Correction method of plane high-precision double-track XYY gantry of SMT high-speed chip mounter

Abstract

The invention discloses a method for correcting a plane high-precision double-track XYY gantry of an SMT (surface mount technology) high-speed chip mounter, which comprises the following steps of: measuring and calculating deviation values of 13 specified coordinate points on an XYY plane by using vision and a specific correction sheet, generating a plurality of correction curves through surface correction, and repeatedly checking an algorithm result, so that the reliability of the correction curves is improved, and the correction precision of the plane high-precision double-track XYY gantry of the SMT high-speed chip mounter is improved. According to the method, the number of correction points is small, the correction time is saved, the stability of equipment is improved, the technical requirements of installation personnel are reduced, the risk that the operation precision is reduced due to mechanical deformation after long-time use is reduced, and the reliability of the equipment is improved. And the positioning deviation under the conditions of shaft bending deformation, displacement and the like caused by insufficient mounting precision or mechanical wear after long-time use is solved.

Description

Technical Field

[0001] The invention relates to the technical field of control systems of SMT high-speed placement machines, and in particular to a correction method for a planar high-precision double-track XYY gantry of an SMT high-speed placement machine. Background Art

[0002] Gantry is the most commonly used mechanical structure in automation equipment. Equipment with low precision and speed can be completed with a single-track XY. The gantry is generally composed of four parts: motor, connector, slide rail, and screw. Equipment with a larger operating area or higher speed requirements often uses a double Y structure, that is, the two tracks in the Y direction use different motors to simultaneously drive the Y axis to move in the same direction at high speed.

[0003] Compared with the single-track XY, ​​the double-track XYY gantry structure has one more axis to be installed, and one more motor for synchronous control. Both the mechanical accuracy and control accuracy are double the test. Although linear motors can be used to replace the lead screw structure in precision machinery to improve mechanical accuracy, the wear and deformation of high-speed equipment is higher than that of general equipment, so the algorithm has to be recalibrated.

[0004] There are many kinds of correction methods for the double-track XYY gantry structure used in SMT high-speed placement machines. The most commonly used methods on SMT high-speed placement machines are mechanical fixture correction, calibration plate correction and calibration column correction. The functions of various methods are similar. They all use a known "ruler" to calibrate an unknown plane. Mechanical fixture correction is a pure hardware method. The XYY axis is manually moved to the origin or a specified position, and then the error is reduced to a reasonable level by manually adjusting the mechanical position angle. This method is effective for origin deviation or angle deviation, but it does not work for the deformation of a certain part in the middle of the three axes of XYY, and it is time-consuming and laborious to adjust, and requires professionals to achieve it. Calibration plates or calibration columns are more often automatically corrected by software. Adding a visual system can more accurately obtain small deviations. Calibration plates are better than calibration columns. They can detect the deviation value of each position of the XYY axis, but high-precision calibration plates are expensive, and large-format calibration plates need to be customized and are easily damaged, which is very troublesome to use. Summary of the invention

[0005] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a correction method for a planar high-precision double-track XYY gantry of an SMT high-speed placement machine, so as to solve the positioning deviation caused by shaft bending deformation, displacement, etc. due to insufficient installation accuracy or mechanical wear after long-term use. The correction is simple, fast and convenient, and the correction speed is fast and the cost is low, avoiding the problem of affecting the production accuracy of the entire equipment.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is: a method for calibrating a planar high-precision double-track XYY gantry of an SMT high-speed placement machine, comprising the following steps:

[0007] Step 1: Place a white calibration sheet on the PCB calibration board as a black background, and make sure that the calibration sheet appears completely on the dual-track XYY gantry with a pixel width of (pixel width W m , pixel height H m ) within the field of view of the mark camera, identify the center coordinates (X1, Y1) and angle of the calibration piece;

[0008] Step 2: According to X difference = (W m / 2-X1)*X direction pixel accuracy, Y difference=(H m / 2-Y1)*Y direction pixel accuracy, move the XYY axis of the dual-track XYY gantry and make the visual center of the mark camera overlap with the center of the calibration film, identify the length L1, width H1 and angle θ1 of the calibration film, and the calculated center coordinate offset is (0, 0);

[0009] Step 3: According to the XYY operation, the smallest coordinate (X min , Y min ) and the maximum coordinate (X max , Y max ) to obtain the coordinates of the four vertices (X min , Y min )、(X max , Y min )、(X min , Y max )、(X max , Y max ), and then calculate the effective plane center point coordinates (X mid , Y mid ),in

[0010] X mid =(X max -X min ) / 2+X min ,

[0011] Y mid =(Y max -Y min ) / 2+Y min ;

[0012] Step 4: Calculate the center coordinates of the four edges from the center coordinates of step 3.

