JPS63219009A - Backlash correction system in full closed loop control - Google Patents

Abstract

PURPOSE:To realize the more precise working through a full closed loop control system by adding a correction pulse to a shaft, pertinent to the change of the moving direction of a pulse distribution, when the moving direction of the pulse distribution changes, and subtracting the portion of the added correction pulse from a following distribution period. CONSTITUTION:When direction deciding means 21X, 21Y discriminate that the moving direction command of an distribution pulse command has been inverted, and it is discriminated that the direction has been inverted, the correction pulse, corresponding to the backlash portion, previously set and stored in memories 22X, 22Y, is added to a normal distribution pulse at first, and outputted, and from the next pulse distribution period, a cancel pulse is outputted in the reverse direction of the command direction so as to amend the corrected correction pulse quantity, in other words, the pulse that the cancel pulse is subtracted from the normal pulse distribution quantity is outputted. The cancel pulse is made to be five pulses in a first distribution period after the inversion of the direction, and to be three pulses in a second period, and two pulses in a third period, and 10 correction pulses are canceled. By processing in such a way, the work in conformity with the command is performed.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a backlash correction method in a full closed loop control type servo mechanism controlled by a numerical control device.

Servo gains controlled by conventional numerical υ control devices are broadly divided into semi-closed loop control methods and full closed loop control methods. As shown in FIG. 4, the semi-closed loop control system uses a control system that responds to the difference between the movement command (command pulse) output from the information processing circuit of the numerical control device and the rotation amount of the servo motor detected by the position detector 104. The position control circuit 100 outputs the voltage, and in response to the output, the servo drive circuit 101 drives the servo motor 102 to drive the ball screw 105 and pole nut 10 of the feeding means.
When the machine (table, etc.) 107 is moved via 6, and the difference between the movement command amount and the movement amount of the servo motor 102 detected by the position detector 104 becomes "0", the output voltage of the position control circuit 100 changes. It becomes "0", and the rotation of the servo motor is stopped and positioning is performed. This semi-closed loop control method determines the position of the machine by detecting the position of the servo motor, and does not determine the position of the machine itself. Therefore, in this semi-closed loop control method,
The positioning accuracy depends on the accuracy of the feeding means such as a ball screw. In other words, the position of the servo motor is accurately positioned at the commanded position by feedback control;
The position of the machine will vary depending on the accuracy of the feeding means such as a ball screw. Note that 103 is a speed detector, and the servo drive circuit 101 controls the speed of the servo motor 102 according to the speed detected by the detector 103.

In recent years, high-precision ball screws have been developed and become popular, and this semi-closed loop control method has virtually no accuracy problems in practice, and most numerically controlled machine tools use this method. In order to obtain even higher precision for ball screws using this method, pitch error correction and back crack correction functions have been developed, and these functions are generally widely used in numerical control devices.

Of these, to explain the back crack correction, as shown in Fig. 6, when a back rack a exists between the ball screw 105 and the ball nut fixed to the table (III machine) 107, to table 1
07 and then change the direction of movement from right to left, the servo motor will idle for 8 minutes during the back push. That is, even if the servo motor rotates according to the commanded amount, the table 107 will not move enough by the distance of the discreet 9a. Therefore, the numerical control device uses a backrack correction function to servo the regular command pulse every time the direction of movement of the machine (table) changes! ! Before giving it to lW4, a correction pulse equivalent to back rack a is given, so that the servo motor rotates slightly by adding the regular command pulse and the compensation pulse, and the machine follows the regular command. It will move by the amount of the pulse, and will be accurately positioned.

However, in large machines where the length exceeds the length that can be manufactured by ball screws, a rack and pinion drive is necessary, but the accuracy of the rack is considerably lower than that of ball screws, so semi-closed loop control is not practical. This may cause problems. Furthermore, the pitch error of the ball screw may change due to changes in temperature, and if very high precision is required, the semi-closed loop control method may not be sufficient.

