Stepper motor tuning

[KaruroChori]

At the moment most motors in the amd5 are running very hot (so much so that I am worried they are even risking to skip steps because of that), and they are generally quite noisy, even at low speed.
I really do not care much about getting 600mm/s with 2m/s^2, for that I have a delta printer.

So my plan was to add a more silent profile on my adm5 :D.

The issue is that I cannot find any information about the motors used in the x-y (I have yet to check the others).
If I transcribed them correctly they should be bj42d22 53v02 but there is not a single datasheet for them online I can find.

The idea was to add a profile for them to https://github.com/andrewmcgr/klipper_tmc_autotune and let the extension do the heavy lifting of configuring the tmc2209.

[jlodew0]

But not any different then standard firmware I assume?
edit: I requested a datasheet from Keli motor group. (seems to be the manufacturer)
Mind you, most steppers kan take a lot of heat. More then you will bare to touch.
If I get an aswer, I will update here.
Since the values come straight out of the klipper config of FF:
[tmc2209 stepper_x]
uart_pin: PC2
interpolate:True
run_current: 0.9
#hold_current:0.3
sense_resistor:0.10
#stealthchop_threshold: 12000

[tmc2209 stepper_y]
uart_pin: PA4
interpolate:True
run_current: 0.9
#hold_current:0.3
sense_resistor:0.10
#stealthchop_threshold: 12000

[tmc2209 stepper_z]
uart_pin: PC5
interpolate:True
run_current: 0.6
hold_current:0.6
sense_resistor:0.10
#stealthchop_threshold: 999999

[tmc2209 extruder]
uart_pin: eboard:PB10
interpolate:True
run_current: 0.8
hold_current:0.8
sense_resistor:0.10
#stealthchop_threshold: 999999

[KaruroChori]

Yes yes, it is the same problem as with the original configuration.
The difference is that we have more freedom now on how to change klipper here, so I am trying to take advantage of that :D

[KaruroChori]

edit: I requested a datasheet from Keli motor group. (seems to be the manufacturer) Mind you, most steppers kan take a lot of heat. More then you will bare to touch. If I get an aswer, I will update here.

Thank you!

[jlodew0]

I wouldn't worry to much in the meanwhile. Most steppers are rated up to 80-100 degrees celsius.
And 50 degrees celsius already feels to hot to touch for longer dan 3 seconds. Actually 54 degrees gives you full skin thickness burn wounds after 30 seconds. (to give an idea that It may feel uncomfortable but it's ok for the stepper).
And if they eventually die after some time, they are dirt cheap these days.

[KaruroChori]

While you are right, the magnetic permeability of the core will decrease with temperature raising as well, limiting the amount of torque the motor can provide.
And since xy motors are in the same enclosure with the rest of the electronics, I would be a bit conservative about temperatures to be honest, to avoid affecting the motherboard lifespan.

[xblax]

If we can find a tuned configuration that works reliably I'm open to changing it. Also thought about enabling stealthchop mode, but on the other hand the hotend cooling fan is so noisy it doesn't really matter, if we can't reduce the fan noise (that's a separate topic I guess).

Tuning the stepper makes sense, not only for noise, but but also for accurate extrusion.

I can check my other printer config later today, that also has Klipper with TMC2209 which I modded myself to the stock board.

For print speeds, I think the stepper settings should work for 300mm/s printing and (maybe) 600mm/s movements. I thought about reducing the excessive acceleration a bit, but not sure what other think about it.

[printer]
kinematics: corexy
max_velocity: 600
max_accel: 20000
max_accel_to_decel:5000
max_z_velocity: 25
max_z_accel: 500
square_corner_velocity: 25

I would probably set max_accel to 10000 and max_accel_to_decel to 2500.

@KaruroChori
You can try to manually install that Klipper extension for auto-tuning. The mod systems has no git at the moment also no systemd, so that auto-installer won't work.

I don't think we need to have that extension in the mod installed by default. We can just add the fine tuned stepper settings after finding them once.

[KaruroChori]

Yeah, honestly my temporary configuration is more like:

[printer]
kinematics: corexy
max_velocity: 150
max_accel: 5000
max_accel_to_decel:2500
max_z_velocity: 25
max_z_accel: 500
square_corner_velocity: 5

with stealthchop enabled, and the current on motors tuned down a bit.

So it is not possible to have a solution fitting everyone :D. But at least we could have some more reasonable default.

As for the auto tuning, I first need the datasheet for the motors. Hopefully @jlodew0 will get a reply back, otherwise I will have to directly measure those parameters myself. So for now it is paused.

[xblax]

I think you are bit to conservative :D That's close to my old "Geetech Mecreator 2" that I modded to Klipper with input shaping and TMC2209. ADM5 kinematics are much more rigid.

