Analysis of PDT noise originating from driving method of a HDD spindle motor. Sangjin Sung, Gunhee Jang, Changjin Lee & Kyungjin Kang

Transcription

1 Analysis of PDT noise originating from driving method of a HDD spindle motor Sangjin Sung, Gunhee Jang, Changjin Lee & Kyungjin Kang Microsystem Technologies Micro- and Nanosystems Information Storage and Processing Systems ISSN Volume 18 Combined 9-10 Microsyst Technol (01) 18: DOI /s

2 Your article is protected by copyright and all rights are held eclusively by Springer- Verlag. This e-offprint is for personal use only and shall not be self-archived in electronic repositories. If you wish to self-archive your work, please use the accepted author s version for posting to your own website or your institution s repository. You may further deposit the accepted author s version on a funder s repository at a funder s request, provided it is not made publicly available until 1 months after publication. 1 3

3 Microsyst Technol (01) 18: DOI /s TECHNICAL PAPER Analysis of PDT noise originating from driving method of a HDD spindle motor Sangjin Sung Gunhee Jang Changjin Lee Kyungjin Kang Received: 30 September 011 / Accepted: May 01 / Published online: 16 June 01 Ó Springer-Verlag 01 Abstract This paper investigates the prominence discrete tone (PDT) noise originating from the driving method of a hard disk drive (HDD) spindle motor with 1 poles and 9 slots. Torque ripple of a HDD is reconstructed by the multiplication of measured back electromotive force (BEMF) and measured switching current. It shows that the frequency components of PDT noise match with those of torque ripple. It also investigates the frequency change of PDT noise due to two driving methods which have the different switching-off periods to detect the zero-crossing of the BEMF. The BEMF has odd harmonics of the number of pole pair, but the current has even harmonics of the number of pole pair due to the switching-off period as well as those odd harmonics of BEMF. We theoretically derive the torque equation in terms of the frequency components of BEMF and switching current. We also verify that the even harmonics of the driving current due to driving method generate the 54th harmonic of torque ripple, and show that the pure sinusoidal BEMF with the 6th harmonic can decrease the 36th harmonic of torque ripple as well as PDT noise in the HDD spindle systems. 1 Introduction A hard disk drive (HDD) is a source of the noise in a computer, and overall sound pressure level or sound power has been used to evaluate noise level of a HDD. However, nowadays the prominence discrete tone (PDT) noise, which S. Sung G. Jang (&) C. Lee K. Kang Department of Mechanical Engineering, Hanyang University, 17 Haengdang-dong, Seongdong-gu, Seoul , Republic of Korea ghjang@hanyang.ac.kr is irritating to human ear at specific frequency, is becoming an important indicator in evaluating the noise in a HDD, and it is becoming strictly regulated. The PDT noise is generated by resonance, manufacturing error and electromagnetic sources. One of the major sources is torque ripple whose characteristics are determined by cogging torque and current switching. Electromagnetic torque is generated by the multiplication of the driving current and the back electromotive force (BEMF). Most of drivers of HDD spindle motors generate the driving current with the sinusoidal shape by pulse width modulation (PWM) with the assumption that BEMF is sinusoidal. However, BEMF is not sinusoidal, and current has additional harmonics mainly generated switching-off period to detect the zero-crossing of BEMF, to identify the rotor position, to energize right phases and to control speed. Therefore, the discrepancy between driving current and BEMF generates other components of torque ripple in addition to cogging torque. It may ecite the HDD spindle system to generate the PDT noise. Many researchers have investigated the sources of the PDT noise in brushless DC (BLDC) motors, which can be divided into three categories: mechanical, aerodynamic, electromagnetic sources (Vijayraghavan and Krishnan 1999). Kim et al. (010) eperimentally identified the PDT noise of BLDC motor caused by a mechanical contact between shaft and bearing. Jung et al. (011) eperimentally showed the aerodynamic PDT noise originated from the polygon mirror in a scanner BLDC motor. However, the mechanical and aerodynamic PDT noise sources have significantly reduced in the conventional type of a HDD with fluid dynamic bearings (FDBs), so that the electromagnetic forces become a dominant noise source (e.g. Bi et al. 003; Jintanawan et al. 010). Many researchers have discussed that there is very close correlation between

