How can I reduce L6472 Stepper Motor Noise due to harmonics?

[LWalk.1]

I have a motor control PCB with x6 L6472PD stepper motor drivers on. 2 of the drivers are being used to drive NEMA17 "hybrid stepper" linear actuators that move a module along a linear guide. The two drive systems are laid out identically, and conveniently the same module can be driven by either the "left" or "right" driver.

Normally our system works perfectly well, but we have observed that on approximately 5% of PCBs, the module vibrates loudly when being moved at about 300-400 sps. There is no particular pattern of "left" or "right". Intriguingly, the module in question does NOT vibrate if is driven from the opposite driver.

Through a series of experiments, I have narrowed down the effect to the driver circuit in particular, to eliminate cabling. I can move an affected PCB through different hardware setups, and the vibration characteristic will always follow the driver circuit.

Using a current probe, I have obtained plots of the current output of an affected driver circuit and compared to a healthy driver circuit. In the time domain, the most observable difference is an ~3% difference in timing, which I have put down to the tolerance of the internal oscillator.

EDIT2: I now am running both motor drivers from the same common oscillator. This removes the timing differences, but the noise and harmonics (described below) remain.

In the frequency domain however, there is a very noticeable difference between the two waveforms. The FFT trace below shows a primary peak at 100 Hz, corresponding directly with the step frequency of 400 sps (4 steps = 1 wave period). In the affected module both odd and even harmonics are present, whereas for the healthy module, only odd harmonics are present.

I have experimented with the control system values not to much effect. The pitch of the vibration reduces with reducing RUN current, but is still significantly louder on an affected versus an healthy driver circuit.

The control settings for the driver are:

TON_MIN = 4 µs

TOFF_MIN = 21 µs

TOFF_FAST = 10 µs

FAST_STEP = 16 µs

TSW = 48 µs

SLEW = 0 (320 V/µs)

CURR_RUN = 968 mA

I have tried varying each of these parameters by ±10 µs without any noticeable affect on the vibrations.

Any ideas what is causing my harmonics? Or if the harmonics are even related to the audible vibrations I'm getting?

EDIT: Motor driver schematic segment (missing connectors):

 

We are driving with 24V

[Cristiana SCARAMEL]

Hello @LWalk.1​ and welcome to the ST Community.

Could you share some waveforms showing the motor current phases with shorter time base?

[LWalk.1]

Here are some waveforms of both affected and healthy waveforms on some shorter time bases.

Affected:

Healthy:

I've placed the cursors to highlight some features of the waveform in each case.

I only have a single current probe, so I cannot capture both phases at the same time. I have performed the FFT analysis from my earlier post and confirmed the frequency content is the same on both phases, on both the healthy and affected side. I could possibly rig something up to measure with a voltage probe if necessary.

[LWalk.1]

I caught some additional waveforms at a 50µs time base:

affected:

Healthy:

There is an obvious difference between the waveforms in all cases... there is a 50 mV DC offset between the healthy and affected driver. I had been putting this down to slight variations in the position of the current probe (it is a hall effect sensor), but it is very consistent, and I am careful not to move the current probe when I change over the drivers.

EDIT: 50 mV on the scope screen. I'm not quite sure what this equates to Amperes.

[Cristiana SCARAMEL]

Hi @LWalk.1​,

the current control looks good, so it could depend on a mechanical resonance of the motor.

[LWalk.1]

Hi @Cristiana SCARAMEL​,

I can't be sure what the resonance of the motor is, but it is the same motor being used in both plots (not the same type of motor, but physically the same exact motor). Since both drivers are driven from the same oscillator, I don't see how physical motor characteristics could make a difference.

Luke