In such a case, a full closed loop control system is used in which a position detector such as a linear scale is attached directly to the table of the machine to provide feedback.

As shown in FIG. 5, the full closed loop control system uses a machine (table) 1 driven by a servo motor 102.
The amount of movement of 07 is detected by a position detector 108 such as a linear scale, and the amount of movement vJ commanded from the information processing circuit of the numerical control device is determined by the amount of movement of the machine! The servo motor is driven so that the difference between lllω becomes "0", and the command movement 4 corresponds accurately to the amount of machine movement, and no backrack problem occurs. That is, even if back crack a exists as shown in FIG. 6, when a command to change the moving direction of table 107 is input, first, even if the servo motor moves by back crack 9a, table 107 will not move. As a result, the feedback pulse from the position detector 108 such as a linear scale is not input to the position control circuit 100.
The position control circuit 100 (the value of the error register in this circuit is still the command movement circle) outputs a voltage according to the command movement ♀, and after completing the movement of the balaclara 9a, the command amount is roughly changed. l111g Servo motor 1 for 1 minute
02 is driven, and the machine (table) 107 also moves by the commanded movement amount. In this full closed loop control method, backrack correction is not necessary, and backrack correction is not performed in the full closed loop control method. .

Problems to be Solved by the Invention As mentioned above, in the full closed loop control system, feedback control is performed by detecting when the machine actually moves in response to a movement command, so backrack correction is not necessary. It looks like.

However, when two or more axes need to move in synchronization, such as when machining circular arcs, when the direction of movement of one axis is reversed, the influence of backrack may occur and form errors may occur. For example, backrack effects occur where the moving direction of one axis changes at the quadrant switching point in circular interpolation.

Figure 3(a).

In (b), in circular interpolation, at the quadrant switching point,
This is an explanatory diagram of the reason why shape errors occur in circular arcs. Considering the case where the center of circular arc 1 is the coordinate origin and the circular arc 1 is cut around 41 o'clock, in the second quadrant of Fig. 3 (a),
As shown by the cutting direction at point P1, both the X-axis and the Y-axis move in the positive direction, and the pulse distribution amounts ΔX and ΔY for each pulse distribution period in the numerical control device are the X-axis according to circular interpolation.

Both Y-axes have positive distribution amounts. However, as processing progresses,
When entering the first quadrant from the second quadrant, as shown in the cutting direction at point P2, the direction is positive with respect to the X axis, but
The direction is negative with respect to the Y axis. When the direction of movement of one axis changes like this, the effect of backrack appears. That is, the pulse distribution m to the error register in the position control circuit of the servo mechanism based on the circularly interpolated distribution period to each axis is distributed according to the circularly interpolated values for the X and Y axes, and the machine If we do not follow the distribution m,
This means that the arc as instructed will not be cut. If there is a back rack, the direction of movement of the axis does not change, in the example above, the X-axis, the axis moves according to the distributed amount, but for the Y-axis, where the direction of movement is reversed, the servo motor moves according to the distributed amount when the direction is reversed. Even if it does, it will be absorbed by the back rack and the machine will not move according to the dispensed amount. FIG. 3(b) is an explanatory diagram showing the state at this time. It is assumed that the movement of the X-axis remains unchanged and only the Y-axis is reversed from the positive direction to the negative direction. On the other hand, it is assumed that pulse distribution of +ΔX is performed and pulse distribution of -ΔY is performed for the Y axis. Also, to simplify the explanation, assuming that the servo motor moves by the amount distributed during one pulse distribution cycle period, pulse distribution is performed from point P2 by ΔX to the X axis and ΔY to Ylkll. The machine should move to point P3, but since the direction of movement has been reversed for the Y-axis, which had been moving in the positive direction until the previous pulse distribution cycle, the Y-axis servo motor Even if it moves by -ΔY, the feeding means will idle by the back rack a of the feeding means for the ball screw/nut or rack/pinion, and the machine will move -(
It moves only by ΔY-a) and moves to point P3',
It will no longer move as instructed, resulting in an error in the machined shape. To explain the relationship at this time in detail, in the above example, ΔX=30. -ΔY=-20. Assuming that a=10 and that the servo motor moves within one pulse distribution cycle by the amount of pulses distributed during that cycle, consider the value of the error register in the position control circuit, the amount of movement of the servo motor, and the amount of movement of the machine. Then, in the pulse distribution period immediately after the movement direction of only the Y axis is reversed, the results are as shown in Table 1 below.

