VELOCITY AND VIBRATIONS MEASUREMENT IN A COMPUTER HARD DISK DRIVE

Transcription

1 VELOCITY AND VIBRATIONS MEASUREMENT IN A COMPUTER HARD DISK DRIVE submitted by Jeswin Ching Department of Mechanical Engineering In partial fulfillment of the Requirement for the Degree of Bachelor of Engineering National University of Singapore Session 2004/2005

2 Abstract A computer hard disk drive (HDD) plays an important role in storage of important data in many industries as well as homes. As the need for faster and increasing capacity but smaller in size of a HDD increases, the reliability of a HDD becomes all the more important, especially preventing errors due to track mis-registration (TMR). In order to avoid TMR due to air induced vibrations of the actuator arm, the flow patterns within a HDD need to be investigated. In this project, flow velocities within a commercial, single platter 2.5 inch HDD are measured using the Laser Doppler Anemometer and the flow patterns are visualized. The effects of varying both the actuator arm s position and the motor s rotational speed on the flow patterns are observed. The vibrations of the actuator arm are measured using the Laser Doppler Vibrometer. Modes of vibrations and the magnitude of each mode are plotted and investigated. From observations, the modes and the magnitude of vibrations change when varying the actuator arm s position and the rotational speed of the motor. An estimated maximum measured TMR for each rotational speed is calculated and compared with the allowable TMR. From this project, it is found that the main cause of flow induced vibrations is due to excitation by vortices that are formed when air flows pass the actuator arm. 2

3 Acknowledgement I would like to express my sincere gratitude to my supervisor of the project, Associate Professor Lim Siak Piang and project collaborator Dr. Yip Teck Hong from Data Storage Institute (DSI) for offering their invaluable advice, knowledge and guidance. Special thanks are also extended to DSI engineers Mr. Albert Tan Chok Shiong for his guidance on the operations of the Laser Doppler Vibrometer. Lastly, I would like to thank all other people who have contributed to the completion of this project in any aspect. 3

4 Table of Contents ABSTRACT... 2 ACKNOWLEDGEMENT... 3 TABLE OF CONTENTS... 4 LIST OF FIGURES... 6 LIST OF TABLES ) INTRODUCTION ) BACKGROUND ) OBJECTIVES ) TECHNICAL APPROACH ) LITERATURE REVIEW AND THEORETICAL BACKGROUND ) AERODYNAMIC FORCES ) EXPERIMENTAL SETUP AND INSTRUMENTATION ) EXPERIMENTAL SETUP ) EXPERIMENTAL PROCEDURES ) Laser Doppler Anemometer (LDA) ) Basic Procedures to Setup the LDA ) Laser Doppler Vibrometer (LDV) ) Basic Procedures to Setup the LDV ) RESULTS AND DISCUSSIONS ) FLOW PATTERN IN HDD ) Streamlines for OD, MD and ID positions at 2800rpm ) Streamlines for OD, MD and ID positions at 5400rpm ) Streamlines for OD, MD and ID positions at 7200rpm ) Effects of Varying Actuator Positions on the Flow Pattern ) Effects of Increased Rotational Speed on the Flow Pattern

5 4.1.5) Applications ) VIBRATIONS OF THE ACTUATOR ARM ) Effect of Varying Actuator Position on the Vibrations ) Effect of Varying Rotational Speed on the Vibrations ) Displacements of Individual Modes of Vibrations ) Applications ) CONCLUSION ) RECOMMENDATIONS ) REFERENCES APPENDIX A EXPERIMENTAL SETUP APPENDIX B DOPPLER EFFECT APPENDIX C LASER DOPPLER ANEMOMETER

6 List of Figures Figure 3.1: The 2.5 inch HDD..14 Figure 3.2: The Grid System...15 Figure 3.3: Conversion of Time Domain to Frequency Domain...18 Figure 3.4: Laser Head of the LDV..19 Figure 4.1: OD at 2800rpm...21 Figure 4.2: MD at 2800rpm..21 Figure 4.3: ID at 2800rpm 21 Figure 4.4: OD at 5400rpm..22 Figure 4.5: MD at 5400rpm..22 Figure 4.6: ID at 5400rpm 22 Figure 4.7: OD at 7200rpm..23 Figure 4.8: MD at 7200rpm..23 Figure 4.9: ID at 7200rpm 23 Figure 4.10: Horizontal Measurement at 5400rpm...27 Figure 4.11: Horizontal Measurement at 7200rpm...27 Figure 4.12: Vertical Measurement at 5400rpm...28 Figure 4.13: Vertical Measurement at 7200rpm...28 Figure 4.14: Horizontal Measurement at ID position...31 Figure 4.15: Horizontal Measurement at MD position.31 Figure 4.16: Horizontal Measurement at OD position..31 Figure 4.17: Vertical Measurement at ID position...32 Figure 4.18: Vertical Measurement at MD position.32 Figure 4.19: Vertical Measurement at OD position..32 6

