EV Motor NVH Testing: ضوضاء, Vibration and Harshness Measurement Guide

بواسطة فريق هندسة EconoTest · الشركة المصنعة لمنصة الاختبار, شنغهاي

Why NVH Has Become Critical for EV Motors

In an internal combustion engine vehicle, engine noise, exhaust sound, and road noise mask the acoustic signature of the electric motor and gearbox. In a battery electric vehicle, those masking sources are gone. Motor whine, gear rattle, and electromagnetic hum become the dominant acoustic experience in the cabin — and increasingly the primary customer quality complaint. NVH (ضوضاء, Vibration and Harshness) testing for EV motors has moved from a desirable-to-have to a mandatory development gate at most major OEMs.

Sources of Motor Noise and Vibration

Electromagnetic Noise

The most significant noise source in PMSM and induction motors is the electromagnetic force between the stator teeth and rotor magnets (or rotor flux). These forces are periodic at the spatial harmonic frequencies determined by the pole and slot combination. For a 12-pole, 36-slot motor, the dominant electromagnetic force occurs at 12× electrical frequency and its harmonics. At 3,000 دورة في الدقيقة (50 Hz electrical), this generates a 600 Hz tone — clearly audible and often described as motor whine.

PWM switching from the inverter adds additional noise at the switching frequency (typically 4–16 kHz) and its sidebands. These high-frequency components are less objectionable to most listeners but can cause fatigue in thin panels and excite resonances in motor housing components.

Mechanical Noise

Bearing noise, rotor imbalance vibration, and gear mesh noise (from the motor’s integral reducer) are mechanical sources. Bearing noise appears at specific frequencies calculated from bearing geometry and shaft speed (BPFO, BPFI, BSF, FTF frequencies). Gear mesh noise appears at tooth-pass frequency (speed × tooth count) and its harmonics. A new, well-aligned motor should have no prominent gear mesh peaks above the broadband floor; elevated gear mesh tones indicate pitch error, eccentricity, or inadequate gear contact.

Structural Resonances

Even small electromagnetic or mechanical excitation forces can produce large noise and vibration levels if they coincide with a structural resonance of the motor housing, stator stack, or mounting bracket. Modal analysis identifies natural frequencies; the design challenge is ensuring no operating-speed harmonic aligns with a mode within ±20% of its frequency.

NVH Test Methods on the Dynamometer

Vibration Signature Analysis (VSA)

Mount accelerometers on the motor housing at bearing locations (drive end and non-drive end) and on the dynamometer gearbox housing if a reducer is present. Run the motor from minimum to maximum speed under constant load, recording vibration time-domain data continuously. Post-process with FFT to produce a waterfall plot (frequency vs speed vs amplitude). This reveals order-tracked components (rising with speed) and fixed-frequency resonances (constant frequency regardless of speed).

VSA on the test bench is the primary tool for production NVH quality control. A go/no-go specification can be defined based on amplitude limits at key frequencies. A motor exceeding the limit at 6× electrical frequency indicates rotor asymmetry; a peak at gear mesh frequency indicates gear quality issue.

Sound Power Level Measurement (ISO 1680)

ISO 1680 is the primary standard for electric motor acoustic noise measurement. It requires either a reverberant test room (sound power from sound pressure level + room constant) or a semi-anechoic chamber (direct free-field measurement). The motor is run at rated load and speed; sound pressure level is measured at a standard measurement surface around the motor. Sound power level L_W is calculated from average SPL across the surface, independent of room geometry.

This test is typically performed in a dedicated acoustics laboratory, not on the main production test bench. لكن, preliminary screening on the production bench using a simpler microphone array can flag outlier units before they reach the acoustic lab.

Order Tracking Analysis

Order tracking records vibration or sound data synchronously with motor speed so that harmonic components are expressed as multiples of rotation speed (orders) rather than absolute frequency. A 2nd-order peak means a vibration at exactly 2× rotational speed, regardless of what that frequency is at any given moment. Order tracking is particularly useful for identifying rotor imbalance (1st order), electromagnetic forces (order = 2P where P = pole pairs), and gear mesh (order = tooth count).

