JPH02132324A - Absolute value encoder - Google Patents

Abstract

PURPOSE:To detect an absolute angle in one turn by using a code detecting element in addition to a code plate which is divided equally into N slits and given 0 or 1 in the respective slits so that all binary-coded values B1-BN forming P continuous slits form different cyclic random number sequences. CONSTITUTION:A P-bit random number code of 1 and 0 is set and this binary- coded value is denoted as B'1. The code value is shifted right by one bit and 0 is added to generate a code value B'2; when B'2not equal to B'1, a (P-1)-bit random code which has 0 added on the left side of the most significant digit bit is generated. Further, the code is shifted right by one bit and 0 is arranged at the most significant digit bit position to obtain a code value B'3; when B'2not equal to B'1 and B'2, 0 is added at the most significant digit bit of the code to generate a (P+2)-bit random code and N-bit random codes are generated similarly. When the code is equal, the code is shifted and 1 is added at the most significant return digit. Thus, the absolute angle is found by P slit detecting elements.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an absolute value encoder using a cyclic random number code.

[Prior Art] Conventionally, an absolute value encoder generally uses a code plate in which a plurality of tracks having different numbers of slits are arranged in a radial direction to constitute a pinary code or a gray code having a desired number of bits. However, in this method, the number of tracks increases in accordance with the number of bits, making it difficult to realize a compact, high-resolution absolute value encoder.

As one method for solving this problem, an absolute value encoder using M-sequence random numbers has been proposed.

As is well known, an M-based random number is a sequence of maximum length that can be created using a shift register.
2k-1 cyclic patterns other than 1 are generated. For example 4
In the case of bits, a code as shown in Table 1 occurs in a circuit using a shift register as shown in FIG. however,
R in Figure 7. ~R3 shall not become O at the same time.

Table 1 The M-sequence random numbers generated in this way are a cyclic pattern of 2k-1 different codes, so for example, a code plate with 0 in the opaque part and 1 in the transparent part placed on the circumference. If you create a pattern and read the pattern of adjacent k bits, the absolute position on the circumference can be determined because that pattern exists only at one location on the circumference.

[Problem to be solved by the invention]

Now, the absolute value encoder generally has a resolution of 2k using a pinary code or a gray code. However, the above-mentioned absolute value encoder using M-sequence random numbers can only achieve a resolution of 2k-1, and has the disadvantage that it is not compatible with general encoders.

The purpose of the invention is to provide an absolute value encoder with arbitrary resolution.

[Means for Solving the Problems] The first absolute value encoder of the present invention is a cyclic encoder that is equally divided into N slits and has different binary code values (B1 to BN) made up of P consecutive slits. Absolute angle within one rotation is measured using a code plate in which each slit is marked with a code of O or 1 to form a random number sequence, and a detection element that reads the codes of P consecutive slits on the code plate. It is something to detect.

FIG. 1 is a flowchart illustrating a method of generating a cyclic random number code in a first absolute value encoder.

First, in step 1, an initial code B°1 consisting of P bits of 0 or 1 is set. These P bits become the first P bits of the random number sequence. Next, in step 2, the initial code B'
Shift r to the right (or left) by 1 bit. Here, in the case of a left shift, the processing shown in parentheses below is performed. O is placed in the most significant bit (or least significant bit) after the right shift (or left shift). This bit becomes the P+1st bit in the random number sequence. In step 3, the newly created P-bit code B'2 and the initial code B'+ are compared. If they are different, go to step 4; if they are equal, go to step 5. In step 4, it is determined whether the length of the random number sequence has reached N, that is, whether the new code is the Nth code. If N, the process goes to step 6; otherwise, the process goes to step 2. Assume that in this way, up to the i-th code B'+ is generated in step 4 and the process goes to step 2. In step 2, this code is right-shifted (or left-shifted) by 1 bit and the most significant (
or the least significant bit) and write the new code as B'
+++. Furthermore, in step 3, it is checked whether B'+ to B'+ have the same code as B'141. If there is no same code, go to step 4, but if there is the same code, set the most significant (or least significant) bit O to 1 in step 5.
Change it to , and go to step 3 again. Again, if the new code B'l+1 and B'+~B''1 have the same code, go to step 5 again, but in this case the most significant (or least significant) bit is already 1. Therefore, the most significant bit of the random number sequence is Shift left (or shift right) until the most significant (or least significant) bit is 0, i.e. work backwards to code B'p where the most significant (or least significant) bit is 0, and its most significant (or least significant) bit The process of returning to step 3 again with the new code B″p set to 1 is repeated.