7 Figure 4.20: Horizontal Displacement at 5400rpm..33 Figure 4.21: Horizontal Displacement at 7200rpm..34 Figure 4.22: Vertical Displacement at 5400rpm...34 Figure 4.23: Vertical Displacement at 7200rpm...35 List of Tables Table 4.1: Estimated Maximum Displacement for Horizontal Measurement..34 Table 4.2: Estimated Maximum Displacement for Vertical Measurement..34 7

8 1) Introduction 1.1) Background Hard disk drive (HDD) is one of the most important components that we rely on when it comes to using a computer. As time passes and as technology improves, HDD companies are able to push the limits of the hard disk to far beyond what was once imagined. With the continuing increase in storage density and spindle rotation speed as well as the decrease in size of a HDD, suspension vibration becomes a significant challenge to deal with in order to avoid track mis-registration (TMR) by the actuator arm and the head from crashing on the disk platter. Therefore, there is a need to better understand the vibration characteristic of the actuator arm due to air flow. This will end up increasing the efficiency and reliability of a HDD. 1.2) Objectives The main aim of this project is to give future researchers an idea on the airflow characteristics in a commercial 2.5 inch, single platter HDD and subsequently to show the effect of the airflow on the vibration of the actuator arm. In short, the objectives of the experiments are:- To investigate the airflow patterns within a HDD spinning at 2800rpm, 5400 rpm and 7200rpm and at 3 different arm positions. To measure the vibration of the actuator arm due to the airflow at 5400 rpm and 7200rpm and at 3 different arm positions. To study and investigate the vibration characteristics. 8

9 1.3) Technical Approach The experiments are done on a commercial 2.5 inch single platter HDD. A Laser Doppler Anemometer (LDA) will be used to measure the airflow properties within the HDD at the following cases. Disk spinning at 2800 rpm Actuator arm positioned at the inner diameter of the disk (ID) Actuator arm positioned at the middle diameter of the disk (MD) Actuator arm positioned at the outer diameter of the disk (OD) Disk spinning at 5400 rm Actuator arm positioned at the inner diameter of the disk (ID) Actuator arm positioned at the middle diameter of the disk (MD) Actuator arm positioned at the outer diameter of the disk (OD) Disk spinning at 7200 rpm Actuator arm positioned at the inner diameter of the disk (ID) Actuator arm positioned at the middle diameter of the disk (MD) Actuator arm positioned at the outer diameter of the disk (OD) These same cases will also apply for the measurement of vibration on the actuator arm except for the disk spinning at 2800rpm. A Laser Doppler Vibrometer (LDV) will be used for this purpose. 9

10 2) Literature Review and Theoretical Background A number of literatures were surveyed to find out whether any study on flow structure within a HDD has been done in the past. It is found that the literature with the closest similarity to this project is a paper done by C.Y. Soong, C.C. Wu, and T.P. Liu [1] entitled Flow structure between two co-axial disks rotating independently in This literature touches on the flow visualization for large co-axial disks rotating at different speeds. Both open disk model system and partially shrouded disk model system were examined. The underlining theory for rotating disk flow is the von Karman Viscous Flow. When fluid comes in contact with a rotating disk, the viscous effect due to the interaction between the disk and the fluid induced a rotation in the motion of the fluid. The no-slip condition dictates that the fluid adjacent to the surface of the disk rotates at the same velocity as the disk. Fluid is being continuously flung out of the system in the radial direction due to the centrifugal force and the lack of a pressure gradient in the radial direction to balance the force. The Law of Continuity demands that continuous introduction of fluid is done to replace the outward moving fluid. In an open system, fluid is introduced from a distance far from the disk through an axial flow towards the center of the disk. As the fluid get nearer to the rotating disk, it starts to experience the viscous effect and the centrifugal force due to the rotating disk. The fluid starts to swirl outward towards the outer diameter of the disk. This cycle goes on continuously until the disk stops rotating. 10

11 The above mentioned study [1] did not completely mimic the conditions of a HDD due to several differences: I. The rotating disk in a HDD is usually small (2.5 inches) as oppose to 200mm used in the study. II. The rotational speed for a HDD is in the region of 5400rpm to 7200rpm as oppose to less than 100rpm used in the study. III. A HDD is completely shrouded whereas the study is done for an open system and partially shrouded system. a. The fluid within a HDD is finite. There is no continuous introduction of fluid into the system as oppose to an open system. b. The fluid is not flung out of the system in the radial direction as oppose to an open system. The fluid is conserved within the system. IV. The presence of the moving actuator arm influences the flow structure in a HDD. Since there are no studies done on the flow visualization within a real HDD, there is a need for experiments to be done on this subject. However, it is not sufficient just to study the flow pattern in a HDD. Therefore, besides the study of flow visualization, the related topic of flow induced vibrations will also be studied in this project. This project will be concentrating on the vibrations on the actuator arm. 11