Table 1 Even with the following pulse distribution period, ΔX=30°−ΔY=
If it is -20, it will be as shown in Table 2.

As shown in Table 2, in the pulse distribution cycle immediately after the movement direction is reversed, the machine is insufficiently moved by 8 back pushes in the Y-axis direction, and this will be corrected in the next cycle. However, this explanation is based on the assumption that the servo gain is very high, assuming that the servo motor is driven by the amount of distribution m during the pulse distribution period.In reality, the servo gain cannot be this high, and the movement The movement of the servo motor is delayed in response to the command, and the movement amount of Balaclar 2a remains in the Y-axis error register for several pulse distribution periods, and Balaclar 2a remains in the Y-axis error register.
The correction for a will be performed over several cycles. As a result, the actual circular arc cutting process is performed at the turning point of the quadrant, as shown by the broken line 2 in FIG. 3(a), in a shape different from that of the commanded circular arc.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a backlash correction method in a full closed loop control system that reduces errors due to backlash in a full closed loop control system.

Means for Solving the Problems The present invention provides that, when the moving direction of pulse distribution for each axis changes, in addition to the normal command distribution pulse, the axis for which the moving direction has changed is By applying a correction pulse corresponding to backlash and subtracting the additional correction pulse from the next pulse distribution cycle, backlash can be corrected even in the full closed loop control system.

When the moving direction of an axis changes in a certain pulse distribution period, in addition to the regular command distribution pulses for the relevant axis, a correction equivalent to packlash is applied, similar to the backlash correction adopted in semi-closed loop control. Add pulses to compensate for backlash. As a result, the correction pulse is absorbed by the backlash, and a movement command corresponding to the normal command distribution pulse is input to the relevant axis of the machine. Then, the relevant axis of the machine will move by this regular command distribution pulse, and the feedback pulse generated by the movement of the relevant axis will be transmitted to the position control circuit of the servo mechanism into which the regular command distribution pulse and correction pulse are input. This will be subtracted from the value in the error register, but since the correction pulse is absorbed by the backlash and the machine does not move by this amount, the correction pulse remains in the error register, and eventually the corresponding correction pulse of the machine will be subtracted by this correction pulse. Acts to move the axis. Therefore, this corrected amount is subtracted from the next pulse distribution period, the corrected pulse amount is set to "0", and the machine is moved as instructed.

Embodiment FIG. 1 is a functional block diagram of an embodiment of the present invention, showing an example of two-axis control of the X-axis and Y-axis. The numerical control device is
Read and decipher the NC program 10 and perform interpolation 12
After performing this, a distribution pulse command is output for each of the X-axis and Y-axis, and in response to the distribution pulse command, the interpolator 13X of each axis
, 13Y are the error registers 15X, 15Y in the servo circuits 14X, 14Y to distribute the distribution pulses ΔX, ΔY according to the commands.
Output each to . The servo circuits 14X, 14Y generate output voltages according to the values accumulated in the error registers 15X, 15Y, and the servo motors 16x.

16Y respectively, and ball screws 17X and 17Y.

Machine 19X via pole nuts 18X and 18Y.

Move 19Y. On the other hand, when the machines 19X and 19Y move, feedback pulses are generated from position detectors 20X and 20Y, such as linear scales provided in the machines, respectively, and the error register 15x is generated.

