DESIGN VERIFICATION OF A SOLID-STATE DRIVE A Thesis Presented to the faculty of the Department of Mechanical Engineering California State University, Sacramento Submitted in partial satisfaction of the requirements for the degree of MASTER OF SCIENCE In Mechanical Engineering by Kenneth Salyphone FALL 2013 © 2013 Kenneth Salyphone ALL RIGHTS RESERVED ii DESIGN VERIFICATION OF A SOLID-STATE DRIVE A Thesis by Kenneth Salyphone Approved by: __________________________________, Committee Chair Akihiko Kumagai __________________________________, Second Reader Stewart Dunlap ____________________________ Date iii Student: Kenneth Salyphone I certify that this student has met the requirements for format contained in the University format manual, and that this thesis is suitable for shelving in the Library and credit is to be awarded for the thesis. __________________________, Graduate Coordinator Akihiko Kumagai Department of Mechanical Engineering iv ___________________ Date Abstract of DESIGN VERIFICATION OF A SOLID-STATE DRIVE by Kenneth Salyphone Moore’s Law has brought upon innovation and technological advances to the semiconductor industry, which has resulted in advancements to Solid-State Drives (SSD). This thesis is a design verification of a Solid-State Drive that utilizes a Peripheral Component Interface Express (PCIe) connection. The design of the drive is modeled using SolidWorks Computer Aided Design (CAD) and modeling simulations provided within SolidWorks simulations. SolidWorks modeling simulations such as Drop, and Flow are used to simulate industry and/or customer standards for the drive. Further verification of the Solid-State drive is performed using Lansmont Shock/Drop machine and a Thermal Wind Tunnel. The empirical data from the verification tests are used to correlate the data acquired from the SolidWorks simulations. The results show that the simulations acquired by SolidWorks and the empirical data from the verification test are consistent. The simulation also proved to be a useful predictive tool when inquiring about possible outcomes in designing products and predicting material behavior, such as in this design verification of a Solid-State Drive. v _______________________, Committee Chair Akihiko Kumagai _______________________ Date DEDICATION vi I would like to dedicate this thesis to my parents, family, and loved ones who have helped on my journey here. Thank you! ACKNOWLEDGEMENTS vii I would like to acknowledge the Mechanical Engineering team at Micron Technology, Inc. for providing great insight and support. Thank you Stew, Conner, Michael and Maria! I would also like to acknowledge Akihiko Kumagai, Sue Holl and the Department of Mechanical Engineering at California State University of Sacramento. Thank you! TABLE OF CONTENTS Page Dedication ................................................................................................................... vii viii Acknowledgements ................................................................................................... viii List of Tables ............................................................................................................... xi List of Figures ............................................................................................................. xii Chapters 1. SOLID-STATE DRIVE ............................................................................................1 1.1 Introduction ................................................................................................ 1 1.2 Background ................................................................................................ 2 1.3 Problem Statement ..................................................................................... 3 1.4 Scope ...........................................................................................................3 2. CASE STUDY 1: DROP TEST SIMULAITON, VERIFICATION, RESULTS AND ANALYSIS ..........................................................................................................4 2.1 Modeling .....................................................................................................4 2.2 Drop Test Simulations ................................................................................5 2.3 Drop Test Simulation Results ...................................................................11 2.4 Drop Test Verification ..............................................................................25 2.5 Drop Test Verification Results .................................................................30 2.6 Drop Test Simulation and Verification Analysis ......................................33 3. CASE STUDY 2: FLOW SIMULATION, VERIFICATION, RESULTS AND VERIFICATION ................................................................................................43 ix 3.1 Flow Simulation ........................................................................................43 3.2 Flow Simulation Results ...........................................................................46 3.3 Flow Verification ......................................................................................50 3.4 Flow Verification Results .........................................................................53 3.5 Flow Simulation and Verification Analysis ..............................................54 4. CONCLUSION .......................................................................................................57 References ....................................................................................................................58 x LIST OF TABLES Tables Page 1. Table 2.1: List of Parts and Material Properties ................................................ 8 2. Table 2.2: X-axis Drop Test Simulation and Verification Analysis Correlation ........................................................................................................ 36 3. Table 2.3: Y-axis Drop Test Simulation and Verification Analysis Correlation ........................................................................................................ 39 4. Table 2.4: Z-axis Drop Test Simulation and Verification Analysis Correlation ....................................................................................................... 41 5. Table 2.5: List of Parts and Thermal Properties .............................................. 44 6. Table 2.6: Global and Volume Goals .............................................................. 47 7. Table 2.7: Average Maximum and Maximum Temperature from Flow Verification ....................................................................................................... 54 8. Table 2.8: Flow Simulation Results (Top) and Verification Results (Bottom) ........................................................................... 54 xi LIST OF FIGURES Figures Page 1. Figure 2.1: SSD Fully Populated/ Detailed Model ............................................ 4 2. Figure 2.2: Simplified Model of the SSD for Simulation Purposes .................. 5 3. Figure 2.3: SSD in Jig ........................................................................................ 