12 2.1) Aerodynamic Forces In order to study the flow induced vibrations, it is necessary to establish a link between the flow and the aerodynamic forces that cause the vibrations. Vibrations occur when the actuator arm intrudes the air flow created by the rotating disk. In this case, the actuator arm behaves like an aerofoil subjected to a flow passing over it. As the flow pass through a solid body, a complex trailing edge wake would likely be created thus disrupting the flow and causes flow separation. Flow separations in turn cause turbulences and the formation of vortices. It is these turbulences and vortices that give rise to aerodynamic forces that contributed to vibrations. To analyze the aerodynamic forces, the investigation has to start from the Navier-Stokes equation of motion which governs mechanics of fluid. Basically, the Navier-Stokes equation stipulates that there are 3 components that govern turbulences in flow: Pressure gradient Kinetic energy Interaction between velocity and vorticity. The third component is the one that relates to the aerodynamic forces created due to flow pass the actuator arm. 12

13 For turbulent flow that is steady in the mean, as in the case in a HDD, the instantaneous pressure, velocity and vorticity can be represented by their mean value and their fluctuating value. Pressure Velocity Vorticity ~ p = P + p u ~ = U + u ~ ω = Ω + ω (1) (2) (3) Since the mean values will produce a constant force, they can be neglected in the analysis because it is hypothesized that they will not induce vibrations. For turbulent flow, the fluctuating pressure is insignificant as it will be equalized by the formation of turbulence. After narrowing down the scope, the only components that contribute to vibration are the fluctuating velocity ( u ) and the fluctuating vorticity ( ω ).Hence, the fluctuating velocity and fluctuating vorticity are the main components that contribute to the flow induced vibrations on the actuator arm. The relationship between these fluctuating components and the aerodynamic force can be seen from the modified Kutta-Joukowski Law [5]. Kutta-Joukowski Law = ρla u ω F (4) Where L = characteristic length ρ = air density A = cross-sectional area perpendicular to the flow 13

14 3) Experimental Setup and Instrumentation 3.1) Experimental Setup For this project, a commercial 2.5 inch single platter HDD is used for all the experiments. The original top cover of the HDD is replaced by a transparent glass cover. The reason for this replacement is because a transparent material is needed to allow the laser beam from the LDA and the LDV to pass through in order to measure the airflow and the vibration within the shrouded HDD. Glass is chosen as the transparent material because of its relatively low reflective index which allows the beam to pass through relatively unhindered thus eliminating errors due to the surface reflections. Actuator arm Figure 3.1: The 2.5 inch HDD Since this experiment requires the use of a LDA (Principle of the LDA will be explained in the Appendix C), so it is also necessary to create an entry point to introduce LDA seeding into the system. Therefore, a seeding hole was created at the left side of the HDD. More information regarding the seeding method will be covered in a later section of this report. 14

15 3.2) Experimental Procedures 3.2.1) Laser Doppler Anemometer (LDA) The LDA was chosen as the equipment to conduct the experiments on the airflow characteristics because it is non-intrusive thus making it the ideal choice for measurement of airflow within a HDD as the test section in a HDD is very small compared to other traditional fluid property testing devices and only the LDA is able to conduct the experiment without the fear of space constrain or the problem of the physical obstruction of the flow that will be caused by using probes. To achieve a better velocity profile of the airflow in the HDD, it is imperative that enough data points are collected within the area of interest. A grid system is used to determine the data points and their coordinates, as shown in Figure 3.1. These coordinates are then input into the LDA which measures them automatically. Figure 3.2: Grid system 15

16 The usage of proper seeding is paramount for the running of experiments that require the LDA. It is only with the participation of a proper seeding that sufficient light can be reflected back into the LDA receiver to give a reading. A suitable seeding should have the following properties: 1) The seeding should be able to be introduced into an enclosed area. 2) The seeding must not condense on the glass cover thus clouding it and block the entry or exit of the laser. 3) Seeding had to be big enough to reflect sufficient light so as to obtain a reading. 4) Seeding had to be small enough so that it would not affect the airflow ) Basic Procedures to Setup the LDA 1) Firstly, the HDD is secured onto a platform horizontal to the ground. 2) The motor of the HDD is switched on and is allowed to run for a while. 3) Seeding is injected into the HDD with a syringe through the entry point at the left side of the HDD. Seeding is periodically injected into the HDD to assure good readings from the LDA. 4) The LDA is switched on and the laser is allowed to stabilize before any readings are taken. 5) To ensure that there is a probe volume, the oscilloscope meter is checked to see whether there are any Doppler signals formed on the screen If a Doppler signal is not read, it is due to two most common reasons. i. There is insufficient seeding introduced into the system ii. The probe volume is either on or too near to either the spinning disk or the glass cover. 16