[0013] At y=Y min The center coordinates on the line (X mid , Ymin ),

[0014] At y=Y max The center coordinates on the line (X mid , Y max ),

[0015] When x=X min The center coordinates on the line (X min , Y mid ),

[0016] When x=X max The center coordinates on the line (X max , Y mid );

[0017] Step 5: Because the establishment of an N-order polynomial regression curve requires N+1 sampling data, and the minimum degree of axis bending requires a quadratic equation, that is, 3 sampling coordinates are required on each curve that needs to be corrected. In order to more accurately determine whether the XY axis has an S-shaped bend, reduce the error introduced during the measurement process, and increase data accuracy, 4 data verification coordinates (X 1 / 4 , Y 1 / 4 )、(X 3 / 4 , Y 1 / 4 )、(X 1 / 4 , Y 3 / 4 )、(X 3 / 4 , Y 3 / 4 ),

[0018] X 1 / 4 =(X max -X min ) / 4+X min ,

[0019] X 3 / 4 =(X max -X min )*3 / 4+X min ,

[0020] Y 1 / 4 =(Y max -Y min ) / 4+Y min ,

[0021] Y 3 / 4 =(Y max -Y min )*3 / 4+Y min ;

[0022] Step 6: From step 3 to 4, get the maximum value Y of the X axis on the Y axis max , median value Y mid , minimum value Ymin The coordinates of the three correction points at the time of max , the middle value X mid , minimum value X min Form a 3*3 matrix and get the absolute coordinates of 9 calibration points.

[0023] Then, the absolute coordinates of four verification points for determining whether the shaft bending degree is C-shaped bending or S-shaped bending are obtained from step 5, and the absolute coordinates of 13 correction points are obtained by combining the 9 bending degree correction coordinates and the 4 bending shape verification coordinates;

[0024] Step 7: Use the nozzle of the SMT high-speed placement machine to grab the calibration sheet and grab the calibration sheet to the pixel width (pixel width W f , pixel height H f ) and is identified by the flying camera to obtain the length L'1, width H'1, angle θ'1 and center coordinates (X'1, Y'1) of the correction piece, and calculate the coefficient of the angle θ'1 divided by θ1 in step 2,

[0025] If θ'1 / θ1 is within the range of 0.99 to 1.01, it means that the nozzle pick-up speed and height are correct. If the judgment result is that there is a difference, readjust the pick-up speed and height until the center position and angle of the calibration sheet are not offset during the pick-up process;

[0026] Step 8: After the flying camera recognizes that there is no offset, the calibration sheet is placed at the absolute coordinate position of the first calibration point through the suction nozzle of the SMT high-speed placement machine, and then the mark camera is used for recognition to obtain the length L"1, width H"1, angle θ"1 and center coordinates (X"1, Y"1) of the calibration sheet, and the coefficient of angle θ"1 divided by θ'1 in step 7 is calculated. If θ"1 / θ'1 is within the range of 0.99 to 1.01, it means that the nozzle placement speed and height are correct. If the judgment result is that there is a difference, adjust the placement speed and height so that there is no angle offset during the placement process;

[0027] Step 9: Calculate and record the difference between the center coordinates of the calibration piece identified by the mark camera at the first calibration point and the absolute coordinates of the calibration point in step 6.

[0028] Center difference △X1=(X"1-W m / 2)*5 / W m -(X'1-W f / 2)*5 / W f ,

[0029] Center difference △Y1=(Y"1-H m / 2)*5 / H m -(Y'1-H f / 2)*5 / H f ,

[0030] (X"1, Y"1) is the center coordinate of the calibration piece placed at the first calibration point captured by the mark camera.

[0031] (X'1, Y'1) is the center coordinate of the calibration piece captured by the flying camera when the calibration piece passes by during placement;

[0032] Step 10: Repeat steps 7 to 9 to obtain the difference values ​​of the remaining 12 calibration points △X2~△X 13 and △Y2~△Y 13 ,

[0033] Calculate the calibration point (X min , Y max ) 11 and the calibration point (X mid , Y max ) 12 The average of △Y avg9 =(△Y 11 +△Y 12 ) / 2, and get the difference between the test value and the theoretical value △Y cmp9 =(△Y9-△Y avg9 ),

[0034] Similarly, we can calculate that when y=Y max The difference between the test value and the theoretical value of other calibration points on the straight line △Y cmp12 =(△Y 12 -△Y avg12 ), △Y cmp10 =(△Y 10 -△Y avg10 ), determine △Y cmp9 / △Y cmp10 / △Y cmp12 Is it greater than 0? If it is greater than 0 and |△Y cmp12 |>|△Y cmp9 |、|△Y cmp12 |>|△Y cmp10 |, then y=Y max When the X-axis is C-shaped, otherwise it is S-shaped.

