JPH01269015A - System for processing position detecting signal - Google Patents

Abstract

PURPOSE:To accurately detect positions even in the vicinity of peak values of the outputs of sensors by providing the pair of sensors, means for modulating pulse width, etc. CONSTITUTION:When sensor outputs eA and eB of a pair of sensors 9A and 9B are inputted to a direction discrimination circuit 34 after the outputs are shaped in waveform to square waves eCO and eDO at comparators 31 and 33, rough pulses are generated. The rough pulses are counted by a rough counter 12 so as to obtain a rough position. Regarding the accurate position, the outputs eA and eM are subjected to PWM modulation at corresponding modulators 35 and 36 and PWM signals (1) and (2) are generated. The PWM signals are inverted by means of inverters 37 and 38 and a total of four PWM signals are obtained. Then parts corresponding to divided each section are selected from the PWM signals. Moreover, values of both of the rough and an accurate counters 12 and 24 are caused to be latched by a register 25 at every period of the selected PWM signals and the stroke position is found from values of both counters 12 and 24.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a processing system for ffi signals such as rotary encoders and linear encoders.

(Prior art) For example, in order to detect the stroke position of a piston rod of a hydraulic cylinder, a magnetic scale is embedded in the piston rod at a predetermined interval, and the output signal from a magnetic sensor attached to the cylinder side is pulsed and counted. Sometimes.

To explain this with reference to Figure ll55 and Figure 6, 51
is a magnetic sensor that outputs a sine wave that corresponds to one period for the minimum opening distance (1st axis) of the magnetic scale due to the stroke of the piston rod.As shown in Figure 5, there are two magnetic sensors set so that the phase is 90" different from each other. be done.

The outputs eA and eB from these sensors are sent to a comparator 52.
The waveform is shaped and converted into a square wave eAo*eao, and this is input to the phase discrimination (also called direction discrimination)/multiplier circuit 5.
3, a forward direction pulse and a reverse direction pulse are generated based on the following logical formula, and these pulses are counted by an up/down counter 54 to detect the stroke position of the screw thread rod.

Positive direction pulse “eAo(↑)shaeIIOteneAo(↓)
・e day 0 +eeo(↑)・eA.

+eeo(↓)・eA.

Reverse direction pulse = exe (↑)・eBO +eAo(↓)・1080 +eeo(↑)・eA.

+eso(↓)・eA.

However, eAo (↑); Pulse generated at the rising edge of eAo esc (↑): Pulse generated at the rising edge of eao (↓); e A Pulse generated at the falling edge eso (↓); Falling edge of eeo The pulse generated at Ao; eAo's inverted signal τno'', eao's inverted signal .; It becomes possible to detect up to I!

(Problem to be solved by the invention) In this type of method, the limit is to divide one cycle of the sensor output into 4, so in order to improve the detection accuracy of the stroke position, the opening distance between 1% and 1% should be shortened. Although it had to be dense, there were limitations in machining to cut the perforations finely, and it was not possible to detect minute strokes.

Therefore, the point where the sensor output (eA, es) intersects with zero potential (zero cross point) is roughly located at position 11 (this is called the "coarse position").
In addition to the ``precise position'' (the fine position divided into a predetermined number of parts between each zero-crossing point), it has been proposed that this precise position is determined by interpolation calculation using a pulse width modulation method. (Machieki 62-99203
issue).

To explain this using the block diagram in Figure 10, 9A, 9B
are a pair of magnetic sensors that are offset from each other by 1/4 of a scale with respect to a magnetic scale (not shown), and sensor outputs e^ and eII shown in FIG. 7 are obtained from these sensors 9A and 9B. can get.

Note that in FIG. 7, the sine wave is represented by a triangular wave for convenience, similar to FIG. 5.

Regarding the detection of the rough position, the inverted signal eg of the sensor outputs eA and eB is waveform-shaped by the comparators IOA and 10B and converted into a square wave signal eAOyeBg, which is input to the direction discrimination circuit 11. A pulse (referred to as a "coarse pulse" since it corresponds to a coarse position) is obtained. Therefore, when this coarse pulse is counted by the coarse counter (a counter that counts coarse pulses) 12, the counted number corresponds to the coarse position. In addition, in the conventional example,
We counted all the rises and falls of eAo+eno, but here it is the rise of eAo.

Only falling edges are counted.