17 6) After getting a clear Doppler signal, the laser is positioned at a certain coordinate on the grid a. A reading is taken at that coordinate and the average velocity at that coordinate is checked. b. The theoretical average velocity of that coordinate is calculated using the equation Velocity (m/s) = radius (m) * angular velocity (rad/s) c. The theoretical answer is checked whether it tallies with the experimental data within a tolerance of 5% d. The laser is shifted to the opposite side of the initial coordinate and this time, the velocity measured should have same magnitude but negative to that of the first velocity reading. 7) The LDA is allowed to acquire data automatically. After data acquisition is completed, raw data is transfer into an Excel spreadsheet for sorting. 8) The sorted data is plotted out using computer software, TecPlot ) Laser Doppler Vibrometer (LDV) To study the vibration characteristics of the actuator arm, a Laser Doppler Vibrometer (LDV) was employed and the result was analyzed with the help of a Dynamic Signal Analyzer (DSA). The LDV is now well established as an effective non-contact alternative to the use of a traditional contacting vibration transducer. The LDV is technically well suited to general application but offer special benefits where certain measurement constraints are imposed. 17

18 The advantages that a Laser Doppler Vibrometer offers are: 1) Non-contact measurement 2) Wide dynamic range for measurement 3) High spatial resolution 4) Small sensor head However, the precautions needed to take note of are: 1) The sensor head must be placed directly opposite the object. 2) The luminous energy of a reflected laser beam must be maintained at a given level. 3) The LDV is susceptible to the effects of oil or water on the object. 4) Extra care must be taken when measuring a rotating object since the LDV is affected by the noise caused by rough surfaces After data has been acquired by the LDV, the analog displacement output is plugged into the DSA to determine the frequency response function through the means of Fast Fourier Transform (FFT). This will yield the resonant frequencies that correlate to the various modes of vibration. The frequency domain is preferred over time domain because the measurement of vibration is easier to analyze in frequency domain while the time domain signal is sum of sinusoids at different frequencies as shown in Figure 3.2. Figure 3.3: Conversion of time domain into frequency domain 18

19 ) Basic Procedures to Setup the LDV 1) Firstly, the HDD is secured onto a platform horizontal to the ground. 2) A reflective tape is applied on the actuator arm. This is done to improve the reflectivity index of the actuator arm so that most, if not all, of the laser beam is reflected back to the receiver. 3) The laser head is adjusted to the position where it received the strongest signal reflected back from the actuator s reflective tape, shown by a signal indicator unit. The laser emitter and receiver are both located at the laser head. Laser Head Figure 3.4: Laser head of the LDV 4) The necessary readings are taken and sent to the DSA unit for further processing before the final result is shown. It is in the DSA unit that the time domain readings are converted into frequency domain by using the Fast Fourier Transform method. The power spectrum method is used to accurately find out the energy content of a signal at a particular frequency. 19

20 For this experiment, the most common frequency domain measurement was used, the power spectrum measurement which displays the square magnitude of the FFT results. It allows one to find how much energy exists at a given frequency or at what frequency energy is present. The power spectrum gives an accurate indication of the energy content of a signal at a particular frequency. On the down side, the phase information is lost in the process but it is not important for this experiment. Some settings are also set to the DSA unit to make sure the result produced is reliable 1) The frequency chosen for the LDV to operate is from 0 khz to 51.2 khz 2) Between the range of given frequencies, a resolution of 800 were selected. This means that 800 frequencies ranging from 0 to 51.2 khz were chosen for the LDV to operate. This was with the intention of producing a better and more accurate result 3) In order to further increase the accuracy of the frequency measurement, a Hanning window and linear averaging were used. The Hanning window is a smoothing window when applies to the result, it greatly reduces any noise in the frequency domain. The averaging process means the DSA unit retains information from previous spectrum result and incorporates each new data set into the average spectrum. For this experiment, each process is linearly averaged over 50 times. 20

21 4) Results and Discussions The following experimental results are separated into the flow measurement and the vibrations measurement of the actuator arm followed by discussion after each part. 4.1) Flow Pattern in HDD 4.1.1) Streamlines for OD, MD and ID positions at 2800rpm Figure 4.1: OD at 2800rpm Figure 4.2: MD at 2800rpm Figure 4.3: ID at 2800rpm 21

