Group 5 Lathe Final Report

Sam Weiss, Stephen Powelson, Josh Gafford, Andrew Marecki, Eric Correll, Johannes Schneider

Introduction
Our main goal for this class was to design and fabricate a desktop lathe capable of achieving <10m precision. Using mechanical engineering design principles, intuition, benchmark experiments and computer analysis, every component of our lathe was carefully analyzed and optimized pre-production to ensure a smooth manufacturing process and successful operation of our final product. In addition, our lathe was designed so that the effect of environmental and operational disturbances would be minimized. The objective of this report is to explain the design process of our desktop lathe, to explain error analysis of the first prototype, to offer solutions to resolve these errors and improve the performance of our lathe, and to reflect upon what this course has taught us concerning precision machine design.

Team Member Roles

Sam Weiss: MathCAD Guru Stephen Powelson: SolidWorks Guru Josh Gafford: FEA Guru Andrew Marecki: Documentation Guru Eric Correll: Machining Guru Johannes Schneider: Budgeting and Measurement Guru Although each team member was the overseer of sorts for his assigned role, every member contributed to the design and manufacturing process in areas not specific to their role. This allowed for every member to be on the same page and the opportunity to contribute equally to the project without being constrained by their specific role.

1. Design, Analyses, Calculations, and Fabrication Approach

During the design phase of our lathe, we felt it extremely necessary to perform in-depth analyses on all critical components of our lathe. Using fundamental principles from statics, dynamics, thermodynamics and mechanical design, as well as finite-element analyses (FEA), we calculated relevant safety factors prior to manufacturing in order to ensure proper functioning of our parts under both operational and environmental disturbances. Figure 1 shows a comparison of the solid model of our lathe with the first manufactured prototype. 1.1 Spindle and Spindle Housing The spindle and spindle housing (Figure 2) was the first sub-assembly that we designed and fabricated. The proper design of this sub-assembly is critical to the lathe performance as any eccentricity in the rotation of the spindle or inadequate stiffness directly translate into errors in the piece being cut.

The spindle and housing as well as the housing interface with the headstock were designed to maximize stiffness in the X, Y (radial) and Z (axial) directions* so as to minimize shaft deflection when cutting forces are applied. Using basic cutting force equations from 2.008, we estimated maximum cutting forces to be on the order of 500 N in both the radial direction (when cutting) and axial direction (when thrusting). Additionally, we estimated that the pulley tensioning force would contribute a maximum of 100 N at the motor end of the spindle shaft. We used MathCAD to generate shear, moment, and minimum-diameter diagrams (see Figures 3a, 3b, and 3c) of our shaft under these loading conditions to design our spindle shaft to deflect a maximum of 0.001 at the chuck. Additionally, fatigue calculations were performed on our shaft design using Goodman criterion to ensure that our shaft would reach its endurance strength (infinite life) under the applied conditions with a safety factor of 1.5. The selection of appropriate bearings was another critical aspect of our spindle design, as the bearings need to support the forces exerted on the shaft. We selected tapered roller bearings because they can sustain both large radial and axial forces encountered during lathe operation, and made our specific bearing selection using relevant equations for loading conditions and bearing life. Two of these bearings were installed in the back-to-back configuration and preloaded via a preload nut and Belleville washer on our shaft to ensure adequate stiffness (which is a function of preload). This preloading mechanism allows us to control the bearing preload (and therefore bearing stiffness) by controlling the torque applied when tightening the nut. The bearings were constrained in a configuration such that thermal expansion of the shaft during operation would not cause the bearings to fail, but only relieve the bearing preload. Additionally, the bearings were placed as close to the loaded ends of the spindle as possible to minimize spindle deflection due to cutting, thrust, and pulley forces. We performed bearing life calculations to ensure that our bearings would be able to last for a minimum of 109 cycles in the configuration that we chose. 1.2 Motor Mount, Transmission The motors used for the lathe are specified to run at a nominal speed of 4500 RPM. To emulate typical cutting speeds necessary for soft stock such as aluminum and brass, we chose to run our lathe at 3000 RPM, meaning we had to incorporate a transmission ratio of 1.5:1. Additionally, we chose to use a v-belt for increased frictional surface-area in order to compensate for the magnitudes of torque we expected from cutting and thrust forces. A particular design issue was that of the tensioning mechanism of our motor mount. We had to design it such that it provided the necessary travel to tension the belt, while being elegant, accessible, and simple. Our solution is shown in Figure 4. We used rubber washers to damp the system and isolate it from the vibrations of the motor, so that these vibrations would not be transferred to the spindle housing. Our tensioning device allows for ~1 inch of travel (center-tocenter distance from motor pulley to spindle pulley). 1.2 Cross-Feed Flexure The crossfeed flexure (Figure 5) serves as a linear bearing to control movement of the cutting tool in the radial direction, by constraining movement in all but this direction. We
*