[0035] Step 11: Repeat step 10 to get y at y = Y min The difference between the measured value and the theoretical value of the calibration point on the straight line △Y cmp4 =(△Y4-△Y avg4 ), △Y cmp2 =(△Y2-△Y avg2 ), △Y cmp5 =(△Y5-△Y avg5 ),

[0036] Determine △Y cmp4 / △Y cmp2 / △Y cmp5 The result shows that the bending deformation is C-shaped or S-shaped. If the results of step 10 and step 11 are consistent, and |△Y cmp4 -△Y cmp9 |<5,|△Y cmp5 -△Y cmp10 |<5, then record the difference of the bending shape and all the calibration points. If the judgment results are different, repeat steps 7 to 9 to obtain another set of test data. Record the valid data after the judgment results are the same.

[0037] Step 12: Put y in Y max , Y mid , Y min The three correction points on the three straight lines establish the Langrange interpolation curves, which are recorded as correction curves P3, P4, and P5 respectively. min , Y min ) to coordinate (X max , Y max ) The calibration curve of the five calibration points on the oblique straight line is recorded as P2, and the coordinates (X max , Y min ) to coordinate (X min , Y max ) The calibration curve of the five calibration points on the oblique straight line is recorded as P1;

[0038] Step 13: According to the Lagrange interpolation formula, we can get

[0039]

[0040] There are 5 calibration point data in the calibration curve P1. Project the X coordinates of these 5 points in the XYY gantry coordinate system to 0, 1, 2, 3, and 4 respectively to reduce the amount of calculation in the operation process. The product from i=0 to i=4,

[0041] l0(x)=(x-1) / (0-1)*(x-2) / (0-2)*(x-3) / (0-3)*(x-4) / (0-4),

[0042] l1(x)=(x-0) / (1-0)*(x-2) / (1-2)*(x-3) / (1-3)*(x-4) / (1-4),

[0043] l2(x)=(x-0) / (2-0)*(x-1) / (2-1)*(x-3) / (2-3)*(x-4) / (2-4),

[0044] l3(x)=(x-0) / (3-0)*(x-1) / (3-1)*(x-2) / (3-2)*(x-4) / (3-4),

[0045] l4(x)=(x-0) / (4-0)*(x-1) / (4-1)*(x-2) / (4-2)*(x-3) / (4-3);

[0046] Step 14: Substitute the coordinates (X max , Y min ) to coordinate (X min , Y max ) The difference △Y measured at the five calibration points on the oblique straight line and the product obtained in step 9 are multiplied to the formula Get the first Lagrangian interpolation curve P1,

[0047]

[0048] Step 15: Repeat steps 13 to 14 to obtain the remaining Lagrange interpolation curves P2, P3, P4, and P5, and save all correction curve formulas.

[0049] P2=l 0(p2) (x)*(-16)+l 1(p2) (x)*(-4)+l 2(p2) (x)*(-2)+l 3(p2) (x)*(-6)+l 4(p2) (x)*(-14),

[0050] P3=l 0(p3) (x)*(-13)+l 1(p3) (x)*(-3)+l 2(p3) (x)*(-14),

[0051] P4=l 0(p4) (x)*(-25)+l 1(p4) (x)*(-2)+l 2(p4) (x)*(-10),

[0052] P5=l 0(p5) (x)*(-16)+l 1(p5) (x)*(-0)+l 2(p5) (x)*(-13);

[0053] Step 16: Correct the center point (X mid , Y mid ) as the origin, make a coordinate system, and assume that the point to be measured Q(x, y) satisfies x>X mid And y>Y midIt can be judged that Q is in the first quadrant, so the calculation formula of the Y coordinate offset of point Q is related to the three correction functions P2, P3 and P4. Then according to the distance formula from point to line Get the distance L from point Q to P2, P3, and P4 respectively QP2 , L QP3 , L QP4 , and then according to the distance ratio, the proportion coefficients of the three calibration curves can be obtained

[0054] a1=1-L QP2 / (L QP2 +L QP3 +L QP4 ),

[0055] a2=1-L QP3 / (L QP2 +L QP3 +L QP4 ),

[0056] a3=1-L QP4 / (L QP2 +L QP3 +L QP4 );

[0057] Step 17: According to the P2, P3, and P4 correction formula ratio coefficients obtained in step 16, the Y coordinate offset of point Q is obtained.

[0058] ΔY Q =a1*P2+b1*P3+c1*P4;

[0059] Repeat steps 13 to 16 to get the X coordinate offset of point Q.

[0060] Δ X Q =a'1*I2+b'1*I3+c'1*I4,

[0061] Save the correction formula of difference X and difference Y, call the correction curve formula to perform position correction before moving, and then move to the corrected coordinates.

[0062] The calibration sheet is a high-precision white ceramic sheet with a length and width of 5x5 mm.