But I limit acceleration in slicer profiles, I have different profiles for high quality and high speed printing. In the profiles I set acceleration and speed for different features of the print. I.e. in high speed mode infill could print at 300mm/s with 10000 accel and outer perimeter with 150mm/s with 5000 accel etc. I think printer config should set reasonable upper bounds.

I believe it's a good idea to add M204 to set lower acceleration in homing_override, i.e. to 5000 before homing. Than slicer profiles can still set higher acceleration for high speed profiles if needed.

[KaruroChori]

You probably don't have neighbors as insufferable as mine :D
It if was up to me, I would not care less about noise.

But yes, I agree. That should be tuned in the slicer, not by placing hard limits to klipper config.

[jlodew0]

Can I use stealthChop mode on an extruder with pressure advance?
(https://www.klipper3d.org/TMC_Drivers.html#can-i-use-stealthchop-mode-on-an-extruder-with-pressure-advance)
Many people successfully use "stealthChop" mode with Klipper's pressure advance. Klipper implements smooth pressure advance which does not introduce any instantaneous velocity changes.

However, "stealthChop" mode may produce lower motor torque and/or produce higher motor heat. It may or may not be an adequate mode for your particular printer.

Lowering run current, decreases torque and can enventually lead to step skipping.

But in all honesty I don't think we have to worry about the temperatures. They even run on the same current in the enclosed m5 pro (so even hotter). On my ender 3 I had replaces my steppers with higher torque ones and even run them with higher current then specs stated. As for the speed, max velocity and acceleration will almost never be hit. And when slicing they get overridden most of the time by slower settings as well.

[KaruroChori]

Can I use stealthChop mode on an extruder with pressure advance?

In my experience with other printers yes, no problem there.

[xblax]

I think on my other printer I have stealthchop disabled on the extruder stepper for that reason. Probably the micro steps should be increased and interpolate disabled.

[KaruroChori]

For reference, my other printers are not direct drive, and they all use relatively larger stepper motors, so torque is plenty. It is possible that with pancake motors having stealth-chop is a bit too close to the sun.

[jlodew0]

It seems like FF has been pretty modest with current settings:
规格书(BJ42D22-53V02)-drawing.pdf
p.s. I just asked and the 1.5A is RMS current.

[jlodew0]

also interesting:
TMC interpolate setting introduces small position deviation¶
The TMC driver interpolate setting may reduce the audible noise of printer movement at the cost of introducing a small systemic positional error. This systemic positional error results from the driver's delay in executing "steps" that Klipper sends it. During constant velocity moves, this delay results in a positional error of nearly half a configured microstep (more precisely, the error is half a microstep distance minus a 512th of a full step distance). For example, on an axis with a 40mm rotation_distance, 200 steps_per_rotation, and 16 microsteps, the systemic error introduced during constant velocity moves is ~0.006mm.

For best positional accuracy consider using spreadCycle mode and disable interpolation (set interpolate: False in the TMC driver config). When configured this way, one may increase the microstep setting to reduce audible noise during stepper movement. Typically, a microstep setting of 64 or 128 will have similar audible noise as interpolation, and do so without introducing a systemic positional error.
https://www.klipper3d.org/TMC_Drivers.html

[jlodew0]

https://klipper.discourse.group/t/hold-current-and-run-current-tmc2209/1171/5

https://www.analog.com/media/en/technical-documentation/data-sheets/TMC2209_datasheet_rev1.09.pdf

[xblax]

Yes the 0.9A seem to be fine. I tried lowering the current to 0.8A and enabling stealthchop mode, but the steppers did not like that all on high speed moves :D I think even on my old printer, I use 0.9A for some of the steppers.

Interpolation should be fine, because I believe that is not a cumulative error, just a tiny "lag". But we could test to increase the micro steps to 32 for a bit improved accuracy.

[KaruroChori]

规格书(BJ42D22-53V02)-drawing.pdf

You got it! Sadly I will be abroad for a while, so I cannot do anything hardware-related.

[jlodew0]

Yes the 0.9A seem to be fine. I tried lowering the current to 0.8A and enabling stealthchop mode, but the steppers did not like that all on high speed moves :D I think even on my old printer, I use 0.9A for some of the steppers.

Interpolation should be fine, because I believe that is not a cumulative error, just a tiny "lag". But we could test to increase the micro steps to 32 for a bit improved accuracy.

the fan makes so much noise that I don't see any benefit in using stealthchop.

[KaruroChori]

Which fan? If you are running the part cooling fan at 100% that clearly overshadows everything else.

I generally do not print with 100% part cooling fan. Even with PLA, it is extremely overpowered. I only raise it above 30% for bridging or on extremely short layers.

The one cooling the hot-end is noisy but still comparable with the noise level of motors running in spread-cycles. So changing them to stealth-chop in my case has a positive impact on noise.