4 1366 Microsyst Technol (01) 18: torque ripple and PDT noise (e.g. Kim et al. 011; Gao et al. 011). Several researchers analyzed the characteristics of the PDT noise due to the driving mode and proposed the driving mode to reduce the PDT noise (e.g. Lin et al. 003; Bi et al. 003; Soh and Bi 009). However, prior researchers did not investigate the characteristics of the PDT noise and torque ripple originating from the sensorless driving methods of a HDD to detect the zero-crossing of BEMF. This paper investigates the PDT noise originating from the driving method of a HDD spindle motor. Torque ripple of a HDD is reconstructed by the multiplication of measured BEMF and measured switching current. It investigates the correlation between the frequency components of PDT noise and those of torque ripple. It theoretically derives the torque equation in terms of the frequency components of BEMF and switching current. It also investigates the frequency change of PDT noise due to two driving methods which have the different switching-off periods to detect the zero-crossing of the BEMF, and then finally it proposes a method to decrease the PDT noise in a HDD spindle system. Characteristics of acoustic noise due to driving methods in a HDD This research analyzes the characteristics of the acoustic noise in a HDD spindle system due to the driving methods. The HDD spindle motor in this research has si pole pairs and nine slots, and it supports two.5 inch disks at 5,400 rpm. Figure 1 shows the control method of the HDD spindle system and the driving method A and B to identify the rotor position by detecting the zero-crossing of BEMF and to energize the right phase of the HDD spindle motor. In Fig. 1, e a, e b, and e c are the BEMF of each phase, and X(A) and X(B) are the signals generated from e a, e b, and e c for the driving method A and B to identify the rotor position and to energize the right phase of the HDD spindle motor. Driving method A and B detect the zero-crossing of the BEMF every 10 and 60, respectively. Figure shows the measured phase voltages controlled by driving method A and B by the eperiment setup as shown in Fig. 3a. The voltages cannot be applied to the whole energized period because the zero-crossing information of the BEMF needs to be detected in order to identify the rotor position for control purpose. Both the input and neutral voltages are simultaneously measured during energized periods. Driving method A detects the BEMF zero-crossing and controls the rotating speed once during 15 electrical degrees. However, driving method B detects the BEMF zero-crossing and controls the rotating speed twice during the front and back 10 electrical degrees. Since the spindle motor has 6 pole pairs, driving method A and B control the rotating speed 18 and 36 times per one mechanical revolution, respectively. Figure 4 shows the eperiment setup to measure the acoustic noise of a HDD in semi-anechoic chamber with background noise of 17.3 db-a. Figure 5 shows their acoustic noises and the overall sound pressure levels of driving method A and B. The overall sound pressure levels are similar, but the characteristics of the PDT noises are different. Driving method A has larger amplitude of the 36th harmonic than driving method B. However, driving method B has larger amplitude of the 54th harmonic than driving method A. 3 Characteristics of current due to driving method Electromagnetic torque ecluding cogging torque can be represented by the multiplication of BEMF and driving current as follows: T ¼ 1 X a;b;c k ðe k i k Þ ð1þ where T,, i k and e k are the electromagnetic torque, the angular velocity, the phase current and BEMF, respectively. The BEMF of the motor was measured by the eperiment setup as shown in Fig. 3a. Once the input voltage of the HDD rotating at 5,400 rpm is shut off, the HDD spindle motor slowly decelerated due to the inertia of the rotor with two disks. The BEMF and the rotating speed are simultaneously measured, and the BEMF at 5,400 rpm is calculated from the fact that the BEMF is proportional to the rotating speed. Figure 6 shows the BEMF at 5,400 rpm and its frequency spectrum. The BEMF has odd harmonics of the number of pole pair. Figures 7 and 8 show the currents and their frequency spectra according to driving method A and B. The currents are measured by the eperiment setup as shown in Fig. 3b. The voltage equation of a BLDC motor can be epressed as follows: V e ¼ L di dt þ Ri ðþ where V, L and R are the voltage, the inductance and the phase resistance, respectively. Equation () shows that the BEMF induces the same harmonics to the current of the motor, so that the 30th and 4nd harmonics of the current are generated from the 30th and 4nd harmonics of the BEMF (Hung and Ding 1993). The 1th, 4th and 48th harmonics are generated from the switching-off period to detect the BEMF zero-crossing, and driving method B generates larger amplitudes on those harmonics than driving method A.

