[sm3]

Hello CNCZoners!
I have build CNC machine to elements of mosaic. I am tuning it now, however I bump into problem I am unable to solve so far.
pieces I am cutting on it on Y axis are consistently about 0.2mm larger than X side.
I am now, while tuning the machine? cutting only a square 10mm side. so, X side is fine - 9.97-10.03 mm, but Y side is about 10.25-10.27.
I was trying to make some changes in stepconf of linuxcnc, but no luck so far.
on the photo is my X axis configuration. Y axis is exactly the same. I was trying to upload second image but for some reason it won't upload.

Any suggestions/recommendations, please?

[routalot]

It appears that your steppers are directly connected to the leadscrews,which eliminates the possibility of belt stretch.What your post doesn't say is whether you were climb cutting or conventional cutting and an easy test is to try the other and measure the test piece. Any difference other than 0.26mm would point to a flexing problem as nothing else will have changed.

[sm3]

you're right - I did not mention it. the reason being is because there is nothing unusual here -table is horizontal. My theory only is that - this machine has been built in my garage. and of course (because my budget for it was extremely low ) 99% of parts - are not of the best quality.
but this consistency tells me that it should be correctable. My theory is that ball screw driving Y axis is not the same as for X axis. Even though both were purchased together. So, I've been trying to make adjustments in stepconf using that screen (attached) - but I like I said - no luck. or maybe my understanding of these settings are wrong?

[routalot]

Did you try cutting in the opposite direction?

[sm3]

no, I haven't. to make it all run in opposite direction I will need to change everything - model file I have with all its lines in it, so, the program I have, which reads this model and generates G-code - would generate G-code for these changes. and even for this simple case I can write G-code manually, I still will need to change entire model to obtain new coordinates values. Why do you think it might matter? Please, explain it to me. What I am trying to do now is to change pitch value for ballscrew for Y axis, leaving this number for X axis unchanged. Any other ideas, please?

[routalot]

I have a good understanding of the process of creating a toolpath and would be interested in learning which software you are using for this part of the process.I am reasonably certain that your steppers will have 200 steps per rotation and that your leadscrews will have a pitch of exactly 5.00 mm.What I believe you have measured is the result of either the gantry flexing or spindle mounts flexing in response to the forces generated by the cutting action.It could even be some of both going on.If this is the case any other values in the stepconf table will give you a different error at any other y position as the flexing won't be proportional to the travel of the machine.By cutting in the opposite direction the forces will be changed and the dimension of the part can be compared to that which was cut in the opposite direction.A picture of the machine's gantry and spindle mount would be helpful.If you would like to verify the rigidity of the tool you can place a dial gauge against the tool and push to see how much movement there is when a moderate force is applied.

[machinehop5]

...I think what you're looking for is in the ini file
https://www.forum.linuxcnc.org/38-ge...comp-confusion

INI Configuration

happy hunting,
DJ

[sm3]

I am using program I have written myself to generate G-code. It is pretty specific because of my specific needs.
My machine is not usual one - it cuts out element of any shape out of hard/brittle materials like glass, stone etc. - anything softer than diamond. it has been designed for flat application - another words it cuts out element of mosaic. I see your point. It could be what you are saying and I am observing something like this. I'll explain. I am using diamond wire for cutting. And even though it is pretty significantly tensed - it is still sometimes bent. If that happens the result will be see on the part, as with gantry keeps moving, the wire will tend to get into straight position and so, the cut won't be correct. I doubt though that this opposite force can be larger than force generated by motor and so, transferred to gantry. Plus, like I said, I see this difference consistently only for Y axis. That's why I thought - maybe it is this particular part fault - they are not after all that precise. everything is made in China.

[sm3]

Thank you DJ - that sounds promising and probably is exactly what I need. Lemme try it.