[0063] Compared with the prior art, the advantages of the present invention are as follows: the present invention uses a specific correction sheet to replace the calibration plate to achieve the same function of the calibration plate and can better adapt to the calibration requirements of various formats. At the same time, the number of calibration points is reduced according to a specific algorithm, making the calibration process faster and more suitable for normal production needs.

[0064] The present invention adopts Lagrange interpolation method to calculate the curve of XYY axis bending, so that the curve interpolation value is closer to the real value than the common polynomial, and the error value of the specified point is corrected according to the curve to increase the mounting accuracy of the high-precision SMT placement machine.

[0065] The present invention uses vision and a specific correction sheet to measure the deviation values ​​of 13 specified coordinate points on the XYY plane, thereby reducing the technical requirements for installers and the risk of reduced operating accuracy due to mechanical deformation after long-term use. At the same time, it can measure whether the two Y axes are not parallel when installed, whether the movements are not synchronized, and other problems.

[0066] The present invention adopts dual vision to perform correction measurement, which reduces the measurement error caused by the jitter of the grasping force during the placement of the correction piece, and makes the measurement data more accurate.

[0067] The present invention adopts a 13-point plane positioning correction method, which has fewer correction points and can save correction time. In addition, the surface correction can generate multiple correction curves, repeatedly check the algorithm results, increase the reliability of the correction curves, and improve the stability of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is the position distribution diagram of 13 correction points of the present invention.

[0069] Figure 2 It is the changing curve in the first interpolation polynomial of △Y of the present invention.

[0070] Figure 3 It is the changing curve in the first interpolation polynomial of △Y of the present invention.

[0071] Figure 4 These are the five Lagrange interpolation curves of △Y of the present invention.

[0072] Figure 5 This is a flow chart of the calibration sheet of the present invention being placed at 13 calibration points.

[0073] Figure 6 This is the step of calculating the △X, △Y Lagrange interpolation polynomials of the present invention.

[0074] Figure 7 This is the calibration curve calling process of the present invention.

[0075] Figure 8 It is the overall flow chart of XYY axis bending correction of the present invention.

[0076] Fig. 9 This is a schematic diagram of an SMT placement machine equipped with a flying camera, a mark camera and a suction nozzle.

[0077] Fig.10 It is the value of ΔX of the present invention under 13-point distribution.

[0078] Fig.11 It is the value of △Y of the present invention under 13-point distribution DETAILED DESCRIPTION

[0079] A calibration method for a planar high-precision double-track XYY gantry of an SMT high-speed placement machine. Figures 1 to 11 As shown, the following steps are included:

[0080] Step 1: Place a white calibration sheet on the PCB calibration board as a black background, and make sure that the calibration sheet appears completely on the dual-track XYY gantry with a pixel width of (pixel width W m , pixel height H m ) within the field of view of the mark camera, identify the center coordinates (X1, Y1) and angle of the correction piece.

[0081] Step 2: According to X difference = (W m / 2-X1)*X direction pixel accuracy, Y difference=(H m / 2-Y1)*Y direction pixel accuracy, move the XYY axis of the dual-track XYY gantry and make the visual center of the mark camera overlap with the center of the calibration film, identify the length L1, width H1 and angle θ1 of the calibration film, and the calculated center coordinate offset is (0, 0).

[0082] Step 3: According to the XYY operation, the smallest coordinate (X min , Y min ) and the maximum coordinate (X max , Y max ) to obtain the coordinates of the four vertices (X min , Y min )、(X max , Y min )、(X min , Y max )、(X max , Y max ), and then calculate the effective plane center point coordinates (X mid , Y mid ), where X mid =(X max -X min ) / 2+X min , Y mid =(Y max -Y min ) / 2+Y min .

[0083] Step 4: Calculate the center coordinates of the four edges from the center coordinates of step 3, at y = Y min The center coordinates on the line (Xmid , Y min ), at y = Y max The center coordinates on the line (X mid , Y max ), when x=X min The center coordinates on the line (X min , Y mid ), when x=X max The center coordinates on the line (X max , Y mid ).

[0084] Step 5: Because the establishment of an N-order polynomial regression curve requires N+1 sampling data, and the minimum degree of axis bending requires a quadratic equation, that is, 3 sampling coordinates are required on each curve that needs to be corrected. In order to more accurately determine whether the XY axis has an S-shaped bend, reduce the error introduced during the measurement process, and increase data accuracy, 4 data verification coordinates (X 1 / 4 , Y 1 / 4 )、(X 3 / 4 , Y 1 / 4 )、(X 1 / 4 , Y 3 / 4 )、(X 3 / 4 , Y 3 / 4 ),

[0085] X 1 / 4 =(X max -X min ) / 4+X min , X 3 / 4 =(X max -X min )*3 / 4+X min ,

[0086] Y 1 / 4 =(Y max -Y min ) / 4+Y min , Y 3 / 4 =(Y max -Y min )*3 / 4+Y min .