As described above, in step 4, the Nth coat B's
In step 6, it is determined whether the code obtained by shifting B'N to the right (or left) by 1 bit can become the initial code B'1, and if they are equal, the process ends.
If they are not equal, proceed to step 5 and a cyclic random number code will be obtained.

In addition, in step 2, shift left (or shift right)
The same cyclic random number sequence can be obtained by adding 1 to the most significant bit (or least significant bit) after In this case, O may be changed to 1 and 1 to 0 in subsequent processing.

The second absolute value encoder of the present invention is a cyclic random number sequence that is equally divided into N slits and has a different P-bit binary code value CB, ~BN) consisting of every M slits and P slits. O or 1 in each slit to form
The absolute angle within one rotation is detected using a code plate on which a code is written and a detection element that reads codes from every M slits and P slits from the code plate.

FIG. 2 is a flowchart illustrating a method of generating a cyclic random number code for a second absolute value encoder.

FIG. 2 shows a method of generating a random number sequence of length N every M numbers.

First, in step l1, a sequence 1.2, . '2. ...+B'Ma1. Also, as a random number sequence, sequence 1.2. .... Sequence M+1
For each bit of sequence 1, 2 .・・・． A sequence of length (M+1)P arranged bit by bit in the order of sequence M+1 is an initial random number sequence. Next, in step l2, 1 for sequence 1
Shift bits to the right (or left). Here, in the case of a left shift, the processing in parentheses will be performed thereafter. The topmost position after right shift (or left shift) (
Or the least significant) bit is set to O, and its binary code is B'M+2. Next, in step 13, the newly created P-bit code B″M+2 is changed to the previous code B″M+2.
'1, B'2,...+ Compare whether it is equal to B'M+1. If there are no equal codes, step 14
If there is an equal code, go to step l6. In step 14, the most significant (or least significant) bit of sequence 1 is placed to the left (or right) of the most significant (or least significant) bit of the random number sequence to create a random number sequence of (M+1)P+1 bits, and a new code is created. It is determined whether or not it is the Nth code. If it is the Nth code, the process goes to step 20; otherwise, the process goes to step 15. In step 15, sequence 2 is changed to new sequence 1, ... sequence M+1 is changed to new sequence M, sequence 1.
is replaced with a new sequence M+1, and step 12 is performed again for the new sequence 1 to search for a new code B'M+3. Also, when going to step 16, it is determined whether the most significant (or least significant) bit of sequence 1 is O, and if it is 0, step l7
If not, go to step 18. step l
In step 7, the most significant (or least significant) bit is changed to 1 and the process returns to step 13. On the other hand, in step 18, the current code is the initial code B'1, B
It is determined whether “2, . In step 19, sequence 1 sequence 2, etc. Sequence M
+1 is shifted to the left (or right) by 1 bit, and the process returns to step 16 again. At this time,
The random number sequence will be returned by M+1 bits. Above step 1618. In loop 19, sequence 1 is shifted left (or right) until the most significant (or least significant) bit of sequence 1 becomes 0 or sequence 1 becomes the initial code. Also, in step 21, it is determined whether the current code is the maximum value that a P-bit binary number can take, and if it is not the maximum value, the process goes to step 12, and if it is the maximum value, it is assumed that a random number sequence cannot be generated by this algorithm and processing is performed. Abort. In step 22, the current code is incremented by 1 and the code is set as a new initial code, and the process goes to step 13 again.

