linear glass scale error?

[beezerlm]

I have recently purchased a cnc mill with fagor linear glass scales and I checked the x axis with a 24" standard and found .028" of error. I don't know much about the scales. Do they need routine cleaning? Can they be calibrated? could they be mounted improperly? I would appreciate any advice I can get.

[Al_The_Man]

I have never known glass scales to have that much error, the reason they use glass as there is minimum (no) expansion, it could be in the way it is being interpreted, or calibrated. Is it cumulative?
If cleaning is required, usually the end plate removes to enable the reading head to be slid out, and using optical tissue with pure alcohol, carfully run the tissue over the scale.
Al.

[beezerlm]

It does not seem cumulative. the first twelve inches it is out .008 and the second it is out .020! Should I Take it apart and clean it?

[Al_The_Man]

It does look cumulative so I would expect the calibration is out (pulses/degree of movement). Generally, if it is on a CNC machine for positioning, a dirty scale will error out the control.
Al.

[beezerlm]

The total error over 24" is .028" The first 12" is .008 The second twelve is .020. This is cumulative? I am confused. I tried adjusting the calibration but the error is greater on one side than the other.

[Al_The_Man]

If there was .020" error over the whole range then it would be due to something like backlash, but if it accumulates with distance then this smacks of calibration, the fact that it is a scale on the final moving assembly, means there should be no mechanical error . so the calibration in counts/inch should be exactly at the scale resolution.
Al.

[ViperTX]

...is the glass scale being flexed in anyway??...

[laserman]

It could be a couple of things, does the cnc have a internal temp. probe and is it working correctly? Are the comp tables turned on? Do you have any mechanical errors in the axis that is showing the error? Is the error repeatable? Is the reader head tight?

[gus]

Is it losing movement or gaining it?

What kind of control is it?

How are you checking the accuracy?

Remember, linear encoders don't actaully count lines, they have a grating that has the same spacing as the encoder, so motion results in a on off the same as the grating period, but an average of like 25 lines or something. Anyway, minor screwups are not going to affect accuracy