ADJUSTABLE GEOMETRIC CONSTRAINTS © 2001 MIT PSDAM AND PERG LABS Why adjust kinematic couplings? KC Repeatability is orders of magnitude better than accuracy Accuracy = f ( manufacture and assemble ) Accuracy Kinematic Coupling Repeatability Adjusted Kinematic Coupling Non-Kinematic Coupling © 2001 MIT PSDAM AND PERG LABS Mated Position Desired Position Accuracy & Repeatability Serial and parallel kinematic machines/mechanisms SERIAL MECHANISMS Rotary 3 Structure takes form of open loop I.e. Most mills, lathes, “stacked” axis robots Kinematics analysis typically easy Linear 1 Rotary 1 Base/ground PARALLEL MECHANISMS Structure of closed loop chain(s) I.e. Stuart platforms & hexapods Kinematics analysis usually difficult 6 DOF mechanism/machine Multiple variations on this theme © 2001 MIT PSDAM AND PERG LABS Photo from Physik Instruments web page Parallel mechanism: Stewart-Gough platform 6 DOF mechanism/machine Multiple variations on this theme with different joints: Spherical joints: Prismatic joints: Planar: 3 Cs 5 Cs 3 Cs Permits 3 rotary DOF Permits one linear DOF Permits two linear, one rotary DOF Ball Joint Sliding piston Roller on plane E.g. Changing length of “legs” Spherical Joint Prismatic Joint Mobile Base/ground © 2001 MIT PSDAM AND PERG LABS Base/ground Parallel mechanism: Variation 6 DOF mechanism/machine by changing position of joints Can have a combination of position and length changes Mobile Base © 2001 MIT PSDAM AND PERG LABS Base Kinematic couplings as mechanisms Ideally, kinematic couplings are static parallel mechanisms IRL, deflection(s) = mobile parallel kinematic mechanisms How are they “mobile”? Hertz normal distance of approach ~ length change of leg Far field points in bottom platform moves as ball center moves ~ joint motions Top Platform Bottom Platform Far-field points modeled as joints © 2001 MIT PSDAM AND PERG LABS Hertz Compliance Model of kinematic coupling mechanism Initial Position of Ball’s Far Field Point dr dl dz dn Final Position of Ball’s Far Field Point Initial Contact Point Initial Position of Groove’s Far Field Point z “Joint” motion r Final Contact Point Final True Groove’s Far Field Point Ball Center Groove far field point © 2001 MIT PSDAM AND PERG LABS Accuracy of kinematic mechanisms Since location of platform depends on length of legs and position of base and platform joints, accuracy is a function of mfg and assembly Parameters affecting coupling centroid (platform) location: Ball center of curvature location Ball orientation (i.e. canoe ball) Ball centerline intersect position (joint) Ball radii Groove Groove Groove Groove center of curvature location orientation depth radii Ball’s center of curvature Symmetry intersect Contact Cone Groove’s center of curvature © 2001 MIT PSDAM AND PERG LABS Utilizing the parallel nature of kinematic couplings Add components that adjust or change link position/size, i.e.: Place adjustment between kinematic elements and platforms (joint position) Adjustment Strain kinematic elements to correct inaccuracy (element size) Ball’s center of curvature © 2001 MIT PSDAM AND PERG LABS Example: Adjusting planar motion Position control in x, y, qz: Rotation axis offset from the center of the ball A A e 180o B B Eccentric left Eccentric right © 2001 MIT PSDAM AND PERG LABS Patent Pending, Culpepper ARKC demo animation y x 1 3 1 2 Input: Output: Actuate Balls 2 & 3 Dy © 2001 MIT PSDAM AND PERG LABS 3 2 Input: Output: Actuate Ball 1 Dx and Dqz Planar kinematic model Equipping each joint provides control of 3 degrees of freedom View of kinematic coupling with balls in grooves (top platform removed) Joint 1 q1 A Axis of rotation Offset, e Center of sphere y x qz q3A Joint 3 © 2001 MIT PSDAM AND PERG LABS q2A Joint 2 z x Vector model for planar adjustment of KC A1 r1c M 1’ M1 r1d r1b r1a rD r2d r3d qz y A3 x M3 r3a r2c r3c r3b r2a M2 M 3’ A2 r2b M 2’ r1a + r1b + r1c + r1d = rD r2a + r2b + r2c + r2d = rD r3a + r3b + r3c + r3d = rD © 2001 MIT PSDAM AND PERG LABS Analytic position control equations Vector equations: r1a + r1b + r1c + r1d = rD r2a + r2b + r2c + r2d = rD r3a + r3b + r3c + r3d = rD System of 6 equations: 0 0 cos q 1A 0 0 sin q 1A 0 cos q 2A 0 0 sin q 2A 0 0 0 cos q 3A 0 0 sin q 3A 1 0 0 1 1 0 0 1 1 0 0 1 L 1B L 1D cos q 1A L 2B L 2D sin q 2A L 3B L 2D cos q 2A x L 3D sin q 3A y L 3D cos q 3A q z L 1D sin q 1A L 1D L 1A cos q 1A L 1C cos q 1C L L sin q L sin q 1D 1A 1A 1C 1C L 2D L 2A cos q 2A L 2C cos q 2C L 2D L 2A sin q 2A L 2C sin q 2C L L cos q L cos q 3A 3A 3C 3C 3D L L sin q L sin q 3A 3A 3C 3C 3D For standard coupling: 120o; offset E; coupling radius R; x y z E 2 sin( q1C q1 A ) sin( q 2C q 2 A ) (q 3C q 3 A ) 3 3E sin( q 2C q 2 A ) (q 3C q 3 A ) 3 1 z1 z 2 z 3 3 © 2001 MIT PSDAM AND PERG LABS qX 1 2 z1 z 2 z 3 3 qY 3 z 2 z1 3 qZ sin(qz) ~ qz: 1 E sin( q1C q1A ) sin( q 2C q 2 A ) (q 3C q 3 A ) 3 R ARKC resolution analysis For E = 125 microns Centroid X Position, 120 Degree Coupling R90 = -1.5 micron/deg 150 True Position Plot R0 = “0” micron/deg 100 Linearized Position Plot C 50 x [micron] q1 0 0 30 60 90 120 150 180 -50 -100 -150 q 1c [degrees] y x q Limits on Linear Resolution Assumptions q1C © 2001 MIT PSDAM AND PERG LABS % Error Lower Limit [ Degree ] Upper Limit [ Degree ] Half Range [ Degree ] 1 75 105 +/- 15 2 70 110 +/- 20 5 60 120 +/- 30 10 47 133 +/- 43 Forward and reverse kinematic solutions ARKC Kinematic Analysis Spread Sheet Do not change cells in red, only change cells in blue PART I: COUPLING CHARACTERISTICS Use this to specify groove angles and input ball rotations and heights used to calculate coupling position in part II q1A q2A 90 1.571 330 degrees radians degrees z1 5 microns 0.0001969 inches z2 500 5.760 radians qA 210 3.665 degrees radians q1C 87.7073 1.531 331.1455 5.780 211.1467 3.685 degrees radians degrees radians degrees radians E 125 0.0049 microns inches RT 57150 2.2500 microns inches z3 q2C qC microns 0.0196850 inches 200 microns 0.0078740 inches Note: RT = LiD PART II: CALCULATED MOVEMENTS This takes input from part I to calculate the position of the top part of the coupling. qZ 0.0000 0.0006 0.000000 radians radians degrees qX -0.004024 radians -4024.4889 radians -0.230586 degrees x 5.0005 0.000197 microns inches qY -0.003031 radians -3030.7040 radians -0.173647 degrees y -0.0014 0.000000 microns inches z 235.0000 0.009252 microns inches PART III: REVERSE SOLUTION Use this to input a desired position and goal seek to solve for the position of the balls DESIRED POSITION POSITION ERROR qZ 0.000 0.00 x 5.000 -0.001 BALL SETTINGS 0.0 rad microns See part I for modified ball angles, the following angles are the difference between groove and ball angles q1 -2.2927 degrees q2 1.1455 1.1467 degrees degrees q y 0.000 ERROR SUM © 2001 MIT PSDAM AND PERG LABS 0.001 2698.1 microns <- x, y, qz <- Use the solver on this value to set it to a value (ideally zero) that is small, but greater than or equal to 0. Low-cost adjustment (10 m) Peg shank and convex crown are offset Light press between peg and bore in plate Adjustment with allen wrench e Epoxy or spreading to set in place Friction (of press fit) must be minimized… z r Top Platform q = 180o 2E Bottom Platform © 2001 MIT PSDAM AND PERG LABS Moderate-cost adjustment (3 micron) Shaft B positions z height of shaft A [ z, qx, qy ] qz ] Shaft A positions as before [ x, y, Force source preload I.e. magnets, cams, etc.. z Shaft A input r Magnet A Magnet B Shaft B input Teflon sheets (x 4) Top Platform qB = -360o Ball Bearing support Bottom Platform © 2001 MIT PSDAM AND PERG LABS e l “Premium” adjustment (sub-micron) z r Run-out is a major cause of error Seals keep contaminants out Dual motion actuators provide: Actuator Air bushings for lower run-out Flexible Coupling Linear motion Rotary motion Seal Air Bushings Seal Shaft Ball Groove Flat © 2001 MIT PSDAM AND PERG LABS Mechanical interface wear management Wear and particle generation are unknowns. Must investigate: Coatings [minimize friction, maximize surface energy] Surface geometry, minimize contact forces Alternate means of force/constraint generation At present, must uncouple before actuation Sliding damage } © 2001 MIT PSDAM AND PERG LABS