Assume that up to the i-th code B'+ is generated in this way and the process goes to step 12. In step l2, the new sequence l is shifted to the right (or left) by 1 bit, and the most significant (or least significant) bit is set to 0, making the new code B'l. Furthermore, in step l3, B”1,・
・・・． Check whether B゜1 has a code equal to B'I+1. If there are no equal codes, go to step 14, but if there are equal codes, go to steps l6 and 1.
At step 7, the most significant (or least significant) bit O is changed to 1, and the process goes to step 13 again. Again, if there is a code equal to the new B'l+I and B''1,...,B'1, we go to step 16 again, but in this case the most significant (or least significant) bit is already 1, so The top of sequence 1 (
Shift left (or shift right) until the most significant (or least significant) bit becomes O, i.e. go back to code B'4 where the most significant (or least significant) bit is O, and in step l7 change that most significant (or least significant) bit. Change the code changed to 1 to a new B
The process of going to step l3 again as 'J' is repeated.

As described above, in step 14, the Nth code B'N is
Suppose that up to B'N-M, ·
..., B'8, that is, sequence 2, ..., sequence M+1 and sequence 1 are shifted to the right (or left) by 1 bit to the most significant (or least significant) bit? Or, determine whether the code B'N+++...+B'lJ+ll+■ with 1 is equal to the initial setting code B゜1,...,B'u++, and if they are equal, the search ends; if they are not equal, If the process goes to step 16 again, a cyclic random number sequence of length N every M bits will be obtained.

Note that the same cyclic random number sequence can be obtained even if 1 is placed in the most significant bit (or least significant bit) after the left shift (or right shift) in step 12. In this case, 0 may be changed to 1 and 1 may be changed to O in subsequent processing.

As described above, according to the present invention, an absolute value encoder with arbitrary resolution can be realized.

[Example] Next, an example of the present invention will be described with reference to the drawings.

FIG. 3 is a diagram showing a code plate of the first embodiment of the absolute value encoder of the present invention.

If we read the black part as ``O'' and the white part as ``1'', we get the following cyclic sequence. If this is read by a detection element as a sequence of 4 adjacent bits R0 to R3, it will be R in Table 2. It is worth noting that one round can be divided into 16 different patterns like ~R3).

Table 2 Table 3 shows the process of generating the cyclic random number sequence in Table 2 using the algorithm shown in FIG. Here, N=1
6, P=4 bits, right shift, and the random number code when reaching step 3 in FIG. 1 is shown each time.

First, ゜'oooo゜゜ is set as the initial random number code B'l (step 1). If we shift this to the right by 1 bit and set the most significant bit to O, then the code for tooth 2 is B'2=”
It becomes ooooo゜゜ (step 2). Since B'1 and B'2 are equal, set the most significant bit to 1 and set the tooth. 3 code B '2 = '1000° is obtained (step 3.5
).

No. 3 code B'2 is No. 2 code B'2
, and since N=16, the most significant bit is shifted to the right by 1 bit, and the tooth . 4 code B '3''
'0100' is obtained (step 3, 4.2). Hereafter, in the same way, teeth. 5 code B '4 = '001
0°, teeth. 6 code B '5 = '0001', No
．． 7 (7) :l-doB', = 'oooo' is obtained. Code B'a is No. Since it is equal to the code B'+ of 1, the most significant bit is set to 1 and No. ．． ．． Code B'a of No. 8 is obtained as No. 3.
is equal to code B'3, and the most significant bit is already 1
Therefore, shift to the left until the most significant bit becomes 0,
That is, the code goes backwards up to the code B'5 with the most significant bit being O, and the code 〈l001〉 with the most significant bit set to 1 is set as the new code B's (step 3.5). This code B'. = '1001' does not have the same item before,
Also, since N=16, shift it to the right by 1 bit, set the most significant bit to O, and code B't, = '0 for N010.
100° is obtained (steps 3, 4.2). In this way, teeth. 36, 16th code B'+6='0001
'' is obtained, so it is determined whether the code ゛゛゜゜゜゛゛゜゜゜゛゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜゜〾.

Since the random number sequence obtained as described above can have any length N, an absolute value encoder with resolution N can be realized by making a code plate by using this as a slit. Furthermore, an encoder can be realized in which an incremental slit with a resolution of N is provided in addition to the random number slit. Furthermore, a multi-rotation type absolute value encoder can be realized by providing means for generating one pulse per rotation on the same code plate or on the same rotation center axis, and by providing means for counting the incremental pulses with a counter.