6 4. Figure 2.4: SSD in the Jig and Block ................................................................. 7 5. Figure 2.5: Drop Test Simulation X-axis........................................................... 8 6. Figure 2.6: Drop Test Simulation of the SSD on the Y-axis ............................. 9 7. Figure 2.7: Drop Test Simulation of the SSD on the Z-axis ............................ 10 8. Figure 2.8: von Mises Stress Plot for the X-axis Drop Simulation ................. 11 9. Figure 2.9: von Mises Stress Plot of the Components for the X-axis Drop Test Simulation ........................................................................................ 12 10. Figure 2.10: von Mises Stress Plot on the Capacitor Lead for the X-axis Drop Test Simulation ........................................................................................ 13 11. 2.11: Equivalent Strain Plot of the Components for the X-axis Drop Test Simulation ........................................................................................ 14 12. Figure 2.12: Equivalent Strain Plot of the Capacitor Leads for the X-axis Drop Test Simulation ............................................................................ 14 13. Figure 2.13: Displacement Plot for the X-axis Drop Test Simulation............. 15 xii 14. Figure 2.14: Displacement Plot of the Capacitor Leads for X-axis Drop Test Simulation ............................................................................ 16 15. Figure 2.15: von Mises Stress Plot of the Components for the Y-axis Drop Test Simulation ............................................................................ 17 16. Figure 2.16: von Mises Stress Plot of the Capacitor Leads for the Y-axis Drop Test Simulation ............................................................................ 17 17. Figure 2.17: Equivalent Strain Plot of the Components for the Y-axis Drop Test Simulation ............................................................................ 18 18. Figure 2.18: Equivalent Strain Plot of the Capacitor Leads for the Y-axis Drop Test Simulation ............................................................................ 19 19. Figure 2.19: Displacement Plot for the Y-axis Drop Test Simulation............. 20 20. Figure 2.20: Displacement Plot of the Capacitor Leads for the Y-axis Drop Test Simulation ............................................................................ 20 21. Figure 2.21: von Mises Stress of the Components for the Z-axis Drop Test Simulation ........................................................................................ 21 22. Figure 2.22: von Mises Stress of the Capacitor Leads for the Z-axis Drop Test Simulation ........................................................................................ 22 23. Figure 2.23: Equivalent Strain of the Components for the Z-axis Drop Test Simulation ........................................................................................ 23 24. Figure 2.24: Equivalent Strain of the Capacitor Leads for the Z-axis Drop Test Simulation ........................................................................................ 23 xiii 25. Figure 2.25: Displacement Plot for Z-axis Drop Test Simulation ................... 24 26. Figure 2.26: Displacement Plot of the Capacitor for the Z-axis Drop Test Simulation ........................................................................................ 25 27. Figure 2.27: Strain Gage Type and Specifications .......................................... 26 28. Figure 2.28: Strain Gage Locations ................................................................. 27 29. Figure 2.29: Vishay System 6000 and Model 6010A Strain Gage Cards ....... 28 30. Figure 2.30: X-axis Drop Test Verification ..................................................... 29 31. Figure 2.31: Y-axis Drop Test Verification ..................................................... 29 32. Figure 2.32: Z-axis Drop Test Verification ..................................................... 30 33. Figure 2.33: Strain Plot of the Components for the X-axis Drop Test ............ 31 34. Figure 2.34: Strain Plot of the Components for the Y-axis Drop Test ............ 32 35. Figure 2.35: Strain Plot of the Components for the Z-axis Drop Test ............. 33 36. Figure 2.36: Strain vs. Time (10^-5 seconds) for the X-axis Drop Test.......... 35 37. Figure 2.37: X-axis Drop Test Results with Strain Gage Oriented in the Y-axis ................................................................................................................ 35 38. Figure 2.38: X-axis Drop Test Results with Strain Gage Oriented in the Z-axis ................................................................................................................ 36 39. Figure 2.39: Strain vs. Time (10^-5 seconds) for the Y-axis Drop Test.......... 37 40. Figure 2.40: Y-axis Drop Test Results with Strain Gage Oriented in the Y-axis ................................................................................................................ 38 xiv 41. Figure 2.41: Y-axis Drop Test Results with Strain Gage Oriented in the Z-axis ................................................................................................................ 38 42. Figure 2.42: Strain vs. Time (10^-5 seconds) for the Z-axis Drop Test .......... 40 43. Figure 2.43: Z-axis Drop Test Results with Strain Gage Oriented in the Y-axis ................................................................................................................ 40 44. Figure 2.44: Z-axis Drop Test Results with Strain Gage Oriented in the Z-axis ................................................................................................................ 41 45. Figure 3.1: SSD in Control Volume for Flow Simulation ............................... 43 46. Figure 3.2: Boundary Conditions ..................................................................... 44 47. Figure 3.3: Heat Sources .................................................................................. 45 48. Figure 3.5: Radiative Surfaces ......................................................................... 46 49. Figure 3.6: Flow Trajectory Plot ...................................................................... 48 50. Figure 3.7: Surface Plot ................................................................................... 49 51. Figure 3.8: Cut Plot .......................................................................................... 50 52. Figure 3.9: Wind Tunnel Verification Test...................................................... 51 53. Figure 3.10: Mapping of Thermocouples (marked by star) ............................. 53 54. Figure 3.11: Percent Difference Equation ....................................................... 