That’s great but there’s no documentation I can find to help understand the logs/info/errors, and regarding the variables, I’ve built my maslow to spec and used recommended frame setup (12x8 anchor bolts in my concrete floor).
I’ve looked at your logfile and put the numbers into the calibration simulator (GitHub - BarbourSmith/Calibration-Simulation). The bottom center point completely screws up the fit. That’s the first point in the array ({bl:2152.16, br:2102.60, tr:2286.20, tl:2285.05}). If I take that point out, I get fitness values of around 0.4, but the calculated anchor positions still vary wildly from run to run. Given the low number of points and all points being in the center, that’s not a surprise.
Still I wonder why that one point is so bad. That brings me to the picture of your setup. You’ve got those skinny pins in the four corners with spacers that elevate the belts above the floor. That looks very much like my first attempt which failed miserably due to bending/flexing of the anchors. Could you try leaving out the spacers so that the force from the belts goes more directly into the concrete?
PS: I just found this in your logfile:
[MSG:INFO: Measuring Frame Flex]
[MSG:INFO: Flex measurement: TLBR: 0.795 TRBL: 76.833]
So yeah, try removing the spacers.
1 Like
From the video it looks like your orientation of the Maslow is facing the wrong direction. It should have the power pointing to the long side of your board. It looks like you have it pointing in portrait mode i.e. to one end.
2 Likes
This is just a guess that the machine makes (it’s not right, but it’s hopefully good enough to start from).
This is super helpful! Can you try rotating the machine 90 degrees (so the dust collector is facing where the camera is). In theory it shouldn’t matter, but I always do mine that way.
I also agree, this looks like an issue for sure 
Thanks.
Tried again. No luck.
- re-oriented Maslow to long side (although other posts say orientation doesn’t matter)
- retract all, extend all twice
- removed spacers from my 3/8 frame bolts that are anchored into my concrete - belt ends are now at floor level.
- calibration settings are:
extend distance 2182.8mm
calibration grid 300x300
grid size 3x3
retraction force 1500
calibration force 1000
failure
Maslow-serial.log (6.8 KB)
I still don’t understand why Maslow automatically calculates my frame size as a square (per the log) when it’s actually a rectangle.
Also, it measured flex while one belt was not fully tight - it was actually quite loose and pulling the machine so the flex measurement couldn’t possibly be accurate.
1 Like
Calibration still reports this:
[MSG:INFO: Flex measurement: TLBR: 5.534 TRBL: 181.980]
With the anchors as they are now, the problem must be somewhere else. No way there is 180mm real flex. Are the belts tightly on the spools? @bar, can this be a hardware issue (encoder?)?
1 Like
Can you put a copy of your maslow,yaml file up. I suspect it may give us a clue to what is going on.
1 Like
maslow-2.yaml (332 Bytes)
ignore the incorrect file name. I had another copy in my computer folder so Mac auto-renamed it when it downloaded
1 Like
Bar wrote:
This is super helpful! Can you try rotating the machine 90 degrees (so the
dust collector is facing where the camera is). In theory it shouldn’t matter,
but I always do mine that way.
It will matter in the calibration area will be orientated wrong with
height/width swapped.
David Lang
1 Like
ultimately didn’t make any difference though
1 Like
That’s not what I was looking for, yaml should be a text file, that looks like an html file
1 Like
That’s what downloaded from my Maslow. I’ll try re-adding a clean yaml from the firmware update and re-calibrating but I’ve done this before too so I don’t have high hopes.
1 Like
boslaw wrote:
- retract all, extend all twice
sorry to be pedantic here, just trying to eliminate everything
the retract all should result in near 0 (<1mm) numbers on all belts
looking at the logs, this is the case
- removed spacers from my 3/8 frame bolts that are anchored into my concrete - belt ends are now at floor level.
what are your Z offsets? how thick is your wasteboard and workpiece that you are
calibrting on?
failure
Maslow-serial.log (6.8 KB)
I still don’t understand why Maslow automatically calculates my frame size as a square (per the log) when it’s actually a rectangle.
remember, it only knows the length of the belts, not the angles of the arms, so
when it starts up, it needs to make some guess and all it has to work with is
the length of the belts, so it starts with a square guess, and as it moves, it
will get more accurate (and in theory, the bigger the area and the more points,
the more accurate it will be)
Also, it measured flex while one belt was not fully tight - it was actually quite loose and pulling the machine so the flex measurement couldn’t possibly be accurate.
[MSG:INFO: Flex measurement: TLBR: 5.534 TRBL: 181.980]
yep, that seems like a problem. Is it pulling the belts tight at each
measurement point?
David Lang
1 Like
I think the machine’s initial settings are so far out (due to the orientation) that it would take forever to work out the correct settings. Trying again with a clean yaml should help. You may also try setting the initial size closer to your actual in the maslow.yaml file do FluidNC menu, Config Items.
this is an extract of the relevant part of mine, you will have to adjust for yours.
1 Like
Started with a clean yaml - same problem.
maslow-2.yaml (5.3 KB)
It continues to test for flex while one of my belts is not tight. Don’t know why it does that. When I retract all my belts fully retract tight against the machine, and when it tests measurement points the belts all tighten up fine.
Spoil board is 5/8" particle board and work board is 1/8" thick so I’m pretty close to the floor. Belts are all attached to the anchor bolts at floor level.
1 Like
Did the calibration finish or did you interrupt it. It can take a while after each stage to process the result (and I mean a long while).
1 Like
This is just a super rough initial guess. It just has to start somewhere.
I usually give my machine a bit of a tug so that the top two belts are tight before starting calibration. Maybe that is a factor?
It says calculating, this may take a while. It pumps some fitness numbers. Then it says fitness too low, restarting. Every Time. And that just happens endlessly. Am I supposed to let that run forever?