University of Delaware and Case-New Holland Senior Design Project 2005 Design Proposal - 10/24/2005

Transcription

1 University of Delaware and Case-New Holland Senior Design Project 2005 Design Proposal - 10/24/2005 Team 11: Erik Pearson Paul Rodriguez Casey Strohmeyer James Woodhouse Dr. James Glancey * Rick Strosser Ed Priepke * Mechanical Engineering Faculty, University of Delaware Instrument/Controls Engineer, Case-New Holland Sr. Project Engineer, Case-New Holland

2 Table of Contents Executive Summary... 1 Introduction... 2 Work Accomplished... 3 Performance Expectations... 7 Prototype vs. Production... 8 Cost Projections... 9 Resource Management Validation In System Rotation Test Out of System Rotation Test Shaft Run-Out Test Linear Displacement Test Testing Conclusions Conclusion Future Plans Appendix A Steer-by-wire Specifications Appendix B Max Displacement of Shaft in Torsion Appendix C Customer Wants Appendix D System Level Overview Diagram Appendix E Technical Drawings Appendix F The Mechanical Linkage in the Windrower Being Removed Appendix G - Assembly Instructions List of Figures Figure 1: Current CNH Windrower With Mechanical Steering Linkage... 1 Figure 2: Change of System from Mechanical Linkage System... 2 Figure 3: BEI NCAPS Sensor Technology... 4 Figure 4: Steering Column... 4 Figure 5: Exploded Assembly View... 5 Figure 6: Assembled Prototype... 6 Figure 7: Assembly view shown in mounting location... 7 Figure 8: Angular Displacement of Shaft in Torsion... 8 Figure 9: Finite Element Analysis of Shaft... 8 Figure 10: Cost Analysis Figure 11: Test Result Sample Figure 12: Metric and Target Values Summarry... 13

3 Executive Summary The University of Delaware and Case-New Holland (CNH) are undertaking a joint venture to develop a steer-by-wire system for CNH windrowers. By replacing the current mechanical steering linkage with a purely electronic system, CNH should be able to reduce their manufacturing costs and offer better performance. The electronic steering capability also enables future integration of electronic guidance systems. The University of Delaware group has worked to develop several sensor and mounting location alternatives. We have chosen a sensor and mounting location and developed a detailed design based on specifications and constraints given by CNH. A fully functioning prototype has been built including an optical encoder which will be used only for lab testing purposes and not be included in the actual production model of the steering system. Using LabVIEW software for data collection; detailed experimental procedures were followed to verify the prototype s capabilities. The error that appeared in the sensor signal was considerably greater than expected but after a number of tests to isolate the source of the error it was found that it was caused by excessive run-out on the shaft extension. Future steps include prototype installation in a CNH Windrower for track testing and manufacturing procedure design. Figure 1: Current CNH Windrower With Mechanical Steering Linkage -1-

4 Introduction Currently the steering system of CNH windrowers uses a purely mechanical linkage. The mechanical linkage conveys the rotation of the steering wheel to the pinion arms on the hydrostatic pumps used to control the drive wheel speed of the right and left front wheels. The mechanical linkage can be seen in Appendix F. Removing the mechanical linkage and replacing it with an electronic sensor and two servos to control the pumps converts the steering system to a steer-by-wire setup. The old and new systems are illustrated in Figure: 2 below. The steer-bywire system we designed consists of a steering wheel position sensor that interfaces with the windrower s current electronic control unit and meets the specifications outlined in Appendix A. By replacing the mechanical steering linkage with a steer-by-wire setup, CNH will be able to reduce manufacturing costs and offer greater functionality to the consumer. The steer-bywire system should allow CNH to hold a competitive edge on the market and increase profit on each unit sold while also increasing the longevity of the current system. CNH can maximize benefits by using a sensor that is already available on the market combined with a simply designed mounting bracket, keeping both materials and labor costs down. An electronic system using non-contacting sensing technology can also reduce the wear in mechanical parts, lowering maintenance cost. Steer-By-wire System Current Mechanical Linkage Figure 2: Change of System from Mechanical Linkage System - 2 -