The coordinate system used in this report is the standard lathe coordinate system, where Z is along the axis of the leadscrew towards the headstock, X is parallel to the floor plane into the tool, and Y is perpendicular to the floor plane

therefore designed our flexure to maximize stiffness in the Y and Z directions but remain compliant in the X direction. Due to the complex geometry of our cross-feed, most of the analysis was done in FEA, however fundamental beam-bending equations were used to establish tolerances for our critical dimensions. When designing this flexure, we used CosmosWorks to ensure that the flexure blades could provide the necessary travel without yielding by a safety factor of 1.7. Dynamic FEA analysis showed us that lowest resonant frequency of the flexure (~160 Hz) was well outside the range of operating rotational frequencies (~50 Hz) of the motor and spindle, so that unwanted vibrations that could adversely affect cutting accuracy would not resonate. Ultimately, there is a tradeoff between X-compliance and susceptibility to lower resonant frequencies; using first-principles and FEA we strove to optimize this trade-off. 1.3 Crossfeed Lead Screw Assembly The crossfeed lead screw assembly (Figure 6) serves to convert rotation of the crossfeed lead screw to translation of the crossfeed flexure and cutting tool in the X (radial) direction. Our design uses nuts preloaded by Belleville washers onto thrust bearings on ether side of the thrust bearing mount. The thrust bearings are recessed within the thrust-bearing mount to avoid the introduction of external particulates into the rolling surface. The lead screw is threaded into a Delrin nut (fixed to the carriage), so rotating the screw causes both it and the thrust bearing mount to translate. The Delrin nut is preloaded with a set screw to prevent backlash. 1.4 Leadscrew Flexure Assembly The lead screw flexure assembly (Figure 7) serves to convert rotation of the lead screw to translation of the carriage in the Z (thrusting) direction. A flexure is necessary in this assembly to couple the motion of the lead screw nut to the carriage. We designed our flexure to maximize stiffness in the Z direction of carriage travel as well as rotation about the Z axis, and to allow compliance in all other directions. Compliance is necessary so as to not over-constrain the carriage, since the rail/bushing interface already fully constrains the carriage. Using FEA, we designed this flexure to allow for up to 3.175 mm of travel in the compliant directions without yielding. Additionally, the flexure was designed to resist axial forces and torques about the lead screw axis typical to normal cutting and thrusting operations with a safety factor of 1.3. The flexure incorporates a vertical arm that connects and couples the movement of two compliant blades so that deformation in the y-direction minimizes rotation about the axis. Additionally, a thin blade on the top of the flexure allows for deformation in the x-direction. The flexure houses a Delrin lead screw nut, which is preloaded with a set screw to eliminate backlash. 1.5 Rails-Carriage Interface The rails are responsible for supporting the carriage and must be sufficiently stiff in order to minimize rail deflection during typical cutting forces. The rail system can be simplified into a set of parallel statically indeterminate beams with a centrally-applied load due to the carriage and some fraction of the cutting force. We used principles of beam bending and FEA analysis to ensure that our rails would experience a central deflection on the order of a 25 m for the maximum case of a 500N applied load.

