The present invention relates to a device for correcting of backlash, which arises on a change in the direction of movement in a drive mechanism comprising a power driven screw and a nut means co-operating with the screw, the nut means being connected to a unit which is movable back and forth along the screw, preferably a printing unit of a printer or a typewriter. The device comprises a pulse generating means of incremental type which in synchronism with the movement of the screw is arranged to generate two pulse trains having the same quadrangular shape but being 90 out of phase. The pulse trains have a period corresponding to the displacement of the movable unit a predetermined distance along the screw and are arranged to be applied to a means which out of the pulse trains determines the position and the direction of movement of the movable unit. It is known in printers in connection with screw drive mechanisms of the kind referred to above to provide an incremental code disc that rotates with the screw and co-operates with a photoelectric device, in the usual way comprising light-emitting diodes and phototraπsistors, to generate two pulse trains being 90 out of phase. Out of the two pulse trains information is derived about the position and the direction of movement of a printing unit movable back and forth along the screw.

In order for the movable unit to be easily moved along the screw a certain play or backlash must be present between the screw and a nut means co-operating with the screw. Usually, the backlash is compensated in a mechanical way, for example the nut means may comprise two nuts which are spring biassed away from each other and towards the flanks of the helical groove of the screw.

The object of the invention is primarily without using mechanical arrange- ments to correct for the play or backlash that necessarily must be present between the nut means and the screw. In accordance with the principle of the invention a measured or estimated value of the existing backlash is stored and used for the correction of control signals which are applied to an electric motor provided for driving of the screw. By this kind of correction the drive arrange- ment can be mechanically constructed such that the nut means will move along the screw with a minimum of friction at the same time as the backlash can be compensated at a higher degree than would be possible with mechanical arrange¬ ments. Correction must take place every time the movable unit changes its direction of movement. If in that connection the backlash is regularly measured

C" ^*~ I

the wear, appearing in the screw drive mechanism as time goes on and involving a gradual increase of the backlash, will be taken into consideration. Variations in magnitude of the backlash caused by wear cannot be equally corrected by mechanical correcting arrangements. A further object of the invention is to achieve correction of the existing backlash in step ^' s corresponding to a quarter of the period of the pulse trains utilized for determining of position and direction of movement.

The objects indicated will be achieved in a device having the characterizing features stated in the appending claims. An embodiment of the invention will now be described with reference to the

^' accompanying drawings. Fig. 1 is a block diagram of a correcting device according to the invention as applied to a printer, shown schematically. Fig. 2 is a detail view showing the block 23 of Fig. 1. Fig. 2a shows the time diagrams relating to the signal level at two inputs X, Y of the block 23. Fig. 3 shows how to combine two pulse trains in order to move a predetermined flank of one of the pulse trains in a selected direction. Fig. 4 is a block diagram showing a device for combining the pulse trains according to Fig. 3. Figs. 5 and 5a show schematically the order in correcting for backlash in both directions of movement of the movable unit shown in Fig. 1. Fig. 6, finally, is a circuit diagram of a useful modifying circuit built up with digital standard TTL- circuits.

Fig. 1 schematically shows a printer 10 comprising a carriage 11, which supports a printing head' 12. The carriage Is supported by supporting shafts 13, 14 and is driven back and forth along the supporting shafts by a drive mechanism comprising a screw 16, driven by a motor 15, and a nut means 17 fixedly connected to the carriage 11. The printing head 12 co-operates with a platen 18, acting as an abutment, for applying characters onto a web, not shown, the web being fed in the space between the printing head and the platen in a direction perpendicular to the direction of the supporting shafts 13, 14.