Also, the hot-end fan is always on while printing and at a constant level, so it is easier to "cancel it out" from the background noise. I find the same is not true for motors running in spread-cycle.

I also set the one in the electronic enclosure around 66%. It takes down a lot of noise, but the airflow seems to be still decent.

[b-fitzpatrick]

Any progress on motor tuning? They're generally OK but quite noisy with heavy VFAs at certain combinations of speed/direction.

[b-fitzpatrick]

FWIW, I've been experimenting with motor current, and they're much quieter at lower current. I'm running PETG, so not really pushing extreme speeds. Machine limits at 300 mm/s, 12000 mm/s², 6000 mm/s² A-to-D, and 5 mm/s SCV.

I kept dropping the current on the X and Y motors, and I started noticing some different sounds at 0.3 A but no layer shifts in my short test. To be safe, I bumped it back up to 0.45 A, which is half of the stock current.

I also increased microsteps from 16 to 64 and then 128, and disabled interpolate. I'll probably drop it back down to 64. Not sure how much effect this had, but Klipper docs recommend increased microsteps as alternative to interpolate.

Sound is pretty decent, now, but there's some rattle during gyroid infill. It goes away if I drop A-to-D down to 3000, but that slows it down quite a bit. I think there's too much play in the linear bearings. I have both a Pro and non-Pro. I went away from the Pro due to the motor noises seemingly being amplified by the enclosure walls, but the linear bearings seem to be tighter on it. I'll have to try the lower motor current on the Pro.

I'm new to motor tuning, so if there's anything to watch out for, please let me know.

[b-fitzpatrick]

128 microsteps are too many. Had a couple of "Timer too close" errors. Printing at 64 microsteps now.

[consp]

Interesting, might want to try that myself!

[KaruroChori]

For my future reference, to tune the registers of the tmc drivers: https://github.com/MRX8024/chopper-resonance-tuner

[KaruroChori]

It would be nice to extend these measurements to speeds beyond 100mm/s to also check how these values behave during fast travel moves. But considering the baseline at 0.45A was quite low, this is still an improvement.
You only added a bit of noise at low speeds, so I consider it a win.

[b-fitzpatrick]

Sadly without patching it is hardcoded

So, what does "patching" entail? Temporarily replacing adx345.py with a modified copy?

[KaruroChori]

In the easiest way possible yes, and just restart klipper.
If we were to make this module more "structural" I would apply a git patch to that file during the installation of the chopper-tuner, but as a one time operation it is a degree of polish which is not worth the effort :D

Do you need assistance editing it? I am awful at python but I should be able to do that :D

[b-fitzpatrick]

No, thanks. I can handle that. Probably won't get to it today, though.

[KaruroChori]

Then I am done for today as well :D.

[b-fitzpatrick]

By the way, with such configuration your are running the motor at 100kHz which is quite above the usual cap for driving them. Ideal values should stay above the audible band and below 50kHz. Going too high can cause issues with thermal losses and radio frequency emission. I would have selected a different configuration, maybe not as silent but more within range for a typical usage.

Thanks, I didn't know that. It looks like the settings I picked for the lower-current config are OK at 23.3 kHz.

[KaruroChori]

Yes it should be ok for most people. If you have kids particularly sensitive they might be on the edge of be annoyed by it. Some pets as well. But it is usually considered ultrasound by most accounts.

[KaruroChori]

About https://github.com/andrewmcgr/klipper_tmc_autotune
My modified branch to work on this printer https://github.com/KaruroChori/klipper_tmc_autotune/tree/flashforge-adm5

To work, tmc.py from extras should be updated to main klipper. Ideally also the other tmc****.py should be patched with the later version.

Testing now.

[KaruroChori]

Ok I got it to work and tune the motors.
Its effectiveness should be tested, but at least it is there.
If we patch tmc.py (from the code I read it is safe to do so) we can easily distribute it alongside klipper for everyone to use in case results with it are found to be positive.

The basic configuration to append is

[motor_constants bjd42d22-53v02]
# Coil resistance, Ohms
resistance: 2.10
# Coil inductance, Henries
inductance: 0.0046
# Holding torque, Nm
holding_torque: 0.40
# Nominal rated current, Amps
max_current: 1.50
# Steps per revolution (1.8deg motors use 200, 0.9deg motors use 400)
steps_per_revolution: 200

[autotune_tmc stepper_x]
tuning_goal: auto
motor: bjd42d22-53v02

[autotune_tmc stepper_y]
tuning_goal: auto
motor: bjd42d22-53v02

[autotune_tmc stepper_z]
tuning_goal: auto
motor: bjd42d22-53v02

[KaruroChori]

I edited the inductance value on the previous post (it was mH in place of H, I cannot read).
If anyone has copied this in their configuration file, please change it to avoid potential damage on your printer.