5 During a discussion between the team and CNH, a list of customer wants was compiled and the metrics used to measure each of these wants were identified. We ranked the metrics and the customers wants by importance to the different customers. We also drafted up target values and constraints in a jointly made specification document attached as Appendix A. This document stems from the standards from the TTA Group 1 on steer-by-wire systems, CNH s own requirements on electrical systems, and benchmarking of the current industry. After reviewing the wants, metrics, and constraints, we were able to draw up a number of designs. Using these designs and a ranking system, we chose the best concept for further design. A shaft and a bracket for holding the various components were developed using a detailed design process. After several re-designs a final prototype design was set for construction. We utilized laser sheet metal cutting and bending equipment at CNH to fabricate and inner an outer bracket to encase and hold steady the sensor, damper, and shaft. CNH rapid prototyping machinery was implemented in the manufacturing of the adjustment collar that couples the shaft to the sensor. We machined our shaft extension from aluminum round stock in the University of Delaware s College of Engineering Student Shop. Work Accomplished The team has developed a solution for the final design that is composed of several subcomponents: rotary position sensor, viscous damper, shaft extension, and a bent sheet metal bracket to hold it all in place. First the team looked into finding an appropriate angular position sensor. We originally chose a magnetic Hall Effect sensor made by Delphi that could measure 1800º of rotation on two independent channels. This sensor met or exceeded most requirements and target values. However, we discovered that Delphi was previously a division of General Motors (GM) that has split off as an independent company. General Motors still holds sole rights to the sensor we were interested in. As a result, we could only acquire the sensor through GM as a replacement part which proved to be prohibitively expensive. Also, when we purchased a sensor through a local dealership, it did not match the specifications document we had received from Delphi 2. All of the difficulties with the Delphi sensor have made it necessary to pursue the previously established contingency plan. After getting in touch with our contact at BEI Duncan Electronics, we obtained the information on a current production Non-Contacting Angular Position Sensor (NCAPS). The NCAPS sensor can meet all design requirements; however, the sensor requires custom design and fabrication, resulting in significant lead time. In order to meet the current schedule, we are using one of the NCAPS sensors that BEI already has in production for our prototype even though it does not have the required redundant outputs. We will continue to use the NCAPS prototype until BEI fabricates our custom sensor. 1 TTA Group refers to an organization that has been working to achieve a list of industry standards in steer-by-wire systems. 2 There was some difficulty in getting part numbers from either Delphi or GM so we think we ended up with a different sensor all together - 3 -

6 Figure 3: BEI NCAPS Sensor Technology The change to the BEI sensor has not altered our plan to mount the sensor in the boot area of the steering column. Although the BEI sensor is slightly larger than the Delphi, we can still mount it and the damper with ample space for the possible future addition of a locking mechanism in the boot. We disassembled the steering column, shown with its outer casing removed in Figure: 4, to remove the shaft in order to get a better idea of how to extend the current shaft to accommodate the various components. For the production model the shaft will be extended with the appropriate profile by the steering column vendor3. However for the prototype we machined a shaft extension that was press fit into the current splined end. The end was cut of and faced flat then a hole was drilled and reamed in the center to accept an interference fit with the shaft extension. Once the two were pressed together another hole was drilled through both and a roll pin inserted to prevent any rotation. Figure 4: Steering Column Figure: 5 shows an exploded view of the assembly. Starting with the top right we have the aluminum shaft extension shown in grey. The white disk below that is the viscous damper. Next is one of the two brackets, shown in blue, with a small red tab used to position the damper and prevent it from rotating. Sandwiched between the two brackets is the BEI sensor. The bottom bracket is shown in red with blue spacers on either side of it. Dimensioned technical drawings of these components can be found in Appendix E. After we solidified the design of the shaft extension, we then began work on the design of the bracket around the shaft and its various components. 3 The vendor has not yet been contacted but CNH has had the shaft custom altered previously and believes this should not be a problem -4-

