TMC9660-3PH-EVKIT: abnormal velocity feedback and response curves with intermittent unexpected high-speed rotation

Hello,

I am evaluating a TMC9660-3PH-EVKIT purchased from DigiKey. I am using TMCL-IDE to configure a 2804 BLDC motor with an MT6701 incremental encoder, but I am seeing abnormal velocity feedback, poor response accuracy, and intermittent unexpected high-speed rotation during the Velocity P&I Wizard.

Hardware and software setup

• TMC9660-3PH-EVAL V1.1
• Landungsbrücke V3.1
• Landungsbrücke firmware V3.11.7
• TMCL-IDE V4.12.0
• TMC9660-3PH-PARAM-EVAL / Parameter Mode
• 12 V laboratory power supply
• Motor: 2804 BLDC, 12N14P, 7 pole pairs, 220 RPM/V
• Phase resistance Rs: 2.55 ohm
• Phase inductance Ls: 0.86 mH
• Flux: 0.0035 Wb
• Rated current: 0.5 A
• Maximum current: 2 A
• Encoder: MT6701 in ABZ incremental mode
• Encoder connected to J502 Encoder 1
• Encoder steps configured as 4096
• Motor tested without mechanical load

I completed the CSA/ADC, Current P&I, Encoder, FOC (ABN), and Velocity P&I Wizard steps. All calculated parameters were applied.

During the Encoder configuration step, the open-loop angle and ABN encoder angle appeared to move in the same direction and remained approximately synchronized.

However, the following problems occur during FOC and velocity-loop testing.

1. Abnormal velocity feedback and poor tracking accuracy

In the Velocity Response graph, RAMP_VELOCITY and TARGET_VELOCITY appear to overlap, but ACTUAL_VELOCITY is very noisy and does not accurately follow the commanded ramp.

In some FOC or Velocity Response tests, ACTUAL_VELOCITY suddenly changes to approximately -7,000,000 to -8,000,000 internal units, although the actual mechanical movement of the motor does not correspond to such a value.

I would like to confirm whether this value is expected and how the internal velocity value should be converted to mechanical rpm.

2. Abnormal Current Response graph

The Current Response graph does not show a clear and stable torque-current response.

In one abnormal test, FOC_VOLTAGE_UQ increased to approximately 16383 and remained there, while the other traces did not show an expected closed-loop response.

In other tests, FOC_VOLTAGE_UQ mainly appeared as large noisy positive and negative spikes.

I am not sure what the expected Current Response graph should look like during this Wizard step.

3. Intermittent unexpected high-speed rotation

The mechanical behavior is not repeatable and does not appear to have a clear fixed threshold:

• At approximately 400 rpm, the motor did not enter abnormal high-speed rotation. It only moved through a limited angle in the forward and reverse directions.
• At approximately 619 rpm, the motor suddenly entered obvious unexpected high-speed rotation.
• At approximately 630 rpm, the motor again did not enter high-speed rotation and only moved through a limited angle.
• At approximately 1000 rpm, unexpected high-speed rotation occurs with a relatively high probability.
• However, values above 600 rpm may sometimes behave normally and sometimes abnormally.

Therefore, the abnormal behavior does not seem to depend only on a fixed rpm threshold.

4. Velocity PI result

The Velocity P&I Wizard calculated the following values:

• Velocity P = 13120
• P Norm = 2
• Velocity I = 0
• I Norm = 3
• Biquad enabled with the Wizard-calculated parameters

I would like to confirm whether Velocity I = 0 is a valid result for this motor and whether the calculated PI or Biquad parameters could cause the unstable behavior.

Questions

1. What mechanical motion is expected during the Velocity Response test: continuous rotation, limited forward/reverse movement, or another motion?
2. What should normal Velocity Response and Current Response graphs look like?
3. Is the noisy ACTUAL_VELOCITY response expected, or does it indicate incorrect feedback scaling or unstable control?
4. Is an ACTUAL_VELOCITY value of approximately ±7 to 8 million internal units normal? How should it be converted to mechanical rpm?
5. Is Encoder Steps = 4096 correct for an MT6701 ABZ output configured for 1024 pulses per revolution? Does TMC9660 expect encoder lines, pulses, or quadrature counts per revolution?
6. Does the rpm value in the Wizard represent mechanical rpm or electrical rpm? Is any additional conversion required for a 7-pole-pair motor?
7. Is Velocity I = 0 a valid tuning result?
8. Could the intermittent high-speed rotation be caused by encoder direction, UVW phase order, electrical-angle alignment, initial rotor position, velocity scaling, PI/Biquad parameters, or another internal state?
9. Which fault flags, status values, parameters, or registers should I record to diagnose this issue?
10. What safe initial current, velocity range, and diagnostic sequence would you recommend for this motor?
11. Are there any known related issues in TMCL-IDE V4.12.0, Landungsbrücke firmware V3.11.7, or the current TMC9660 configurator?

I have also submitted an official ADI technical support case:

Case number: CS-594569-L3B4H6

Thank you for your help.

Figure 1. Velocity Response at 400 rpm — no unexpected high-speed rotation

Figure 2. Velocity Response at 619 rpm — unexpected high-speed rotation and ACTUAL_VELOCITY drops to approximately -7.6 million internal units

Figure 3. Current Response at 619 rpm — FOC_VOLTAGE_UQ reaches approximately 16383 and remains saturated

Figure 4. Velocity Response at 630 rpm — no unexpected high-speed rotation

Video 1. Actual motor behavior during the abnormal 619 rpm test