Accuracy solution

While setting up my maslow, I noticed there was an accuracy error of about 4mm.


It seemed to me that the error was linear in every quadrant of the square.

To compensate for the error I made a program that adjusts all the X-Y coordinates of the NC file into a new NC file.

When I ran the new NC file, the offset error was close to 1mm. (and I think this could get smaller if my measurements were better)

I did only 1 cut with this, but for now it seems to work fine.

If anyone is interested, you can find the software here: https://github.com/tdesign81/Maslow-Post-Processor

7 Likes

Nice work!!

There is a version of this built into the machine too

These settings:

Will let you automatically scale any input gcode linearly automatically.

2 Likes

I know about the scale settings.

But the linear error seems to be different for every quadrant.
Also there is an error on the X axis when just moving on the Y axis.
For example:
If you draw a line on the X axis with y =0
and you draw a line on the x axis with y=500
you will see that the lines will drift apart (y).
So you have to compensate y for x movements too.

I know the adjustments are better done in the maslow but i didnt want to mess around in the firmware.

2 Likes

Very cool. What are you using to mark the paper with the M4?

A marker diameter 8. It’s not the best, I have to make something better for it.

I have a few posts on this, but I think this one gets to the heart of the accuracy issue and why it’s not linear: X & Y Accuracy Analysis: Belt Stretch

I’ve had to take a break from this research, but next up I want to work with @bar to create a setting in the software that can compensate for belt stretch and keep testing. Hopefully I can get back to it soon!

3 Likes

Hi all,

Question about the x and y scaling option:
I ran my first cut yesterday. Cut out a rectangle of 2000mmx38.6mm
Y isn’t off, but the x measured 1982mm after cutting. What value do I enter into scale x window to compensate?
Do I enter the scaled value 1982/2000x100=99,1 (0.991) or 2000/1982x100=100,908 (1.009)

My 3D printer (klipper/fluidd) asks for the first value to account for filament shrinkage and automatically compensates so the dimensions are accurate again.

Thanks!!

2 Likes

I think you want the 1.009 value a you are trying to increase size. But I’m not 100.009% certain.

Dano

1 Like

:rofl: Thanks Dano!
I’m just going to try that and see what happens!!
:oncoming_fist:t3:

1 Like

I’m sorry for the slow reply!

The input gcode will be scaled by that value so since your part is coming out smaller than we want it to I think that 1.009 is in fact correct.

X axis Scale correction 1.009 worked out fine!
Designed a quick test stool in Fusion and let Maslow work its magic!
I set up the slots and pockets with a 0.1mm tolerance. For the seat pockets that turned out a bit tight. Had to do some manual “post processing” to make ik fit. But for a first project I am more than happy with the results!


3 Likes

Excellent work! I’m loving following along with your exploring of how we can improve the system accuracy, but it’s also good to see that you are able to use the machine as is too :grin:

1 Like