On the other hand, regarding precise positioning, one period of sensor output eB (section ■-■ in Figure 7) is divided into two parts (■-■
The precise position is obtained by performing interpolation calculation using Be for the first half period and eB for the second half period. Specifically, the sensor output eB and the inverter 13
The inverted output eg is input to the corresponding modulator (consisting of a comparator) 14.
M) Then, a pair of PWM signals (1) and (2) are obtained.

In this case, adopting a sine wave as a modulation signal means that ee
, f3e is part of a sine wave (not a straight line), so
By using the same sine wave, the duty ratio (ratio of high level time to one period of eM) of the PWM signal approximately matches the value on the interpolation line eB shown in Fig. 8 (e
This is because it is theoretically proven that, in an ideal state where the frequency of

This proof is as follows using Figures 8 and 9.

In addition, FIG. 8 shows the half period of the sensor output ea and the modulation signal e,
FIG. 9 is a partially enlarged view of FIG. 8, with the amplitudes and shakers being matched and superimposed.

Since the sensor output e8 is a sine wave, e a = (1+5in(or)x)/2
= (1), and x(xi
When calculating the value of
) ・=(2) is obtained.

On the other hand, the straight line eI interpolated to eB is e r = x + (1/2) ・
Since it can be expressed as (3), x=x from equations (2) and (3)
When i, e+ (e+!) is 811=(1/r)sin-' (2e ai-1)+
(1/2) ...(4).

Next, the duty ratio of the PWM signal is determined.

Let x=xi be the center of the period! ! The modulation signal evI is eM
= (1/2)l(1-5in((nr)(x
-xi))1...(5) It can be expressed as follows. However, it is assumed to be n"l.

Considering the ideal case of n→ω, x at the intersection of es and e
JI! When IExcl is calculated (Fig. 9, 11, eM=e
si and esi=(1/2) X((1-sin((nr)(xc 1-xi))1,
', xe 1 =xi-(1/f)sin-'
(2e B i -1)...(6) Also, from Fig. 9, XC2 = (xi-(1/nil + (1/r) sin-' (2e a i -1)-(
7) It is easy to see what will happen.

Therefore, the duty ratio σi of PWMq is as follows: σi=[(xi xc + )+(xc 2−(x
i-(1/n)))+t(xi+(1/n)-xi))
] / 1(xi + (1/ n) - (xi - (1/ n
))1=((2/n r )sin-'(2esi-
1)+(1/n)l(2/n)-'=(1/ff)sin-' (2e 8 i-1)
+(1/2) ...(8) Formula (4
) and (8), it is clear that the duty ratio σi of the PWM signal is equal to the value eIi on the interpolation line. In other words, due to PWM modulation, the characteristic of the PWM signal duty ratio with respect to the sensor output is linearized as shown in FIG. 12. This concludes the proof.

From the pair of PWM signals (1) and (2), a multiplexer 17. An AND circuit (circuit 2) is created, and this is the precision counter (counter that counts the precision pulses) 2.
It counts as 4. Incidentally, FIG. 11 shows the output waveforms of each section in FIG. 10.

However, near the sensor output taking the maximum value or the minimum value (both together referred to as the peak value), it does not form a straight line as shown in Figure 12. Therefore, near the peak value, the sensor output and the PWM signal The duty ratios do not correspond accurately and the accuracy of position detection decreases.

This is because when matching the amplitudes of the sensor output and the modulation signal and the vibrator as shown in Figure 8, a small relative deviation (matching error) that could not be matched causes the PWM to change in the vicinity of the peak value.
This is because it greatly affects the signal duty ratio, and if you try to increase the resolution near the peak value, you can accurately measure each amplitude of both signals.

You will have to match the swingers.

This invention has been made in view of the above-mentioned problems in the prior art, and it is an object of the present invention to provide a method that allows position detection to be performed accurately and stably, regardless of alignment errors, even near the peak value of the sensor output. purpose.

(Means for Solving the Problems) The present invention provides a pair of sensors that are provided relatively shifted in the displacement direction and output a sine wave signal corresponding to a position scale, and a pair of sensors that output sinusoidal signals corresponding to a position scale, A means for performing pulse width variation II (PWM) using a high frequency sine wave as a modulation signal (e M ), and a means for dividing one cycle of the sensor output so as not to include the vicinity of the peak value of the sensor output (for example, by 1/4 cycle). (dividing into four sections) means,
The apparatus included means for selecting a portion corresponding to each divided section from the pair of pulse width modulated signals (PWM signals), and means for converting the selected pulse width modulated signal into a position signal.

