How Resonance Data is Used

Transcription

1 Leading Edge Suspension Resonance Control and Technology DISKCON AP, March 2007

2 Introduction Increasingly, suspension resonance characterization and process control have become critical factors in drive development and performance Demands for precise, high frequency characterization of complex resonance profiles require advancements in resonance metrology This presentation reviews resonance measurement challenges, along with the progression and performance of newly developed resonance measurement technology 2

3 How Resonance Data is Used Resonance is a functional parameter and provides valuable information: An open loop response that can be used to design the servo loop during product development Data provides feedback for suspension manufacturing process setup and control Frequency Response Function (FRF) Gain (db) Gain Phase 1T Sway 1 Gimbal 1T 2T Sway Phase (Deg) Frequency (Hz) 3

4 Resonance Measurement Challenges Creating an accurate and repeatable measurement system for all resonance modes (bending, torsion and sway) for frequency and gain Increasing the measurement range from 30 khz to 50 khz Reducing tester-to-tester variation Improving software data analysis Reducing measurement time to improve tester productivity Improving ease of use for operator Keeping cost of system low 4

5 Key Factors That Affect Resonance Measurement Sensors Z-height variation: Z-height calibration methods: System to system variation Mount block (MB) design: Within system variation System-to-system variation Fixture mode interactions: Modes from MB interface Mode from shakers: Splitting of modes Reduced gains Software analysis 5

6 Sensor Limitations 6

7 Why Accelerometers Are Not Adequate Accelerometers need to be mounted on the system which increases interfaces and mass to induce modes Accelerometers have a nonlinear sensitivity response, which poses two problems: Is not an accurate method of measurement unless sensitivity factors are corrected over the measured range Have fixture modes below the desired 50 khz level 7

8 Z-height Variation 8

9 Z-height Calibration Variation Required spec was ±5 µm (~ ) Error budget analysis showed that the calibration error was around ±35 µm (~ ) with 95% CI 12.9% 1.5% 8.3% 33.1% 2.7% 41.6% CB to CB variation Parallelism Of CB Disk Runout Disk Deflection Dial Gauge errors Others W notch Variation in Z-height affects system-to-system variation specifically for torsion modes Gain [db] Z-height [inch] 9

10 Mount Block (MB) Variation Tolerance stack-up analysis of the MB interface showed that variation of 74 µm (1 Std. Dev.) can be observed based on print specs: Analysis performed on 14.5 mm part Variation estimated at dimple location Does not include part variation 50% of variation is from the MB interface 10

11 Fixture Mode Interaction and Reduction 11

12 MB Interface Modes Major modes are caused by the MB s additional mass and its interface to the tooling fixture and clamping Animation based on actual laser measurements made on the accelerometer adapter and MB (not including part) 12

13 Shaker Mode Effects 13

14 Shaker Mode Interaction Causes mode splitting due to shaker mode at 17 khz FRF Data Axial mode of shaker causing rounding of sway peak Input Data Input 0.8 V rms Input 0.1 V rms 14

15 Summary of Findings So Far Use of a non-contact measurement method for input and output displacements is recommended for lower fixture modes and an improved measurement range MB design contributes significantly to Z-height variation, which in turn causes increased torsion mode variation Shaker and MB modes present in the part mode region interact to cause reduced repeatability of frequency and gain measurement 15

16 High Performance Resonance Tester (HPRT) Design Objectives 16

17 System Requirements 50 khz test capability Disk speed to 20 k RPM Elimination of part shaker mode interactions Improved repeatability and reproducibility in frequency and gain Sound automated peak marking 50% reduction in test cycle time Improved operator usability features 17

18 Improved Sensor Capability 18

19 Laser Vibrometer... A Better Choice Provides a non-contact measurement method and can be mounted outside of the measurement loop Measures up to 2 MHz with high accuracy 19

20 Improved Z-height Repeatability 20

21 Integrated Shaker Design Create an integrated shaker and MB system to: Reduce mode-causing interfaces Reduce mass Reduce Z-height variation Improve system-to-system repeatability Mount Block 21

22 Shaker Design 22

23 Design Philosophy Pursued shaker design to eliminate fixture modes that are past critical frequencies of parts to avoid mode interaction Elimination of fixture modes through compensation will not include part and shaker mode interactions Mode Free shaker Comped Shaker 23

24 Mode Reduction and Design Space Determination Initial Baseline Design #1: Verification of integrated shaker design (no mode up to 15 khz) DRT Design #1 Gain (db) Design #2: No mode up to 25 khz, with offset sway frequency -10 Freq (Hz) 24

25 Mode Reduction and Design Space Determination (Cont.) Gain (db) DRT Design # Freq Hz Design #3: High sway frequency matches, with increased modes 8dn FRF Comparison Gain Reference Design# Design #4: High sway frequency matches, with modes eliminated Frequency 25

26 Software Improvements 26

27 Improved Peak Marking Miss-marking of peaks was a common problem due to noisy data and not knowing where to mark peaks Filtered data Noise Filtering Algorithm Created a proprietary noise filtering algorithm Used windage overlap to guide operators to mark peaks Windage Windage Overlap Resonance 27

28 Improved Visual Marking of Peaks Named modes in windage and gain for better visual verification 28

29 Improved Visual Marking of Peaks Named modes in windage and gain for better visual verification Overlapping frequency tolerances 29

30 Improved Visual Marking of Peaks Named modes in windage and gain for better visual verification Overlapping frequency tolerances One-sided frequency tolerances 30

31 Improved Z-height Testing Ability to remark peaks and view an entire test Can run a bode and full z-height test back-to-back without unloading the part 31

32 Improved Windage Data Analysis 2.5 Windage spike removal: More accurately mark peaks Spikes are due to sensor errors and do not represent drive condition Cumulative windage measurement included Displacement (nin Pk) Displacement (nin Pk) Frequency (Hz) Unfiltered windage Frequency (Hz) Unfiltered windage Windage with spikes removed 32

33 Current Status Part 5dn on 3 systems 12 systems deployed across different programs and multiple sites HPRT 105 HPRT 106 HPRT 111 Data Offset to Show Similarity Frequency range Spindle speeds (direction) Part skew angle Part length Fixture modes Slider size Disk diameter Cycle time HPRT 1 50 khz (CW & CCW) Variable in X & Y 7 35 mm No modes up to 40 khz 50%, 30%, 20%, hybrids 95 mm & 65 mm 2.7 minutes (est.) 33

34 Conclusion HTI has developed a resonance tester that meets industry requirements for range, accuracy and speed through system optimization and software improvements HTI is deploying this technology on new products HTI continues to improve resonance measurement technology to meet future industry needs and challenges for improved product performance 34