Diagnosing ±2 RPM Speed Ripple on SINAMICS S120 1FK7

1. Problem Overview: SINAMICS S120 with 1FK7 Servo Motor Speed Ripple

When commissioning a SINAMICS S120 booksize drive system with a CU320-2 PN Control Unit driving a 1FK7 synchronous servo motor under no-load conditions, the actual speed display (r63) typically shows a periodic deviation from the setpoint. A common field observation is a speed window of ±2 RPM around a 1000 RPM setpoint, which raises the question of whether the fluctuation represents an actual mechanical speed error, a measurement artifact, or a control-loop issue.

This article documents the diagnostic procedure and root-cause analysis for this behavior on a system with the following components:

The key engineering question during commissioning is: is ±2 RPM a normal encoder-derived measurement artifact, or does it indicate a real speed fluctuation requiring remediation? Field tests at Siemens branch offices, comparing a 1FT7 motor (2048 pulses/revolution, ±0.15 RPM) and the 1FK7 motor (512 pulses/revolution, ±2 RPM) on the same drive platform, demonstrate that the bulk of the observed ripple is encoder-resolution dependent and not a control-loop malfunction.

2. Understanding r61 and r63: Two Distinct Speed Signals

The SINAMICS S120 firmware computes the actual speed in two separate paths with different post-processing characteristics. Misidentifying which parameter is being displayed or traced is a common source of confusion during commissioning.

The relationship between r61 and r63 is governed entirely by p1441:

• If p1441 = 0 (default in many configurations): r63 = r61. Both parameters display the raw encoder value.
• If p1441 > 0: r63 is a low-pass-filtered version of r61, with the time constant in milliseconds set by p1441.

The raw value r61 carries every artifact of the incremental encoder evaluation chain: sinusoidal signal maxima and minima, analog zero crossings, and the quantization effect of the pulse count multiplied by the fine-resolution bits. This is why even a perfectly tuned and well-commissioned drive loop will show oscillation in r61 when traced with high time resolution. The oscillation is not a fault; it is a faithful representation of the encoder's discrete output.

3. Encoder Pulse Calculation and Resolution

The encoder pulse number stored in p408 is the line count per mechanical revolution of the motor's integrated encoder disc. The fine resolution in p418 determines the bit count of the interpolation stage that divides each analog sin/cos period into discrete increments inside the Sensor Module (SMx) or CUA32.

For the 1FK7083-2AC71-1RG1 motor referenced in this analysis:

For the comparison 1FT7 motor (2048 pulses, 11-bit fine resolution), the total is 2048 × 2048 = 4,194,304 increments per revolution — exactly four times the resolution of the 1FK7. This resolution ratio (4:1) directly explains the observed speed-band ratio on r61 (1FT7 ≈ ±0.15 RPM vs 1FK7 ≈ ±2 RPM, a factor of ≈ 13; the additional factor beyond the 4× pulse-ratio advantage comes from the 1FT7's lower cogging torque, which removes a separate motor-dependent ripple contribution).

The position deviation visualized with the LSM (Limit-Switch / Motion) function in the trace tool yields the angular error directly:

Position deviation [°] = (Δinc / (p408 × 2^p418)) × 360°

A 100-increment deviation on a 1FK7 (p408 = 512, p418 = 11) equals 100 / 1,048,576 × 360° ≈ 0.034° — an extremely small angular error that does not produce visible mechanical vibration or affect workpiece quality in any but the most demanding precision applications.

4. Root Causes of Speed Ripple in 1FK7 Systems

The ripple observed in r61 has multiple contributing sources. To isolate them, trace r61 at a slow setpoint (60 RPM = 1 Hz mechanical = 1 rev/s) and apply a Fast Fourier Transform (FFT) to the captured buffer. This separates the dominant frequency components into distinct bins, each associated with a different physical source.

5. Diagnostic Workflow: SINAMICS Trace and FFT Analysis

The following step-by-step procedure isolates the source of speed ripple on a 1FK7 system. It is suitable for execution with STARTER or SINAMICS Startdrive commissioning software.