7 Figure 5: Exploded Assembly View Initial bracket designs were wider and attached to the top surface of the yoke. However the required bends were not practical for typical bending methods. The mounting hardware was moved to the side and the bottom bracket was narrowed (slots allow sensor to overhand edge slightly) to accommodate the necessary hardware. Slots or oversized holes were put into each bracket and the yoke to allow the entire system to be adjustable and to accommodate for machining tolerances. A combination of horizontal slots in the yoke and vertical slots in the top bracket allow for top bracket adjustment in either direction. Oversized holes in the bottom bracket left it free for adjustment in the horizontal and vertical directions. Oversized holes are feasible only in the bottom bracket, because it is sandwiched between the top bracket and the spacers. An extension was added to the horizontal face of the top bracket to place a damper stopping tab. This tab was mounted with an oversized hole to make it adjustable in any direction in the horizontal plane. CNH laser cut and bent the brackets and also cut the spacers. The collar that adapts the shaft profile to the sensor was created from a hard plastic using rapid prototyping equipment at CNH. The shaft extension was turned down and milled in the student shop at the University of Delaware. Additionally a mounting bracket for the optical encoder used for testing as well as a coupler for the encoder and shaft extension were fabricated in the student shop as well. After all fabrication was completed the prototype was ready for assembly. First the new shaft had to be press fit into the existing shaft. Then, we slid the damper up the new shaft followed by the collar and the top bracket. Next, we placed the sensor and it s mounting bracket were added. The spacers were placed between the bracket and the yoke and it was all securely bolted together with M6 carriage bolts. Finally, we mounted the optical encoder to the mounting bracket and installed the shaft coupler. The completed assembly is shown in Figure: 6 below. The entire - 5 -

8 system was placed on an aluminum frame that can support the steering column in any of its possible positions. Existing Column Shaft New Shaft Spacers Inner Mounting Bracket Outer Mounting Bracket Viscous Damper BEI Sensor Collar to Mate New Shaft and Encoder Shaft Encoder Mounting Bracket Optical Encoder for Testing Figure 6: Assembled Prototype Before the prototype was completed the testing software development began. LabVIEW was chosen as the development package to communicate with both sensors and record test data. The sensor from BEI was chosen to have a Controller Area Network (CAN) communication protocol for ease of integration with the electronics already in use in the windrower. In order to use LabVIEW with the supplied Softing PCMCIA CAN card the driver functions are called directly from the dynamically linked libraries. The encoder output signal is a square wave where each pulse corresponds to 1/1200 of a revolution. This signal was captured using a National Instrument data acquisition card. Both signals were then converted to angles and logged in real time

9 Performance Expectations Figure 7: Assembly view shown in mounting location This system design should meet or exceed all previously set requirements within the project scope. The required parts for the system can all be simply machined at CNH or by the steering column manufacturer, purchased from a manufacturer, or are already provided within the current steering column. The sensor needed for the unit will be in compliance with all target values for accuracy, resolution, and drift and design constraints such as redundancy and voltage requirements because BEI plans to produce a custom sensor specifically for our application. The bracket was designed to meet all performance specifications, holding all the other components in place properly and also scoring high on a new want of ease of assembly. Keeping in mind the abilities of CNH s laser metal cutting tools and sheet metal forming tools, a simple to assemble, two piece bracket system was designed. This bracket is the key to keeping assembly time to a minimum. A shaft extension was found to be required early in the design process in order to attach the sensor and damper as needed. The shaft was design to be a 3/8 diameter aluminum piece with ¼ flats as this will be a perfect slip fit into the viscous damper. BEI stated that they will be able to manufacture the sensor to fit this shaft design as well. An engineering stress/strain analysis was conducted to verify that the design of the shaft will not affect the accuracy of the overall system. The angular displacement at the end of the shaft is directly proportional to the torque applied which is 2 Newton-meters maximum and the length of the shaft which will be at most 4 centimeters. The angle is inversely proportional to the second polar moment of inertia of the shaft and the modulus of rigidity of aluminum. The carried out calculation can be seen in entirety in Appendix B and it reveals a maximum angular displacement of E -5 degrees. This small - 7 -