5 Microsyst Technol (01) 18: Fig. 1 Control method of the HDD spindle system and the driving method A and B by detecting the zero-crossing of BEMF Fig. Measured phase voltages of the driving method (a) A and (b) B Fig. 3 Eperiment setup to measure (a) the phase voltage and the BEMF, and (b) the driving current due to driving method 4 Characteristics of torque ripple due to driving method Figures 9 and 10 show the torque ripples and their frequency spectra due to driving method A and B calculated from the measured BEMF and current in Eq. (1). Driving method A generates larger amplitude of the 36th harmonic of torque ripple than driving method B, but driving method B generates larger amplitude of the 54th harmonic of torque ripple than driving method A. The frequency characteristics of the torque ripples are very close to those of the PDT noises. Since driving method B controls the rotating

6 1368 Microsyst Technol (01) 18: Fig. 4 Eperiment setup to measure the acoustic noise of the HDD due to driving method speed 36 times per revolution, the driving method B generates smaller amplitude of the 36th harmonic of the torque ripple than driving method A. After the BEMF and the current are decomposed by using the Fourier series, the torque can be epressed as follows: TðtÞ m;n ¼ E mi n 1 ðm nþp þ cos 3N p 1 ðm þ nþp þ cos 3N p cosððm nþtþ cosððm þ nþtþ ð3þ where m, n, E m and I n are harmonics and amplitudes of BEMF and current respectively, and N p is the number of pole pairs. Equation (3) shows which harmonics of the BEMF and the current contribute to the harmonics of torque ripple. Table 1 shows the harmonics of torque ripple induced by the harmonics of BEMF and current. The 54th harmonic of torque ripple is generated from the 1th, 4th and 48th harmonics of the current. Therefore, the driving method B generates larger amplitude of the 54th harmonic of the torque ripple than the driving method A, because the driving method B generates larger amplitudes of the 1th, 4th and 48th harmonics of the current than the driving method A, as shown in Fig. 8. Table 1 shows that the 30th and 4nd harmonics of the BEMF affect all harmonics of the torque ripple. Pure sinusoidal BEMF without the 30th and 4nd harmonics is Fig. 5 Sound spectra due to the driving method (a) A and (b) B of the HDD

7 Microsyst Technol (01) 18: Fig. 6 a Measured BEMF and b its frequency spectrum of the HDD Fig. 7 Measured currents due to the driving method (a) A and (b) B of the HDD Fig. 8 Frequency spectra of the measured current due to the driving method (a) A and (b) B of the HDD Fig. 9 Measured torque ripple due to the driving method (a) A and (b) B of the HDD

8 1370 Microsyst Technol (01) 18: Fig. 10 Frequency spectra of the measured torque ripple due to the driving method (a) A and (b) B of the HDD Table 1 Harmonics of torque ripple generated from the harmonics of BEMF and current Harmonics of current Harmonics of BEMF 6th 30th 4nd 6th 3 E 6I 6 3 1th 3 E 6I 1 cosð18tþ 3 4th 3 E 6I 4 cosð18tþ 3 30th 3 E 6I 30 cosð36tþ 3 4nd 3 E 6I 4 cosð36tþ 3 48th 3 E 6I 48 cosð54tþ 3 E 30I 6 E 30I 1 cosð36tþ 3 E 4I 6 cosð18tþ 3 E 30I 4 cosð54tþ 3 E 4I 4 E 30I 30 3 E 30I 4 cosð7tþ 3 E 4I 4 E 30I 48 cosð18tþ 3 cosð36tþ E 4I 1 cosð54tþ cosð18tþ E 4I 30 cosð7tþ E 4I 48 cosð90tþ Fig. 11 Torque ripple of the driving method (a) A and (b) B due to pure sinusoidal BEMF Fig. 1 Frequency spectra of torque ripple of the driving method (a) A and (b) B due to pure sinusoidal BEMF