15Y, and the error registers 15X and 15Y are subtracted. As a result, the values of the error registers 15X and 15Y become "0", and the servo motors 16X and 1
The drive of the machine 6Y is stopped, and the machines 19X and 19Y are stopped at the commanded positions.

The above is no different from the numerical control method in conventional full closed loop control. In the present invention, in the above method, the direction determining means 21x determines that the moving direction command of the distribution pulse command is reversed.

21Y, and when it is determined that the direction has reversed, a correction pulse corresponding to the backlash set and stored in the memories 22X and 22Y is added to the relevant distribution cycle and output, and then the next pulse and distribution From the cycle, a cancel pulse is output in the opposite direction to the command direction, that is, subtracted from the regular pulse distribution amount so as to correct the corrected pulse interval. In the example shown in Fig. 1, the correction pulse is 10, the cancel pulse is 5 in the first pulse distribution cycle after the direction is reversed, and the second pulse is 5.
The correction pulse 10 is canceled by setting it as "3" in the second pulse distribution cycle and "2" in the third pulse distribution cycle. The correction pulse 1 and the cancel pulse are set according to the gain of the servo mechanism used.

For example, if the gain is very large, the response is quick as described above, so after outputting a correction pulse, it may be set to generate a cancel pulse that cancels all the correction pulses in the next pulse distribution cycle.

The setting of the cancellation pulse that cancels this correction pulse may be determined by experiment, but in this embodiment, the correction pulse is considered to be a step input, and the cancellation pulse is set by approximating the first-order lag corresponding to the step input. The relationship between the set gain value of the servo mechanism and the suspension of the cancel pulse is set as shown in Table 3. In addition,
In Table 3, L is the correction pulse amount.

Table 3 shows that when the gain of the servo mechanism is large, the added correction pulse ff1L is completely removed as a cancel pulse L in the first pulse distribution period following the pulse distribution period in which the correction pulse is added. Also, if the gain is small and the response of the servo mechanism is slow, this is considered to be a -order delay, and it is assumed that the correction pulse is output in response to the correction pulse ff1L between the pulse distribution period to which the correction pulse is added and the next pulse distribution period. In the next pulse distribution cycle, the cancel pulse that is assumed to have been output between the previous pulse distribution cycle and the current pulse distribution cycle is output as a cancellation pulse in the next pulse distribution cycle. , assuming that the servo mechanism responds with a first-order delay, and after adding the correction pulse, the correction pulse ff1L is delayed by one pulse distribution period.
This cancels the first-order delayed output pulse amount corresponding to the first-order delay.

By processing in this way, even if the direction of the distribution pulse changes at the change of quadrants during circular interpolation during arc cutting as shown in Figure 3 (a), backlash can be added by the correction pulse. After that, the correction pulse ff1L corrected according to the gain of the servo system is canceled, so that the shape error shown by the broken line in Fig. 3(a) does not occur, and the machining is performed as instructed, that is, the circular arc. can be cut.

FIG. 2 is a flowchart of an embodiment when a numerical control device with a built-in computer implements the method of the present invention.

