Printer Model: Tronxy Veho 1000-16 2E
MCU / Printerboard: Tronxy STM32F446ZET6
Host / SBC Linux Mint 22.1 - AMD A6-3600 APU - 6Gb RAM
klippy.log klippy.log (132.1 KB)
Issue:
The printer has a print size of 1000x1000mm on XY and 1600mm on Z. It is a Cartesian printer. Since I installed the Klipper firmware I have a problem with small layer shifting on the Y axis only. It does this on all prints, I generated the gcodes with Cura and with PrusaSlicer. The problem does not depend on the slicer because on other printers with klipper I have no problems. I have tried reducing the acceleration from 500 mm/s^2 to 100 mm/s^2 and the right angle speed from 5 mm/s to 1 mm/s, unfortunately the problem remains. I have tried doubling the current supplied to the two Y-axis nema 23 stepper motors without getting any improvement. With the original firmware Marlin never had any problems with layer shifting.
Pictures might help to explain the problem better unfortunately I cannot upload them as I am a new user.
I find the ripples in the walls more annoying.
Are you running the stepper drivers in StealthChop or SpreadCycle mode?
I’m on the phone so checking the log is not that easy.
When it is in Y only it comes from the bed movement.
Are the ripples in the same axis only?
How tight is the beds belt?
Can you show us some photos of the printer and especially of the bed?
Hi, thank you for your reply. It is not a bed slinger, but the Y axis is still heavier than the X axis.
confirm that the lines are there only on the Y axis.
The belts are well tensioned, I have tried different tensions and with marlin there were never any problems.
The printer does not have TMC drivers, but it has the drivers you can see in the picture:
Make sure that the step_pulse_duration conforms to the requirements of the external stepper drivers.
Apart from this, the issue looks hardware related to me.
Hi, thank you for your reply. I looked up step_pulse_duration values for external drivers and they are usually higher than the 2µs used by Klipper.
I will give it a try with higher values.
It seems strange to me that the problem is due to the hardware because with Marlin the printer worked correctly.
Recently, I helped several people setup servos. step pulse duration is already mentioned, I also suggest checking the pulse inversion.
It might happen that your driver expects an inverted pulse - so you can check it by setting up the ! to the step pin.
A simple check is that your stepper should be able to move precisely by steps.
Like:
microsteps = 16
rotation_distance = 60
So, you have 60 / 200 / 16 = 0.01875 per micro step.
If you do the FORCE_MOVE to that distance in one direction and then the other direction, it should step in one and then in the other.
In my understanding, V_{ref} or “voltage between GND and the V_{ref} pin” has been a way to set the stepper driver’s current (I_{RMS} or I_{max}) on old stepstick modules.
This has largely been replaced by other means to set the driver’s current.
It works!
Thank you very much!!!
This layer shifting problem was solved by setting in the [stepper_y]: step_pulse_duration: 0.000035
Probably 0.000035 seconds is also too much, maybe 0.000005 would be enough for this driver.
Klipper by default sets 0.000002 seconds for non-TMC driver. Some guidance on step_pulse_duration can be found in the Marlin firmware under: #define MINIMUM_STEPPER_PULSE in the Configuration_adv.h file.
Now I am going to set the input shaping as well to eliminate ghosting.
However there is not much ghosting but VFA in your prints.
Even on your first photo those ripples are beautifully stacking/aligning across the shiftet parts.
This can result from motor resonances or maybe your belts or rails.
Although here now offtopic you could try printing a larger hollow tower with different speed bands along its Z axis.
If the ripples change in their spacing or alignment the root cause are the motors/settings of the drivers, if not it might be your kinematic system.
Hi, actually it is not ghosting, but VFA. I think the problem is kinematics because these artifacts were there even with Marlin firmware. Anyway, I will try to print a test with different speeds as the height in Z increases. Thank you.