55 xv 1 CHAPTER 1 SOLID-STATE DRIVE 1.1 Introduction: “The number of transistors incorporated in a chip will approximately double every 24 months.” --Gordon Moore, Intel co-founder The statement above, known as Moore’s Law, has defined the progression of the products created within the semiconductor industry for the past 40 years (Moore’s Law 2013). One of those products is the Solid-State Drive (SSD). The Solid-State Drive is a data storage device that uses integrated circuit assemblies as memory to store data persistently. However, a more common form of data storage device is and has been the Hard-Disk Drive (HDD). SSDs have no moving mechanical components, which distinguish them from the common and traditional HDDs that contain spinning disks, and movable read/write heads. Compared with electromechanical disks such as the HDDs, SSDs are typically more resistant to physical shock, run more quietly, have lower access time and less latency. However, the price of an SSD is typically 7 to 8 times more expensive than that of an HDD. SSDs use NAND-based flash memory, which retains data without power. For applications requiring fast access, SSDs may be constructed from dynamic random access 2 memory (DRAM) (Ekker 2009). These NAND-DRAM combination SSDs employ capacitors to maintain data after power loss. 1.2 Background: SSDs originated during the 1950s with two similar technologies: Magnetic Core Memory and Charged Capacitor Read-Only Store (CCROS) (Rent 2010). Magnetic core memory was the predominant form of random-access computer memory for 20 years (1955 to 1975). It uses tiny magnetic toroids (rings), the cores, through which wires are threaded to write and read information (Magnetic-Core Memory 2013). Core memory technology advanced considerably through extensive use by NASA and up until 1990 on most mission-critical defense systems, due to its reliability and data durability. In the mid 1950s, as the transistor was emerging from IBM research, IBM developed their first bulk solid-state nonvolatile memory called the Charged Capacitor Read-Only Store (CCROS). It was the first true SSD and was the predecessor to today’s EPROMS, EEPROMS, and Flash memory devices. Today’s SSDs offer exponentially more storage and speed, and continue to evolve as the advancements in memory, controllers, and other core components come to market (Rent 2010). On December 2009, Micron Technology announced the world’s first SSD using 6 gigabits per second SATA interface (Jansen 2009). 3 1.3 Problem Statement: For the past four decades, the semiconductor industry has experienced Moore’s Law and its ability to forecast that transistors in chips will double approximately every 24 months. This law is evident in storage devices, specifically SSDs, and its ability to increase storage space while keeping the same form factor. The form factor is designation for the size of the drive, and ranges from a small form factor, which is the size of a 2.5-inch HDD, to what the industry call a full height drive. The progression of increased storage space, also known as an increase density, per form factor, has led to the need for SSDs to be designed and verified in a way that best service the client and client needs. In this paper the design and verification of a half height half length SSD is investigated from a mechanical engineering perspective. 1.4 Scope: The scope of the thesis will consist of using SolidWorks CAD to model the SSD with its components populated on the Printed Circuit Board (PCB). With the SSD modeled in SolidWorks, simulations performed on the model are used as a predictive tool to predetermine any situations that may pose problems. The simulations performed on the model are SolidWorks Drop Test and Flow Simulation. In addition, the simulations will be verified by an actual drop and flow test so the results can be correlated. 4 CHAPTER 2 CASE STUDY 1: DROP TEST SIMULAITON, VERIFICATION, RESULTS AND ANALYSIS 2.1 Modeling: The SSD is modeled using SolidWorks 2013 CAD software. This involves drawing all the components in the SSD. The SSD being modeled includes two PCB boards that holds the DRAM, NAND, Controller, Capacitors, Resistors, and board-toboard connector. The SSD model is drawn and assembled to its exact specifications and every component is fully detailed. Figure 2.1 shows the model assembled with all the components in the drive. Figure 2.1: SSD Fully Populated/ Detailed Model. 5 2.2 Drop Test Simulations: The drop test simulation is a simulation that SolidWorks provides in order to determine stresses, strains and or displacements at specific locations on the model. In order to perform the drop test simulation the model is simplified and any detailed components are drawn with simple geometric shapes. For example, the solder balls are removed from DRAM, NAND, Controller, Board-to-Board connectors, smaller capacitors and transistors also removed, and the components are simply mated to the PCB to represent solder connection methods. Figure 2.2 shows the assembly as a simple model. Figure 2.2: Simplified Model of the SSD for Simulation Purposes. To replicate how an SSD is dropped while in a server, a jig must be created. This jig is a replication of how the SSD mounts in the server. Figure 2.3 shows the SSD 6 mounted in the jig. The drop test simulation is performed on the x, y and z-axis to determine the stresses, strains and displacements on the drive. Figure 2.3: SSD in Jig. The drop test simulation on the x-axis requires the SSD in the jig mounted on the drop test block according to its x-axis. Figure 2.4 shows the SSD properly mounted in the jig and drop test block. 7 Figure 2.4: SSD in the Jig and Block. With the SSD in the jig and mounted to the drop test block the setup can be performed. The setup begins by defining all the materials that closely, if not exactly, resemble the make-up of the components in the SSD, the jig and drop test block from a mechanical material property. The mechanical material properties are the yield strength, modulus of elasticity etc. Figure 2.5 lists the parts and the materials of the parts. 8 Table 2.1: List of Parts and Material Properties. Next, the drop test simulation parameters need to be defined. That is, the drop is being simulated at a height of 0.5m from its centroid and dropped at 9.81m/s^2. Figure 2.5 shows the setup of the drop test simulation with the arrow pointing toward the direction of the drop and the SSD mounted in the Jig according to its x-axis. Figure 2.5: Drop Test Simulation X-axis. The drop test simulation is also conducted on the y and z-axis of the SSD. The drop test simulation setup remains the same as far as the material of the parts, and 9 procedure of the drop; however, the orientation of the SSD changes. Figure 2.6 and 2.7 shows the drop test simulation of the SSD on the y and z-axis, respectively. Figure 2.6: Drop Test Simulation of the SSD on the Y-axis. 10 Figure 2.7: Drop Test Simulation of the SSD on the Z-axis. 11 2.3 Drop Test Simulation Results: The simulation was performed on the x, y and z-axis and the results show stresses, strains and displacements on the SSD at 1000 microseconds. In order to view the stresses, strains, and displacement from the simulation the block is hidden from the results view. The x-axis simulation results were good, in that, it showed what would be realistic results. Figure 2.8 shows the von Mises stress results, and stress occurs on many areas of the SSD. Upon inspection of the von Mises stress plot, there are noticeable stresses in the front and back of the SSD in the range of 355 kPa to 100 MPa. Figure 2.8: von Mises Stress Plot for the X-axis Drop Test Simulation. Further inspection of the von Mises stress plot using probes validates the stress values. Figure 2.9 shows a probe on the PCB, NAND, DRAM and bracket connection. The stresses are 5.6 MPa for the solder joint connection on the PCB, 18.1 MPa for the NAND, 18 MPa for the DRAM and the bracket connector had a stress of 51 MPa. These 12 stresses are below the yield strength of each component, which means they will not undergo any plastic deformation. The leads on the capacitor are also worth looking at. Figure 2.10 shows the von Mises stress and there is noticeable stress due to the SSD being dropped on the x-axis, with the stress on the lead of the capacitor 1GPa on the left lead and 1.3 GPa on the right lead. There is quite a bit of stress on the capacitor leads because of the small size compared to the capacitor body. It is a safe assumption to make that the leads will be experiencing high stress/strain values, and will need to be monitored closely. Figure 2.9: von Mises Stress Plot of the Components for the X-axis Drop Test Simulation. 13 Figure 2.10: von Mises Stress Plot on the Capacitor Lead for the X-axis Drop Test Simulation. The strain results are also examined, but specifically with the probe on the NAND, DRAM, PCB, bracket connector and the leads on the capacitor. Figure 2.11 is the equivalent strain on the NAND, DRAM, PCB and bracket connector. Strain is a unit less measure because it is the change in length at a point of measure divided by the total length. The strain values are 191.7 on the NAND, 292.3 for the DRAM, 395.3 for the PCB and 1694 for the bracket connector. Figure 2.12 shows the equivalent strain for the leads of the capacitor, which are 6.5e-2 and 3.5e-2. High strain values are also seen in the leads, especially in the bend, because of the orientation of the drop and the size of the leads compared to the capacitor body. 14 Figure 2.11: Equivalent Strain Plot of the Components for the X-axis Drop Test Simulation. Figure 2.12: Equivalent Strain Plot of the Capacitor Leads for the X-axis Drop Test Simulation. 15 The displacement of the SSD due to the x-axis drop simulation is also a key interest. The results quantify the amount of distance parts of the SSD move. Figure 2.13 illustrates the displacement, and a probe on the SSD pinpoints the actual displacement value. In the figure, the maximum displacement on the SSD is in the capacitor, as it displaced 3.9 mm. This occurrence is within reason due to the orientation of the drop and the size if the leads connected to it. The overall range in which the SSD displaces seems to be in line with what is expected. The movement the SSD experiences during impact is consistent with what was observed during drop. The displacement of the capacitor leads are also looked at, and shown in Figure 2.14. The results show the leads are displacing 1.4 mm from the left lead and 1.3 from the right lead. Figure 2.13: Displacement Plot for the X-axis Drop Test Simulation. 16 Figure 2.14: Displacement Plot of the Capacitor Leads for X-axis Drop Test Simulation. Next, the results for the drop simulation on the y-axis are examined. The von Mises stress, equivalent strain and displacement are plotted and probed at significant locations. The von Mises stress is plotted in Figure 2.15, and the stresses are probed at the same locations as in the x-axis results. The locations are the NAND, DRAM, PCB, bracket connector, and the capacitor leads. Probing the location of the NAND, DRAM, and bracket connector, the plot shows that the stresses are 27 MPa, 7 MPa, 7.88 MPa and 62.8 MPa, respectively. The von Mises stress is also probed for the capacitor leads. Figure 2.16 shows the results of the probed capacitor leads and the values are 843 MPa from the left lead and 1.3 GPa from the right lead. 17 Figure 2.15: von Mises Stress Plot of the Components for the Y-axis Drop Test Simulation. Figure 2.16: von Mises Stress Plot of the Capacitor Leads for the Y-axis Drop Test Simulation. 18 The equivalent strain is probed at the same locations as the stress and in Figure 2.17 the results are plotted. The strain at the probed locations is 204.5 at the NAND, 97.3 at the DRAM, 157.7 at the solder joint connection on the PCB and 1558 at the bracket connector. The equivalent strains are also looked at in the capacitor leads. Figure 2.18 is a plot of the probed capacitor leads. The results of the probe show the capacitor leads experiencing 8722 and 9916 from the left and right leads, respectively. Figure 2.17: Equivalent Strain Plot of the Components for the Y-axis Drop Test Simulation. 19 Figure 2.18: Equivalent Strain Plot of the Capacitor Leads for the Y-axis Drop Test Simulation. The displacement is also probed on the y-axis drop to determine the distance the SSD moves. Figure 2.19 is a plot of the probed areas with the displacement values shown. The probed areas show a displacement of 0.83 mm at the NAND, 0.87 mm at the DRAM, 0.69 mm at the PCB, and 0.93 at the bracket connector The displacement of the capacitor leads are probed, and the left lead is displaced 0.96 mm while the right is displaced 0.81 mm (Figure 2.20). 20 Figure 2.19: Displacement Plot for the Y-axis Drop Test Simulation. Figure 2.20: Displacement Plot of the Capacitor Leads for the Y-axis Drop Test Simulation. The results of the z-axis drop test simulation are finally looked at, in which the von Mises stress, equivalent strain, and displacement are probed at the NAND, DRAM, 21 Controller, bracket connector, and the capacitor leads. The von Mises stress results are plotted in Figure 2.21 and the results show the probed areas stated above. The von Mises stresses at the probed locations are 7 MPa at the NAND, 4 MPa at the DRAM, 4 MPa at the solder joint of the Controller, and 56 MPa at the bracket connector. The von Mises stress at the capacitor leads are shown in Figure 2.22 and the stresses are 162 MPa for the left lead and 57 MPa for the right lead. Figure 2.21: von Mises Stress of the Components for the Z-axis Drop Test Simulation. 22 Figure 2.22: von Mises Stress of the Capacitor Leads for the Z-axis Drop Test Simulation. The results for the equivalent strain are probed at the same location on the SSD as the stress results. Figure 2.23 shows a strain plot and the results of the probed areas. The strain results of the probed areas are 75 at the NAND, 22 at the DRAM, 186 at the solder joint of the PCB and 961 at the bracket connector. The strain on the bracket is high and seems to correlate to the nature of the drop test simulation. The strain in the capacitor leads were also probed and found to have strain values of 519 and 573 in the left and right leads, respectively (Figure 2.24). These values of strain on the leads seem good because of the orientation of the drop simulation and the size of the leads. 23 Figure 2.23: Equivalent Strain of the Components for the Z-axis Drop Test Simulation. Figure 2.24: Equivalent Strain of the Capacitor Leads for the Z-axis Drop Test Simulation. Then the displacement of the components is probed to get an idea of how much displacement is occurring. Again, the probed areas are the NAND, DRAM, solder joint 24 area of the Controller, the bracket connector, and the leads of the capacitor. Figure 2.25 shows the displacement in the probed areas and they are 0.6 mm in the NAND, 0.88 mm in the DRAM, 1.0 mm in the solder joint area of the Controller, and 1.0 mm in the bracket connection. The capacitor leads also experience displacement due to the drop, and the results are displacements of 0.25 and 0.22 mm in the left and right leads, respectively (Figure 2.26). Figure 2.25: Displacement Plot for Z-axis Drop Test Simulation. 25 Figure 2.26: Displacement Plot of the Capacitor for the Z-axis Drop Test Simulation. 2.4 Drop Test Verification: The drop test verification is an actual test method performed on the SSD mounted in the jig and block of a Lansmont M23 Drop Test Machine. This test verifies the simulation performed in SolidWorks, in terms of determining the stresses, strains and displacements of the SSD. However, prior to placing the SSD in the jig and Lansmont, the SSD is wired with several Vishay uniaxial strain gages so that strain measurements can be taken during the impact. Two types of strain gages are used, and the specifications can be seen in Figure 2.27. The placements of the strain gages, shown in Figure 2.28, are located on the DRAM, NAND and on the solder joint connection of the Controller on the PCB. The strain gages are oriented normal to the y and z-axis. Therefore, there will be a strain gage oriented normal to the y-axis on a DRAM, and a strain gage oriented normal to the z-axis on a second DRAM. The orientation placement of the strain gage is also used on the NAND and controller. The placement of the strain 26 gage on the PCB behind the solder joint of the controller is an industry standard test. This test determines the strain on solder joint connection according to IPC/JEDEC-9703 Mechanical Shock Test Guidelines for Solder Joint Reliability. Figure 2.27: Strain Gage Type and Specifications. 27 Figure 2.28: Strain Gage Locations. The strain data are logged via a Vishay System 6000 data logger using multiple Model 6010A strain gage cards to register the strain gage data (Figure 2.29). The Vishay System 6000 features a data acquisition rate up to ten thousand samples per second per 28 channel, with a maximum throughput of two hundred thousand samples per second. This acquisition rate is optimal for conditions involving dynamic forces. Figure 2.29: Vishay System 6000 and Model 6010A Strain Gage Cards. The Lansmont M23 Drop Test Machine performs the drop and holds the SSD and jig. The drop test verification is conducted following the same conditions as in the simulation. That is, the drop height will be 0.5 meters from the centroid and dropped at 9.81 m/s^2. Figure 2.30, 2.31 and 2.32 show the SSD and Jig in the Lansmont in the x, y and z- axis, respectively. 29 Figure 2.30: X-axis Drop Test Verification. Figure 2.31: Y-axis Drop Test Verification. 30 Figure 2.32: Z-axis Drop Test Verification. 2.5 Drop Verification Test Results: With the drop test simulation completed, the SSD is put through a verification test to determine the actual strain values. Moreover, the strain gage data acquired came from seven strain gages, due to equipment constraints. The strain gages used to acquire the results came from the solder joint connection strain gages in the y and z axis orientation, the DRAM in both axes, the board-to-board connection solder joint area in both axes and the NAND in the z axis orientation. Having performed the drop verification test for the x, y and z-axis with strain gages mounted on the various locations, the Vishay System 6000 data logger was able to capture, at 10000 samples per second, the results of the test. The result of the x-axis drop is presented in a plot in Figure 2.33. The figure shows the amount of strain vs. time that the SSD experiences during its initial impact to its final resting position. In the plot, the SSD experiences three modes of strain and this 31 is evident by the three strain peaks in the plot. The initial impact is seen by the initial strain peak, while the second peak measured the components of the SSD experiencing more strain and the third peak is where the components of the SSD experienced the most strain. The duration of time from the initial impact to the final impact was roughly 0.2 seconds, with the maximum strain value of the initial impact being 2000 from the strain gage mounted to test the solder joint connection oriented in the y-axis. The maximum strain in the third peak of the plot is 6000 and is experienced by the same strain gage testing the solder joint connection, with the gage oriented in the y-axis. The solder joint connection of Controller experienced the maximum strain in compression. The plot also distinguishes maximum strain values for the other components, but for simplicity, the maximum strain is being reported for the component experiencing the strain. Strain vs. Time X-axis Drop 6000 Solder joint Z axis 2000 Solder joint Y axis 0 -2000 1 109 217 325 433 541 649 757 865 973 1081 1189 1297 1405 1513 1621 1729 1837 1945 Strain (Unitless) 4000 DRAM Y axis DRAM Z axis B2B connector Z axis -4000 B2B connector Y axis NAND Z axis -6000 -8000 Time (10^-5 seocnds) Figure 2.33: Strain Plot of the Components for the X-axis Drop Test. 32 The strain plot of the components for the y-axis drop test is shown in Figure 2.34. The results show multiple strain measurements on the components of the SSD; however, the maximum strain in the initial impact mode and maximum peak strains are looked at. The result of the plot show a maximum strain value of 468 in the initial impact mode experienced by the solder joint connection, followed by 6000 in the second event. The duration of time from initial impact to final resting position is 0.02 seconds. Strain vs. Time Y-axis Drop 1000 Solder joint connection Z axis 0 Solder joint connection Y axis 1 16 31 46 61 76 91 106 121 136 151 166 181 196 211 226 Strain (Unitless) 2000 DRAM Y axis -1000 DRAM Z axis -2000 B2B connector joint Z axis -3000 B2B connector joint Y axis -4000 NAND Z axis -5000 Time (10^-5 seconds) Figure 2.34: Strain Plot of the Components for the Y-axis Drop Test. The strain values of the components for the z-axis drop test are shown in Figure 2.35. The plot in the figure reports the stain measurement for the components of the 33 SSD. The plot shows multiple strain measurements at multiple locations on multiple components, but the maximum strain at the initial impact and overall maximum strain is examined. The maximum strain at the initial impact region is 429 and is the result of the solder joint behind the Controller, with the strain gage oriented in the z-axis. The maximum strain occurs 0.0016 seconds by the same strain gage and component at roughly 3000. Strain vs. Time Z- Axis Drop 1500 1000 0 -500 1 127 253 379 505 631 757 883 1009 1135 1261 1387 1513 1639 1765 1891 2017 Strain (Unitless) 500 -1000 Solder joint connection Z axis Solder joint connection Y axis DRAM Y axis DRAM Z axis -1500 B2B connector joint Z axis -2000 B2B connector joint Y axis -2500 NAND Z axis -3000 -3500 Time (10^-5 seconds) Figure 2.35: Strain Plot of the Components for the Z-axis Drop Test. 2.6 Drop Test Simulation and Verification Analysis: Having acquired the results from the drop test simulation and the drop test verification it is important to see how well the results correlate. Correlating the results will help determine the accuracies of the simulation as a predictive tool for the 34 verification test. Section 2.5 detailed the results of the verification test for the drop of the axis with plots of strain versus the total time of the drop, from initial impact to rest. For simplification of the simulation and the verification, the analysis will examine the first 1000 microseconds, which is the initial impact to 1000 microseconds after impact. The values after 1000 microseconds will be compared and analyzed. Figures 2.36, 2.37 and 2.38 are the results from the verification and the simulation as a plot. Figure 3.22 show the results obtained when the SSD experiences the moment of initial impact to 1000 microseconds after the initial impact. The rows in Figure 2.36 list the components with the strain gage and its orientation. The columns in the figure are the strain data. Column 12 will be used when correlating measurement with the simulation data for the x-axis drop. Table 2.2 puts the results in a spreadsheet to be analyzed. The figure shows the component relationship from the verification test and the simulation, in terms of strain measurement/orientation and the element value associated to the verification measurement. The figure also displays the strain results from the verification and simulation, and shows a percent difference. Further examination of the data in the figure shows inconsistencies in the data correlation, ranging from 3.0 to 146 percent differential. Reasoning behind this inconsistency will be explained at the end of the section and in the conclusion of the paper. 35 Strain vs. Time X-axis Drop Strain (Unitless) 200 0 -200 -400 -600 -800 -1000 -1200 -1400 1 2 3 4 5 6 7 8 9 10 Solder joint connection Z axis -4 10 -15 18 -18 12 0 -9 24 -30 -95 -1208 Solder joint connection Y axis -10 9 -5 -1 7 -13 20 -21 23 -11 -10 -299 DRAM Y axis -10 5 -2 -2 2 -7 12 -14 12 -8 DRAM Z axis -16 -23 -23 -20 -20 -16 -24 -13 -27 -15 -20 B2B connector joint Z axis 4 -15 -11 2 -31 21 -43 16 11 12 -2 -154 55 -25 -16 -103 -499 B2B connector joint Y axis 16 0 0 10 -16 24 -22 24 -9 -1 -29 -212 NAND Y axis 19 21 21 16 16 18 17 22 19 18 21 Figure 2.36: Strain vs. Time (10^-5 seconds) for the X-axis Drop Test. Figure 2.37: X-axis Drop Test Results with Strain Gage Oriented in the Y-axis. -13 36 Figure 2.38: X-axis Drop Test Results with Strain Gage Oriented in the Z-axis. Table 2.2: X-axis Drop Test Simulation and Verification Analysis Correlation. Analysis of the drop test simulation and verification for the y-axis at the moment of initial impact to 1000 microseconds after impact has concluded the data set and plots in Figure 2.39 for the verification test, and in Figures 2.40 and 2.41 for the simulation. Column 11 of Figure 2.39 will be used when correlating data with the simulation. Moreover, Table 2.3 is an analysis correlation of the data including the percent difference for the strain measurement. Looking at the percent difference of the strain 37 data from the simulation and the verification test, the results seem to correlate a little better; however, still a bit of a discrepancy. The range of percentage difference varies from 1.6 to 55 percent. This variance is concerning, but reasons for this occurrence are explainable. Strain vs. Time Y-axis Drop Strain (Unitless) 200 150 100 50 0 -50 -100 -150 -200 -250 -300 -350 1 2 3 4 5 6 7 8 Solder joint connetion Z axis 3 12 -181 -209 36 1 -48 -4 -233 -279 -311 Solder joint connection Y axis -3 20 -145 -214 16 19 -53 6 -195 -268 -284 DRAM Y axis -4 12 -32 -82 -16 18 -21 5 -46 -98 -97 DRAM Z axis -4 -24 -22 32 18 -30 3 -20 -21 26 16 B2B connector joint Z axis -5 -9 84 99 -29 -7 22 -3 110 128 142 B2B connection joint Y axis 11 -30 37 137 12 -36 40 -29 64 134 130 NAND Y axis 11 7 6 18 12 5 12 5 12 18 Figure 2.39: Strain vs. Time (10^-5 seconds) for the Y-axis Drop Test. 9 10 11 19 38 Figure 2.40: Y-axis Drop Test Results with Strain Gage Oriented in the Y-axis. Figure 2.41: Y-axis Drop Test Results with Strain Gage Oriented in the Z-axis. 39 Table 2.3: Y-axis Drop Test Simulation and Verification Analysis Correlation. The z-axis drop test simulation and verification has resulted in data plots shown in Figure 2.42, 2.43, and 2.44. The figures show the strain data from the location of the strain gage. In Figure 2.42, the results of the drop verification test are shown from the moment of initial impact to 1000 microseconds after impact. Column 12 is the data set used to correlate the simulation results and the verification test results. Table 2.4 is a spreadsheet of the correlation of data between the simulation and the verification test results at 1000 microseconds after impact. The percent difference between the strain data of the simulation and the verification test is conducted as a way to better analyze the data. The percent differences range from 0.35 to 79.2 percent. Again, this inconsistency is of concern. 40 Strain vs. Time Z-axis Drop Strain (Unitless) 400 300 200 100 0 -100 -200 -300 -400 -500 1 2 3 4 5 6 7 8 9 10 11 12 Solder joint connection Z axis 10 -51 -21 16 -47 -47 -15 35 -52 -3 -38 -429 Solder joint connection Y axis 12 -42 -30 23 -36 -44 -22 29 -37 -19 -1 -361 DRAM Y axis -1 -2 12 4 DRAM Z axis -17 -16 -13 -17 -19 -15 -15 -18 -25 -13 -17 -12 2 -21 B2B connector joint Z axis -23 32 B2B connector joint Y axis -8 21 7 NAND Y axis 21 11 29 -4 -16 -34 4 -15 -14 30 2 -14 -39 -13 16 15 5 27 9 20 24 -21 -26 42 -32 9 271 -15 21 -2 11 188 27 6 25 15 -28 Figure 2.42: Strain vs. Time (10^-5 seconds) for the Z-axis Drop Test. Figure 2.43: Z-axis Drop Test Results with Strain Gage Oriented in the Y-axis. 41 Figure 2.44: Z-axis Drop Test Results with Strain Gage Oriented in the Z-axis. Table 2.4: Z-axis Drop Test Simulation and Verification Analysis Correlation. Upon analysis of the drop test simulation and the verification, and the data not correlating in terms of consistent percent differences, this is still considered a success. While the outcome is not officially within means of a 90 percent success rating, the data acquired and the results were acceptable. 42 The inconsistent results in data can be attributed to many things like the mesh of the simulation was not fine enough, the model overly simplified, or the assumed material was not an close enough to the real thing. The drop verification test could also be attributed to the results being a bit off. Things like the strain gage malfunction could have caused error in measurements. Ultimately, the results show that trying to determine a dynamic force or in this case, a dynamic strain, is tricky. In static conditions strain measurements are easily reproducible and the measurement is more accurate, while in dynamic conditions the results in terms of measurement data can be inconsistent and hard to reproduce. Alternatively, possible solutions for acquiring consistent data sets can be done by starting the drop at lower heights and correlating that data with the simulation to determine if in fact the strain gages can handle dynamic measurements. 