Duet 3 1XD board Bed leveling problem

Philippe71

For a special machine I need to drive 2 x Z motors with hi-current (5 or 6 A/24V)
I have made a temporary installation of the new Expansion 1XD with 2 External drivers of Z motors connected on it.
Duet 1XD drive the 2 Zmotors. It works, the Z motors are OK, Z homing is OK (I just had to disconnect the Enable on the external drivers).

The problem is my bed leveling does not work any more...

On this high temperature machine, we have special bed leveling, because we have to deport the system from the heated chamber .

We use an electromagnet that pushes a probe into the chamber. When the Bed rises, it pushes the sensor and a blade (in red, in the drawing above), because the electromagnet is not very powerful. A standard low cost optical sensor (https://www.aliexpress.com/item/32817734162.html) placed in a cooled area gives the information of Z.
This system works since more than 2 years, with Repetier firmware and RRF3 on your Duet3 since 8 months

With the 1XP, the sensor gives random errors, sometime during the first probe, sometime after 2 or 3 probe, sometime in the middle of the G29...
![Error: Z probe was not triggered during probing move](

If I deploy the probe by M401, the Z_Probe info switch to 0, it looks OK
!(
If I retract the probe by M402, the Z_probe info switch to 1000, it look OK too.

I try with the optical sensor of bed probe connected to MotherBoard 6HC (in config.g)
M558 P9 C"0.io8.in" H8 F200 T3000 ; set Z probe type to switch and the dive height + speeds
M557 X15:285 Y15:235 P6 ; define mesh grid

and connected on the 1XP
M558 P9 C"122.io0.in" H8 F200 T3000 ; set Z probe type to switch and the dive height + speeds
M557 X15:285 Y15:235 P6 ; define mesh grid

For deploying the Z probe I use the out8 port, the electromagnet requires 300mA
The deployprobe.g :
M950 P1 C"0.out8" Q5000 ; allocate GPIO port 8 to expansion board out8 port, 5000Hz for reducing vibration of Electromagnet
M42 P1 S0.8 ; set 80% PWM on GPIO port 1, avoid to too much heat the Electromagnet

The retractprobe.g :
M42 P1 S0

I tried to add ferrites on the optical sensor cable... same random results.

Due to hi-temperature and thermal dilations, I have to do a bed leveling before each print.
In my print gcode, I have a G32 that call the bed.g :

M561 ; clear any bed transform
T0 ; Activate T0 head (the one with the sensor)
M401 ; deploy Z probe (run deployprobe.g file)
G1 X150 Y125 Z20 F1000 ; move head in the middle of bed before establish a Z=0 datum using the Z probe
M400 ; wait current move to finish
G30 ; establish a Z=0 datum using the Z probe
G1 X150 Y125 Z20 F300 ; move head in the middle of bed
G29 ; probe the bed and enable compensation
M402 ; retract Z probe (run retractprobe.g file)
M376 H10 ; Set bed mesh compensation taper to 10mm (bed mesh error are compensated on 10mm)
G1 X150 Y125 Z60 ; Move the head down in the middle of bed

Note :

The same configuration worked fine without the 1XP board.
On 6HC Firmware is : MB6HC version 3.1.1 running on Duet 3 MB6HC v0.6 or 1.0 (standalone mode) and Board EXP1XD : firmware 3.1.0+ (2020-06-03b1) on 1XP

Do I made error, or you have any ideas or things to test ?

Philippe71

For information, I placed an oscilloscope on the input signal of the board (122.io0.in), the signal seems to be correct.
I switch M401/M402 every 0.5sec!

In fact the G30 works every time, it's the G29 that's the problem, now it's almost always at the second point I'm getting the message :

dc42

Does the probe deploy and retract once per G29 command, or once per probe point?

There is a maximum rate at which sensors on an expansion board can send data to the main board, so adding a settling time delay may help. See the M558 R parameter.

Philippe71

Does the probe deploy and retract once per G29 command, or once per probe point?
The probe deploy & retract once per G29 command... The error message appear after the first G29 point during the probe of the second point just after the probe is triggered .

