Version 1.15 Testing and Feedback

Troubleshooting
1

I ran a calibration with V 1.15 today and remembered to remove any bit and reduce the Z all the way down. I used a 9x9 grid. I set my belt extension to 2000, but left the other config items to default.

Calibration values upon successful calibration:

Calibration values:
Fitness: 2.952430440738816
Maslow_tlX: -26.5
Maslow_tlY: 2292.3
Maslow_trX: 3597.3
Maslow_trY: 2305.6
Maslow_blX: 0.0
Maslow_blY: 0.0
Maslow_brX: 3598.8
Maslow_brY: 0.0

I noticed the belts getting very tight as the sled moved near the anchor points, but didn’t witness any grinding.

Attached is a blueprint of my table (I can’t promise that the anchors are all in plane). Diagonal measurements show this plan is within 15mm accuracy, but as the results show, it isn’t a perfect rectangle.

Screenshot 2025-12-05 154737
Screenshot 2025-12-05 1547371144×824 7.67 KB

.Questions:
-) What does the Maslow_tlX: -26.5 mean from the calibration results?
-) To get the aspect more accurate do I update the Yaml values to match my measurements, or should I move the anchor points to obtain a square rectangle?
-) Recommend any other adjustment?

2

Excellent questions!

This means that the top left anchor point is 26.5mm to the left of the bottom left anchor point.

The bottom left anchor point is (0,0) and all of the other anchor points have their coordinates relative to that point.

This shouldn’t matter. The internal math doesn’t depend on the anchor points being in a square shape.

The real question is why are the anchor point locations that the machine is finding different from the ones you are measuring (for example 3,623.8mm vs 3584.5mm along the top)?

I just ordered a super precise tape measure and I’m working on trying to figure out exactly what is going on there.

1 Like
3

I just added a laser to my kit. (Black Friday) Should the measurement be to the exact centers of the anchor posts? If so, I’ll CAD that in and repost. I used a tape and loyal helper Gina to measure the original points posted.

4

Yup! That’s the spot.

From what I was reading the laser tape measures are actually lower precision than the steel ones, but I might be wrong about that

1 Like
5

I’ve measured using a Bosch (I do not work for them) laser with a claimed accuracy of 1/16" over 100’. I think I have it down to < 5mm accuracy checking a diagonal.

When I CAD out the rectangle, do I assume the bottom X is true to the coordinate system? It seems that your math(s) align to this since you explained a Y anchor delta. Thanks

6

BTW, I agree that the human eye and a tape can go way past 1/16" accuracy.

7

Screenshot 2025-12-08 114501Y axis
Screenshot 2025-12-08 114501Y axis769×540 4.74 KB

Measured anchor points aligned to X Axis showing Y axis drift. A regular rectangle is connecting (0,0) with the upper right anchor point.

8

Yes, that is right

9

Do you by any chance have a copy of the log from when this was computed?

I would like to take a look at the measurements that the machine took and then step through the calculation process and see what we can learn

10

Maslow-serial.log (48.2 KB)

Attached

1 Like
11

I’m playing around with this data and it is super helpful! Thank you for sharing that

12

I’m trying to work back from this to figure out what the actual anchor point locations would be from this to compare to the calculated ones.

Do you by any chance have those diagonal measurements? That would be the easiest way of computing everything I think

13

Scott wrote:

Maslow-serial.log (48.2 KB)

so, the stock calibration came up with:

Calibration values:
Fitness: 2.952430440738816

running this through the alternate calibration shows some points having bad
measurements and the results off significantly

Maslow_tlX: -26.5 vs 18.0
Maslow_tlY: 2292.3 vs 2292.0
Maslow_trX: 3597.3 vs 3617.0
Maslow_trY: 2305.6 vs 2305.4
Maslow_blX: 0.0
Maslow_blY: 0.0
Maslow_brX: 3598.8 vs 3624.0
Maslow_brY: 0.0

is there any way for you to measure the 6 distances between the anchors so we
can put them through a manual calculator and see which set of numbers is more
plausible?

David Lang

Adaptive Init:
x1: 2140.3 mm, x2: 3215.6 mm, y2: 2139.8 mm, x3: 1070.1 mm, y3: 2139.8 mm

Outliers at indices: [1, 8, 16]
Threshold: 1.230 mm

Dismissed Outlier Points:
Index 1: {bl: 2258.99, br: 2027.61, tr: 2017.10, tl: 2266.84} (BR error: 1.25
mm, BL error: 1.06 mm)
Index 8: {bl: 1907.13, br: 2307.15, tr: 2366.82, tl: 1994.52} (BR error: 0.16
mm, BL error: 1.42 mm)
Index 16: {bl: 2070.43, br: 2080.34, tr: 2205.32, tl: 2212.91} (BR error: 0.33
mm, BL error: 1.46 mm)

Full Fit:
x1 (B): 3624.0 ± 0.5 mm
x2 (C): 3617.0 ± 0.5 mm, y2 (C): 2305.4 ± 0.7 mm
x3 (E): 18.0 ± 0.2 mm, y3 (E): 2292.0 ± 0.7 mm

Residual sum: 306944.531 mm²

Cleaned Fit (no outliers):
x1 (B): 3624.0 ± 0.4 mm
x2 (C): 3617.0 ± 0.5 mm, y2 (C): 2305.4 ± 0.7 mm
x3 (E): 18.0 ± 0.2 mm, y3 (E): 2292.0 ± 0.7 mm

Residual sum: 283903.450 mm²

14

(0,0) to upper right corner is 4363+10 considering the anchors are 10mm or 3/8 in diameter.

Top left to lower right is 4283+10.

As mentioned, I have not lasered the plane of the table to see if the frame has warped. I suspect there may be a dip in the top right anchor, but the measurements are line of sight.

I’m thinking I should print some bumpers to fit the anchor points to ensure the laser is reading center to center. I can do that if needed

1 Like
15

Thank you!

16

hmm, those diagonals with the drawing earlier show a pretty bad match

image
image1920×712 106 KB

it shows the 2nd diagonal off by 77mm compared to what it calculates from the other 5 measurements.

1 Like
17

Scott wrote:

I’m thinking I should print some bumpers to fit the anchor points to ensure
the laser is reading center to center. I can do that if needed

If you are having trouble measuring center to center, just put hex bolts in the
anchor holes and turn them so the flats are perpendicular to your measurement
and measure left edge to left edge (hook the tape measure on the outside of one
bolt and measure to the near side of the other bolt)

David Lang

3 Likes
18

Thanks. I think what I have is reasonably close. But will run another set of measurements tomorrow. Right now it is like catching a fish by hand to get a reading. Either I see the laser or the display

1 Like
19

This is an excellent idea. I’m going to copy that

If you have a chance to I would really appreciate it. Your calibration measurement data is exceptionally clean so it is really really good for simulating what is going on and trying to match up the generated measurements with real world measurements

1 Like
20

I’ll get back to you with validated data. I’ll measure in both directions to confirm.