43 CHAPTER 3 CASE STUDY 2: FLOW SIMULATION, VERIFICATION, RESULTS AND ANALYSIS 3.1 Flow Simulation: Next, a flow simulation is performed on the SSD to determine the nature of the flow, temperature distribution, and maximum temperatures, across and through the SSD inside a control volume during simulated operating conditions. A simplified model is used when performing the flow simulation (Figure 2.2). The model is placed in a control volume so that an internal flow simulation can be used to determine set goals for inputted data. The flow simulation model in Figure 3.1 shows the SSD in a volume controlled enclosure. Figure 3.1: SSD in Control Volume for Flow Simulation. 44 Prior to the goals being established and defined, data such as Solid Materials, Boundary Conditions, Heat Sources, and Radiative Surfaces must be set for the system. The Solid Materials are the components of the SSD and are defined using thermal properties of the material selection for the component. Table 2.5 shows the components, material selection and thermal property associated with it. Table 2.5: List of Parts and Thermal Properties. Next, we look at the Boundary Conditions that are comprised of the inlet and outlet of the control volume. The inlet has a boundary condition that describes airflow being forced through at a constant velocity of 1.0 m/s while the outlet is being exposed to the environment at 101325 Pa and an ambient temperature of 22 degrees Celsius. Figure 3.2 shows the SSD in the control volume with the boundary conditions. Figure 3.2: Boundary Conditions. 45 The Heat Sources are inputted into the simulation. The Heat Sources are the heat generated from the components on the SSD. There are four heat-generating components, the NAND at 19W total, DRAM at 5W total, Controller at 5W, and the Power at 2W total. Figure 3.3 shows the heat generation from these components. Figure 3.3: Heat Sources. The Radiative Surfaces are the components of the SSD being defined as either the metallic or the non-metallic surface. The metallic surfaces are the capacitor, bracket, screws, pem nut and heat exchanger. The non-metallic surfaces are the remaining components that make-up the SSD. Figure 3.4 show the metallic surface highlighted in light blue while the non-metallic surfaces are colored. 46 Figure 3.4: Radiative Surfaces. With the data inputted in the flow simulation, the goals can be set. The goals are desired results that are set for specific finding. For this simulation, there are two goals being desired and they are the Global and Volume Goals. The Global Goals look at desired results from inside the control volume as a whole, while the Volume Goals are specific to the components of the SSD. Global Goals desired are the maximum temperature inside the control volume, specifically the SSD, and the maximum velocity that flows inside and out of the control volume and through the SSD. The Volume Goals desired are the maximum temperature of the NAND, DRAM, Controller, Inductors, Voltage Regulators, and Capacitor. 3.2 Flow Simulation Results: The results of the flow simulation were based on the goals defined in the simulation setup, along with plots in the figures of this section. Table 2.6 shows the results of the Global and Volume Goals. The Global Goals looks at the total system and determines the component with the maximum temperature. The velocity that flows through the system is a result of the global goal as well. The results in the table also 47 show the maximum temperature in the SSD and its individual components. The maximum velocity was also determined and displayed in the results of the table. The table also shows that the component with the maximum temperature due to heat generation was the power components, at 67.79 degrees Celsius. This temperature is normal for the power components, which normally operates at 125 degrees Celsius, but not more than 150 degrees Celsius. The results that are important for the analysis of the flow simulation and verification, are the thermal behavior of the NAND, DRAM, and Controller (FPGA), in terms of average and maximum temperature. The maximum temperature of the NAND, DRAM, and Controller (FPGA) are 66.88, 67.59, and 55.88 degrees Celsius, respectively. Table 2.6: Global and Volume Goals Results. The flow simulation also gave visual results of the velocity flow in the control volume as it pertains to the SSD. Figure 3.5 is a plot of the velocity profile in the control volume. It displays the flow trajectory along with the path throughout the control volume and within the SSD. The flow trajectory is constant and smooth in the inlet region, but displays noticeable change when it is impacted by larger components, such as the heat sink or the board-to-board connector. The larger components slow the flow down, and as 48 the flow is redirected, the velocity increases. Eddies are apparent as the flow exits the SSD through the bracket. The dark blue circles are where they are occurring, most notably in the upper left region. Figure 3.5: Flow Trajectory Plot. The simulation results also display a surface plot of the components and its temperature. From the plot, the temperature range is between 33 and 63 degrees Celsius. This temperature range is from the components in the SSD. This plot shows the temperature due to the heat being generated from the component (Figure 3.6). 49 Figure 3.6: Surface Plot. The cut plot shows an image of the temperature and flow through the SSD and it is seen as combination of the flow and surface plot. The solid colors are the temperature of the component, while the arrows are the direction of the flow. The cut plot looks at the heat generation and takes into account the velocity at the inlet and provides an image of the direction of the temperature flow. The temperature flow is moving from the inlet to the outlet. Figure 3.7 is the cut plot of the flow simulation. 50 Figure 3.7: Cut Plot. 3.3 Flow Verification: The flow verification is used to verify the flow simulation performed in SolidWorks. This test will verify how the temperatures of the components in the SSD are varying. The process of the flow verification involves mounting components of the SSD with thermocouples, namely the NAND, DRAM, and Controller. An Agilent 34972A LXI Data acquisition logger will log the data from the thermocouples during the test. Then the SSD is placed in a wind tunnel that will channel airflow through the SSD, while undergoing four sequenced read/writes and four random read/writes over 90 minutes per read/write, with a total run time of 720 minutes. The wind tunnel uses a Thermonics T2500 Precision Temperature Forcing System to induce forced air into the tunnel at a constant velocity of 1.0 m/s, with the ambient temperature at 22 degrees Celsius. The test system is shown in Figure 3.8. The Thermonics Temperature Forcing System provides 51 airflow through the larger end of the triangular shaped box, and channeled into the smaller box section where the flow will pass through the SSD mounted with thermocouples. The Agilent data logger will capture the temperature change and the data will be exported into Excel. SSD mounted with thermocouples inside Thermonics Temperature Forcing System Figure 3.8: Wind Tunnel Verification Test. The SSD sits in the wooden box at the right of the wind tunnel (Figure 3.8) and is mounted with numerous thermocouples on the NAND, DRAM and Controller. Figure 3.9 shows the mapping of the thermocouple on the SSD during the flow verification test. 52 Thermocouples are placed on twelve NAND components, two DRAM components and the single controller, which are distinguished by the red star in the map. 53 Figure 3.9: Mapping of Thermocouples (marked by star). 