10 displacement is far smaller than the resolution we plan on our sensor providing, so angular displacement affects are negligible. θ = θ = Tl JG degrees E Figure 8: Angular Displacement of Shaft in Torsion Also a Finite Element Analysis was carried out on the shaft using SolidWorks and the results can be seen in Figure 9. The warmer colors symbolize areas of greater static displacement, with brightest red areas showing approximately 9 E -5 inches. Prototype vs. Production Figure 9: Finite Element Analysis of Shaft The current BEI prototype sensor will not be used in the final manufactured version. When the time comes to manufacture the final system, all necessary parts will be available. Due to the circumstances presented in constructing a steer-by-wire system with a prototype sensor, some aspects of the assembly will differ from the production model. The prototype sensor has a continuous 360º rotational capability and one independent CAN bus output. The final sensor is designed to provide two independent outputs. Due to time constraints, we have completed our testing and proofing of concept on the prototype sensor, which BEI already has provided. We demonstrated that we can produce an angular displacement measurement with any degree of rotation. The sensor is also supposed to demonstrate redundancy. The sensor used in prototyping only has one independent output and, therefore, cannot demonstrate redundancy. This constraint can only be demonstrated in the custom designed sensor that will be used in production. CNH also requested that its steering column provider replaces the internal shaft with one that will meet our specifications. However, to prove that this shaft design will work, we constructed a shaft in multiple pieces, mimicking the future shaft. The future shaft will be one solid assembly. The designs presented at this time will be for prototyping only. The prototype version will have less strength than the intended production version. This means that if the - 8 -

11 prototype can meet all specifications, then the manufactured version will exceed them. For testing purposes, we mounted a high resolution optical encoder with an independent bracket on the extension. These additions will not be present in the production model but are necessary for testing purposes on the prototype. All the differences between prototyping and production previously mentioned are necessary to overcome the above said hurdles. While the actions taken do limit some testing capabilities, the prototype will prove the general concept for production. Cost Projections There are two different costs associated with his project. One cost is for the prototype and verification of design ability, while the other deals with the projected expenses for manufacturing the concept for production. Two sensors ordered from BEI costs $ per unit when purchased one at a time. One sensor is for our design team to use for in house testing and one is for the engineers at CNH to use for preliminary testing. The sensors purchased from BEI in bulk of thousands will run in the range of $60.00 to $70.00 per unit. Sheet steel for the mounting bracket will cost under $5.00 to make one bracket in both production and for prototyping. Aluminum rods also cost well under $5.00 for prototyping. Prototype testing and concept validation will have additional costs of testing equipment such as a stand to mount the steering column while testing and an optical encoding truth sensor. Overhead for equipment along with the price of labor at CNH runs for about $ per hour. Lab testing at CNH consists of component level analysis and H.A.L.T. (Highly Accelerated Life Testing). The testing and component building of the prototype at CNH should take roughly 10 hours. The total estimated cost of producing a prototype including executing various tests, is estimated at around $19, The costs of the designed system for production add up to an expected $75.00 per unit. The target price for this project was set at $90.00 by CNH. A completed tabulation of all cost projections can be found below in Figure

12 Prototyping Cost Estimates Sensor for Prototyping: $ Sensor for CNH Preliminary testing: $ Mounting Stand: $50.49 Shaft Extension: $0.00 Bracket for Mounting Sensor: $10.00 Optical Encoder (Truthing Sensor): $0.00 Testing: $17, est. Miscellaneous Hardware: $10.00 Total Estimated Cost: $ Unit Production Cost Estimates Sensor: $70.00 est. Shaft Change: $5.00 est.. Brackets for Mounting Sensor: $5.00 est. Miscellaneous Hardware: $5.00 est. Steering Column Depopulation: -$10.00 est. Total Estimated Unit Cost: $75.00 Figure 10: Cost Analysis Resource Management Resources from both CNH s Technical Center and the University of Delaware including information, machinery, personnel, stationary, hardware and work space were beneficial in the creating and implementing of a proper problem solution. The Technical Center in New Holland, Pennsylvania served as the site for all sponsor and team meetings throughout the course of the project. The problem solving and design work was mostly all done at the University of Delaware by our team. CNH took responsibility for manufacturing most of the parts. Laser sheet metal cutting and bending machinery CNH has at their facility was utilized to make the bracket system. High-end rapid prototyping equipment at CNH took care of making the collar that holds the sensor to the shaft. The shaft extension piece was built and press fit into the existing column shaft at the UD machine shop. The mounting stand to hold the entire column during testing and demonstration and the mounting bracket for the optical encoder was also machined in the UD shop. Validation In System Rotation Test The intention of this test is to show accuracy and repeatability. The set up consists of the BEI sensor sharing a shaft with the optical encoder (truth sensor) inside the steering column. The optical encoder is mounted directly below the BEI sensor. A motor powered by a variac is mounted to the steering wheel to turn the wheel for multiple revolutions at an approximately constant rate. A LabVIEW program reads the outputs from both sensors and compiles the angle