[0087] Step 6: From step 3 to 4, get the maximum value Y of the X axis on the Y axis max , median value Y mid , minimum value Y min The coordinates of the three correction points at the time of max , the middle value X mid , minimum value X minA 3*3 matrix is ​​formed to obtain the absolute coordinates of 9 correction points. Then, from step 5, the absolute coordinates of 4 verification points for judging whether the shaft bending degree is C-shaped bending or S-shaped bending are obtained. The 9 curvature correction coordinates and the 4 curvature shape verification coordinates are combined to obtain the absolute coordinates of 13 correction points.

[0088] Step 7: Use the nozzle of the SMT high-speed placement machine to grab the calibration sheet and grab the calibration sheet to the pixel width (pixel width W f , pixel height H f ) and is identified by the flying camera to obtain the length L'1, width H'1, angle θ'1 and center coordinates (X'1, Y'1) of the correction piece. Calculate the coefficient of angle θ'1 divided by θ1 in step 2. If θ'1 / θ1 is in the range of 0.99 to 1.01, it means that the picking speed and height of the nozzle are correct. If the judgment result is that there is a difference, readjust the picking speed and height until the center position and angle of the correction piece are not offset during the picking process.

[0089] Step 8: After the flying camera recognizes that there is no offset, the correction sheet is placed at the absolute coordinate position of the first correction point through the suction nozzle of the SMT high-speed placement machine, and then the mark camera is used for recognition to obtain the length L"1, width H"1, angle θ"1 and center coordinates (X"1, Y"1) of the correction sheet, and the coefficient of angle θ"1 divided by θ'1 in step 7 is calculated. If θ"1 / θ'1 is in the range of 0.99 to 1.01, it means that the placement speed and height of the suction nozzle are correct. If the judgment result is that there is a difference, adjust the placement speed and height so that there is no angle offset during the placement process.

[0090] Step 9: Calculate and record the difference between the center coordinates of the calibration piece identified by the mark camera at the first calibration point and the absolute coordinates of the calibration point in step 6.

[0091] Center difference △X1=(X"1-W m / 2)*5 / W m -(X'1-W f / 2)*5 / W f ,

[0092] Center difference △Y1=(Y"1-H m / 2)*5 / H m -(Y'1-H f / 2)*5 / H f ,

[0093] (X"1, Y"1) is the center coordinate of the calibration piece placed at the first calibration point captured by the mark camera.

[0094] (X'1, Y'1) is the center coordinate of the calibration piece captured by the flying camera when the calibration piece passes by during placement;

[0095] Step 10: Repeat steps 7 to 9 to obtain the difference values ​​△X2~△X of the remaining 12 calibration points. 13 , △Y2~△Y 13 , calculate the correction point (X min , Y max ) 11 and the calibration point (X mid , Y max ) 12 The average of △Y avg9 =(△Y 11 +△Y 12 ) / 2, and get the difference between the test value and the theoretical value △Y cmp9 =(△Y9-△Y avg9 ),

[0096] Similarly, we can calculate that when y=Y max The difference between the test value and the theoretical value of other calibration points on the straight line △Y cmp12 =(△Y 12 -△Y avg12 ), △Y cmp10 =(△Y 10 -△Y avg10 ), determine △Y cmp9 / △Y cmp10 / △Y cmp12 Is it greater than 0? If it is greater than 0 and |△Y cmp12 |>|△Y cmp9 |、|△Y cmp12 |>|△Y cmp10 | then y=Y max When the X-axis is C-shaped, it is considered to be S-shaped.

[0097] Step 11: Repeat step 10 to get y at y = Y min The difference between the measured value and the theoretical value of the calibration point on the straight line △Y cmp4 =(△Y4-△Y avg4 ), △Y cmp2 =(△Y2-△Y avg2 ), △Y cmp5 =(△Y5-△Y avg5 ), determine △Y cmp4 / △Y cmp2 / △Y cmp5 The result shows that the bending deformation is C-shaped or S-shaped.

[0098] If the results of step 10 and step 11 are the same, and |△Y cmp4 -△Y cmp9 |<5,|△Y cmp5-△Y cmp10 |<5, then record the bending shape and the difference of all correction points.

[0099] If the judgment results are not the same, repeat steps 7 to 9 to obtain another set of test data, and record the valid data after the judgment results are the same.

[0100] Step 12: Put y in Y max , Y mid , Y min The three correction points on the three straight lines establish the Langrange interpolation curves, which are recorded as correction curves P3, P4, and P5 respectively. min , Y min ) to coordinate (X max , Y max ) The calibration curve of the five calibration points on the oblique straight line is recorded as P2, and the coordinates (X max , Y min ) to coordinate (X min , Y max )The correction curve of the five correction points on the oblique straight line is recorded as P1.