22 4.1.2) Streamlines for OD, MD and ID positions at 5400rpm Figure 4.4: OD at 5400rpm Figure 4.5: MD at 5400rpm Figure 4.6: ID at 5400rpm 22

23 4.1.2) Streamlines for OD, MD and ID positions at 7200rpm Figure 4.7: OD at 7200rpm Figure 4.8: MD at 7200rpm Figure 4.9: ID at 7200rpm 23

24 4.1.3) Effects of Varying Actuator Positions on the Flow Pattern From Figure 4.1 to Figure 4.9, it can be observed that flow pattern within a HDD changes as the position of the actuator arm varies from OD, MD to ID. The general trend for speeds of 2800rpm and 5400rpm is that the change in flow pattern is gradual as the actuator arm moves from the OD position to the MD position. The flow maintain in an almost concentric swirl about the center of the disk with minimal flow separation. However, the flow pattern gets erratic due to large flow separation as the actuator arm reaches the ID position, especially around the outer radius of the disk. However, an anomaly is observed when the disk rotates at 7200rpm. Contrary to the previous two cases, the flow pattern is very chaotic when the actuator arm is at the OD position. Furthermore, as the actuator arm moves inward to the MD and the ID position, the swirl becomes more concentric. The experiment for 7200rpm was repeated several times to verify this anomaly and the pattern still remained the same. One possible explanation is because the fluid accelerated as it flow pass the actuator arm thus inhibited the flow from separation ) Effects of Increased Rotational Speed on the Flow Pattern From observations of Figure 4.1 to Figure 4.9 and comparing the flow patterns at different rotational speed but same position of the actuator arm, it can be observed that the flow gradually becomes more erratic as the speed increases from 2800rpm to 7200rpm. This behavior is expected because flow becomes more turbulent as the Reynold s Number, which is a function of velocity, increases. 24

25 As the rotational speed increases from 2800rpm to 5400rpm, it is observed that flow separation happens earlier, i.e. nearer to center of the disk, for all three positions of the actuator arm. For flow at 7200rpm, the pattern becomes very chaotic for the OD position. Again, an anomaly is detected at the ID position where the flow is surprisingly concentric and neat ) Applications With a better understanding of the flow patterns and flow structures within a HDD, some improvements can be made to existing or future HDD designs. One particular area that will benefit from this is the study of heat transfer. Efficient heat transfer for a HDD is all the more important for portable HDD commonly used in handheld devices like PDAs and Apple s ipods. Portable HDD rely on natural convection for cooling because manufacturers cannot attach cooling fans into such devices. Air transfers heat through convection as it flows within a HDD. So, by analyzing the flow pattern, heat sinks can be placed at appropriate areas to maximize heat transfer. Another area that will benefit is the location for the air filter. The function of the filter is to clean the air of dust and debris. As such, the filter needs to be placed at a location where the bulk of the air will flow to as determined by the streamlines. 25

26 4.2) Vibrations of the Actuator Arm For this experiment, the vibrations measurements of the actuator arm are done for two directions, the horizontal direction and the vertical direction. The horizontal measurement measured the vibrations modes of the actuator arm in the direction parallel to the disk s surface. The vertical measurement measured the vibrations modes in the perpendicular direction to the disk s surface. These results include the vibrations contributed by the spindle motor and the voice-coil motor since it impossible to completely isolate the vibrations due to the motors in a HDD. However, measurements are done to identify the vibrations by the motors alone in the absence of flow. 26

27 Voltage (mv) rpm Horizontal ID MD OD Motor Frequency (Hz) Figure 4.10: Horizontal measurement at 5400rpm rpm Horizontal Voltage (mv) ID MD OD Motor Frequency (Hz) Figure 4.11: Horizontal measurement at 7200rpm 27

28 Voltage (mv) rpm Vertical Frequency (Hz) ID MD OD Motor Figure 4.12: Vertical measurement at 5400rpm Voltage (mv) rpm Vertical Frequency (Hz) ID MD OD Motor Figure 4.13: Vertical measurement at 7200rpm 28