For determining the position of the printing head along a line on the web there is provided an incremental code disc device 19 comprising a code disc 39 rotating with the screw 16 and a photoelectric device 40 co-operating with the code disc. The code disc device emits two pulse trains which are of equal quadrangular shape but being 90° out of phase. Via a line 20 the pulse trains are led to a block 21 comprising a device for correcting of backlash. In the block 21 the pulse trains are modified and subsequently transferred to a block 23 via a line 22. The block 23 comprises a device which out of the pulse trains derives information about the position and direction of movement of the printing head 12. The block 23 has two output terminals which emit signals on conductors 24,

operate a position counter 26 to count up and down, respectively. The position counter is connected to a block 28 via a line 27. The block 28 comprises a control device connected to operate the motor 15 via a line 29. For the guidance of the backlash correcting device in block 21 a value of the backlash between the screw 16 and the nut means 17 must be provided. The backlash value can be measured or produced in some other way. In the example a block 30 is provided which comprises a backlash measuring device which can be of the kind described in the International patent application No. PCT/SE82/00177. Regularly, the backlash measuring device measures the appearing backlash and stores the measured value which via a line 31 can be transferred to the backlash correcting device in the block 21. For the superior control of the various activities in the printer 10 a microprocessor is provided which is represented by a block 32, the control function of which being represented by an arrow 33.

Fig. 2 shows the block 23 which comprises a device for decoding as to position and direction of movement of the pulse trains A and. B generated by the code disc device 19. For the sake of simplicity it is assumed that the two pulse trains have passed the block 21 without being changed. The block 23 has two input terminals X and Y, respectively, to which the pulse trains A and B., respectively, are applied. In Fig. 2a the time diagrams of the signal levels on input terminals X and Y are shown. For the calculation of the position of the carriage 11 during the movement to the right in Fig. 1, the counter 26 is operated to count up by a signal emitted on the conductor 24 each time a positive flank appears on the input Y at the same time as the input X is high. This is illustrated by the upward directed arrows above the time diagram referred to by X in Fig. 2a. As the carriage moves to the left the position counter is correspondingly operated to count down by a signal emitted on the conductor 25 each time a negative flank appears on the input Y at the same time as the input X is high. The points, of. time at which counting down occurs are indicated by the downward directed arrows below the time diagram referred to by Y. The directions of movement of, the carriage are indicated by arrows R and L, respectively. A comparison between the inputs X and Y indicates that counting up and counting down, respectively, of the position counter 26 occurs in the the same positions during movement to the right and movement to the left, respectively, which means that the negative effect of the backlash has been eliminated in a simple way. In the following description of the backlash correcting device, counting up and down, respectively, of the position counter takes place with the periodicity indicated. However, the backlash is compensated in steps which amount to a quarter of the period by relative displacement of the pulse trains in a way v/hich

will be described more in detail below. The value used for correcting purposes, which could be estimated or measured, of course also has the higher resolution that will be required. For example, in using the measuring device according to the International patent application No. PCT/SE82/00177 referred to above, the pulse trains according to the time diagrams X and Y in Fig. 2a may be combined such that every change in any of the pulse trains generates a pulse which activates the measuring device. Hence, measuring will be performed four times during the period of either of the pulse trains, the measuring times being evenly distributed, i.e. displaced one quarter of the period relative to one another. With reference to Fig. 3, the theory of the correcting of backlash will now

^" be described. The two pulse trains to be modified in the block 21 have been designated A and B, respectively. These pulse trains can now be combined In four different ways in such an order that every change from one combination to a succeeding one will cause the displacement of one predetermined flank of pulse train B 90° or one quarter of the period. In the uppermost line designated a there are shown pulse trains which arrive in the block 21 from the code disc device 19. The first combination means that the pulse trains remain unchanged and these will become reference pulse trains throughout the following description relating to the order to be chosen for the different combinations in order for a predetermined flank in the pulse train B to be moved first four steps of 90 to the right and then one step to the left. The different combinations are indicated in the lines a - f and in the pulse train B the positive flanks have been numbered 1 - 10. The flank 8 has been specially marked and the influence on the position of this flank by the different combinations will now be studied. In order to effect the first displacement of the flank 8 one step of 90 , the pulse trains A and B are to change place and furthermore the pulse train A is to be inverted. This is shown in line b where the pulse trains A and B have been modified accordingly into B and A. Then, the pulse trains B and A are modified, as shown in line c, such that B is inverted and A remains unchanged, and moreover the pulse trains change place and hence give the result A, B as outputs from the block 21. In line d the next change is shown, which means that the pulse train A is replaced by B and the pulse train B is replaced by A. The last one of the four steps to the right in Fig. 3 will be obtained according to line e by the pulse train B being replaced by A and the pulse train A by B whereupon the order according to line a has been re-established. As a completion, line f shows that modifying of the pulse trains according to line d causes a displacement of the flank 8 one step to the left. By performing the different combinations in the order represented by the lines a - e in a cyclically repeated course it will be possible to have the flank 8