There is a maximum rate at which sensors on an expansion board can send data to the main board, so adding a settling time delay may help. See the M558 R parameter.
Tried with R 0.2, R0.5 R1.0 same result...

I use G32 for testing the Z probe
My bed.g file :
; bed.g
; called to perform automatic bed compensation via G32
;
M561 ; clear any bed transform
G1 X150 Y125 Z12 F1000 ; move head in the middle of bed before establish a Z=0 datum using the Z probe
G30 ; establish a Z=0 datum using the Z probe
G29 ; probe the bed and enable compensation
M376 H10 ; Set bed mesh compensation taper to 10mm (bed mesh error are compensated on 10mm)
G1 X150 Y125 Z60 ; Move the head to Z60

My ZProbe configuration in config.g
...
M558 P8 R0.5 C"122.io0.in" H7 F200 T3000 ; set Z probe type to switch and the dive height + speeds
G31 X0 Y0 Z3.0 ; set default Z probe trigger value,
M557 X15:285 Y15:235 P6 ; define mesh grid
...

Philippe71

A short Video explaining better :
Bed Leveling Problem
You can see the first G29 point is OK and the problem appear during the second point.

dc42

Thanks. Do you think it is probing to the full depth allowed by the dive height? Or it the probe triggering when it should and stopping the Z motion, but another part of the firmware hasn't registered that the probe triggered?

Philippe71

It seems the probe triggering when it should and stopping the Z motion, but another part of the firmware hasn't registered that the probe triggered.
( on the second point, I see the probe trigerring on the oscilloscope)

The config is : Z motors, Z home Sensor, and Bed leveling on the 1XD board

Philippe71

I tried with the Z probe connected to the motherboard (0.io6.in), it works better but the G29 fail about 10-15 probing...

It seems the signal is different between the first Zprobing and second z probing of G29

The first Z probing

The seconds and more Z probing

Philippe71

Sometime the signal is like that and the probing fails

Philippe71

I have checked manually the bed level (in case of the bed was too much inclinated), it is at about 0.3mm error.

I have also checked the probe sensor by sending M401/M402 at 100ms delay, it looks OK !

I'm confounded, any idea or thing to test ??

dc42

The signal from the probe certainly looks wrong. Is it varying depending on the XY position of the probe? Are you certain that you have set the dive height large enough, in case the probe isn't travelling down (or the bed up) quite far enough?

dc42

PS - another possibility: perhaps the Z axis is skipping steps in such a way that on each probe in the G29 sequence, the probe starts a little higher than last time.

Philippe71

@dc42

Thanks, I will try to increase dive height, just for summarize :

1 - The Z Motors are connected on the 1XD board with 2 external drivers
https://www.omc-stepperonline.com/download/DM556N-L.pdf

If I connect the Probe on the 1XD board :

The signal is OK on the G30 probe
The signal is OK on the first G29 probe, then wrong on the second probe

If I connect the Probe on the 6HC board :

The signal is OK on the G30 probe
The signal is OK on the first G29 probe, then reduce on other probes and after about 10-15 probe the signal is wrong.

2 - If I connect the Z Motors and Z Probe on the 6HC (Same dive height), ZProbe works (but I have other problems with brakes of Z motors)

Philippe71

@dc42
I do not see or hear any Z skipping steps, Z movements are smooth and regular.

Philippe71

Increased dive height to 15mm > the G29 fails on the twelfth probe
(Z motors on the 1XD, and ZProbe on the 6HC)

Philippe71

I made an other test :
Connect the 2 Z-motors to the drivers of the 6HC board, and changing only the configuration of the Z stepper motors... (No change in the others parameters of config.g)
Z probing works fine ! no error, bed is fully probed

Philippe71

100 probe points works too...

Philippe71

The problem with this configuration is :
Z motors needs brakes, it works fine in normal use. But if I push Emergency Stop button, the bed goes down quickly...
Is there a way to activate brakes when pushing Emergency Stop button ? (before doing the Emergency Stop ?)

dc42

Are the brakes the sort that are normally engaged, and you have to supply power to them to release them?

dc42

PS - the reason why it didn't work using the external stepper drivers is most likely that the M569 timings you are using are insufficient, so that on each G29 probe, the bed drops a little.