Klipper with closed loop steppers

General Discussion
1

Hello everyone,

I entertain the idea of using klipper with closed loop axis control. I did a little research, but I ended up a little confused. In my understanding it should not be a problem to use closed-loop drivers instead of open-loop drivers. I dont see that Klipper would even have to know about that. Is there anything I am missing?

Has anyone implemented a closed-loop setup? I am using NEMA23 steppers on my bigger machine.

During my research I read about salmon skin effects on closedloop setups. Can this be avoided with currently available systems?

Thanks in advance!

2

See Mechaduino experiment

3

I think this is mainly a question of the quality of the closed loop system. I am myself considering a similar move, but I would prefer actual servos (JMC has some affordable even in NEMA 17 form factor). Such systems are often used in CNC milling machines which have a higher requirement for the accuracy compared to 3D printers. While milling a metal surface, any unwanted movement will always create visible artefacts, even if it is only 1/100mm. Of course one important point is how much the angular error of the motor translates into a linear error on the axis: A CNC milling machine usually uses ball screws where you need to turn the motor more for the same linear movement compared to belt drives. (My machine is a combined 3D printer and CNC mill, and I have ball screws but with a relative high pitch…)

1 Like
4

Not having dug too deep into this topic but as far as my understanding goes there are two philosophies:

Mechaduino approach as per above link:

  • Klipper runs on the dedicated stepper MCU board
  • Klipper manages PID, open / closed loop etc.
  • Only roughly developed and not fit for daily use

STEP / DIR approach

  • Toolboard runs own firmware
  • Klipper and toolboard does not know each other nor do they interact
  • Klipper only provides STEP and DIR signals, which are processed by the toolboard
  • Toolboard is responsible for controlling the stepper motor, microstepping etc.

I cannot comment on either as I have no experience with.

1 Like
5

I’m running MKS S42C for the x and y of a CoreXY. They are defined as A4988 and get the step and dir from the printers mainboard a BTT GTR. Works absolutely fine.

3 Likes
6

Did you need to short the SLP & RST pins on the stepstick adapter?
Apparently A4988 drivers need to have these two pins shorted, but I’m not entirely sure about the MKS SERVO42C or SERVO42D.

7

I’ve been told by people smarter than me in both the Armchair Heavy/Engineering and creapunk Discord servers that servos, in Klipper’s current state, are not super reliable due to how Klipper’s clock is implemented and the lack of clock sync between MCUs - apparently this aspect of the firmware is also too pervasive throughout the coding framework, so a simple plugin/PR probably won’t be able to address this.

Apparently, we’re better off running servos on RepRap firmware instead.

To be clear, Klipper will still in-fact work with servos, but we won’t be using them to their fullest potential, and there is a chance for the positional data on the servo pcb to lag behind that of the MCU(s).

8

Hello @x_jaydn !

Thank you for your input. Due to the age of this thread and the ongoing development, this thread is closed.