JPH0257846B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to an absolute rotary position detection device called an absolute rotary encoder. As shown in Fig. 1, the absolute address type rotary position detection device encodes each coordinate position and writes it as a bright and dark pattern on a rotor (code plate) 1 attached to an input shaft 5. Holding the slit plate (to improve detection accuracy) 4,
By arranging the light emitting diode 3 and the photodiode 2 to face each other, position information corresponding to each coordinate position can be obtained from the photodiode 2 as the code plate 1 rotates due to the movement of the object to be detected. As the code written on the code plate 1, a single distance code (unit distance code) has conventionally been used. This means that only one bit always differs between two sets of adjacent codes.
The reason for using such a code is that if you use a code in which two or more bits change at the same time, there is a possibility that when detecting the code, an inversion of one bit is detected, but other bits have not yet been inverted. This is because it always occurs and incorrect code is output. The most well-known of these single metric codes is the Gray code. Gray code is pure 2
It is often used because it has a one-to-one correspondence with the decimal code and mutual conversion is easy. However, in the rotating code plate 1, in order to make the relationship between the first code and the last code a single distance code, the code plate 1
There is a constraint that the number of divisions must be 2 ⁿ (n is an integer) (If 2 ⁿ , the first code is "000...0", the ² nth code is "100...
1). Therefore, in many conventional rotary encoders commercially available, the number of equal divisions, such as 1024 divisions, 2048 divisions, and 4096 divisions, is a power of 2. However, for users, such an odd number of divisions is not desirable, and 1000
There is a tendency to prefer numbers without fractions, such as divisions or 3600 divisions. Conventionally, there have been two methods used to manufacture devices with such a divided number: (1) a method using a synchronizing signal, and (2) a method using a special code. Among these methods, method (1) is as follows. FIG. 2 shows the boundary between the code at address 0 and the code at address 999 of a code board divided into 1000 (the diagonally shaded part represents "1"). Adjacent codes from address 0 to address 999 all differ by only one bit, but when returning from address 999 to address 0, three bits become "1" as shown in this figure.
→Changes to “0”. Since it is virtually impossible to make these changes occur completely simultaneously, a synchronization signal Z is used in addition to bits ²⁰ to ²⁹ .
When this signal Z is "1", a code equal to address 0 is output regardless of other bits. This detection circuit is shown in FIG. 3. In FIG. 3, light emitted from a light source (light emitting diode) 3 passes through a code plate 1 and a fixed slit plate 4, and is irradiated onto a photoelectric element (photodiode) 2 at a position corresponding to the pattern of the code plate 1. The signal is passed through an amplifier 6 and an AND circuit 7, converted into a decimal number by a code converter 8, and outputted via an output buffer 9. The code plate 1 includes a track 1z dedicated to the synchronization signal Z, and the light obtained through this track 1z is received by a photoelectric element 2z and sent to an amplifier 6.
The signal is amplified by z, inverted by an inverter 10, and applied to an AND circuit 7. Therefore, when the synchronizing signal Z is obtained, the output becomes 0 (ie, the output at address 0) regardless of other bits. Therefore, each bit signal ²⁰ to ²⁹ changes from “1” to “0”.
As shown in Figure 2, if the position at which the bit signal changes is slightly shifted to the right from the true boundary P ₀ to the position P ₀ ' and overlapped with the synchronizing signal Z, each bit signal 2 ⁰ to 2 Position where ⁹ changes from “1” to “0”
Even if there is a variation in P ₀ ', it is not a problem as long as it is on the right side of the position where the synchronizing signal Z rises. Furthermore, at the position where the synchronization signal Z falls, each bit signal has already been
Since ²⁰ to ²⁹ have all fallen to 0, the output continues to be 0 even after the synchronization signal Z falls at position Pz. In the manner described above, ambiguity in reading can be eliminated. However, there is a problem in that an extra track 1z and a detection system are required to obtain the synchronization signal Z. In other words, when dividing by 1000 as in this example, the minimum required number of bits is 10 bits, but in reality 11 bits are required to obtain the synchronization signal Z. Therefore, in terms of the difficulty in manufacturing the code plate and the number of detection systems, it becomes equivalent to the case of 2048 divisions. Furthermore, although a detailed explanation of the above method (2) using a special code will be omitted, it has the same problem as the method using a synchronization signal in that it requires extra bits. This invention was made in view of the above points,
The present invention aims to provide a position detection device that can obtain an arbitrary division number using the minimum necessary number of bits without using synchronization signals or special codes. The basic idea of this invention is to focus on the fact that in a single-distance code string such as a Gray code, there are codes other than adjacent codes that have a single-distance code relationship, and to By removing intervals appropriately to form a circulating single-distance code string of an arbitrary number of codes (that is, a single-distance code in all parts), and patterning it into a rotor (code plate). , to obtain an arbitrary number of divisions. In other words, for example, if ²⁴ numbers from 0 to ²⁴ -1 in decimal notation are converted into Gray code, they will be as shown in Table 1 below. Looking at this, we can see that those that have a single distance code relationship are: It turns out that there are things other than those that are next to each other.

