Looking for help better understand Delta kinematics, and why I can't get repeatability.

125 views
Skip to first unread message

Christopher Karg

unread,
Jun 1, 2020, 10:21:09 PM6/1/20
to Delta robot 3D printers
I am trying to calibrate my delta printer with mixed results. I have squared the towers and leveled the bed to the best of my available means. From there I am using David Crocker's Delta printer least-squares calibration calculator found at https://escher3d.com/pages/wizards/wizarddelta.php with G30 probing from 5mm above bed height to get the height error. I am using a StrainEffector V1.3, Azteeg X5 mini V2, and Haydn Huntley's magball delta arms (288.27)

Where I am have problems is repeatability I have pasted below a copy of commands I send that the results.

SENDING:G28
SENDING:G0 X0.00 Y0.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
SENDING:G0 X0.00 Y100.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
SENDING:G0 X86.603 Y50.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
SENDING:G0 X X86.603 Y-50.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
SENDING:G0 X0.00 Y-100.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
SENDING:G0 X-86.603 Y-50.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
SENDING:G0 X-86.603 Y50.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
SENDING:G0 X0.00 Y0.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
Z:5.0500
Z:4.9850
Z:4.9850
Z:4.9825
Z:4.9825

Z:3.5300
Z:3.5275
Z:3.5275
Z:3.5275
Z:3.5275

Z:4.9425
Z:4.9575
Z:4.9550
Z:4.9575
Z:4.9575

Z:5.0300
Z:5.0325
Z:5.0375
Z:5.0400
Z:5.0475

Z:4.7975
Z:4.7975
Z:4.8000
Z:4.8050
Z:4.8025

Z:5.1025
Z:5.1025
Z:5.1025
Z:5.1050
Z:5.1050

Z:5.0150
Z:5.0125
Z:5.0150
Z:5.0100
Z:5.0150

Z:4.9550
Z:4.9575
Z:4.9575
Z:4.9600
Z:4.9675

Not much of a difference but still a difference between the first

G0 X0.00 Y0.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
(Z:5.0500
Z:4.9850
Z:4.9850
Z:4.9825
Z:4.9825)
and the last

G0 X0.00 Y0.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
(Z:4.9550
Z:4.9575
Z:4.9575
Z:4.9600
Z:4.9675)

Then the difference when I G28 between them.

SENDING:G28
SENDING:G0 X0.00 Y100.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
SENDING:G28
SENDING:G0 X86.603 Y50.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
SENDING:G28
SENDING:G0 X X86.603 Y-50.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
SENDING:G28
SENDING:G0 X0.00 Y-100.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
SENDING:G28
SENDING:G0 X-86.603 Y-50.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
SENDING:G28
SENDING:G0 X-86.603 Y50.00 Z5.00, G30 F150, G30 F150, G30 F150, G30 F150, G30 F150
Z:5.0625      (Z:3.5300)
Z:5.0600      (Z:3.5275)
Z:5.0600      (Z:3.5275)
Z:5.0625      (Z:3.5275)
Z:5.0650      (Z:3.5275)

Z:5.7650      (Z:4.9425)
Z:5.9450      (Z:4.9575)
Z:6.0525      (Z:4.9550)
Z:6.0450      (Z:4.9575)
Z:6.0475      (Z:4.9575)

Z:5.0850      (Z:5.0300)
Z:5.0900      (Z:5.0325)
Z:5.0925      (Z:5.0375)
Z:5.0925      (Z:5.0400)
Z:5.0925      (Z:5.0475)

Z:4.8975       (Z:4.7975)
Z:4.9075       (Z:4.7975)
Z:4.9000       (Z:4.8000)
Z:4.9100       (Z:4.8050)
Z:4.9125       (Z:4.8025)

Z:5.1025       (Z:5.1025)
Z:5.1225       (Z:5.1025)
Z:5.1250       (Z:5.1025)
Z:5.1250       (Z:5.1050)
Z:5.1275       (Z:5.1050)

Z:5.3350       (Z:5.0150)
Z:4.9925       (Z:5.0125)
Z:4.9300       (Z:5.0150)
Z:4.9275       (Z:5.0100)
Z:4.9300       (Z:5.0150)


I have also tried sending G90 before I started to ensure I am using absolute positioning, but still getting the kind of results I have shown above. Any help or insights would be greatly appreciated.

Thanks,
  Christopher


config-override
config.txt
FIRMWARE.CUR

David Crocker

unread,
Jun 2, 2020, 5:28:31 AM6/2/20
to Delta robot 3D printers
Some observations:

1. As the results are different with G28 between the probe points, I suspect you may have some backlash in your motion system. To test for backlash, try this:

G0 X25 Y0 Z5, G0 X0, G30 F150
G0 X-25 Y0 Z5, G0 X0, G30 F150

Repeat the above several times. The idea is to approach the probe point from opposite directions and see whether they give different results.

Then try the same again but approach the probe point from opposite Y directions.

Then try the same again but with a probe point away from the centre (adjust the X and Y coordinates you start from to be about 25mm above and below the probe point).

If the probe heights are different on alternate runs (corresponding to approaching the point from different directions), you have backlash. This is usually most apparent when probing well away from the centre. Haydn's magball joints have very low backlash provided that you lubricate the balls (I use a little silicone grease). Cheap linear rails may have backlash (I have read that they can sometimes be improved by dismantling and repacking with new balls), so may wheeled carriages.

2. What does the F parameter in the G30 command mean for Smoothieware? I guess it's probing speed, but in mm/sec or mm/min? Recommended probing speed for the Smart Effector is 1200mm/min assuming that the firmware latency to the Z probe triggering is no more than a millisecond or so.

3. As the Azteeg X5 uses a LPC1769 processor, it should be possible to run RepRapFirmware on it, then you would have auto calibration. See https://github.com/gloomyandy/RepRapFirmware.

HTH David
Reply all
Reply to author
Forward
0 new messages