We selected 0.75 steel stock after FEA studies showed this to be sufficiently stiff. The carriage rides on the rails via oil-impregnated brass bushings. The front rail supports two bushings mounted directly on the carriage, while the back rail supports a flexure-mounted bushing (Figure 8). This flexure is necessary because while it constrains the rotation of the carriage about the X-axis, it is compliant in X translation and Y-axis rotation, and therefore does not over-constrain the carriage. Without this compliance, any slight misalignment in the rails would cause the carriage bushings to wedge over the range of Z direction travel. FEA analysis and principles from beam bending were used to select the blade thickness of the carriage flexure. 1.6 Toolholder and Kinematic Coupling Kinematic couplings are fixtures designed to exactly kinematically constrain the part in question. Our group opted to incorporate a kinematic coupling into our tool holder (Figure 9) to allow us highly repeatable positioning of our tool. Our coupling uses 3 radial v-grooves milled into the top of our crossfeed flexure that mate with three hemispheres sunk into the bottom of our tool holder. Each sphere has two contact points with a v-groove, for a total of 6 contact points which constrain all 6 degrees of freedom. Hertzian contact between the spheres and the grooves allows for significant stiffness (that increases as a function of applied force) in all directions. Because the cutting forces are concentrated at the toolholder, a kinematic coupling is an ideal solution to increasing stiffness. In the extreme case of a tool plunge, the added stiffness of the kinematic coupling as a function of force is pivotal in preventing toolholder failure. Using MathCAD, we estimated the stiffness of our kinematic coupling prior to manufacturing. Although repeatable tool positioning is typically not a major concern with manual lathes, this ability would be highly desirable in a CNC machine when programming tool offsets. 1.7 Lathe HTM Model To predict the deformation of our lathe under typical cutting, thrust and pulley forces, we modeled the structure as a network of masses and springs in series and parallel. Applied thermal and physical loads, as well as stiffnesses of each component, bolted joint, and epoxy interface were estimated and plugged into a series of transformation matrices describing the deformation of the point of interest relative to a designated ground point. The product of these matrices over the entirety of our lathe provided a resultant matrix of deformation we could expect during normal operation. In considering nominal cutting conditions (50 N thrust/cut/plunge loads), our HTM model estimated the following deformations (i.e. displacement of the part due to an applied cutting load at the tool tip): 138.81 m (0.0055) in Z 86.1 m (0.0034) in Y -121.54 m (0.0048) in X

2. Lathe Performance, Error Analysis and Solutions

2.1 Preload Tests We empirically determined optimum preload of our spindle bearing by experimentally determining the stiffness of the Belleville washer, establishing a relationship between nut rotation and preload force (Figure 10), and applying known torques to the spindle. We adjusted the preload for several iterations and determined the relationship between preload and friction, runout, and endplay. The results are shown in Figure 11. We found that optimum preload (minimizing runout and friction) was achieved with 1/6th turn past finger-tight of the preload nut. 2.2 Spindle Error Spindle error measurements were taken by rotating the spindle/chuck assembly with a cylinder clamped in a chuck and measuring the amount of eccentricity of a point on the cylinder with a capacitance probe. Our first spindle assembly yielded eccentricity on the order of 100 microns (see Figure 12). We also noted that our first assembly had a high amount of frictional resistance when turning, which led us to believe that both this friction and the large error was caused by either eccentricity in the shaft or non-concentric bearing races, due to manufacturing error in the shaft and/or housing. The spindle assembly has since been remade and fitted with a new chuck; after shimming the chuck we reduced the error of our spindle to ~50 microns. 2.3 Crossfeed Flexure Error To measure error in the crossfeed flexure movement we used a coordinate measurement machine (CMM). The machine measures the positions of three spheres fixed to the flexure as the flexure is advanced through its entire range of motion. The locations of the three spheres define the plane of the flexure and allow us to determine both translational and rotational error in the flexures movement. We defined error as deviation from only X axis movement (the nonconstrained direction), per inch of X axis travel. Our flexure was found to have extremely small errors, with a Z error of -6.6 m and Y error of 27.7 m. 2.4 Carriage Lead Screw Travel Error A similar process as detailed above was repeated to determine carriage lead screw travel error, this time while advancing the carriage in the Z direction with the lead screw. We defined error as deviation from only Z axis movement (the non-constrained direction), per inch of Z axis travel. We found an X error of 56.13 m and a Y error of -14.5 m. The relatively large X error is a result of misalignment of the front rail with the lead screw axis, which can be resolved by shimming the rail to align it. 2.5 Taper Error We ran our lathe and cut a rod to a diameter by running the Z axis while keeping the X axis fixed. Measuring the diameter of the rod revealed that we had a diameter taper of 0.288 mm over 3.8 cm of travel. This is due to misalignment of the spindle axis with the lead screw axis. Given the taper degree and the diameter of the bolt circle of the housing-headstock interface, we calculated that by inserting a shim of 0.48 mm on the back-side bolt, between the housing and headstock, we should be able to properly align the spindle. After inserting the shim, we successfully eliminated the taper.