29 4.2.1) Effect of Varying Actuator Position on the Vibrations For both directions of measurements, there are some differences in the modes (peaks) of vibrations and the level of excitation (indicated by the height of the peaks) as the actuator arm moves from the OD position to the ID position as shown from Figure 4.10 to Figure For the horizontal measurements, three modes of vibrations that are common for both rotational speeds can be detected at frequencies of 600Hz, 2.6kHz and 4kHz. The peak at 2.2kHz for speed of 5400rpm is mainly excited by the spindle motor as the height of the peak remains constant for all three actuator positions. It can also be observed that there is an extra peak at 1.3kHz that is only present in the ID position at speed of 5400rpm. Referring back to section 4.1.2, the flow pattern for the ID position is the most erratic. So, the high turbulence of the flow could be a reason as to presence of the peak at 1.3kHz. For speed of 7200rpm, the peak at 2.9kHz is excited by the spindle motor. However, as the level of excitation is not constant for different actuator position, it is hypothesized that the flow also excites the actuator at this frequency. For the vertical measurements, four modes of vibrations can be identified at frequencies of 700Hz, 1.3kHz, 2.6kHz and 4kHz. Again, the modes at 2.2kHz and 2.9kHz are detected, which are due to excitation by the spindle motor at 5400rpm and 7200rpm respectively. For speed of 5400rpm, it is observed that the level of vibrations by the motor is higher when there is no flow. In this case, the damping effect of the flow outweighs the excitation effect on the actuator arm. 29

30 4.2.2) Effect of Varying Rotational Speed on the Vibrations As expected, the level of excitation (indicated by the height of the peaks) increases as the velocity of the flow (indicated by the rotational speed) increases. One of the reasons is that, as the velocity of the flow increases, its kinetic energy also increases. So, the actuator arm is being excited by a flow that has more energy thus resulting in the increased in the level of excitation. Another reason is due to the increased turbulence of higher velocity flow. As flow separations occur more often in higher velocity flow, vortices are formed more frequently, thus causing increased number of aerodynamic forces that excite the system at additional modes of vibrations. So, increasing the rotational speed not only increases the level of vibrations but increases the number of modes also as shown by Figure 4.14 to Figure

31 Voltage (mv) Horizontal ID Freq (Hz) 5400rpm 7200rpm Figure 4.14: Horizontal measurement at ID position Voltage (mv) Horizontal MD Freq (Hz) 5400rpm 7200rpm Figure 4.15: Horizontal measurement at MD position Voltage (mv) Horizontal OD Freq (Hz) 5400rpm 7200rpm Figure 4.16: Horizontal measurement at OD position 31

32 Voltage (mv) Vertical ID Freq (Hz) 5400rpm 7200rpm Figure 4.17: Vertical measurement at ID position Voltage (mv) Vertical MD Freq (Hz) 5400rpm 7200rpm Figure 4.18: Vertical measurement at MD position 3 Vertical OD Voltage (mv) rpm 7200rpm Freq (Hz) Figure 4.19: Vertical measurement at OD position 32

33 4.2.3) Displacements of Individual Modes of Vibrations In order to have a better understanding of the displacements cause by the vibrations of the actuator arm in a HDD, the displacement frequency response graphs are plotted. It is important to note that Figure 4.20 to Figure 4.23 only shows the displacements contributed by each mode of vibrations and not the total vibrations of the actuator arm. However an estimate can be made to obtain the total displacement by summing up all displacements of each mode. This estimation gives the maximum displacements possible for that particular speed and arm position because the assumption is that there is no phase difference between all the modes and the amplitude of each mode super positioned upon each other at some point in time. Displacement (nm) rpm Horizontal Frequency (Hz) ID MD OD Motor Figure 4.20: Horizontal displacements at 5400rpm 33

34 Displacement (nm) rpm Horizontal Freq (Hz) ID MD OD Motor Figure 4.21: Horizontal displacements at 7200rpm Displacement (nm) rpm Vertical ID MD OD Motor Freq (Hz) Figure 4.22: Vertical displacements at 5400rpm 34

35 7200rpm Vertical Displacement (nm) ID MD OD Motor Freq (Hz) Figure 4.23: Vertical displacements at 7200rpm Table 1: Estimated Maximum Displacement for Horizontal Measurement Speed (rpm) Displacement (nm) Actuator Position ID MD OD Motor Table 2: Estimated Maximum Displacement for Vertical Measurement Speed (rpm) Displacement (nm) Actuator Position ID MD OD Motor

36 4.2.4) Applications Track mis-registration (TMR) is a prominent problem in many HDD and is becoming more severe as the track density of a HDD increases. With the demand for more data storage capacity for a 2.5 inch HDD and faster rotational speed to improve seek time, active vibration control for the actuator arm need to be implemented, or a new and improved actuator arm need to be designed. In order to make the necessary improvements on the actuator arm to reduce or eliminate TMR, the vibrations characteristics have to be understood beforehand. The frequency response graphs for the horizontal measurements allow design engineers to identify the modes of vibrations of the actuator arm when subjected to the motor and the air flow thus allowing them to design adaptive feedforward control algorithms to attenuate TMR by reducing or eliminating certain modes of vibrations. Similarly, the vertical measurements provide information so that engineers can design against large vibrations that may cause the head to crash onto the rotating disk. The displacements graphs allow comparison between the magnitude of each mode of vibrations against the allowable TMR for a specify track density. Track density is determined by the number of track-per-inch (TPI) and the distance between adjacent tracks is known as the track pitch. Currently, most HDD have track density of more than TPI or 100 ktpi, with some server HDD reaching as high as 500 ktpi. For HDD with a track density of 100 ktpi and 500kTPI, the track pitch is 159 nm and 51nm respectively. However, industrial standards state that the allowable TMR for the actuator arm is only 10% of the track pitch. So, in practice, the maximum allowable TMR is 36

