Y axis steps/mm seem not making difference when changing the value

[murdock62]

Hello,

I have a version with TMC2208. Firmawre version 17.

When I change the steps per mm on the y axis and increase this value the displacement of this axis is not more important. On the X axis it works normally.

Many thanks for your help.

[murdock62]

New informations,

I think I have identified a small problem with the firmware. (Tests carried out on firmware versions 1.0.17 and 1.0.10 with the same results)

Wishing to precisely calibrate my D3 (TMC2208), I wanted to carry out a fine calibration of the axes.
Neither one nor two:

• I fix my fixing device for my caliper on the Y axis.
• I'm making a home.
• I send via the console a G91 to switch to relative coordinates.
• I just started to push my caliper to "make contact".
• I put my caliper to 0.
• I send a G1 Y70 F1800.
• Verdict 69.81.

A quick little calculation: ((70 / 69.81) * 80) = 80.217. I prefer unlike the rule to round down and between 80.21 in steps per mm of the y axis.

I resume my tests:

• I'm making a home.
• I send via the console a G91 to switch to relative coordinates.
• I just started to push my caliper to "make contact".
• I put my caliper to 0.
• I send a G1 Y70 F1800.
• Verdict 69.81!? !!

Recalculate: ((70 / 69.81) * 80.21) = 80.428. I enter 80.42 in the steps per mm of the y axis.

I resume my tests:

• I just started to push my caliper to "make contact".
• I put my caliper to 0.
• I send a G1 Y70 F1800.
• Verdict 70.18!? !!

New calculation: ((70 / 70.18) * 80.42) = 80.213. I enter 80.21 in the steps per mm of the y axis.

I resume my tests:

• I just started to push my caliper to "make contact".
• I put my caliper to 0.
• I send a G1 Y70 F1800.
• Verdict 70.00 ??

Well let's admit ... I redo a Home I redo my verification:

• I send via the console a G91 to switch to relative coordinates.
• I just started to push my caliper to "make contact".
• I put my caliper to 0.
• I send a G1 Y70 F1800.
• Verdict 69.81! ??????????????????????? !!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!

And there I understand !!!

When the values ​​are not applied there has always been a de facto homicide.
Neither one nor two I re-enter my value (80.21) but this time directly by gcode (M92 Y80.21)
The machine responds directly to "settings stored".

Without doing home I resume my tests:

• I just started to push my caliper to "make contact".
• I put my caliper to 0.
• I send a G1 Y70 F1800.
• Verdict 70.00

My conclusion (to be compared with other tests, on other machines and firmware versions to exclude certain experimental bias:
When home, the machine does not retrieve the value of steps per mm in the eeprom but directly in the firmware, thus resuming the default value. (80)

Am I the only one to get this result?

[murdock62]

Good morning all,

The problem is solved.

I started from the sources of version 1.0.16 (only version available on TENLOG's GITHUB)
In the file "Marlin_Main.cpp" at line 2689 you have this:

int Y_step_per_unit = axis_steps_per_unit[Y_AXIS];

this is used to save the current value of the steps per mm of the Y axis in a variable then what during the "home" phase
this axis is temporarily assigned to the independent management of the second motor of the Z axis.

Only problem this variable is of type integer and the steps per mm can be decimal.
So if to have a perfect displacement I had put 80.24 steps per mm for the Y axis as long as I do not "home"
no worries only here if I do "home" sela will enter my 80.24 in my integer variable (which will bring it back to 80)
and so I come back to a bad calibration of my Y axis ...

To correct the problem, just replace this line with this one:

float Y_step_per_unit = axis_steps_per_unit[Y_AXIS];

It's not much, but it can really be a problem!

It would be a good idea for the engineers at Tenlog to fix this in version 1.0.17. Thanks to them !

Good prints!