2.6 Lathe Stiffness We tested the stiffness of various components on our lathe by using a spring force gauge to apply a known force in the x, y, and z-directions. Using a dial indicator, we recorded how much the component of interest deformed under the applied load, and graphed the results in order to derive the relative stiffness. Figures 13 and 14 present the results of our stiffness tests at various points on our lathe and the associated linear fits. Additionally, the data from each graph are summarized in Table 1 in terms of the relative stiffnesses of each part. Table 1 shows that the y-direction stiffnesses of both the toolholder and the chuck are significant; this is necessary for the function of our lathe, as the greatest forces will be experienced by the chuck and the toolholder in the y-direction during normal operations. For the toolholder, stiffnesses in the x and the z-directions are almost identical, which is inherent in the geometry of our kinematic coupling (refer to Figure 9). Z-direction stiffness is crucial, as it will determine the precision of face-cutting operations; additionally, x-direction stiffness resists the thrust force imparted by the part on the tool in a plunge-operation. Under expected normal cutting operations, our resultant stiffnesses are adequate. By comparing empirical results with our estimated deflections using HTMs (Section 1.7), we see that actual deflections experienced by the lathe were of the same order of magnitude (but a little less) than estimated values, lending empirical validity to our HTM model. 2.7 Kinematic Coupling Repeatability We used the CMM, referenced to a point on the toolholder, to measure the precision and repeatability of our toolholder. We found that our toolholder was repeatable to within 1.95 m in the x-direction, 2.58 m in the y-direction, and 3.05 m in the z-direction. These values were calculated with 95% confidence using statistical methods. It is evident that we are well within our expected range of precision (10 m). 2.8 Plunge Test In order to simulate a worst-case scenario, we swiftly plunged our tool into a piece of round diameter steel stock. Although ideally we would like the belt to slip, the friction between our belt and the pulleys at lowest tension was so great that the motor stalled before the belt actually slipped. Other than this, no harm was done to the lathe. We can potentially solve this problem by replacing our v-belt with a standard flat belt, which will reduce the friction exerted by the belt on the pulley. A comparison of expected results vs. actual machine performance can be found in Table 2 in the appendices.

3. Reflections/Conclusion
We have certainly learned a great amount about precision machine design over the course of this semester. It is both conceptually and physically challenging to design and integrate numerous components that are capable of precision machining on the order of 10 m. Additionally, many environmental factors (ambient temperature, external vibration) and

operational factors (generated heat, vibration due to motor, ball-bearings, etc.) can result in an amplified response in a precision machine; it was important to keep these factors into consideration and design elegant solution to address each. This is possibly the largest eyeopening aspect of mechanical design, namely, the great number of factors that can contribute adversely to machine performance. A lesson that we took away from this project is the applicability of equations to simulate realworld conditions. Although equations can provide conservative, first-order estimates of machine performance, it is very difficult to accurately model the system performance. Especially in precision machining, numerous non-quantifiable factors can contribute adversely to the performance of our device. Keeping this in mind, we made sure to keep our estimates very conservative, and used high factors-of-safety to compensate for the unpredictable real-world variability that our machine would surely be exposed to. During the remainder of the course, we will focus on fixing the errors with the proposed solutions. Noting these errors and attributing them to a source was a valuable learning experience; it shows that even though everything may work in a CAD model of a machine, the actual finished product can easily deviate from this model. In the domain of precision machine design, even small deviations can have drastic negative impacts on a machines ability to cut precise parts.