1. Set a low speed setpoint. Command 60 RPM (1 Hz mechanical). This slows the mechanical frequency components and separates the encoder harmonics in the FFT spectrum. Do not exceed 200 RPM during the diagnostic run.
2. Configure the trace buffer in STARTER / Startdrive:
   • Trigger mode: free-running (no trigger)
   • Duration: 1000 ms
   • Sampling time: 125 µs (current controller cycle, p0115[0])
   • Signals to record: r61 (raw actual speed), r479 (actual position value)
3. Record r61 in the time domain. Inspect the waveform. Expect a high-frequency riding oscillation (encoder-line frequency) superimposed on a near-constant average value (the commanded speed). Note the peak-to-peak amplitude.
4. Apply FFT to the r61 buffer. Identify the dominant spectral peaks. Look for:
   • 512 Hz (PPR fundamental of the 1FK7 encoder)
   • 1024 Hz and 2048 Hz (sin/cos harmonics)
   • 18 Hz (cogging at 18 slots × 1 rev/s)
   • 1 Hz and 2 Hz (mechanical / coupling misalignment)
5. Record r479 actual position in parallel. Use the trace tool's mathematical LSM function to derive the deviation from an ideal linear position curve. Convert the resulting peak deviation from increments to mechanical degrees using the formula in Section 3.
6. Record r2169 (monitoring-smoothed speed) for comparison. Set p2153 = 1 ms to enable the monitoring PT1 filter; trace r2169 in the same window at the same scale. The difference between r61 and r2169 represents the encoder-noise contribution alone, isolated from any mechanical or motor-related disturbance.
7. Interpret the FFT spectrum. If the 512 Hz peak dominates and the 18 Hz peak is small, the encoder is the dominant source. If the 18 Hz peak dominates, the motor cogging is the dominant contributor and a 1FT7 swap is the recommended remediation. If non-periodic disturbances appear, check cable routing and shielding (Section 8).

6. Position Deviation Analysis Using the LSM Function

The r479 parameter is the raw actual position value in encoder increments (range: 0 to p408 × 2^p418). Unlike r61, r479 is not differentiated over time; therefore, it does not amplify encoder quantization noise into the speed domain. r479 is the proper diagnostic signal when the application is sensitive to position errors (machining, indexing, printing registration).

Using the trace tool's mathematical functions, plot the deviation of r479 from a mathematically ideal linear position ramp over the trace window. This deviation, expressed in increments, converts directly to mechanical angle:

Δθ [°] = (Δinc / (p408 × 2^p418)) × 360°

For a 1FK7 with 1,048,576 inc/rev, a peak deviation of 100 inc equals 0.034° of mechanical angle. Even a peak deviation of 1000 inc (0.34°) — an order of magnitude larger than typical — is invisible on most machined surfaces and is irrelevant for general positioning tasks. This is the meaningful error to assess on the workpiece, not the speed display.

If the application requires sub-arc-minute position accuracy, the proper response is to upgrade the motor (e.g., to 1FT7 with 2048 PPR) or to add an external high-resolution encoder on the load side (load-side position control via SMC30 or similar Sensor Module), not to chase the r61 speed display.

7. Smoothing Strategies: Monitoring Filter vs. Control-Loop Filter

Two distinct smoothing mechanisms exist in the SINAMICS S120 firmware, and the choice between them has direct consequences for closed-loop behavior. Selecting the wrong filter can degrade the achievable bandwidth and introduce new oscillation modes downstream.

Recommended diagnostic procedure:

1. Leave p1441 at the value established during commissioning (typically 0 or a small value such as 0.1 ms).
2. Set p2153 = 1 ms to activate the monitoring filter.
3. Trace r61 (raw) and r2169 (monitoring-smoothed) simultaneously at identical vertical scale.
4. Compare the two waveforms. The difference represents the encoder-noise component alone, isolated from any control-loop interaction.

A 1 ms PT1 low-pass filter has a corner frequency of f_c = 1 / (2π × T) ≈ 159 Hz. At 512 Hz (the encoder-line fundamental), the attenuation is approximately 20·log₁₀(√(1 + (512/159)²)) ≈ 20·log₁₀(3.32) ≈ 10.4 dB per stage. Higher harmonics (1024 Hz, 2048 Hz) are attenuated more strongly. A 1 ms time constant is sufficient to remove the visible encoder ripple from r2169 without introducing perceptible lag into the operator display.