【table】

[Table] For example, if there is a single distance code relationship with 0 (0000), in addition to the adjacent 1 (0001) and 15 (1000),
There are 3 (0010) and 7 (0100). In other words, since this is a relationship in which only one bit differs from the relationship of a single distance code, if a code is represented by n bits, there are n codes that have a relationship of a single distance code for each code. do. Figure 4 shows the first
Among the 16 Gray codes in the table, those that have a single distance code relationship are shown by connecting them with lines. For two codes having a single distance code relationship, even if the code sandwiched between them is removed, the entire code string remains a single distance code string. For example, in Fig. 4, 2 (0011) and 13 (1011), which have a single distance code relationship with each other,
3, 4,... with the codes 14, 15, 0, 1 removed.
..., 12, 13, 2, 3, ... codes are also single distance code strings. Therefore, by appropriately selecting the position to be removed, it is possible to create a single distance code string with an arbitrary number of codes. For example, in Figure 4
To create 10 code strings from 16 code strings, 6 codes need to be removed, so [13,
14, 15, 0, 1, 2], [1, 2, 3, 4, 5,
6], or remove consecutive parts such as [15, 0, 3, 4, 7, 8], [1, 2, 5, 6,
9, 10] and so on. The method described above is to pick up the ones related to a single distance code one by one and select and remove only the necessary number of them. However, if the total number of codes is large and there are many to remove, it is necessary It is difficult. As can be seen from FIG. 4, there is a certain regularity in the relationship between single distance codes, so if this regularity is found, selection can be easily made. As a regularity, for example, in a Gray code, it can be said that objects located at symmetrical positions from the center of the code string have a single distance code relationship. For example, in the example shown in Figure 4, the center of the code string is between 7 and 8, so each Gray code of 6 and 9, 5 and 10, etc., located at a centrally symmetrical position, is a single gray code. It is related to the distance sign. Therefore, by taking advantage of this property, the number of divisions to be set (e.g. 1000) is subtracted from the initial number of codes (e.g. 1024), and the extra number (24) is paired up two by two from the center of the code string. (remove 12 codes before and after the center of the code string), a single distance code string consisting of the number of codes corresponding to the number of divisions is constructed. Therefore, in this invention, assuming that the desired even number of divisions is N, 2 ⁿ⁺¹ −N=2X is selected from among 2 ⁿ⁺¹ Gray code strings satisfying 2 ⁿ⁺¹ > ^N > 2 ^n. -From X
The code string in the interval between 2 ⁿ +1-1 is removed. This allows any even number of divisions to be set (where 2 ⁿ⁺¹ −N=
In the case of 4+8m (m=0,1,2,3,...),
This coincidentally coincides with the code string obtained by the method described in Japanese Patent Laid-Open No. 54-51436, which was known before the filing of the present application, and therefore is excluded from the scope of the claims. ). This method will be explained in detail below. Now, when the number of divisions to be set is N (an even number), find an integer n that satisfies 2 ⁿ⁺¹ > N > 2 ⁿ (1). Also, obtain a pure binary code string from 0 to 2 ⁿ⁺¹ -1. Also, find X that satisfies the following function. 2 ⁿ⁺¹ −N=2X (2) Since 2 ⁿ⁺¹ and N are both even numbers, X is always an integer. Here, from 2 ^{n+ 1 pure binary code strings from 0 to 2 n+} 1 -1, a 2 ⁿ -1st to 2 ⁿ -Xth numbers b 2 nth to 2 ^{n +} X- Remove each of the X items up to the first. Then, the following N code strings are obtained. 000...0 (0th) 000...1 (1st) ~ ~ 011...0 (2 ⁿ -X - 1st) 100...1 (2 ⁿ +X) ~ ~ 111...1 (2 ^{n +1} - 1st) N pieces (3) By converting each of these codes into Gray codes, N unit distance code strings are obtained. In other words, when the number of divisions is N, the above (1)