displaced an arbitrary number of steps of one quarter of the period to the right. The corresponding displacement to the left will be achieved by performing the different combinations in the opposite order. The designation +1 indicates the points of time for the four displacement steps in the positive direction, i.e. to the right. The designation -1 correspondingly indicates the time at which the dis¬ placement of one step to the left takes place according to line f.

Fig. 4 shows the different circuit variants that in principle will be needed for performing the combinations shown in lines a - d of Fig. 3. Lines a - d correspond to the circuit variants 0 - 3 in Fig. 4. The two pulse trains being applied to the block 21 are designated A and B as before, whereas the pulse trains transferred to the block 23 from the block 21 are designated X and Y, respect¬ ively. Moreover, inversion of any of the pulse trains is indicated by a filled ring.

The circuit variant 0 means that the pulse trains A and B pass the block 21 unchanged and, accordingly, X will be equal to A while Y will be equal to B. In the variant 1, A and B are to change place and A is to be inverted so that X will become B and Y will become A, compare line b in Fig. 3. The variant 2 means that A and B are to be inverted and, hence, X will , become A while Y will become B, see line c in Fig. 3. The circuit variant 3, finally, means that the pulse trains A and B change place and that the pulse train B is inverted so that X becomes B while Y becomes A. The arrows designated "+1" indicate the order of the different circuit variants 0 - 3 in moving an arbitrary positive flank in the pulse train B to the right in Fig. 3 in steps of 90° or one quarter of the period. The arrows "-1" indicate correspondingly the order in moving to the left. By S . and S, in the block 21 (Fig. 1) those inputs are designated which control the selection of circuit variants. These inputs are connected to the block 32 in a way not indicated.

Fig. 5 schematically shows a circuit arrangement for performing the different combinations of the pulse trains A and B. Pairs of rectangles 0,0 - 3,3 are connected to the inputs A, B to which the corresponding pulse train is applied, either directly or via an inverting circuit represented by a filled circle. The rectangles are arranged in two groups 0 - 3, 0 - 3, one of which is connectible to an output X for a correspondingly designated pulse train and the other is connectible to an output Y for a pulse train having the same designation. The circuit arrangement is such that the rectangles of one pair are simultaneously connected to the respective output, so that one rectangle is connected to X whereas the other is connected to Y. In the example shown in Fig. 5, the rectangles 1 are provided with arrows connecting the rectangles to the outputs X

and Y, respectively. For example, the arrows can be constituted by movable contacts in a switch.