9 Microsyst Technol (01) 18: Table Comparison of the amplitudes of major harmonics of the torque ripples due to the BEMF shape Harmonic of torque ripple assumed to be applied to Eq. (1). Figures 11 and 1 show the torque ripples and their frequency spectra with the pure sinusoidal BEMF, and Table shows the comparison of the amplitudes of major harmonics. The 36th harmonic of the torque ripples is almost suppressed and most of amplitudes are decreased. In this case, the driving method A is better method to decrease torque ripple than the driving method B, because the amplitudes of the harmonics of torque ripple generated by the driving method A are smaller than those generated by the driving method B. It also shows that the sinusoidal shape of BEMF is an efficient method to reduce the 36th harmonic (6th harmonic of the number of pole pairs) irrespective of driving method. 5 Conclusions Amplitude (mnm) Measured BEMF Pure sinusoidal BEMF Reduction (%) Driving method A 18th th th nd Driving method B 18th th th nd This research investigates eperimentally the characteristics of PDT noises in a HDD due to driving method. Additional harmonics of currents are generated by the driving methods to detect the BEMF zero-crossing and to control the speed of the HDD spindle system. They affect the characteristics of the torque ripple and the PDT noises. We also show that the pure sinusoidal BEMF is an efficient method to reduce torque ripple and PDT noise irrespective of driving method. This research can contribute to the reduction of the PDT noises in a HDD. Acknowledgments This research was performed at Samsung- Hanyang Research Center for Precision Motors sponsored by Samsung Electro-Mechanics Co. Ltd. References Bi C, Jiang Q, Lin S, Low TS, Mamun AA (003) Reduction of acoustic noise in FDB spindle motors by using drive technology. IEEE Trans Magn 39: Gao F, Lin WZ, Lee CW, Phyu HN, Mou JQ, Ong EH (011) Motor noise generation and prediction for high spinning speed hard disk drive. Microsyst Technol 17:81 89 Hung JY, Ding Z (1993) Design of currents to reduce torque ripple in brushless permanent magnet motors. IEE Proc B 4:60 66 Jintanawan T, Sillapapinij A, Suwankawin S (010) Eploration of vibro-acoustic characteristics of HDD spindle motors using electro-magnetic frequency-swept ecitation. Microsyst Technol 16: Jung YH, Jang GH, Kim MG, Cinar Y (011) Reduction of aeroacoustic noise originated from a high-speed polygon mirror scanner motor of a laser beam printer utilizing air-flow analysis. Microsyst Technol 17: Kim MG, Jang GH, Lee CJ, Lim DO (010) Eperimental identification of abnormal noise and vibration in a high-speed polygon mirror scanner motor due to mechanical contact of plain journal bearing. Microsyst Technol 16:3 8 Kim KS, Lee CM, Hwang GY, Hwang SM (011) Effect of the number of poles on the acoustic noise from BLDC motors. J Mech Sci Technol 5:73 77 Lin S, Jiang Q, Mamun AA, Bi C (003) Effect of drive modes on the acoustic noise of fluid dynamic bearing spindle motors. IEEE Trans Magn 39: Soh CS, Bi C (009) Sensorless optimal sinusoidal brushless direct current for hard disk drives. J Appl Phys 105: 07F118 07F118-3 Vijayraghavan P, Krishnan R (1999) Noise in electric machines: a review. IEEE Trans Ind Appl 35: