Toolhead drifts in -x direction when TEST_RESONANCES

General Discussion
1

Issue: Toolhead drifts in the negative X direction while running TEST_RESONANCES for both x or Y axis

Clues:

  • The steppers have an odd sound when the toolhead drifts
  • negative X drift means that both steppers are synchronized
  • shaper_calibrate graph shows multiple minor peaks where drift is occurring. 120Hz peak is the true natural frequency of the machine.

Hardware:

  • Mellow Super infinity HV
  • BTT tmc5160 Pro @ 48V

Driver Config:

[stepper_x]
step_pin: PE2  # Drive0
dir_pin: !PC5
enable_pin: !PF11
microsteps: 32
rotation_distance: 40
endstop_pin: tmc5160_stepper_x:virtual_endstop ##PC0     #PG12  # IO0
position_endstop: 0
position_max: 120
homing_speed: 70
homing_retract_dist: 0
homing_positive_dir: false

[tmc5160 stepper_x]
cs_pin: PC4
spi_bus: spi3
diag1_pin: ^!PG12
interpolate: False
run_current: .8#1   # for 52mms max hold is 1.060 
sense_resistor: 0.075
driver_SGT: 1#1 @1.2amp

Klippy Log:

klippy.log (1.1 MB)

shaper_calibrate_y
shaper_calibrate_y800Ă—600 135 KB

2

Very strange. Never seen this on my printers.

Actually no clue, so just spitballing: The resonance forces are higher than the holding torque of the stepper? So kind of “skipped steps” in this direction.
Would it change with higher run_current and / or microsteps?

3

I bumped the micro steps to 1.3 and the current from 1 up to 1.3. Neither seemed to help though.

4

Check if changing belt tension changes the outcome. From the video appears that one loops is tighter than the other.

5

This is a known hardware issue with Mellow boards. You need to use a custom step pulse duration. Add the line below to all of your tmc5160 sections.

step_pulse_duration: 0.000004

6

I should also note this will cap your maximum speed. Some people have reported success with a step pulse as low as .000001, but even with that value you will be limited to about 500mm/s at 32 microsteps as you currently have configured.

7

Is it only the Mellow Super infinity HV or a broader scope of boards?

8

@mental actually just received some boards from Mellow for testing, but as far as I know all Mellow boards are affected. They added some circuit protection between the drivers and the MCU which is somehow limiting the speed.

1 Like
9

Belt tensions are balanced and are good, not too loose not too tight.

Seriously? what’s the point of having a HV board if you can’t go fast…

I guess that explains why I capped out at 1300mm/s …

I was hoping to hit 2000mm/s. Are there any work arounds for this?

10

You can experiment with shorter step pulse durations, but ultimately your best option will be to reduce your microsteps. If you still can’t reach the speeds you want, your only choice is replacing the board. This is not a problem on other HV boards like the Octopus Pro.

11

That’s unfortunate. My only regret was not buying from aliexpress so that I could write a review and warn others.

I have already torn out the board but now I need to figure out Powerbrrr on an old SKR 1.3.

Thanks for chiming in, you saved me a lot of time.

12

It might be worth experimenting before ditching the board. The speed I quoted is based on a user post on the Mellow Discord. It doesn’t sound like he tested beyond 1000mm/s. Doing some theoretical math the board should be capable of much more as long as microsteps are kept at 16. Here’s the post.

I just wanted to share with you my findings regarding the step_pulse_duration issue with klipper and the FLY protection. If you set your step_pulse_duration to 0.000004, you will never be able to run speed over 600mms with a 16microstepping. I was able to tune down to step_pulse_duration: 0.000001 with no drift on motors, and was then able to reach 1000mm/s. :slightly_smiling_face: I might try to go a bit smaller on pulse lenght and try to find the minimum we can use. But that might be useless. 1000mms is plenty enough unicorn speed haha

13

That’s helpful, thanks for sharing that. We’ll see, there are numerous other issues with the board aside from that are a major annoyance. Maybe I’ll save this board for one of my other printers. It’s a little cramped trying to fit this into the V0