FIG. 4 is a circuit diagram of a random number code generation circuit that generates the 4-bit random number code shown in Table 2.

This circuit consists of a shift register 31 and an exclusive OR circuit 3.
2, a NOR circuit 33, an inverter 3435, an AND circuit 36,' 37, and an OR circuit 38. Among the patterns generated in the circuit shown in FIG. 7, 1000→0001
Add O to 1000 → 0000 → 00
A circuit is added so that it becomes 0l. That is,
When all three bits of RO-R2 are O, the next shift input is determined according to the value of R, and otherwise the same pattern as the circuit of FIG. 7 is generated.

Adjacent 4 bits are detected using the code plate shown in FIG. 3, and the data and the output R of the circuit shown in FIG. 4 are obtained. .about.R3 until they match, the position of the code plate is uniquely determined by the number of pulses of the shift clock. In this way, an absolute value encoder with a resolution of 2k can be realized.

FIG. 5 is a diagram showing a code plate of an absolute value encoder according to a second embodiment of the present invention.

As before, if you read the black part as ``0'' and the white part as ``1'', you will get a circular random number sequence for one round. 1 as shown in Figure 5.
If it is read as a 4-bit number sequence Ro''R3 every third time, it can be seen that one round can be divided into 16 different patterns as shown in R.-R3 in Table 4.

Therefore, even if such a code plate is used, an absolute value encoder can be realized as before. In this case, unlike the first embodiment, the detection elements are arranged every other bit.
The physical feasibility is significantly improved compared to the embodiment.

Table 4 Table 5 shows the process of generating the cyclic random number sequence in Table 4 using the algorithm shown in FIG. Here, N=1
6, M=1, P=4 bits, right shift, the random number code when reaching step 13 in FIG. 2 is shown each time.

First, as an initial code, No. 1.No. 2゜'＋
000゛ as sequence 1 and B'+," o o o o
” is set as number sequence 2 and B゜2. Number sequence 1 is shifted to the right by step 12, and code No. 3 is set as B'3.
shall be. When this is determined in step 13, there is no equal code, so the process goes to step 14, but this is N=16
Since it is not the number, go to step 15, swap sequence 1 and sequence 2, and go to step 12 again. new sequence 1
Shift to the right for No. 4 court B'4, but when this is determined in step 13, it is equal to B'2, so step 16. No. 17. 5 code B'a
is obtained. When this is determined in step 13, this time it is equal to B'+, so the process goes to step 16 again, but since the most significant bit is 1, the code returns to code B'2 where the most significant bit is O.

In step 17, No. 2 is selected as the new B'2. 6 is made and judged in step 13, it is equal to B'+, so the process goes to step 16 again. This time too, the most significant bit is 1, but at the same time the initial code is B'2, so step 21. No. 22 as a new code. B'2 of 7 is made and the determination is made again in step 13. Repeat this from now on.

In this way, NO. 120 and 16th code B'
+6 is obtained, so when the judgment in step 20 is made, the tooth. 121, teeth. Like l22, initial code B'+B
' Since it is equal to 2, the desired random number sequence has been obtained.

FIG. 6 is a diagram showing a code plate of a third embodiment of the absolute value encoder of the present invention.

As before, this sets the black part to "0" and the white part to "
1", you will get a cyclic random number sequence for one round. 2 as shown in Figure 6.
A sequence R of 4 bits every other piece. R in Table 6 if read as ~R3. It can be seen that one round can be divided into 16 different patterns as shown in ~R3.

Therefore, even if such a code plate is used, an absolute value encoder can be realized as before. In this case, unlike the second embodiment, the detection elements are arranged every two bits, so the first
The physical feasibility is significantly improved compared to the embodiment.

Table 7 shows the process of generating the cyclic random number sequence in Table 6 using the algorithm shown in FIG. Here..., N=
16, M=2, P=4 bits, right shift case, second
The random number code when reaching step l3 in the figure is shown for each time.

