Dual z drivers wrong bed mesh compensation

[DanyloT]

Basic Information:

Printer Model: Tronxy X5SA PRO
MCU / Printerboard: SKR 2, SKR 3 EZ
Extruder: Iverntech dual drive extruder
Bed leveling sensor: Tronxy TR sensor
Firmware: OctoKlipper
klippy.log (695.2 KB)

Describe your issue:

A month ago I switched second Z stepper to fifth driver, responsible for second extruder, in order to use Z_TILT for more precise leveling. For the last few weeks I was printing parts using 0.6 nozzle and TPU 98A, everything seems to work great in such setup. But, recently, I started the tests with 1.2 nozzle and immediately inconsistent flow issue appeared. After many attempts to solve that issue, I’ve found out that the problem is connected to bed mesh compensation and dual z drivers. I’ve run two tests: 1) Both z steppers connected to one driver; 2) Dual z drivers and correction mesh full of 0. In both cases problem disappeared.
After that I’ve checked whether both z steppers take in consideration correcting values, and yes they do, so the conclusion is they are not doing crrection correctly.
Also, I’ve tried different BED_MESH_CALIBRATE procedures, in order to gain correct compensation mesh, but it doesn’t seem to be a case, since with normal procedure and single driver, everything seems to be ok.
What is more, during the test I’ve used 4 printers, one of which have newer SKR 3 EZ, and all of them have same issue.

All of that being said, I believe that the root of my problem is the wrong compensation process. Am I right? Is there any solution to such problem?

[EddyMI3D]

Hello @DanyloT !

Can you please attach the klippy.log to your next post?.
It was asked for when you opened this thread.

@DanyloT

Firmware: OctoKlipper

BTW: OctoKlipper is no firmware.
The firmware is Klipper.
OctoKlipper is a support plugin for Klipper running on OctoPrint.

[DanyloT]

Hello @EddyMI3D !
Thank you for the reply.
I’ve added klippy.log to the original post.

[EddyMI3D]

I asked you to attach the klippy log to your new post, not to the original one for reasons:

• Supporters and frequent readers always get presented the newest posts in a thread. When a thread gets long, scrolling back can be tedious.
• If it takes some time to solve the issue, you may be asked for a new klippy.log. If you put that one in the original post too, no one can distinguish the logs.

Anyhow. to your issue:

If you have two active Z axes, you only have stepper_z set to the virtual_endstop:

[stepper_z]
step_pin = PA15
dir_pin = !PA8
enable_pin = !PD1
microsteps = 16
rotation_distance = 16
endstop_pin = probe:z_virtual_endstop
position_max = 406
position_min = -2
homing_retract_dist = 5

[stepper_z1]
step_pin = PD11
dir_pin = !PD10
enable_pin = !PD13
microsteps = 16
rotation_distance = 16
endstop_pin = probe:z_virtual_endstop  # <----------------

What kind of probe do you have?

[DanyloT]

Regarding kind of probe, its inductive.
Concerning setting up virtual_endstop, I’ve made the change you mentioned.
Also, I just finished the test in which instead of launching leveling (Z_TILT_ADJUST) after soaking, I run it at first, since that kind of sensors are more vulnerable to heat, and it seems to be a potential solution, so I’m going to run few more tests during the day to be sure. I will post the results by the end of the day.

[DanyloT]

Regarding changing leveling and bed mesh procedure, sadly, it hasn’t solved problem fully. Inconsistent is less visible on the brim, but there are still issues, especially in the center. It seems that I’m just getting wrong compensation mesh. I attach a picture of the latest print.

[EddyMI3D]

To me, this appears more an extrusion issue that a bed mesh issue.

[DanyloT]

You are right, sorry for not mentioning the whole process I passed to reach the current point of solution search. I will start from the beginning. After switching to 1.2 nozzle, first print was terrible with huge gaps, which gives the idea of inconsistent flow.

So basically I started to change Cura setting responsible for flow, I’ve tried values between 97% - 116%.Of course, with higher flow problem was less visible, but for sure not solved. Next, many attempts I was trying to change Layer Height and Line Width parameters, sometime changing all three at once (flow, height, width).
The next idea was impact of second z stepper connected to fifth driver, originally designed to be used for second extruder, since mentioned issue hasn’t been noticed before the switch, but I haven’t found any information about that.
So I’ve started to look for visible mechanical problems during print, and one thing drew my attention, a slight difference between steppers moves during mesh compensation. After that I’ve checked if both couplings were tight. Following idea was incorrect compensation moves.
Then, I made the test with compensation mesh filled with 0, and surprisingly quality turned out to be much better. That’s short story about how I reached current stage. Since I have only a few months of partial work time experience with 3D printers, probably I’m totally wrong and searching in the wrong direction?

[Sineos]

@DanyloT

After switching to 1.2 nozzle

Are you sure that your extruder / hotend is actually capable of supplying a 1.2mm hotend? You should not underestimate how quickly the extrusion volume is rising, especially on higher speeds

[DanyloT]

Regarding extruder / hotend capability, yes, those printers were used in the past for one specific print with 1.2 nozzle, although the form was less complicated and smaller in diameter.

[DanyloT]

Thank you everyone for help, the solution turned out to be easier than expected, the main root of the problem was wrong compensation mesh, the reason for that is the wrong bed calibration procedure. So basically, long preheating have an impact on the precision of TR sensor which causing some deviation in z_tilt, after what, bed_mesh_calibrate, which also fully depend from sensor, is launched. As a result, wrong compensation mesh. New procedure is next: preheat bed for 5 min (without preheating hotend), z_tilt, bed_mesh_calibrate. Before print: launch z_tilt then preheat.