3.4 Flow Verification Results: The results of the flow verification test were gained by placing the SSD, mounted with thermocouples in a wind tunnel, while an Agilent data logger acquired the data, shown in Table 2.7, while the Thermonics temperature-forcing system provided constant velocity into the wind tunnel. Moreover, the thermocouples were mounted onto the NAND, DRAM, and Controller while the data logger acquired the results of the flow verification test. The flow test required the components endure four sequenced read/write and four random read/write at a time of 90 minutes per read/write. Table 2.7 shows the average maximum and maximum temperature of the NAND, DRAM, and Controller component from the flow verification test. These results show the average temperature of the twelve NAND components, the two DRAM components and the Controller, over the total time duration of the flow test. The acquired results are the average maximum temperature of 56 degrees Celsius for the NAND, 64 degrees for the 54 DRAM, and 56 for the Controller. In addition, the maximum temperatures for the components are 66.5 degrees for the NAND, 65.1 degrees for the DRAM, and 56 degrees Celsius for the Controller. These results show that the SSD runs at optimal thermal conditions when enduring stresses such as read/write functions over a period time. Table 2.7: Average Maximum and Maximum Temperature from Flow Verification Test. 3.5 Flow Simulation and Verification Analysis: The results from the flow simulation and the verification test are compared and analyzed to determine how well the results correlate. For the purpose of this analysis, the average temperature and maximum temperature of the NAND, DRAM and Controller will be looked at from the flow simulation, and compared with the flow verification. The results from the flow simulation and verification test are shown in Table 2.8, and a comparison is formulated. Flow Simulation Results: Verification Results: Table 2.8: Flow Simulation Results (Top) and Verification Results (Bottom). 55 Looking at the average temperature of the NAND, DRAM and Controller from the flow simulation, the respective values are 66.88, 67.60, and 55.89 degrees Celsius. These values are then compared to the average temperature of the flow verification test and those values are 56, 64, 56 degrees Celsius. There are discrepancies in the average temperature from the flow simulation and the verification test. A percent difference formula is used to determine the variation of these discrepancies by using Figure 3.10, where Ts is the temperature of the simulation and Tv is the temperature from the verification test. The percent difference for the average maximum temperature of the NAND, DRAM and Controller are, 19.0, 3.8, and 0.2 percent, respectively. It can be said that the basis for the high percent difference of the NAND is due to the sample size used to determine the average maximum temperature. The sample size for the verification test is twelve NAND components, while the simulation takes into account all sixty-four NAND components. While the NAND percent difference came out a bit high, but in the acceptable range, the DRAM and Controller resulted in a very acceptable percent difference values. The simulation and the verification test seem to correlate well in terms of analyzing the average maximum temperatures. Figure 3.10: Percent Difference Equation. However, it is also important to correlate the maximum temperature from the flow simulation versus the flow verification test and calculate the percent differences, as well. The maximum temperature results are key indicators to how the components in the SSD 56 are operating under the conditioned read/write sequence. The maximum temperature result also indicates that the component is generating its most power under the read/write sequence. As the maximum temperatures of simulation and verification test are examined, Table 2.8 shows that the maximum temperature for the NAND, DRAM and Controller are 66.89, 67.60, and 55.89 degrees Celsius for the simulation, and 66.5, 65.1, and 56 degrees Celsius for the verification. From a visual inspection they look spot on, but using the percent difference equation, the percent differences are 0.5, 3.8 and 0.2 percent. The percent differences further validates the visual inspection, and says that the simulation is almost a perfect indicator for the verification test. Furthermore, the results of the flow simulation and the flow verification test correlate well with each other. 57 4. Conclusions: Moore’s law has been an integral part of the rapidly growing semiconductor industry, bringing about massive increases in density per size to the chip sets over time. With the increase in density of components and the expanding market to provide memory, Solid State Drives need to undergo steps to ensure that they are created to handle the mechanical rigors they may encounter. That is where mechanical design, simulation, and verification are essential to the process of product development. As shown with the content in this paper. Modeling and performing simulations is an extremely useful predictive tool to aid in the development of products. The verification process enhances the tool by being able to correlate the data and analyzing its solutions. While the results of the drop simulation and verification test did not correlate as well as the flow simulation and verification test, the process was ultimately successful. When in the design phase of products, there are no prototypes to test. The SolidWorks simulation tools are accurate enough to aide in the design process to get a prototype that performs as intended, and reduce design iterations to the final product. Upon further review and understanding of trying to measure dynamic loading, in this case strain from a drop test, considerations must be made in order to produce consistency in the results being acquired. Future goals will be to reproduce such consistent data sets and to bring consistent results between simulations and verification of a drop test so that the simulation tool can be a more accurate predictive tool. 58 Reference: Ekker, Neil "Solid State Storage 101: An introduction to Solid State Storage." Jan. 2009. SNIA.org. Retrieved 20 Aug. 2013. IPC/JEDEC-9703. “Mechanical Shock Test Guidelines for Solder Joint Reliability.” Mar. 2009. Jansen, Ng. "Micron Announces World's First Native 6Gbps SATA Solid State Drive." Dec. 2009. DailyTech.com. Retrieved 25 Aug. 2013. “Magnetic-Core Memory.” Wikipedia.com. Retrieved 23 Aug 2013. “Moore’s Law.” Intel.com. Retrieved 25 Aug. 2013. Rent, Thomas M. “Origins of Solid State Drives.” March 2010. Storagereview.com. Retrieved 23 Aug. 2013.