Find n and X that satisfy equation (2), substitute them into n and X in equation (3), and convert them into Gray code. The fact that the code string obtained in this way becomes a unit distance code string is proven as follows. Let each digit of the pure binary code be Bn, Bn- ₁ , ..., B ₁ , B ₀ , and each digit of the corresponding Gray code be Gn, Gn- ₁ , ..., G ₁ , G ₀ . (Bn~Bo, Gn~Go are represented by "1" or "0"). It is known that the following logical formula holds true between Gn~Go and Bn~Bo. Gn=Bn Gk=Bk+1・+ ₊₁・Bk (however, k=0, 1, ..., n- ₁ )} (4) 2 n ⁺¹ pure 2s from 0 to 2 ⁿ⁺¹ -1 It is well known that when formula (4) is applied to each digit of the base code, the resulting Gray code string becomes a unit distance code string (since there are a power of 2 Gray codes, the last (and the first one are also in a unit distance code relationship ^{), the Gray code of 2 n -}
It is only necessary to prove that the Gray codes of the th (one after the section to be removed) are in a single distance code relationship (that is, they are equal except for one bit). So, first, if we express the 2 ^nth and 2 ⁿ -1th numbers in pure binary numbers, we get the following form. Here, from equations (1) and (2) above, X is 0<X<2 ^n-1 , so if X is expressed in pure binary, becomes. Therefore _, 2 ⁿ +X is the sum of equation (5) _and The digit marked as 1
It becomes the one that was made. Also, 2 ⁿ -X-1 is the result of subtracting X from equation (6), so of the 1 column in equation (6), X _o-1 to X ₀ in equation (7) are 1 The digit marked with is set to 0. From the above, it can be seen that in pure binary numbers, the 2 ⁿ +X-th and 2 ⁿ -X-1 digits are the 0 and 1 digits of each digit inverted. Here, in equation (4), if we invert the 1s and 0s of the pure binary numbers Bn to _B0 , the corresponding Gray code will have the most significant bit Gn inverted and the remaining bits
Gk (k=0, 1,..., n- ₁ ) does not change.
That is, it can be seen that the 2 ⁿ +Xth and 2 ⁿ -X-1st bits are all equal Gray codes except for the first bit. From the above, it has been proven that the conversion of equation (3) into a Gray code results in N unit distance code strings. To put it simply, the meaning of equation (3) is 0
In 2 ⁿ⁺ 1 Gray code strings from number 2 ^{n + 1 - 1} , the center of the code string (between 2 ⁿ - 1 and 2 ⁿ )
Using the fact that objects at symmetrical positions have a single distance code relationship, from the center position of the Gray code string, 2 ⁿ -1 and 2 ⁿ , 2 ⁿ -2 and 2 ⁿ +1
By removing numbers 2 ⁿ - 3 and 2 ⁿ + 2, etc. in pairs, and making the remaining number equal to the number of divisions N you are trying to set, you can This is to form a distance code string. FIG. 5 shows a portion of a code plate constructed using the Gray code string obtained according to the above theory. In this case, according to equations (1) and (2), n=9,
X=12. Therefore, 2 ⁿ -X-1 = 499 2 ⁿ + Therefore, since only the most significant bit ²⁹ changes between the 499th and 524th, a single distance code relationship is obtained. FIG. 6 is an example of configuring a 1000-division rotary encoder using the code plate shown in FIG. 5. In FIG. 6, the code plate 1 is a glass disk on which the pattern shown in FIG. 5 is drawn. 4 is a fixed slit plate that serves to improve detection accuracy. A light source (light emitting diode) 3 and a photoelectric element (photodiode) are placed between the code plate 1 and the fixed slit plate 4.
2 are arranged facing each other, and the light reception signal of the photoelectric element 2 is inputted to the storage element 10 via the amplifier 6.
The memory element 10 converts the Gray code output obtained from the pattern of the code plate 1 into a BCD code. The relationship between addresses and outputs at each position on the code plate 1 is shown in Table 2. The reason why the memory element 10 is used here is that if a general logic IC were used to construct a circuit that performs the conversion shown in Table 2, it would be extremely complicated and impractical.

【table】

Claims (1)

[Claims] 1 Let N be the desired even number of divisions, 2 ⁿ⁺¹ >N>
2 ⁿ (However, 2 ⁿ⁺¹ −N=4+8m (m=0, 1, 2,
3,...) are excluded. ) out of 2n ⁺¹ Gray code strings satisfying 2n ⁺¹ −N=2X, from ²ⁿ −X to ²ⁿ
A rotary position detection device comprising a rotor pattern in which code strings in sections sandwiched by +X-1 are removed to form a single distance code string of the desired number of divisions.