Analysis of the electromagnetic acoustic noise and vibrations of a high-speed brushless DC motor

Transcription

1 Analysis of the electromagnetic acoustic noise and vibrations of a high-speed brushless DC motor J. Le Besnerais, Q. Souron, E. Devillers EOMYS ENGINEERING, 1

2 A. Introduction Square-wave-driven BLDC motor Surface PM, p=2 pole pairs Concentrated tooth winding, Zs=12 stator slots Very noisy! Objectives: theoretical, numerical and experimental analysis of vibroacoustic behaviour at no-load 2

3 B. Theoretical analysis: mechanical vibrations Bearing imperfections create vibrations proportional to rotational frequency f r Irregularities of the ball cages create vibrations at 0.36 f r 3

4 B. Theoretical analysis: magnetic vibrations Maxwell forces apply to both stator and rotor, they are proportional to airgap flux density squared Flux decomposition in permeance / magnetomotive force for harmonic analysis: wave of frequency f and wavenumber r noted (r,f) Lowest positive wavenumber: r=gcd(zs,2p)=4 at multiple of 2f s Pulsating radial & tangential (cogging) forces: r=0 at multiples of LCM(Zs,2p)f R 4

5 B. Theoretical analysis: magnetic vibrations Highest force has a wavenumber r=2p=4 Due to high speed operation this force wave can 5

6 C. Numerical analysis: introduction MANATEE electromagnetic & vibro-acoustic software Use of fast & accurate subdomain models for electromagnetics Simulation are carried up to 20 khz in a few seconds 6

7 C. Numerical analysis: ideal case Stator elliptical mode (2,0) found at 8 khz Rotor first bending mode (1,0) found at 2 khz Main radial force and vibration occurs at r=4 f=2f s Radial force spectrogram and radial vibration spectrograms: 7

8 C. Numerical analysis: ideal case The spectrum is less rich than the experimental one There is no strong resonance contrary to experiments, and no excitation of the elliptical mode (2,0) The ideal case is not realistic: eccentricities, pole displacement must be included 8

9 C. Numerical analysis: 10% dynamic eccentricities 9

10 C. Numerical analysis: rotor vibration due to cage defaults 10

11 C. Numerical analysis: rotor pole displacements 11

12 C. Numerical analysis: all imperfections 12

13 D. Experimental analysis Run-up with microphones & accelerometers «Spatiogram technique»: the radial vibration waves are filtered according to their wavenumber to identify both frequencies f and wavenumbers r 8 accelerometers are used -> up to r=4 can be identified (Shannon) 13

14 D. Experimental analysis Rich vibration spectrum with sidebands at 0.37 f r Resonance with mode 1 identifed as rotor bending mode with ODS at 2 khz Resonance with mode 2 identifed as stator elliptical mode with ODS at 8 khz 14

15 D. Experimental analysis r=1 r=2 Mode 1 Spatiograms confirm theoretical & numerical results Strong presence of wavenumber 1 is due to assymetrical pole placement and dynamic eccentricities 15

16 D. Experimental analysis r=4 16

17 E. Conclusions MANATEE software allows to quickly identify the vibroacoustic impact of imperfections in terms of frequencies and wavenumbers Spatiogram experimental technique allows characterizing the vibration waves responsible for noise in terms of frequency, wavenumber and rotation direction Studied BLDC is noisy due to uneven pole spacing and dynamic eccentricity 17

18 Thank your for your attention, any questions? EOMYS ENGINEERING 18