[0101] Step 13: According to the Lagrange interpolation formula, we can get

[0102]

[0103] There are 5 calibration point data in the calibration curve P1. Project the X coordinates of these 5 points in the XYY gantry coordinate system to 0, 1, 2, 3, and 4 respectively to reduce the amount of calculation in the operation process. The product from i=0 to i=4,

[0104] l0(x)=(x-1) / (0-1)*(x-2) / (0-2)*(x-3) / (0-3)*(x-4) / (0-4),

[0105] l1(x)=(x-0) / (1-0)*(x-2) / (1-2)*(x-3) / (1-3)*(x-4) / (1-4),

[0106] l2(x)=(x-0) / (2-0)*(x-1) / (2-1)*(x-3) / (2-3)*(x-4) / (2-4),

[0107] l3(x)=(x-0) / (3-0)*(x-1) / (3-1)*(x-2) / (3-2)*(x-4) / (3-4),

[0108] l4(x)=(x-0) / (4-0)*(x-1) / (4-1)*(x-2) / (4-2)*(x-3) / (4-3);

[0109] Step 14: Substitute the coordinates (X max , Y min ) to coordinate (X min , Y max ) The difference △Y measured at the five calibration points on the oblique straight line and the product obtained in step 9 are multiplied to the formula Get the first Lagrangian interpolation curve P1,

[0110]

[0111] Step 15: Repeat steps 13 to 14 to obtain the remaining Lagrange interpolation curves P2, P3, P4, and P5, and save all correction curve formulas.

[0112] P2=l 0(p2) (x)*(-16)+l 1(p2) (x)*(-4)+l 2(p2) (x)*(-2)+l 3(p2) (x)*(-6)+l 4(p2) (x)*(-14),

[0113] P3=l 0(p3) (x)*(-13)+l 1(p3) (x)*(-3)+l 2(p3) (x)*(-14),

[0114] P4=l 0(p4) (x)*(-25)+l 1(p4) (x)*(-2)+l 2(p4) (x)*(-10),

[0115] P5=l 0(p5) (x)*(-16)+l 1(p5) (x)*(-0)+l 2(p5) (x)*(-13);

[0116] Step 16: Correct the center point (X mid , Y mid ) as the origin, make a coordinate system, and assume that the point to be measured Q(x, y) satisfies x>X mid And y>Y mid It can be judged that Q is in the first quadrant, so the calculation formula of the Y coordinate offset of point Q is related to the three correction functions P2, P3 and P4. Then according to the distance formula from point to line Get the distance L from point Q to P2, P3, and P4 respectively QP2 , L QP3 , L QP4, and then according to the distance ratio, the proportion coefficients of the three calibration curves can be obtained

[0117] a1=1-L QP2 / (L QP2 +L QP3 +L QP4 ),

[0118] a2=1-L QP3 / (L QP2 +L QP3 +L QP4 ),

[0119] a3=1-L QP4 / (L QP2 +L QP3 +L QP4 );

[0120] Step 17: According to the P2, P3, and P4 correction formula ratio coefficients obtained in step 16, the Y coordinate offset of point Q is obtained.

[0121] ΔY Q =a1*P2+b1*P3+c1*P4;

[0122] Repeat steps 13 to 16 to get the X coordinate offset of point Q.

[0123] Δ X Q =a'1*I2+b'1*I3+c'1*I4,

[0124] Save the correction formula of difference X and difference Y, call the correction curve formula to perform position correction before moving, and then move to the corrected coordinates.