13 of each sensor into a table of data. A Matlab script is then used to process the data and create a graph to compare the two curves. An example of our data output can be seen in Figure:11. If the BEI sensor is at least as accurate as the encoder, as expected, then the BEI curve should fall completely with in the error bounds on the encoder line. This would appear in the middle figure which is plotting the difference between the two lines. We would expect this difference never to exceed ±0.5º. If the BEI sensor has a repeatable reading, the error will be dependant solely on the angular position and not the path history (eg the number of rotations). Repeatability was observed during testing, but further analysis should be performed once accuracy specifications are met. It is evident here that we are experiencing considerably more error than expected by the system specifications. One of the first potential issues that was recognized is that the optical encoder used was incremental not directional but the BEI sensor reports the absolute angle. This means that as long as the wheel was turned clockwise the signals stayed together but any counterclockwise rotation causes them to diverge. The third graph shows the numerical derivative of the signal verifying that the BEI never records a negative change so this issues was avoided by constant rotation. Figure 11: Test Results Sample Out of System Rotation Test This test is intended to isolate the source of error to be mechanical or electrical (within the linage and assembly or within the programming). The set up consists of the BEI sensor again

14 sharing a shaft with optical encoder. The BEI sensor is pressed between weights to ensure that it cannot rotate. The Optical encoder is secured in a similar fashion. If the source of error is mechanical, the previous error would not show up in this test. If the source of error is electrical, the error would remain the same. Initially, we found approximately the same magnitude error, which would indicate an electrical error. However, when we moved the two sensors closer together along the shaft, we found that the error actually decreased to approximately ±2.5º. We also noticed a visible deflection in the shaft. The decrease in error as the two sensors moved closer together also indicated a deflected shaft. It then became apparent that a test for shaft deflection was necessary. Shaft Run-Out Test This test examines the straightness of the shaft. It is performed by removing all components from the shaft and replacing the BEI sensor with a dial indicator. The dial will then measure the run-out of the shaft. We performed this test for the shaft mounted inside the system and for the shaft used in the out of system rotation test. A run-out of TIR was found for the system shaft. However this could only be measured on the remaining round faces of the shaft and not a full rotation so conventional total run-out could be worse. The shaft used for the out of system test was measure in a similar manner while mounted in the chuck of a lathe. The run-out on this shaft was marginally better. Linear Displacement Test The intention of this test is to determine the maximum possible displacement of the BEI sensor rotor. The sensor is to be clamped to an incrementally movable base while the shaft is fixed. The base is then to be moved in a single direction and the displacement is to be recorded. The diametric extremes found were approximately 0.01 inches. By comparing the results of the shaft run-out test and the linear displacement test, it is clear that the sensor is absorbing 0.01 inches of the shaft run-out and an additional 0.01 inches of shaft run out are causing the sensor to pivot about its anti-rotation pin. While performing this test, it was noted that the BEI sensor read an angular displacement of approximately one degree for a linear displacement of 0.01 inches. This verifies that ±0.5º of the error is due to the linear displacement of the rotor. Testing Conclusions The performed tests indicate a deflected shaft to be the main source of error. Due to time constraints and available facilities, further testing to conclusively verify this suspicion could not be completed. Future testing should consist of a tighter geometrically toleranced shaft to verify that the system will meet all necessary specifications once this problem is resolved. Conclusion With the final prototype built and tested, we have found that there is some error in our final results making us unable to meet all of our target values. Errors in design and testing methods were located and reduced until we could come closest to our goals. Most metrics were met or could be met in the future with the custom sensor built by BEI used in our system. Our production cost goal was met in the final project and actually was surpassed by $ Resolution and accuracy were the main missed target values in our final testing as can be seen in the figure below and highlighted in red. To improve accuracy modifications would need to be

