Faceting/Linearization leading to poor surface finish

I agree completely with what spumco said and would like to expand on it. I have seen some faceting caused by a combination of extreemly short linear segmented CAM generated code and poorly chosen G64 settings. If multiple very short linear segments are used to approximate a curve, and a large G64 Pn setting are combined, sometimes the result will give a faceted tool path because G64Pn will assume "n" also for the Q setting. If Qn is less than the length of multiple lines of linear code in a row the interpreter will turn them into a single line. The result will be a bunch facets of approximately n length. The answer is to specify a value for Q less than the sum of two of your CAM program's shortest line segments instead of letting the Q be assumed to be equal to P.

Have you reached out to Tormach for this? I know they had released updated servo configuration files for this issue at some point, I am unsure if it's after when this tread was created or not.

They haven't posted them but I have seen them on the FaceBook group(s) in the files section, also I made a pull request on their right to repair github to add all of them so people like yourself have access to them but they have not merged it as of yet.

Its not gcode related. Im using the conversational and moving in straight lines. But, it doesnt matter if its a straight line because I also see it in interpolated moves. It doesnt change with feeds or speed changes or end mill changes. Its always there.

I have reached out & got the updated servo files. But they said it will only help with i templates moves. If im seeing thisnproblem in a straight line then there is a mechanical issues.

I have reached out to clear path & the point me to tormach.

I have set it up per their instruction, its there. I set it up tighter or looser, its there, ive replace contact bearings on the axis, its there.

I autotuned the servos, helped maybe a little. But still there.

At this point I gotta think its a limitation/configuration of the servos that I dont readily have access too. By accident as much as I have messed with, it should have made a difference one way or the other. But it seems to be the same no matter what.

BSAT wrote:

It doesnt change with feeds or speed changes or end mill changes. Its always there.

Does the rastering change with those changes?
Might want to try with a single or dual flute on some aluminium with the same high spindle speed and different feed rates to see if the distance changes, in general, if it does change would point to servo/tuning issues, if it does not would mean mechanical issue.
Would be nice to also do same feed rate, but vary the spindle speed.
Thinking of it, at what spindle speed are you cutting and what mill? I might have missed that.

BSAT wrote:

At this point I gotta think its a limitation/configuration of the servos that I dont readily have access too. By accident as much as I have messed with, it should have made a difference one way or the other. But it seems to be the same no matter what.
 

If:

• the machine is ok mechanically
• the autotune is sorted
• you have the appropriate servo settings from Tormach
• The G-code is 'clean' (no faceting in the code)
• Your M64 Pnnn is set appropriately, and
• Other Tormach owners report the same issue

Then it seems like the only thing left to do is modify the OEM steps-per-unit and see if that changes things.

You have the 'enhanced' version of the servo - that is, it's capable of higher input resolution than the 'normal' version (p/n ends in -RLN).  The ELN is capable of up to 6400 steps per rev, while the RLN is only capable of 800ppr.

The Tormach 1100MX configuration appears to use SCALE = 10000.0 in the INI file.  (tormach_mill_base.ini)
Tormach_mill_clearpath_ecm1.hal:
Assuming a 5TPI ballscrew, it looks like the Clearpaths are set to ~2000 pulses per rev.  You can verify this in Clearpath MSP software.

If that's the case, you can set the Clearpaths to 6400ppr input resolution, and edit the INI file to account for this increased resolution.

You may need to edit the INI file further to account for the decreased time between pulses:
I don't know what the hardware is that's driving the 1100MX, but if it's a Mesa board it will have zero problems outputtting a signal at that resolution & max speed.

If you get issues you can turn down the Clearpath resolution a bit, re-adjust INI file, and try again.

All of the above is free, easily reversible,and safe to test.  A simple, short jog after startup will indicate if 1" jog looks like about 1".  If everything appears to be OK, you can fine-tune the INI scale value if you find that the distance commanded isn't exactly the distance traveled.

Test short-ish moves at full rapid speed once you have the scale calibrated.  Do it with the spindle running as well to see if any electrical noise is affecting the increase pulse frequency.

Then test the cutting performance again.  WIth a 3x increase in servo resolution (a 3x decrease in distance per input pulse) the servo should both move and sound smoother.

If there is still no change in surface finish, and there are no mechanical deficiencies, then the problem may simply be the 1100MX isn't rigid enough or Tormach has changed/edited something inside the trajectory planner that causes this faceting.