Claims

1. A method for calibrating a planar high-precision double-track XYY gantry of an SMT high-speed placement machine, characterized in that: The following steps are included: Step 1: Place a white calibration sheet on the PCB calibration board as a black background, and make sure that the calibration sheet appears completely on the dual-track XYY gantry with a pixel width of (pixel width W m , pixel height H m ) within the field of view of the mark camera, identify the center coordinates (X1, Y1) and angle of the calibration piece; Step 2: According to X difference = (W m / 2-X1)*X direction pixel accuracy, Y difference=(H m / 2-Y1)*Y direction pixel accuracy, move the XYY axis of the dual-track XYY gantry and make the visual center of the mark camera overlap with the center of the calibration film, identify the length L1, width H1 and angle θ1 of the calibration film, and the calculated center coordinate offset is (0, 0); Step 3: According to the XYY operation, the smallest coordinate (X min , Y min ) and the maximum coordinate (X max , Y max ) to obtain the coordinates of the four vertices (X min , Y min )、(X max , Y min )、(X min , Y max )、(X max , Y max ), and then calculate the effective plane center point coordinates (X mid , Y mid ),in X mid =(X max -X min ) / 2+X min , AND mid (And max -AND min ) / 2+Y min ; Step 4: Calculate the center coordinates of the four edges from the center coordinates of step 3. At y=Y min The center coordinates on the line (X mid , Y min ), At y=Y max The center coordinates on the line (X mid , Y max ), When x=X min The center coordinates on the line (X min , Y mid ), When x=X max The center coordinates on the line (X max , Y mid ); Step 5: Because the establishment of an N-order polynomial regression curve requires N+1 sampling data, and the minimum degree of axis bending requires a quadratic equation, that is, 3 sampling coordinates are required on each curve that needs to be corrected. In order to more accurately determine whether the XY axis has an S-shaped bend, reduce the error introduced during the measurement process, and increase data accuracy, 4 data verification coordinates (X 1 / 4 , Y 1 / 4 )、(X 3 / 4 , Y 1 / 4 )、(X 1 / 4 , Y 3 / 4 )、(X 3 / 4 , Y 3 / 4 ), X 1 / 4 =(X max -X min ) / 4+X min , X 3 / 4 =(X max -X min )*3 / 4+X min , AND 1 / 4 (And max -AND min ) / 4+Y min , AND 3 / 4 (And max -AND min )*3 / 4+Y min ; Step 6: From step 3 to 4, get the maximum value Y of the X axis on the Y axis max , median value Y mid , minimum value Y min The coordinates of the three correction points at the time of max , the middle value X mid , minimum value X min Form a 3*3 matrix and get the absolute coordinates of 9 calibration points. Then, the absolute coordinates of four verification points for determining whether the shaft bending degree is C-shaped bending or S-shaped bending are obtained from step 5, and the absolute coordinates of 13 correction points are obtained by combining the 9 bending degree correction coordinates and the 4 bending shape verification coordinates; Step 7: Use the nozzle of the SMT high-speed placement machine to grab the calibration sheet and grab the calibration sheet to the pixel width (pixel width W f , pixel height H f ) and is identified by the flying camera to obtain the length L'1, width H'1, angle θ'1 and center coordinates (X'1, Y'1) of the correction piece, and the coefficient of angle θ'1 divided by θ1 in step 2 is calculated. If θ'1 / θ1 is within the range of 0.99 to 1.01, it means that the nozzle pick-up speed and height are correct. If the judgment result is that there is a difference, readjust the pick-up speed and height until the center position and angle of the calibration sheet are not offset during the pick-up process; Step 8: After the flying camera recognizes that there is no offset, the calibration sheet is placed at the absolute coordinate position of the first calibration point through the suction nozzle of the SMT high-speed placement machine, and then the mark camera is used for recognition to obtain the length L"1, width H"1, angle θ"1 and center coordinates (X"1, Y"1) of the calibration sheet, and the coefficient of angle θ"1 divided by θ'1 in step 7 is calculated. If θ"1 / θ'1 is within the range of 0.99 to 1.01, it means that the nozzle placement speed and height are correct. If the judgment result is that there is a difference, adjust the placement speed and height so that there is no angle offset during the placement process; Step 9: Calculate and record the difference between the center coordinates of the calibration piece identified by the mark camera at the first calibration point and the absolute coordinates of the calibration point in step 6. Center Difference △ X1=(X"1-W m / 2)*5 / W m -(X'1-W f / 2)*5 / W f , Center Difference △ Y1=(Y"1-H m / 2)*5 / H m -(Y'1-H f / 2)*5 / H f , (X"1, Y"1) is the center coordinate of the calibration piece placed at the first calibration point captured by the mark camera, and (X'1, Y'1) is the center coordinate of the calibration piece captured by the flying camera when the calibration piece passes by during the placement process; Step 10: Repeat steps 7 to 9 