37 15.9nm and 5.1nm for 100 ktpi and 500 ktpi HDD respectively. Comparing the value for the 100 ktpi with values from Table 1, it can be observed that the maximum displacements for 5400rpm are slightly over the allowable TMR while the value for 7200rpm are well over the allowable value. However, it is important to note that the displacements in Table 1 are an over-estimation of the real values. 37

38 5) Conclusion The following has been achieved by this project: 6) The flow pattern within a commercial 2.5 inch HDD with the disk rotational speed of 2800rpm, 5400rpm and 7200rpm with three different positions of the actuator arm. 6) The frequencies of the various modes of vibrations of the actuator arm and the displacements of each individual mode as well as an estimation of the maximum displacements. Through this project, the operations of the Laser Doppler Anemometer and the Laser Doppler Vibrometer have been mastered. The interaction of the fluctuating components of the velocity and the vorticity have been identify as the main source of flow induced vibrations. Some of the observations obtained are that flow patterns get more erratic as the rotational speed increases and actuator arm experiences the most aerodynamic forces in the MD position as shown by the level of vibrations. The frequency response graphs for the vibrations of the actuator arm are plotted and examined. It is observed that varying the position of the actuator arm has an effect on the modes of vibrations and the magnitude of vibrations. Increasing the rotational speed also affect the vibrations, mainly the magnitude but it also introduced extra modes of vibrations of the actuator arm. Estimation is done to obtain the maximum displacements of the actuator arm due to vibration and the values are compared to the allowable TMR for track density of 100kTPI. For speed of 5400rpm, the displacements are marginally higher than the allowed TMR. 38

39 6) Recommendations After conducting the experiment and collected all the data needed, the following recommendations were suggested for future research so that study of the flow patterns and vibration characteristics can be improved. 1) The size of the LDA s probe volume should be reduced to less that 1mm in size so that velocity measurements can be done as near to the rotating as possible. Throughout the experiments, due to limitation of the probe volume, the measurements are done at a height of 14mm above the rotating disk. As a result, attempts to study the flow profile at height lower than 14mm have failed. 2) Non-reflective coating should be applied to the components within the HDD, especially the disk and the actuator arm, in order to reduce noise signals due to flaring. 3) Vibrations measurements should be done using the Scanning LDV. Scanning LDV can scan multiple points in a short time thus resulting in a better real time analysis of the modes frequencies. 4) Computational modal analysis should be done to identify the resonance modes and their respective deformation shapes. 39

40 5) References 1. C.Y. Soong, C.C. Wu, T.P. Liu and Tao-Ping Liu; Flow structure between two co-axial disks rotating independently, Experimental Thermal and Fluid Science 27 (2003) p Yunfeng Li and Roberto Horowitz; Active Suspension Vibration Control with Dual Stage Actuators in Hard Disk Drives, Proceedings of the American Control Conference, June Tan Chong Han; Investigation of Form Drag within Computer Hard Disk Drive, Final Year Project, National University of Singapore, 2003/ Tan Jin Sheng; Advanced Aerodynamics in a Computer Hard Disk Drive, Final Year Project, National University of Singapore, 2003/ H. Tennekes and J.L Lumley; A First Course in Turbulence, The MIT Press, Chapter 3, pg

41 Appendix A Experimental Setup 41

42 D1: The Laser Doppler Anemometer (LDA) D2: Probe volume at laser crossing 42

43 D3: The Laser Doppler Vibrometer unit D4: The Dynamic Signal Analyzer 43

44 Appendix B Doppler Effect 44

45 Doppler Effect in LDA and LDV If an acoustic, radio or light wave of a specific frequency is beamed at a moving object; the frequency of the wave reflected from the moving object differs in proportion to the velocity of the object. This phenomenon is known as Doppler shift or the Doppler Effect. The relationship between the frequency at which the wave is beamed and the frequency at which the wave echoes is outlined as: 1. If the object is approaching the source, the reflected frequency is higher than the emitted frequency 2. If the object is receding from the source, the reflected frequency is lower than the emitted frequency The difference between the emitted frequency and the reflected frequency relates to the velocity of the moving object. Normally, the difference in frequency becomes larger as the velocity of the object increases. The measurement of the both the LDA and the LDV are based on this principle. When laser light is beamed at a moving object, the frequency of the laser light reflected from the object differs from the original frequency of the emitted laser light due to this Doppler Effect. The instruments consider the amount of Doppler shift generated. Given that the change in frequency is f D, the velocity of the moving object is V, the wavelength of the emitted laser light is λ, and the angle between the direction in which laser light runs and the direction in which the object moves is θ, then the following equation holds. 45