15 made to the shaft, which we consider the main source of error. As mentioned the production model will incorporate a single piece steel shaft hopefully alleviating most of this error. Metric Target Results Unit Price $90 $75 Accuracy ±1 ±2.5 Resolution ±0.5 ±2.5 Future Plans Temperature -40 F 185 F TBD by CNH Figure 12: Metric and Target Values Summary The successfully built and tested prototype shows that our solution is close to being fit for production and meets most target values or comes within close proximity. Target values not met, prove that the system may need some re-work and trouble shooting. After locating sources of error, which we feel are mostly in the shaft, the next step is the turnover of the prototype and project to CNH for the finalization if the steer-by-wire system assembly. CNH can take the prototype and install it into one of their windrowers that already has the previously existing mechanical linkage steering system removed and replaced with electric servo motors attached to the steering pumps. The test windrower can then be tested for electronic safety requirements and eventually field and track tested. Assuming the system passes all field tests preformed by CNH, a production model based from the prototype can be designed. Slight changes in the design will include the steering column manufacturer producing one continuous shaft including the extension we designed to hold the damper and sensor. In addition to reducing number of parts, which eases assembly, it should also ensure that the shaft is perfectly straight which will make our system more accurate. The bracket system might also be changed as well as the fastening and spacing techniques. One thing that will change slightly is the sensor provided by BEI. The production model sensor will be built specifically for this application. It will be smaller than the currently used sensor, contain two independent output channels to meet redundancy requirements, have an IP 67 sealing to keep foreign materials from affecting internal electronic and mechanical equipment, and have a hole that perfectly matched the profile of the shaft, making the need for a collar to couple the BEI sensor and shaft obsolete. All changes are to make the product easier and less expensive to build in bulk quantities. Manufacturing one complete shaft should mitigate the alignment problems we encountered, getting the shafts perfectly on one axis

16 Appendix A Steer-by-wire Specifications Temperature Requirements: Constraint: -40 F 185 F Size: Constraint: The sensor(s) must fit within the confines of the steering column and not require any alterations to the exterior casing of the column or must fit in the space below the cab with out altering the existing structure. Target: <5720mm³ Redundancy: Constraint: Two independent measuring channels Supply Voltage: Recommended: 5.00V Practical Limit: non-regulated 9V Electrical Current Requirements: Constraint: <3 Amps Resolution: Constraint: ± 2 Target: ± 0.5 Output requirements: CAN Frequency Range: <100Hz Analog: Voltage Output Range: 0-5V Reading Range: V Reliability: Constraint: <0.2% failure in first 1000 hrs of operation Accuracy: Constraint: ±2% Target: ±1% Drift: Target: <2% over full temperature range Price: Target <$90 Electromagnetic Compatibility (EMC): Constraint: >100 v/m Target: >200 v/m

17 Appendix B Max Displacement of Shaft in Torsion Tl θ = JG θ = angle of l = maximum length of shaft G = modulus of rigidity 4 4 πd π J = = = 8.081E l = 3.5inches =.0889 meters G = 26.8E (4)*(.0889) θ = 6 (8.081E )*(26.8E θ = E θ = 9.408E 9 5 displacement T = torsion in shaft J = second polar moment of inertia about z - axis through center T = 4 newton meters Al Pa 6 radians degrees 9 ) 6 θ = θ = Tl JG degrees E

18 Appendix C Customer Wants

19 Appendix D System Level Overview Diagram

20 Appendix E Technical Drawings (All dimensions are in mm and not to scale.) Shaft Extension

21 Damper

22 Top Bracket

23 Collar

24 Sensor

25 Bottom Bracket

26 Appendix F The Mechanical Linkage in the Windrower Being Removed Shown circled in red is the linkage coming down from the steering column in the cab of the windrower and extending back to the hydrostatic pumps which vary wheel speed

27 - 25 -

28 Appendix G - Assembly Instructions Slide damper up shaft until it meets the stop. Slide collar along the shaft until it meets the damper. o Be sure not to push the collar too snugly against the damper. Place sensor on top of the bottom bracket. o The wire harness attachment should be facing downward o The attachment should also be on the side of the slanted vertical bracket faces. Loosely fasten the damper stop tab to the top bracket. Place the top bracket on top of the sensor. o The damper stop tab should be opposite to the side of slanted vertical faces on the bottom bracket. Slide assembly up shaft until the top bracket meets the damper. o The side of the subassembly with the damper stop tab should be facing the side of the steering column with hydraulic cylinders. o Assure that the collar is aligned with the inner portion of the sensor. Align spacers between bottom bracket and yoke. o Ensure clearance for hardware. Insert hardware and loosely fasten. Adjust assembly to ensure proper horizontal and vertical alignment. Tighten all hardware