8. Cable Routing, Shielding, and EMI Considerations

In addition to the deterministic encoder-related ripple, field installations can introduce noise coupled into the encoder feedback cable from the parallel power cable. This produces non-periodic disturbances superimposed on r61 that differ from the deterministic sin/cos ripple described above. Coupling-related ripple is typically bursty, varies with load, and may correlate with the power module switching edges.

For further reading on speed fluctuation sources in industrial drives, refer to peer-reviewed analysis such as the IET paper Analysis of voltage fluctuation impact on induction motors, which documents how supply-side disturbances produce additional spectral content in the stator current that couples into the speed signal through the torque-producing current loop.

If p1300 (open-loop/closed-loop control mode) is set to a sensorless open-loop mode (e.g., p1300 = 0, 1, 2, 20, or 22 in V/f variants), the actual speed is reconstructed from the motor model rather than the encoder. In that case, the encoder-ripple source disappears entirely and the observed ±2 RPM is replaced by a much smaller model-error component. Verify p1300 with the commissioning engineer before assuming encoder-dependent ripple.

9. Verification and Acceptance Criteria

The acceptance test for a 1FK7 system should distinguish three separate criteria. Conflating them leads to unnecessary re-commissioning or motor swaps.

A practical field check: load the motor with a small inertia disc or coupling and observe r61. If the ripple magnitude remains unchanged, the source is encoder electronics. If the ripple reduces dramatically when load inertia is added, the source is dominantly mechanical or motor cogging. This test takes less than five minutes and does not require disconnecting the encoder cable.

10. 1FK7 vs 1FT7 Comparison and Motor Selection Guidance

For applications where low encoder noise is critical, motor selection matters as much as drive commissioning. The 1FK7 and 1FT7 series serve different market segments within the Siemens synchronous servo portfolio.

If the application cannot tolerate ±2 RPM in the display but does not require sub-degree position accuracy, the proper response is to clarify which signal is being observed (r61 vs r63 vs r2169) and apply the appropriate monitoring filter (p2153 = 1 ms) — not to substitute the motor. If the application requires sub-degree position accuracy or sub-arc-minute following error, the 1FT7 is the correct platform choice.

11. Sampling Rate, Trace Window, and FFT Bin Resolution

Diagnosing speed ripple correctly requires the trace configuration to match the frequency band of interest. Insufficient sampling rate aliases the high-frequency encoder harmonics into the low-frequency spectrum, producing misleading FFT results.

For the standard ±2 RPM ripple diagnostic on a 1FK7, the 1000 ms / 125 µs configuration is sufficient. The 8000 samples per channel yield a 1 Hz FFT bin, cleanly separating 1 Hz (mechanical), 18 Hz (cogging), and 512 Hz (encoder fundamental) into non-overlapping bins.

12. Diagnostic LED States on the CU320-2 PN

When commissioning a CU320-2 PN, the on-board LEDs provide coarse status information that complements the trace analysis.

If the RDY LED is not steady green during the ripple diagnostic, resolve the drive-object fault state before drawing conclusions from trace data. Fault codes can be read from the fault buffer (parameter r0945 / r0947) and cross-referenced against the SINAMICS S120 List Manual fault-code table.

13. Safety Function Interaction with Trace Recording

The SINAMICS S120 trace tool operates in a non-intrusive observation mode; it does not modify the closed-loop control path or disable any safety functions. However, several precautions apply during fault-state diagnostics:

• STO (Safe Torque Off): When STO is active (e.g., via terminals X131 or PROFINET PROFIsafe), the drive is in torque-free state and r61 reflects zero speed. Trace records captured during STO are not representative of normal operation.
• SS1 (Safe Stop 1): During SS1, the drive follows a controlled deceleration ramp. Trace records show the deceleration profile, not steady-state ripple.
• SBC (Safe Brake Control): The brake control outputs are independent of the trace tool. Ensure the motor is mounted safely (mechanically braked or coupled) when running the diagnostic at non-zero speed.
• Active faults: If the drive is in a fault state, the trace buffer may be invalidated. Clear the fault, acknowledge the drive, and re-record.

For more detailed safety integration guidance, refer to the SINAMICS S120 Safety Integrated Function Manual available via Siemens Industry Online Support and the 1FK7 motor documentation in the same portal.

14. Frequently Asked Questions