First, as the initial code, No. 1. ..., No.
3, "1000°" is the sequence 1 and B'+, "'100
1"゜ as sequence 2 and B゜2,"oooo゜゜ as sequence 3
and B'3.

Shifting the sequence 1 to the right in step 12 results in a code B'4 with an interval of 4. If this is determined in step 13, there is no equal code, so the process goes to step 14, but since this is not N=16th, the process goes to step l5. In step 15, sequence 2 is replaced with a new sequence 1, sequence 3 is replaced with a new sequence 2, sequence 1 is replaced with a new sequence 3, and the process returns to step 12.

Shifting the new sequence 1 to the right results in the code B'5 for teeth, 5, which is determined in step 13.
' Since it is equal to 4, step 16. No. 17. 6 code B's is obtained. When this is determined in step 13, there is no equal code this time, so step 14.
Number sequence 1 with 15. Sequence 2. Replace the sequence 3 and go to step 12 again. In this way, No. When code B'6 of 8 is obtained, it is determined in step 13 that there is the same code, so the process goes to step 16 again, but since the most significant bit is 1, the process returns to code B's where the most significant bit is 0. Tooth as new B'3 in step 17. 9 is made and judged in step 13, it is equal to B'+, so the process goes to step 16 again. This time too, the most significant bit is 1, but at the same time the initial code is B'3, so step 21.
No. 22 as a new code. B'3 of 10 is created and the determination is made again in step 13. Repeat this from now on.

In this way, No. Since the 16th code B°16 is obtained at 8 4 8, when the determination in step 20 is made, No. 849, No. 850, No.
Initial code B'1, B'i like 851
Since it is equal to B'3 (not shown), the desired random number sequence has been obtained.

In each of the above embodiments, only the low-rise type slit system has been described, but it can also be applied to a linear type, and as shown in Fig. 8, in this case, N-P slits Needless to say, the absolute position can be determined within the range of .

Furthermore, although the above explanation is for an optical encoder, in the case of a magnetic encoder, L and O are N. Needless to say, the magnetic pole pattern can be changed to the S pole.

[Effects of the Invention] As explained above, the present invention uses an algorithm to shift codes by M bits and add O or 1 so that all codes are different. When using the slit pattern of the M-sequence random number code, an absolute-value encoder with a resolution of 2k can be realized by using a circuit that generates a pattern that combines the M-sequence random number code and O. It has the effect of realizing

[Brief explanation of the drawing]

FIG. 1 is a flowchart showing a method for generating a cyclic random number code in the first absolute value encoder of the present invention;
The figure is a flowchart showing a method of generating a cyclic random number code in the second absolute value encoder of the present invention, FIG. 3 is a diagram showing the code plate of the absolute value encoder of the first embodiment of the present invention, and FIG. FIG. 5 is a circuit diagram of a random number code generation circuit that generates the 4-bit random number code shown in Table 2.
FIG. 6 is a diagram showing the code plate of the absolute value encoder according to the third embodiment of the present invention, and FIG. 7 is a circuit for generating M-sequence random numbers. circuit diagram,
FIG. 8 is a diagram showing a code plate of a linear type encoder. 1~6.11~22...Step, 3l...
...Shift register, 32...Exclusive OR circuit, 33...NOR circuit, 34. 35...Inverter, 36. 37・
...AND circuit, 38...A circuit.

Claims (1)

