

Title: Implement bounding box

Title: Disable connection monitoring while uploading files

Title add a button to move the router to a bit change height

Title: save belt lengths when idle (same rules as Z axis)

Title: add option to relax belts before extending

Title: automate grid size calculations “stay in the green”

Title: Automatically compute calibration grid size based on anchor points

1 Like
3

Is maslow supposed to move around the outline prior to starting cut? That did not happen, went straight into cutting mode for both tests.

Sorry missed the button for trace boundary, will try again

Displays offset for circle and job bounds look incorrect.

Circle.wmv (202.9 KB) 270mm

Job Bounds:
Size: 270.1 × 255.6 mm
Z: -5.0 to 4.0 mm (9.0mm range)

Koala.wmv (259.9 KB) 270 x 200mm

Job Bounds:
Size: 282.0 × 316.1 mm
Z: -2.0 to 2.0 mm (4.0mm range)

4

the overhead shots look correct, for the shots at an angle, it looks correct, but the blue outline is at a different height than the black cuts, so it looks offset.

could you double check the bounds?

and when you trace the boundary, does the maslow move around the box?

5

Circle partially outside bounding box.

With Koala stayed within boundaries

6

Ian Abbott wrote:

Circle partially outside bounding box.

With Koala stayed within boundaries

can you get an overhead shot of the circle showing it going outside the bounding
box, then post it, the gcode, and the boundaries so I can scold the AI and get
it to fix it.

and did the button to move seem to match the box?

David Lang

7

This looks like the wrong download links.

Should it carry over a power reset?

8

Ian Abbott wrote:

This looks like the wrong download links.

yep, try this one (the last build response from copilot includes a link to it)

Should it carry over a power reset?

It’s supposed to. any time that it would have saved the Z position before, it is
now saving the belt lengths.

I suspect it’s doing it too much, so we’ll have to change it to only save when
it’s in a state where the belts are tight. but until we test…

David Lang

9

I moved the Z so pencil was touching, then clicked the boundary box button, Maslow moved down from the home position (the diagonal line), then drew the box below and to the right. I raised the Z axis and ran the circle job. I have highlighted the intersections between “boundary box” and circle.

File:

circle.nc (1011 Bytes)

20251004_160940
20251004_1609401920×864 93.9 KB

Job Bounds:
Size: 270.1 × 255.6 mm
Z: -5.0 to 4.0 mm (9.0mm range)

10

Title: save belt lengths when idle (same rules as Z axis)

Well maybe it is but it still displays NaN for X & Y, so it’s not reloading. Can it be put on a button to save position and then a restore button on power up to recover?

11

I’d like to try a bit longer to have it be completly automatic.

give this one a try

David Lang

12

Ian Abbott wrote:

I moved the Z so pencil was touching, then clicked the boundary box button,
Maslow moved down from the home position (the diagonal line), then drew the
box below and to the right. I raised the Z axis and ran the circle job. I have
highlighted the intersections between “boundary box” and circle.

it wasn’t properly handling arcs, give this one a try

Thanks for doing this testing.

David Lang

13

I loaded both requested updates at the same time (feeling lucky). Circle fixed details correct.

Not so good on this:
Save belt position worked, but as I had done a complete shutdown last night, retracting belts etc I did the Retract, Extend, Apply Tension, which worked but it used the values it had previously saved for the starting position i.e. recorded last night which meant everything was off by the difference between where I had parked it last night and where it started from today.

I couldn’t reset the Z home position from the button on the Maslow menu (had to go to menu left of the FluidNC and click Zw)

After testing the circle I powered off the Maslow and then back on again. Maslow was locked, so unlocked and moved Z OK. Could not move maslow, get pop-up

“Maslow local says
Maslow is not ready to move [OK]”

So problems
1: Need to reset values when we do the REAT Dance
2: Unlock movement after power up
3: Enable home Z button

Thank you for all your work on these problems David

14

Ian Abbott wrote:

I loaded both requested updates at the same time (feeling lucky). Circle fixed details correct.

fantastic.

Not so good on this:

Save belt position worked, but as I had done a complete shutdown last night,
retracting belts etc I did the Retract, Extend, Apply Tension, which worked
but it used the values it had previously saved for the starting position i.e.
recorded last night which meant everything was off by the difference between
where I had parked it last night and where it started from today.

I couldn’t reset the Z home position from the button on the Maslow menu (had
to go to menu left of the FluidNC and click Zw)

After testing the circle I powered off the Maslow and then back on again.
Maslow was locked, so unlocked and moved Z OK. Could not move maslow, get
pop-up

I’m doing a bunch of changes here:

only save data to nvm when in state READY_TO_CUT or RETRACTED (these are the
only states when we know the belts are tight and the belt lengths valid)

when the data is loaded from nvm, if the belt lengths are zero, change the state
to RETRACTED otherwise change the state to READY_TO_CUT and disable the alarm

add a way to mark the nvm data as stale/invalid (possibly set to -1??)

do not load belt lengths from nvm if it is stale/invalid

when moving out of the states READY_TO_CUT or RETRACTED, check if the belt
lengths are stale/invalid, if not, set them to stale/invalid

add log_debug statements to report when the belt lengths are saved, restored, or
marked stale/invalid and the state at the time (one log message per event with
all the data in the one line)

let’s see if this addresses the issues you are having.

David Lang

1 Like
15

new version of the PR to save belt lengths (avoid retract/extend across power
cycles)

David Lang

16

Retains X Y

Maslow-serial (79).log (1.9 KB)

maslow
maslow1152×648 40.6 KB

values across power cycles, but still unable to move

17

new version of firmware to auto-calculate the grid size

it was doing it’s calculations before measuring the first 6 points, so it was based on what was in the file, not what was measured.

new firmware to try at Clean up trailing whitespace from PR #349 and fix calibration grid timing · BarbourSmith/FluidNC@18fab56 · GitHub

18

hmm, I’m not seeing the logging that I’m expecting.

can you go to the fluidnc parameters (the left dot) and make sure the logging is set to debug instead of info? I told it to log the stuff as debug to reduce the noise. let’s see if that logs anything more

19

this version of the belt-length patch adds some logging to say if it loaded the lengths or not, and why not

image
image1128×727 39.5 KB

this messagelevel is where you need to change ‘info’ to ‘debug’ and save it

20

Screenshot 2025-10-05 120437
Screenshot 2025-10-05 1204371013×543 21.4 KB

Maslow-serial (80).log (2.6 KB)