Duet2 WiFi generating polygons instead of circles

[flopes]

@flopes This is actually the config.g when I enlarged the part 10x and tricked my printer to print 10x smaller with success.

[flopes]

@flopes This is the g code that fails:
....
G01 Z0 F6.
G01 X.087 Y-.064 Z0 F4.
G03 X.087 Y.064 Z0 I0 J.064 F4.
G01 X0 Y.064 Z0 F4.
G03 X0 Y-.064 Z0 I0 J-.064 F4.
G01 X.043 Y-.064 Z0 F4.
.....

[dc42]

@flopes the minimum length for segments that make up arcs is currently set at 0.1mm. We should perhaps revise that value for situations with high steps/mm, low feed rates and small arc radius.

[flopes]

@dc42

Hi
Is there a quick fix, or something easily changeable in the firmware to get to produce arches of 5microns?
My printer is only 10cm by 10cm so I don't mind sacrificing the maximum length if that means increasing resolution
Thanks

[flopes]

@flopes The first obround is 100micron tall by 300 wide

[flopes]

@flopes Our feed is 6mm/min!

[flopes]

@flopes Also there is some uncertainty on the position of some features of about 25micron.
We cut panels with 100 devices like this

[flopes]

@flopes Some features are aligned most of time. IT IS NOT STEPPER MISSING STEPS.

[flopes]

@flopes Most devices cut like this
Any help on the matter would be greatly appreciated

[T3P3Tony]

@flopes one option to try would be to setup the config as if the printer was 10* bigger, then set the microstepping to 64 instead of 8, and the steps/mm to an appropriately adjusted value. The idea being that the limit of 0.1mm segments would then be 0.01mm in reality due to the 10x scaling.

[flopes]

@T3P3Tony
Scaling 10:1 is how I have been operating. Do you know what is the minimum positioning resolution of the Duet2?
I know It responds to G21G91G1X0.001 with 1 micron displacement, but would the motion control allow for such fine positioning? Could I laser drill a grid 5micron by 5micron?

My tests show about 25micron uncertainty on the position of a cut, but once the cut starts, it can be very precise (5micron?)

[T3P3Tony]

@flopes you are limited by the step resolution. If you have been operating at a 10:1 scaling then switch to a 100:1 as a work around for now.

[flopes]

@T3P3Tony Thanks. I will run a test

[flopes]

@T3P3Tony
It seems to work fine. Polygons are gone at 100X. Error in mispositioning improved as well. May be good enough. Please advise when you will increase board resolution
Thanks
Fernando

[T3P3Tony]

@flopes thanks for the update. I don't have a timescale for changing the arc segmentation resolution right now, possibly it needs to be configurable as while you are controlling motion in the micrometer range, there are other uses working in the meter range!

[flopes]

@dc42
Hi
Can you instruct me to increase the resolution so it does smaller radiuses? I need 30x more resolution than standard, or 3 micron radius. My machine is super slow, with a small pritable area of 100x100mm, so I am not concerned about the other end of the spectrum. Thanks

[dc42]

@flopes please can you clarify what are your actual X and Y steps/mm, before you do the x10 or x100 conversion?

BTW the reason for the late response is that this thread was marked as "solved" so I was ignoring it. Posting to a thread marked as solved is not a good idea . You should either marked the thread as "unsolved" again (which you can do if you are the thread author) or start a new thread.

[flopes]

@dc42 Done. Any tips on where I should start?

[droftarts]

@flopes said in Duet2 WiFi generating polygons instead of circles:

@dc42 Done. Any tips on where I should start?

Not sure what you mean. @dc42 asked you to clarify what are your actual X and Y steps/mm, before you do the x10 or x100 conversion. Maybe post your current config.g and firmware version. As far as I'm aware there has been no change in the arc segmentation resolution in the latest firmware versions.

Ian

[dc42]

@flopes I've reduced the minimum arc segment length from 0.1 to 0.02mm in 3.5.0-beta.4. Perhaps that will go some way to resolving this for you? I'm still waiting to know what steps/mm you are using.