As a side-effect of using highly configurable and advanced stepper drivers such as TMC2660 it makes it necessary to sometimes tune things based on the actual hardware used. This also goes for reducing noise of the motors. This depends on a lot of factors amongst others are motor current, resonance of motor mount, current micro-step position on the motor end as well as a lot of parameters set to the stepper drivers.

TMC2660 have multiple parameters to set that can have an influence on the sounds of motors. These are all combined in the so called Chopper Control Register aka CHOPCONF that is set to the drivers by RRF. This register is made up of the following individual values:

All separate values are encoded in a bit mask and sent as one value from RRF to the stepper drivers.

There are some inter-dependencies on these values but most notable is

HSTRT + HEND <= 15

Most of the time users report very silent steppers while they are moving but unpleasant noises when drivers are standing still. There are various methods implemented in RRF like idle-current-reduction that should address these but sometimes it can be necessary to do a more fine-granular tuning on one or more motors to achieve the lowest noise-level.

The most influence on stand-still noise has been found in the TOFF parameter so that is what this guide will focus on. Of course you are free to also change all other parameters but they have been found to make at most a very marginal impact.

As mentioned before tuning these parameters is highly hardware specific and even if one printer uses the same type of motors for all axes it will probably also be necessary to tune the parameter(s) for each motor separately. Also sometimes the default value already is the best.

To modify the TOFF parameter one uses the M569 command, more specifically the C parameter of this command.

Starting with RRF 2.02 the M569 command has been extended with a new Fnnn parameter to set the TOFF value directly without using the rather hard-to-read CHOPCONF register bitmask.

This requires RepRapFirmware 2.0 or later.

Only tune one motor at a time with no idle-current-reduction. Do it as follows

1. Activate the motor to be tuned by either homing it (for a moving axis) or moving it in case of an extruder motor (you might have to disable cold extrusion prevention with M302 P1 but then only do a very tiny move like G1 E0.01 to prevent grinding of filament - that will be sufficient to activate the motor)
   1. If necessary disable all other motors by issuing M84 <motor designations go here>
2. Disable idle-current reduction for the motors with M906 I100

Next step is to actually modify the value in the CHOPCONF register of the driver for the motor being tuned.

It follows a table of all values with all variables kept at their RRF default value except for TOFF

All of these commands need to be extended to address the motor being tuned by adding the apporiate Pnnn parameter where nnn corresponds to the driver number starting at 0. So usually X=0, Y=1, Z=2, E0=3, E1=4, etc. So a full command to set the Y axis motor to a TOFF value of 8 would be M569 C65976 P1 or M569 F8 P1 using RRF 2.02 or later.

Nothing else needs to be added. The effect (if any) will be immediately after issuing the command.

Generally TOFF values above 10 will give a very unpleasant high-pitched noise but you can of course still try all values to be sure.

Once you have found the best setting for this motor amend the M569 command in your config.g with the Cnnn|Fnnn value you found is the most silent (it is advised to add the corresponding value of TOFF as a comment to that line in case you are using the Cnnn parameter). Then go on with the next motor.

As said before the other variables in CHOPCONF seem to have very little effect on stand-still noise but if you still want to test these there exists a tool that lets you configure all variables and creates the appropriate value for the M569 Cnnn parameter at chopconf-generator (with pre-compiled releases at the Releases page).

Note that this tool also allows output of the result as hexadecimal value but ironically RRF 2.0 does no longer accept hexadecimal input values.

It was found to make a very subtle difference on very slow movements to tune these two parameters. In tests best combination where always with

HEND <= HSTRT

but your mileage may vary so feel free to try any combination you want.

RRF < 2.02: Using the chopconf-generator from above and a small batch script can provide a list of all possible combinations and will make sure that only valid combinations are created.

RRF >= 2.02: You can use the new M569 Ynn:nn:nn parameter to tune these values. The first number is HSTRT, the second is HEND. The third number is optional and is to set HDEC. Note that this value has to be given as 0..3 corresponding to 16, 32, 48 and 64 respectively.

Blanking time can be set via M569 Bnn and takes the value 0..3 corresponding to 16, 24, 36 and 54 respectively.

As a result of TOFF being a major factor of stand-still noise RRF 2.02 introduced a separate parameter Fnn for M569 command to only adjust the value of TOFF without having to deal with hard-to-read bitmasks.

A lot of information can be found at the following sources:

• https://forum.duet3d.com/topic/5392/does...
• TMC2660 Datasheet
• Parameterization of spreadCycle™