Observations over Delta Height

garlicbread

Hi,
Theres a couple of things I've spotted recently
This is to do with the calibration of a delta printer and perhaps some changes to the output of the configurator.

I'm using an Anycubic Kossel Linear Plus, Duet Wifi, latest firmware 2.01
The printer has a magnetic probe that clips directly underneath the nozzel at around 16.6mm dowards on the Z axis

the config files I'm using can be seen here

https://github.com/grbd/GBD.3DPrinter.Kossel.Anycubic/tree/master/DuetBoard/Config/configv2

Delta Height

The first thing I've spotted which is quite important relates to the delta height
It might ether be a case of an error in my G-Code, or possibly a software bug, or maybe a missing feature I'm not sure
But I think the G30 autocalibration doesn't set the overall delta height automatically in relation to the end stops
Then again perhaps it's not designed to do that.

https://duet3d.dozuki.com/Wiki/Calibrating_a_delta_printer#Section_M665_command

This seems to suggest that you should get the delta height within 5mm of the bed before starting the calibration
so perhaps this is where I was going wrong

First as a starting point I used the following to setup the delta parameters

M665 R134.4 L271.5 B116 H300
M666 X0 Y0 Z0

Now the delta height is actually around 289.877 instead of 300 for info
If I run the auto-calibration / G32, then the end result will look something like

Diagonal 272.463, delta radius 134.676, homed height 299.998, bed radius 116.0, X 0.652°, Y 0.751°, Z 0.000°

This is still close to 300mm the original value for the height
I think it's adjusting it based on the tilt of the bed etc, but not in relation to the end stops
It's not a big deal but I think it could catch someone out.

The way I get around this currently is to

Home machine
Attach the probe
heat the bed to aroung 60 degrees (just my own preference ro compensate for thermal expansion of the bed)
issue G30 S-1
repeat the home / G30 S-1 a couple of times to get a couple of values in the G-Code Log
raise the head by about 5mm, so the probe isn't touching the bed anymore
read the offset value in the G-Code logs, which in my case is 26.723
Since I know the z probe offset is 16.6 (via a paper test I've done before)
26.723 - 16.6 = 10.123 difference
so 300 - 10.123 = 289.877 as the new delta height
Lets set the new height with M665 H289.877
Next re-home the machine
Run delta-autocalibration
run "M500" to store the new settings into config-override.g

At this point if I home the machine then bring it down and do the paper test, it's spot on.
Unless this is supposed to be part of the existing G30 calibration
perhaps this should be automated via a new option for G30?
so instead of setting Z0, setting the delta height in relation to the end stops via a single probe to the center of the bed

Auto Home

I've tweaked is the g-code within deltahome.g
https://github.com/grbd/GBD.3DPrinter.Kossel.Anycubic/blob/master/DuetBoard/Config/configv2/homedelta.g

This is a little less agressive and give more repeatable results when probing the bed with G30 S-1 after a homing cycle
compared to the the one from the configurator

G30 / bed.g

For bed.g I used the configurator
However I needed to add a single G30 to the beginning of the file
https://github.com/grbd/GBD.3DPrinter.Kossel.Anycubic/blob/master/DuetBoard/Config/configv2/bed.g

to set the Z0 correctly
otherwise it resulted in the probe crashing into the bed
this might have been due to the delta height being incorrect at the time I'm not sure
but it might be an idea to add it into the config generated by the configurator as a safety margin.

dc42

The auto calibration adjusts the homed height to make the average of the three endstop corrections zero. Please note, the M665 "homed height" is the height of the nozzle above the bed when the 3 endstop switches are triggered. The actual effector height after homing is normally 5mm less than this, because the standard homedelta.g file drops the effector 5mm before centering it.

When you execute a G30 Xnn Ynn Z-99999 command in bed.g, the nozzle moves to the specified XY position with Z equal to the dive height plus the Z probe trigger height.

So the procedure you are using should be completely unnecessary, unless something isn't working as intended. However, you should not set the homed height in config.g to 10mm higher than you know the actual homed height to be, because that will cause probing to start too low unless you also increase the dive height.

As the diagonal rod length was adjusted in the run whose results you gave in your post, I guess you used 7- or 9-factor calibration. It's nearly always better to use 6- or 8-factor calibration.

HTH David

garlicbread

Thanks for the info
Based on what you've said I suspect it's ether because I've set the initial delta height too high, or because the homing cycle has moved the head down by 5mm after homing (this was in the output from the configurator).

It's given me some ideas for testing when I get back
It's possible that initial G30 command I added at the top is interfering.

If I can figure the source out, what I might look into is something like a G30 S-3 command which would just do an initial rough delta height measurement based on a single probe point (the latest Marlin already does this)

The idea being that you wouldn't need to worry if the initial height that had been punched in was more than 10mm than the actual height.
To avoid the risk of probe collisions for someone that was setting up they're machine for the first time. Then the 6 factor which would run afterwards in bed.g would do the fine tuning of the height.

dc42

You can already allow for setting the homed height too high by increasing the M558 dive height parameter when running auto calibration for the first time. This is what we recommend in the wiki page on auto calibrating a delta printer.

The comment in config.g about setting the homed height deliberately greater than the actual value is several years out of date. That was for when manual calibration was used.

garlicbread

I've discovered like you've said during the homing cycle, the axis was moving down by 5mm.
This was the main thing throwing out the height calibration.
G30 assumes that its starting right at the top, once I took this out everything was fine.

I added an update to the docs here since the default output from the configuration tool is to move down by 5mm while homing

https://duet3d.dozuki.com/Wiki/Calibrating_a_delta_printer?revisionid=HEAD#Section_Homing_the_machine

garlicbread

This post is deleted!

garlicbread

yep you were also right about M558
putting G30 at the top of the bed.g was a bad idea of mine, as it threw out the delta height. Finally got it calibrated now
Thanks for all your help