Key Metrics and Pass/Fail Criteria

Metric	Typical EV Motor Specification	Standard / Method
Sound power level at rated load	<65–75 dB(أ) depending on power class	ISO 1680
Housing vibration (velocity RMS)	<2.8 mm/s RMS (ISO 10816 Class I)	ISO 10816-1
Gear mesh order amplitude	OEM-specified; typically <0.5 m/s² at mesh order	Order tracking, OEM spec
Electromagnetic noise order	<50 dB(أ) SPL at pole × slot harmonic	Sound pressure map, OEM spec
No-load idle vibration	<1 mm/s RMS	ISO 10816-1

Test Bench Requirements for NVH Testing

NVH testing imposes additional requirements on the dynamometer test bench beyond standard performance testing:

• Vibration-isolated base plate: the test bench must not inject dynamometer vibration into the motor housing. Air mounts or elastomeric isolators are required between the bench base plate and the building floor
• Low-noise coupling: flexible couplings between the motor output shaft and the torque transducer must be NVH-neutral. Standard jaw couplings introduce tooth-mesh noise at coupling element frequency; disc couplings or membrane couplings are preferred for NVH testing
• Background noise floor: the test room background noise should be at least 6 dB below the motor’s lowest expected noise level. أ 60 dB(أ) motor requires a <54 dB(أ) background — achievable in a treated industrial room but not an open factory floor
• Speed sweep capability: the dynamometer must execute a smooth, controlled speed sweep from idle to maximum speed in 60–120 seconds for waterfall plot acquisition. Step changes cause transient vibration that contaminates the spectrum

الأسئلة المتداولة

What is the difference between NVH and acoustic emission testing?

NVH (ضوضاء, Vibration and Harshness) is a broad term covering all sensory aspects of mechanical noise: airborne sound (ضوضاء), structural vibration (اهتزاز), and the perception of roughness or transients (harshness). Acoustic emission (AE) testing is a specific non-destructive testing technique that detects ultrasonic stress waves from crack propagation or surface contact — a completely different discipline. In the motor testing context, NVH testing is always the intended term for quality-of-sound testing.

How is EV motor NVH different from traditional motor NVH?

Traditional industrial motors are assessed against ISO 10816 vibration severity limits and ISO 1680 sound power standards at fixed rated operating points. EV motor NVH must cover the full operating range (0 to maximum speed and torque) because the vehicle operates across this range continuously. Additionally, EV applications require psychoacoustic metrics (tonality, roughness, sharpness) alongside physical dB measurements, because cabin occupants perceive tonal sounds (like motor whine) as more annoying than broadband noise at the same dB level.

Can NVH testing be integrated with efficiency and performance testing?

نعم, and this is the most cost-effective approach. A test bench with accelerometers and a microphone array already installed can run vibration signature analysis as a background measurement during the efficiency map acquisition — no extra test time required. The VSA data validates mechanical health while the torque/speed/power data quantifies performance. Full acoustic testing to ISO 1680 requires a separate acoustic environment, but production-line vibration screening integrates with the performance test with minimal overhead.

What sampling rate is needed for EV motor NVH measurement?

To capture PWM switching harmonics (4–16 kHz fundamental, up to 50th harmonic at 200–800 kHz), data acquisition must run at 200 kHz minimum, with anti-aliasing filters set to half that. For gear mesh and electromagnetic harmonics in the audible range (20 Hz–20 kHz), 51.2 kHz sampling rate (standard for audio-grade DAQ systems) is sufficient. Most EV NVH programs use a two-channel approach: audio-grade DAQ (51.2 kHz) for the audible band and high-speed oscilloscope or specialized EMI receiver for the PWM-switching range.

Does a louder motor always mean lower quality?

Not necessarily — but tonal noise (a single prominent frequency, like a motor whine) is disproportionately annoying compared to broadband noise at the same overall dB level. A motor that is 2 dB louder in overall level but has no tonal components often sounds more acceptable than a quieter motor with a prominent 800 Hz whine. NVH assessment must include psychoacoustic analysis (tonality index, loudness in sone) alongside simple dB measurements. The EV OEM NVH targets are increasingly written in psychoacoustic units rather than physical ones for this reason.