Duet 3 SBC : Need explanation on multi Z homing

Donpi

Hi,

In my printer I have two independent z, so I tried this :
https://duet3d.dozuki.com/Wiki/Bed_levelling_using_multiple_independent_Z_motors

But I dont anderstand well how it work and if I'm doing the things right.

The easy part I affect the motors in my config.g with

M584 X0.2 Y0.3 Z0.1:0.4 E0.0

What i don't get clearly:

The thing with the M671 command.
Should I give the probe point like this ?

M671 X0:300 Y150:150 S1

Or the real leadscrews position after X and Y are homed (where the head can mechanicaly not go) like this ?

M671 X-55:355 Y150:150 S1

The part with M208.
In the example it is said X-20 but the M208 is only X-5.
Is that ment to set the mechanical limite of the head ?

And the part in bed.g

G30 P0 X20 Y100 Z-99999
G30 P1 X180 Y100 Z-99999 S2

Now we probe at X20 so what is the purpose of setting a print area at X-5 ?
And what is the point with the S2 parameter ? Is this the motor id ?

My print area is of 300*300 et perfectely centered between the two leadscrews. They are both separated by 410mm.

My actual configuration is

;config.g
M584 X0.2 Y0.3 Z0.1:0.4 E0.0
M671 X-55:355 Y150:150 S1
M208 X-10:310 Y-35:305 Z0:300

;bed.g
G28 
M401 
G30 P0 X5 Y150 Z-99999 
G30 P1 X295 Y150 Z-99999 S2 
M402

Will it compensate correctely or am I probing to far away from the leadscrews?

And what append if I do a mesh bed compensation after that ?

As allways, bad english level bla bla bla.
I hope you anderstand me well engough.

Bye
Juan

dhusolo

After homing manually jog the nozzle to your first probe point. It should be the farthest point along X min the probe can reach and note the coordinates. That's where P0 would be. Jog to the fathest side on X max the probe can reach and that would be P1. With 2 steppers Y doesn't matter but X does.

M671 is the actual positions of the leadscrews. the 1st point will be negative because it's outside of the print area. But make sure the first coordinate in M671 is the stepper motor that is wired to the board in Z0.1

This is my bed.g file that gets called when running G32
I have 3 independent lead screws whish is why I have P0-P2.
M557 defines the probing boundary and P4 probes 4 points on X and Y so 16 total.
It homes, probes each point and adjusts for compensation. It repeats that 3 times than does the actual bed leveling probing.

M561               					; clear any bed transform
G90 								; absolute positioning 
G1 Z10 H0							; Making sure we're not going to hit the side of glass
G91
G28									; home all axis bb
M400								; Wait for move to finish
M401 P1								; Deploy probe
M557 X65:328 Y21:291 P4
G30 P0 X55 Y161 Z-99999			; Probe near left lead screw position -65,166
G30 P1 X328 Y301 Z-99999			; Probe near right rear lead screw position 364,25
G30 P2 X328 Y17 Z-99999 S3 			; Probe near right front lead screw position 364,25
G28
M401 P1								; Deploy probe
G30 P0 X55 Y161 Z-99999			; Probe near left lead screw position -65,166
G30 P1 X328 Y301 Z-99999			; Probe near right rear lead screw position 364,25
G30 P2 X328 Y17 Z-99999 S3 			; Probe near right front lead screw position 364,25
G28
M401 P1								; Deploy probe
G30 P0 X55 Y161 Z-99999			; Probe near left lead screw position -65,166
G30 P1 X328 Y301 Z-99999			; Probe near right rear lead screw position 364,25
G30 P2 X328 Y17 Z-99999 S3 			; Probe near right front lead screw position 364,25
G28
M401 P1								; Deploy probe
G29									; Probe the bed and save height map to file
M402 P1								; Retract probe

Donpi

@dhusolo Hi, why do you do the homing/probing 3 times ?

oliof

Multiple runs help to see that the process converges.

Donpi

@oliof but how ?

oliof

In most cases, lead screw positions are approximated so adjustments aren't perfect

In my case, after the the first adjustment, I run again and see a smaller adjustment. I.e. in the first run 0.5mm, and in the second run 0.03mm. Whith the third run there usually is no tilt to adjust for anymore since the remaining difference is less than the resolution of the z probe.

NexxCat

As a side question to this.

Say you are using a kinematic 3 point bed; do you specify the leadscrew positions as where the leadscrews physically are, or where the kinematic pivot is? It would seem that in this situation, you would want to specify the pivot point, then probe as close to those locations as possible?

Phaedrux

@NexxCat said in Duet 3 SBC : Need explanation on multi Z homing:

you would want to specify the pivot point, then probe as close to those locations as possible?

Yes, that makes sense. However, in practice, it isn't really critical because you're likely going to run 2 or 3 cycles anyway to make sure it's converged, however the closer on the first pass the better.

NexxCat

@Phaedrux said in Duet 3 SBC : Need explanation on multi Z homing:

@NexxCat said in Duet 3 SBC : Need explanation on multi Z homing:

you would want to specify the pivot point, then probe as close to those locations as possible?

Yes, that makes sense. However, in practice, it isn't really critical because you're likely going to run 2 or 3 cycles anyway to make sure it's converged, however the closer on the first pass the better.

Perfect, thanks!

dhusolo

@NexxCat I had better results with the pivot point vs the leadscrew position and probe as close to it as possible

dc42

Where you probe doesn't matter if the bed is flat, although obviously measuring the tilt is more accurate if the probe points are widely separated.

You can probe more points than you have leadscrews.

NexxCat

@dc42 said in Duet 3 SBC : Need explanation on multi Z homing:

Where you probe doesn't matter if the bed is flat, although obviously measuring the tilt is more accurate if the probe points are widely separated.

You can probe more points than you have leadscrews.

So you could, for example, probe a 3x3 over the entire bed (9x G30 probe points) and still use S3 on the final one to have the three lead screws correct based on a larger data set?

NexxCat

@NexxCat said in Duet 3 SBC : Need explanation on multi Z homing:

@dc42 said in Duet 3 SBC : Need explanation on multi Z homing:

Where you probe doesn't matter if the bed is flat, although obviously measuring the tilt is more accurate if the probe points are widely separated.

You can probe more points than you have leadscrews.

So you could, for example, probe a 3x3 over the entire bed (9x G30 probe points) and still use S3 on the final one to have the three lead screws correct based on a larger data set?

To answer my own question. You can indeed do this!