[sm3]

Hello Everyone!
just finished cutting yet another piece and it came out much better.
So, I was cutting a square with side 10mm. out of machine it came out 10.1 x 10.07.
it is not perfect, but main thing - I was able to adjust just one axis - Y, as before it was stubbornly and consistently cutting 10.25 - 10.27.
Now with this numbers changed to 10.07 - I think, I have achieved what I needed.
So, I changed in stepconfig number for leadscrew pitch from 5.00 to 5.05. This in turn decreased number of generated steps for motor.

@DJ - I haven't tried making changes to ini file. Reading documentation it is not clear to me will that BACKLASH number be applied to each axis equally or it can be specified for each particular axis? if it is one-for-all - it is not going to work in my case.


Thank you Everyone, who offered his advises and help!

[machinehop5]

@DJ - I haven't tried making changes to ini file. Reading documentation it is not clear to me will that BACKLASH number be applied to each axis equally or it can be specified for each particular axis? if it is one-for-all - it is not going to work in my case.

...yes, can be specified each axis. Open your ini file.... linuxcnc/configs/yourmill/yourmill.ini then scroll down to AXIS 1 (if that is your Y axis) look for BACKLASH = .xxxx ..if its not on the List ..add it after MAX ACCEL line and use one space before and after the = click Save file. The numbers will most likely be negative for BLC using a minus sign.

enjoy,
DJ

[sm3]

Do you think, DJ, it would be better than adjusting ballscrew pitch number?
I am planning to check it again tomorrow and cut yet another part from the same file, to see how stable this change. I expect it to be same and I could not achieve anything close to it modifying any other values in stepconf.
Perhaps, I am wrong here, but 1. this pitch adjustment works. and it make sense to me, because I can explain it - not well made, cheap ballscrews (made in China), so, its precision is far from ideal and it would be reasonable to expect them to perform differently. and difference is not that dramatic - it's just for me makes difference.
BACKLASH is a bit different as I understand it - it is kind of reverse force applied to axis. it is inertia force. it could play significant role when you move your axis fast enough, which is not my case. movements on my machine are very, very slow - because of material I am dealing with and because of cutter type I am using for it. I am using feed rate - 0.09. I wish I could go faster, but experiments told me - I can't.
Do you think I am wrong in what I have said?

[machinehop5]

Do you think, DJ, it would be better than adjusting ballscrew pitch number?

If, its Backlash adjust Backlash...if it's Pitch adjust the Pitch....

Perhaps, I am wrong here, but 1. this pitch adjustment works. and it make sense to me, because I can explain it - not well made, cheap ballscrews (made in China), so, its precision is far from ideal and it would be reasonable to expect them to perform differently. and difference is not that dramatic - it's just for me makes difference.
BACKLASH is a bit different as I understand it - it is kind of reverse force applied to axis. it is inertia force. it could play significant role when you move your axis fast enough, which is not my case. movements on my machine are very, very slow - because of material I am dealing with and because of cutter type I am using for it. I am using feed rate - 0.09. I wish I could go faster, but experiments told me - I can't.
Do you think I am wrong in what I have said?

I agree...what ever works is good enough when dealing with CNC machines.

Cheers,
DJ

[sm3]

)
BTW, I think, I could be fixed at the very beginning when I was tested each axis. However, I found it kinda difficult, because the way it is implemented in linuxcnc.
you can choose for testing to move axis either to both +/- direction or one. However, I would expect it to move in one direction up to the point and stop there, but it moves immediately back. so, how can measure how much exactly it has moved?! the best I could do is to set the speed it was moving with to very low. Still, measuring device's numbers are running on its display and it is very difficult to catch the maximum number it has reached.

[routalot]

Try cutting a different sized piece.Then measure it to see if the machine cuts all sizes correctly.

[sm3]

yep! Thank you!
but again - how do you tune axis on linuxcnc? Am I missing something? Do you have any special tool for it? technique? Or, perhaps I did not see something in linuxcnc and my complaint is not valid at all?