[Claims] Each slit is divided equally into 1.N slits, and each slit is set with 0 or In an absolute value encoder that detects an absolute angle within one rotation using a code plate on which a code of 1 is written and a detection element that reads the codes of P consecutive slits of the code plate, An arbitrary P-bit random number code to be represented is initialized, this binary code value is set to B'_1, and in the first work step, this initial random number code is shifted to the right or left by one bit (here, If you select left shift, perform the processing in parentheses below)
, place 0 in the place that will become the new most significant (or least significant) bit, and check whether the binary code value B_'2 of the second P bit obtained here is the same as the initial code value B'_1. If they are not the same, 0 is added to the left (or right) of the most significant (or least significant) bit of the P-bit random number code, making it a P+1-bit random number code, and in the second work step, 0 is added to the left (or right) of the most significant (or least significant) bit of the P-bit random number code. Shift (or left shift) and place a 0 where it will be the most significant (or least significant) bit at this stage, and the resulting binary code value of the third P bit, B'_3, becomes B' Determine whether they are the same as _1 and B'_2, and if they are not the same, add 0 or 1 to the left (or right) of the most significant (or least significant) bit of the P+1 bit random number code, and create a P+2 bit random number code. This process is repeated until an N-bit random number code is obtained, and at each step, when the same binary code value is found, the most significant (or least significant) bit becomes 0. Shift the random number code left (or right) until
At this stage, 0 in the most significant (or least significant) bit is converted to 1 again, and it is determined whether the binary code value is the same as the one from the initialization stage. If not, the right shift (or left shift) is performed. The absolute value is characterized in that the slit is configured using a method of generating a sequence of N cyclic random numbers, in which the operation of shifting to the left (or shifting to the right) is performed if they are the same. encoder. 2. In the absolute value encoder, it is equally divided into N slits, each slit has a code of 0 or 1, and N random number codes consisting of P consecutive slits (N = 2^P) are M sequences. It is characterized by comprising a code plate written to form a cyclic random number code that is a combination of a random number code and a code of all 0s, and a shift register circuit that generates data having the same configuration as the cyclic random number code. Absolute encoder. 3. P bit binary code value (B_1 to B
A code plate in which a slit code of 0 or 1 is written on each slit to form a cyclic random number sequence in which all _N) are different, and a code of P slits every M is read from the code plate. In an absolute value encoder that detects an absolute angle within one rotation using a detection element, M+1 different number sequences 1, number sequence 2, ..., number sequence M+1, which are represented by P arbitrary 1's and 0's, are initialized. Set these binary code values as B'_1, B'_2,...
, B'_M_+_1, and each bit of sequence 1, sequence 2, ..., sequence M+1 is arranged bit by bit in the order of sequence 1, sequence 2, ..., sequence M+1 as a random number sequence of length (M+1)P. Initialize the sequence, and in the first work step, shift the initial code of sequence 1 by 1 bit to the right or left (if you select left shift, perform the processing in parentheses below). , the number sequence in which 0 is newly placed in the most significant (or least significant) bit is set as the new number sequence 1, and the binary code value of the Mth+2nd P bit obtained here is B'_M
___+_2 is B'_1, B'_2, ..., B'_M_+
Determine whether it is the same as __1, and if it is not (M+1
) Add 0 to the left (or right) of the most significant (or least significant) bit of P random number sequences, making (M+1)P+1 random number sequences, and make sequence 2 a new sequence 1, ..., sequence M+1
Perform a swapping operation to make the new sequence M and sequence 1 the new sequence M+1, shift the new sequence 1 to the right (or left) by 1 bit, and place 0 in the most significant (or least significant) bit. Then, the Mth obtained here
+3rd P bit binary code value B'_M_+_3
is the same as B'_1, B'_2, ..., B'_M_+_2, and if not (M+1), the left side (or
(or on the right side) to create a random number sequence of (M+1)P+2 bits, is repeated until N random number codes are obtained, and when the same binary code value is found at each step,
The current code is the initial setting code B'_1, B'_2,...
, B'_M_+_1, sequence 1 until the working stage where the most significant (or least significant) bit of sequence 1 becomes 0.
, sequence 2, ..., shift the sequence M+1 and the random number sequence to the left (or shift right) and return, convert the 0 in the most significant (or least significant) bit to 1, and the obtained 2 Determine whether the evolved code value is the same as the binary code value from the initial setting stage, and if not, perform the right shift (or left shift), and if the same, perform the left shift (or right shift). The current code is the initial setting code B'_1, B'_2, etc.
, B'_M_+_1, if the current code is the maximum possible value of the P-bit binary code, set the current code to 1.
The slit is configured by N cyclic random number codes generated using a method in which the increased code is used as a new initial setting code and the process starts over from the above-mentioned determination step, and if the value is the maximum value, generation of a random number sequence is deemed impossible and the process ends. An absolute value encoder featuring: 4. A linear type absolute value encoder in which a part of the N cyclic random number codes is cut off and the code plate is linear.