First, in the memory of the numerical control device (as well as setting a correction pulse corresponding to backlash a), the number of outputs of cancel pulses and correction pulses are set according to the gain of the servo system as shown in Table 3. The pulse to be canceled from the pulse distribution period after addition, that is, the cancel pulse amount is set in the memory. Then, the numerical control device performs the processing shown in Figure 2 before or after performing the pulse distribution process for each pulse distribution period for each axis, and cancels the process before or after performing the pulse distribution process for each pulse distribution period. It is determined whether the value of the counter C in which the number of pulses to be output is set is "0" or not (step 81). If it is rOJ, it is assumed that there is no cancel pulse command, and the process moves to step S4 to determine the pulse distribution period in question. It is determined whether the command at is changed from the direction of movement of the command in the previous pulse distribution cycle (step S4), and if it is not changed, this backlash correction process is ended. Then, if this process is performed every pulse distribution cycle, and now the moving direction of the current cycle has changed from the moving direction of the previous cycle,
The process moves from step S4 to step S5, and the correction pulse ff1 corresponds to the backlash amount a set from the memory.
Read L and output it in the corresponding period (step S5
). Then, the number of cancel outputs set in the memory is set in the counter C (step 86). the result,
In the next pulse distribution cycle, since the value of counter C is not "0" in step S1, the process moves to step S2, reads the cancel pulse data stored in the address position corresponding to the value of counter C, and reads the read data. Based on this, a cancellation pulse with a sign inverted from that of the current regular distribution pulse is output. That is, for example, when the number of cancel pulse outputs is set to "3", the value of counter C will be "3" in the next pulse distribution cycle after outputting the correction pulse.
3", but when the value is "3", the first cancellation pulse amount is read, and in the example in Table 3, 0.5411 is read, the sign is inverted, and the moving direction by the regular distribution pulse is determined. In contrast, the cancellation pulse in the opposite direction is 0.5411
Output. Next, step S to subtract 1 from the counter C.
3) Next, the process moves to step S4, but normally the sign does not reverse during the multiple pulse distribution period in which the moving direction is reversed, so the processing within this pulse distribution period is completed, and the counter is not activated in the next pulse distribution period. Since the value of C is "2", the process moves from step S1 to step S2 again, and the counter C
Read the cancellation pulse amount 0.3181 corresponding to
The sign is inverted and output (step S2), and "1" is subtracted from the counter C (step 83). As a result, the value of the counter C becomes "1", so in the next pulse distribution cycle, the third cancel pulse ff10.141L is read, its sign is inverted, and output.
"1" is subtracted from (step 83). At this point, the value of the counter C becomes "0", so in the next pulse distribution cycle, since the value of the counter C is "0" in step S1, the process does not proceed to step S2, and the cancel pulse m output process is performed. Otherwise, the process moves to step S4.

As described above, the correction pulse is output every time the sign of the pulse distribution command for each axis is reversed (every time the direction of movement is reversed), and the correction pulse is output at each cycle from the next cycle until the set number of pulse distribution cycles. Each time, the set cancel pulse is output with its sign inverted. As a result, shape memory errors due to backlash are corrected, and it is possible to process the shape as instructed.

Effects of the Invention As described above, the present invention eliminates partial shape errors due to backlash that occur when the moving direction of axes is reversed during interpolation in which multiple axes are to move synchronously in a fully closed loop control system. Since shape errors are eliminated by correcting backlash, more accurate machining can be performed using a full closed loop control system.

[Brief explanation of drawings]

FIG. 1 is a functional block diagram of an embodiment of the present invention, and FIG. 2 is a flowchart of backlash correction processing of the same embodiment.
Figures 3 (a) and (b) are illustrations of shape errors that occur during arc machining using the full closed loop control method, Figure 4 is an illustration of the semi-closed loop control method, and Figure 5 is the full closed loop control method. FIG. 6 is an explanatory diagram of backlash correction. 1...Circular arc, 2...Shape error, 16X, 16Y, 102...Servo motor, 19X,
19Y, 107... Machine, 20X, 20Y, 108... Position detector. , 2 to 2 mouths χ 3 mouths (α) 30th (bl

Claims (3)

[Claims] (1) In the numerical control method of the full closed loop control method, when the moving direction of pulse distribution for each axis changes, a correction pulse corresponding to backlash is applied to the axis whose moving direction has changed in addition to the regular command distribution pulse. A backlash correction method using full closed loop control that subtracts the added correction pulse from the next pulse distribution cycle. (2) The backlash correction method in full-closed loop control according to claim 1, wherein the cancellation pulse corresponding to the correction pulse is reduced over one or more pulse distribution periods in accordance with the servo gain. (3) The cancellation path considers the correction pulse as a step input and approximates the first-order delay with respect to the step input,
A backlash correction method in full closed loop control according to claim 2, wherein the backlash is reduced in each pulse distribution period.