4.

Figures and Tables

Figure 1: Lathe CAD model (left) and finished lathe prototype (right)

Figure 2: Spindle assembly cutaway view

600 400

Shear (N)

200 0 200 400 600 0 2 4 6

Position (in) Y X
50

(a)

Moment (N*m)

40 30 20 10 0 10 0 2 4 6

Position (in) Y X Total

1.5

(b)

Diameter (in)

0.5

0.575

1.151

1.726

2.302

2.877

3.453

4.028

4.604

5.179

5.755

6.33

Position (in) Minimum Diameter (no torque) Minimum Diameter (with torque) Actual (Inputted) Diameter

(c)

Figure 3: (a) Shear, (b) Moment, and (c) Minimum Diameter diagrams for spindle shaft under normal operating conditions.

Figure 4: Motor Mount and Housing Assembly

Figure 5: Top view of crossfeed flexure, tool holder, and crossfeed assembly

10

Figure 6: Crossfeed lead screw assembly

Figure 7: Lead screw flexure assembly

11

Figure 8: Rail flexure and bushing

Figure 9: Kinematic coupling for tool holder

12

Nut Rotation vs Preload Force

250

200 Preload Force (lbf)

150

100

50

0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 Nut Rotation (radians)

Figure 10: Preload Force vs. Nut Rotation (based on stiffness of Belleville washer)

Preload Force vs friction, runout, endplay

2 friction torque (in-lbs) runout (thou) 1.6 end play (thou) 1.8

1.4

1.2

1 0.8

0.6

0.4 0.2

0 0 5 10 15 20 25 30 35 preload torque (in-lb)

Figure 11: Frictional torque, runout and endplay vs. preload torque

13

Figure 9: Stiffness test results (and linear fits) for chuck deflections
150 Relative Deflection (microns) 100 50 0 0 -50 -100 -150 Angle (degrees) Initial Spindle Test Spindle Rotated 180 Deg Average Relative Deflection 50 100 150 200 250 300 350 400

Spindle Error Measurements

Figure 12: Initial Spindle Error Measurement Results

Toolholder Deflections
200 180 160 Deflections (microns) 140 120 100 80 60 40 20 0 0 20 40 60 Load (N) 80 100 120 y = 0.3049x - 4.5884 2 R = 0.9823 Tool X Tool Y Tool Z Linear (Tool Z) Linear (Tool X) Linear (Tool Y) y = 1.8664x - 1.1994 2 R = 0.9841 y = 1.9034x + 6.0476 R = 0.9953
2

Figure 13: Stiffness Data at Toolholder

14

Chuck Deflections
90 80 70 Deflections (microns) 60 50 40 30 20 10 0 -10 0 20 40 60 Load (N) 80 100 120 y = 0.2805x - 3.81 2 R = 0.9389 Chuck X Chuck Y Chuck Z Linear (Chuck X) Linear (Chuck Y) Linear (Chuck Z) y = 0.3452x - 7.62 2 R = 0.9481 y = 0.8199x - 3.81 2 R = 0.9256

Figure 14: Stiffness Data at Chuck

Table 1: Stiffness values at toolholder, chuck, and for the lathe as a whole. Stiffness [N/mm] Toolholder Chuck 536 1120 3280 3565 525 2897

Direction X Y Z

Lathe 372 1708 445

Table 2: Expected Results vs. Actual Machine Performance Expected Performance <10 m <2 m 10 m over 2 cm -121.54 m 86.1 m 138.81 m Belt Slips Actual Performance <50 m <3 m <1 m over 2 cm -134.41 m 29.27 m 112.36 m Motor Stalls

Spindle Precision Kinematic Coupling Repeatability Taper Error Total Deflection (X) Total Deflection (Y) Total Deflection (Z) Tool Plunge (Worst-Case)

15