The order relating to the combinations of Fig. 5 is shown in Fig. 5a, in which clockwise rotation corresponds to positive correction or displacement of the flank 8 to the right in Fig. 3. Anticlockwise rotation corresponds to negative correction or displacement to the left in Fig. 3. The displacement to the right and left, respectively, in Fig. 3 corresponds in the example to a virtual displacement of the carriage 11 and the printing head 12 in opposite directions, i.e. positive correction corresponds to a virtual displacement to the left and vice versa. In order to explain this state of opposition more in detail, reference is also made to lines e and f in Fig. 3 which lines show the negative correction. It is assumed that the carriage has been moved to the right and then stopped. During the movement the position counter 26 (Fig. 1) has been operated to count up to a value corresponding to the position in which the carriage has been stopped. Upon a subsequent movement of the carriage to the left a backlash correction must take place, which means that the position counter is to count up through a number of additional positions which correspond to the backlash. In order to perform the counting up of the position counter, a negative correction must take place, i.e. the flank 8 in Fig. 3 has to be moved to the left. Looking at a point immediately to the left of the position 8 in the lower of the two pulse trains in line e, the pulse counter in this point has counted only the position 7. In line f, negative correction has taken place with one step of one quarter of the period causing the position 8 to be displaced to the left of the point of observation in line e. This means that the position counter now indicates the position 8 and the desired counting up has been performed. In case of a backlash amounting to three steps of one quarter of a period, upon shift in the direction of movement from the right to the left, correction will take place according to the order 0, 3, 2, 1, as seen in Fig. 5a. In the same way shift from the left to the right of the direction of movement will result in a correction according to the order 1, 2, 3, 0. As indicated above, correction will take place in steps of 90 or a quarter of. the period while the position counter 26 counts up or down in steps corresponding to one complete period. Even if a correction step will not always influence the position counter, every such step will always be taken into account by the lower of. the pulse trains of Fig. 3 being correspondingly displaced relative to the upper pulse train. In Fig. 6 there is shown an example of how to realize the circuit arrange¬ ment of Fig. 5 by the use of four TTL-circuits. In order to avoid correction during the change in any of the pulse trains A and B, the circuit arrangement includes circuits for synchronization by the use of a two-phase clock circuit^ii

well known for those skilled in the art. A clock pulse source, not shown, emit clock pulses on a conductor 34. This conductor is connected to two clock puls inputs designated CP, , CP„ of a TTL-circuit generally denoted by 35 and being o the type 74109. This circuit comprises two flip-flops each having one non-invert ing output Q, , Q ₂ and one inverting output Q, , Q ₂ . The pulse train A appears o the output Q, and the inverse A of this pulse train appears on the output Q, . I the same way the pulse train B appears on the output GL and the inverse B of this pulse train appears on the output Q ₂ . The outputs are connected to a TTL-circuit 36 being of the type 74153 and comprising a double selector each having four inputs and one output. One of the selectors connects a selected one of the four ^' inputs to an output X while the other selector connects a selected one of the four inputs to an output Y. All eight inputs of the circuit 36 are connected to the outputs Q, , Q, ; Q„, Q„ of the circuit 35 in accordance with the pattern shown in Fig. 5. Corresponding inputs of the two selectors of the circuit 36 have been given the same designations 0,0 - 3,3 as used in Fig. 5. However, to the reference numerals the letter A has been added, the letter referring to signal levels at two inputs A, , A ₂ , to which are applied signals S _Q , S, . The last-mentioned signals determine which of the inputs A,0 - A,3 of the respective one of the two selectors in the circuit 36 is to be connected to the output' X and the output Y, respect- ively. In practice, A, , A ₂ can take the binary values 00, 01, 10, 11, which in the order indicated would correspond to (A,0j A,0), (A,l; A,l), (A,2j A,2), (A,3; A,3). Hence, A, , A~ = 00 means that the signals at the inputs A,0 are transferred to the outputs X and Y, respectively.

The two-phase clock synchronization indicated above involves the provision of an additional TTL-circuit 37 of the type 74109 to control that the signals S _Q , S, , determining the shape of the pulse trains at the outputs X, Y of the circuit 36 will not appear at the inputs A, , A« of this circuit at the times when the changes take place in the pulse train A or B. The inputs A, , A„ are connected to non-inverting outputs Q,, GL which belong to the respective flip-flop of the double circuit 37. The signals S-, S, are applied to inputs D-,, D. on the respective flip-flop circuit and the clock pulse inputs CP,, CP^ are connected to the conductor 34 via an inverter circuit 38 of the type 7404.

The function of the device of Fig. 6 is the same as described with reference to Fig. 5 and should not need to be further commented. Moreover, for guiding purposes the different combinations of the pulse trains A, B that appear on the inputs of the circuit 36 have been indicated in Fig. 6. The two-phase synchroniza¬ tion entails the inactivation of the clock pulse inputs CP-,, CP. when the clock pulse inputs CP, ,. CP ₂ are activated.

( O K ^" WI