46 D1: the Doppler Effect As shown in the figure above, the frequency of the reflected laser light is f 0 +f D. Since the wavelength λ of laser is extremely stable, the Doppler frequency f D and velocity V are proportional. Normally, the angle θ is set at 0 (only the component of reflected light parallel to incident light is detected) and, therefore, the velocity of the object moving in the direction of the emitted laser light can be determined by measuring f D. The frequency of laser light is so high, however, that it is extremely difficult to measure directly. For this reason, f D is normally determined by allowing the emitted laser light (f 0 ) to interfere with the reflected laser light (f 0 +f D ). 46

47 Appendix C Laser Doppler Anemometer 47

48 Measurement Principles of the LDA. The Laser Doppler Anemometer, or LDA, is a widely accepted tool for fluid dynamic investigations in gases and liquids and has been used as such for more than three decades. It is a well-established technique that gives information about flow velocity. The LDA s non-intrusive principle and directional sensitivity make it very suitable for applications with reversing flow, chemically reacting or high-temperature media and rotating machinery, where physical sensors are difficult or impossible to use. It requires tracer particles in the flow. The LDA s main advantages are: Non intrusive No calibration required Velocity range 0 to supersonic One, two or three velocity components simultaneously Measurement distance from centimeters to meters Flow reversals can be measured High spatial and temporal resolution Instantaneous and time averaged 48

49 D2: LDA principle (Dantec Dynamics) The basic configuration of an LDA consists of: a continuous wave laser, transmitting optics, including a beam splitter and a focusing lens, receiving optics, comprising a focusing lens, an interference filtre and a photodetector, a signal conditioner and a signal processor. A Bragg cell is often used as the beam splitter. It is a glass crystal with a vibrating piezocrystal attached. The vibration generates acoustical waves acting like an optical grid. D3: The Bragg cell used as a beam splitter (Dantec Dynamics) 49

50 Probe Volume The output of the Bragg cell is two beams of equal intensity with frequencies f 0 and f D. These are focused into optical fibres bringing them to a probe. In the probe, the parallel exit beams from the fibres are focused by a lens to intersect in the probe volume. D4: The probe and the probe volume (Dantec Dynamics) The probe volume is typically a few millimeters long depending on focal length of the lens used. The light intensity is modulated due to interference between the laser beams. This produces parallel planes of high light intensity, so called fringes. The fringe distance d f is defined by the wavelength of the laser light and the angle between the beams: d f = λ 2sin( θ / 2) 50

51 Each particle passage scatters light proportional to the local light intensity. Flow velocity information comes from light scattered by tiny "seeding" particles carried in the fluid as they move through the probe volume. The scattered light contains a Doppler shift, the Doppler frequency f D, which is proportional to the velocity component perpendicular to the bisector of the two laser beams, which corresponds to the x axis shown in the probe volume. The scattered light is collected by a receiver lens and focused on a photo-detector. An interference filter mounted before the photo-detector passes only the required wavelength to the photo-detector. This removes noise from ambient light and from other wavelengths. Signal processing The photo-detector converts the fluctuating light intensity to an electrical signal, the Doppler burst, which is sinusoidal with a Gaussian envelope due to the intensity profile of the laser beams. The Doppler bursts are filtered and amplified in the signal processor, which determines f D for each particle, often by frequency analysis using the robust Fast Fourier Transform algorithm. The fringe spacing, d f provides information about the distance traveled by the particle. The Doppler frequency f D provides information about the time: t = 1/f D Since velocity equals distance divided by time, the expression for velocity thus becomes: Velocity V = d f * f D 51

52 Determination of the sign of the flow direction D5: Doppler frequency to velocity transfer function for a frequency shifted LDA system (Dantec Dynamics) The frequency shift obtained by the Bragg cell makes the fringe pattern move at a constant velocity. Particles which are not moving will generate a signal of the shift frequency f D. The velocities V pos and V neg will generate signal frequencies f pos and f neg, respectively. LDA systems with frequency shift can distinguish the flow direction and measure zero velocity. Seeding particles Liquids often contain sufficient natural seeding, whereas gases must be seeded in most cases. Ideally, the particles should be small enough to follow the flow, yet large enough to scatter sufficient light to obtain a good signal-to-noise ratio at the photo-detector output. Typically the size range of particles is between 1 µm and 10 µm. The particle material can be solid (powder) or liquid (droplets). 52