to obtain the difference of the remaining 12 calibration points. △ X2~ △ X 13 as well as △ Y2~ △ Y 13 , Calculate the calibration point (X min , Y max ) △ Y 11 and the calibration point (X mid , Y max ) △ Y 12 The average △ Y avg9 =( △ Y 11 + △ Y 12 ) / 2, and get the difference between the test value and the theoretical value △ Y cmp9 =( △ Y 9-△ Y avg9 ), Similarly, we can calculate that when y=Y max The difference between the measured value and the theoretical value at other calibration points on the straight line △ Y cmp12 =( △ Y 12-△ Y avg12 ), △ Y cmp10 =( △ Y 10-△ Y avg10 ),judge △ Y cmp9 / △ Y cmp10 / △ Y cmp12 Is it greater than 0? If it is greater than 0 and | △ Y cmp12 |>| △ Y cmp9 |、| △ Y cmp12 |>| △ Y cmp10 |, then y=Y max When the X-axis is C-shaped, otherwise it is S-shaped. Step 11: Repeat step 10 to get y at y = Y min The difference between the measured value and the theoretical value at the calibration point on the straight line △ Y cmp4 =( △ Y 4-△ Y avg4 ), △ Y cmp2 =( △ Y 2-△ Y avg2 ), △ Y cmp5 =( △ Y 5-△ Y avg5 ), judge △ Y cmp4 / △ Y cmp2 / △ Y cmp5 The result shows that the bending deformation is C-shaped or S-shaped. If the results of step 10 and step 11 are consistent, and | △ Y cmp4-△ Y cmp9 |<5,| △ Y cmp5-△ Y cmp10 |<5, then record the difference of the bending shape and all the calibration points. If the judgment results are different, repeat steps 7 to 9 to obtain another set of test data. Record the valid data after the judgment results are the same. Step 12: Put y in Y max , Y mid , Y min The three correction points on the three straight lines establish the Langrange interpolation curves, which are recorded as correction curves P3, P4, and P5 respectively. min , Y min ) to coordinate (X max , Y max ) The calibration curve of the five calibration points on the oblique straight line is recorded as P2, and the coordinates (X max , Y min ) to coordinate (X min , Y max ) The calibration curve of the five calibration points on the oblique straight line is recorded as P1; Step 13: According to the Lagrange interpolation formula, we can get There are 5 calibration point data in the calibration curve P1. Project the X coordinates of these 5 points in the XYY gantry coordinate system to 0, 1, 2, 3, and 4 respectively to reduce the amount of calculation in the operation process. Multiply the products from i=0 to i=4, l0(x)=(x-1) / (0-1)*(x-2) / (0-2)*(x-3) / (0-3)*(x-4) / (0-4), l1(x)=(x-0) / (1-0)*(x-2) / (1-2)*(x-3) / (1-3)*(x-4) / (1-4), l2(x)=(x-0) / (2-0)*(x-1) / (2-1)*(x-3) / (2-3)*(x-4) / (2-4), l3(x)=(x-0) / (3-0)*(x-1) / (3-1)*(x-2) / (3-2)*(x-4) / (3-4), l4(x)=(x-0) / (4-0)*(x-1) / (4-1)*(x-2) / (4-2)*(x-3) / (4-3); Step 14: Substitute the coordinates (X max , Y min ) to coordinate (X min , Y max ) measured at the 5 calibration points on the oblique straight line △ Multiply Y by the product obtained in step 9 to get the formula Get the first Lagrangian interpolation curve P1, Step 15: Repeat steps 13 to 14 to obtain the remaining Lagrange interpolation curves P2, P3, P4, and P5, and save all correction curve formulas. P2=l 0(p2) (x)*(-16)+l 1(p2) (x)*(-4)+l 2(p2) (x)*(-2)+l 3(p2) (x)*(-6)+l 4(p2) (x)*(-14), P3=l 0(p3) (x)*(-13)+l 1(p3) (x)*(-3)+l 2(p3) (x)*(-14), P4=l 0(p4) (x)*(-25)+l 1(p4) (x)*(-2)+l 2(p4) (x)*(-10), P5=l 0(p5) (x)*(-16)+l 1(p5) (x)*(-0)+l 2(p5) (x)*(-13); Step 16: Correct the center point (X mid , Y mid ) as the origin, make a coordinate system, and assume that the point to be measured Q(x, y) satisfies x>X mid And y>Y mid It can be judged that Q is in the first quadrant, so the calculation formula of the Y coordinate offset of point Q is related to the three correction functions P2, P3 and P4. Then according to the distance formula from point to line Get the distance L from point Q to P2, P3, and P4 respectively QP2 , L QP3 , L QP4 , and then according to the distance ratio, the proportion coefficients of the three calibration curves can be obtained a1=1-L QP2 / (L QP2 +L QP3 +L QP4 ), <h2 style=";text-align:left;direction:ltr">a2 = 1-L<h2 style=";text-align:left;direction:ltr"> QP3 <h2 style=";text-align:left;direction:ltr"> / (L<h2 style=";text-align:left;direction:ltr"> QP2 <h2 style=";text-align:left;direction:ltr"> +L<h2 style=";text-align:left;direction:ltr"> QP3 <h2 style=";text-align:left;direction:ltr"> +L<h2 style=";text-align:left;direction:ltr"> QP4 <h2 style=";text-align:left;direction:ltr"> ), a3=1-L QP4 / (L QP2 +L QP3 +L QP4 ); Step 17: According to the P2, P3, and P4 correction formula ratio coefficients obtained in step 16, the Y coordinate offset of point Q is obtained. Δ Y Q =a1*P2+b1*P3+c1*P4; Repeat steps 13 to 16 to get the X coordinate offset of point Q. Δ X Q =a'1*I2+b'1*I3+c'1*I4, Save the correction formula of difference X and difference Y, call the correction curve formula to perform position correction before moving, and then move to the corrected coordinates.

2. The method for calibrating a planar high-precision double-track XYY gantry of a SMT high-speed placement machine according to claim 1, characterized in that: The calibration sheet is a high-precision white ceramic sheet with a length and width of 5x5 mm.

Images