(Function) When sine wave signals from a pair of sensors are pulse width modulated with the same sine wave modulation signal, the duty ratio of the resulting pulse width modulation signal is approximately the value on the straight line interpolated to the sensor output. , where the sensor output is linearized.

In this case, since the sensor output near the peak value is not used, this method is not affected by the deviation (matching error) that occurs when matching the amplitude and loss of the sensor output and the modulation signal (Cew). As a result, position detection can be performed stably with high accuracy even near the peak value.

(Embodiment) FIG. 1 is a block diagram of an embodiment shown in correspondence with FIG. 10, and the same components as in FIG. 10 are given the same reference numerals. The arrangement of the pair of magnetic sensors 9A and 9B with respect to the magnetic scale is also the same as in the previous application, and the sensor outputs eA and eta are shown in the upper part of FIG. 2. However, also for the sensor output in FIG. 2, the sine wave is represented by a triangular wave for convenience.

In the previous application, one cycle of the sensor output was divided into two parts each half a cycle for interpolation calculation, but in this case, the influence of the matching error of amplitude and loss is large near the peak value of the sensor output, so the peak value Perform interpolation calculations while avoiding neighbors 1.
In other words, as shown in Fig. 2, one period T of the sensor output is divided into four sections of 1/4 period each, excluding the vicinity of the peak values of the sensor outputs eA and es (■~■, ■~■
, ■~■, Φ~■), select a part of the pair of sensor outputs e^, e a corresponding to each section (■~■ interval and ■~■
The section is eA, and the sections from ■ to ■ and from ■ to ■ are eB).

Therefore, in this example, eA and en% or e, and eB
The intersection point (■, ■, ..., ■, ... in the figure) is the rough position, so the waveforms of eA, ee%e^, and (3m) are shaped by each comparator 31 and 33 to form a square wave eeoeeDO. , these are input to the direction discrimination circuit 34 to produce coarse pulses, which are counted by the coarse counter 12 to obtain the coarse position.

On the other hand, regarding the precise position, 8A, ee is the corresponding modulation 1
PWM modulated by B35.36, PWM signal (1)
, (2) and invert them with inverters 37 and 38 to create a total of four PWM signals (1), (1), (
2), we obtain (2). In addition, each PWM signal (1), (2
) means an inverted signal of PWM signals (1) and (2).

Then, a portion corresponding to each divided section is selected from these four PWM signals. Here, as is clear from Fig. 2, the PWM signal selected corresponding to each section (this is referred to as the "selected PWM signal") is represented by one cycle between 0 and 0, Section → PWM signal (1), ■～■
section → PWM signal (2), section from ■ to ■ → PWM signal (1), and section from ■ to ■ → PWM signal (2). Switching is performed by each selection signal (eE O, ey o
*eG 11 we No ) as shown in the following table.

Note that the selection signal is generated by the Ronok circuit 39 which inputs 8eo*8DO. Fig. 2 also shows e Co *eo O.

Next, the method of converting the duty ratio of the selection PWM signal into a count value is the same as in the previous application, and the method for converting the duty ratio of the selection PWM signal into a count value is
The clock pulse (CK L pulse) passing through the AND circuit 21 is controlled using the M signal as a date signal, and the passed CKL pulse is counted by a precision counter 24 to be converted into a count value. Note that the resolution for counting the duty ratio is determined by the frequency divider 23 placed in front of the precision counter 24.
Set by.

In order to obtain the stroke position from the values of the coarse counter 12 and the fine counter 24, the values of both counters are latched in the register 25 every cycle of the selection PWM signal. As the latch signal (LATCH signal), the clock pulse cp that first appears after the rise of the square wave signal eMo obtained by waveform-shaping the modulation signal e2 by the comparator 19 is extracted and used (see FIG. 11).

Furthermore, since it is necessary to reset the branching unit 23 before counting, a reset signal (RESET signal) is obtained by passing the clock pulse Cρ following the LATCH signal and the coarse pulse through the OR circuit 22. In addition, RESE
The reason why the coarse pulse is included in the T signal is that the sensor output and the modulation signal are asynchronous, so if the coarse pulse comes in the middle of the modulation signal, the contents of the fine counter that is counting will be carried over after the coarse pulse. This is because errors occur in interpolation.

On the other hand, the precision counter 24 must be set to an appropriate initial value before counting, instead of being cleared to zero as in the previous application.
According to the present invention, as shown in FIG. 3, the minimum value of the PWM signal duty ratio is 0.25, and the maximum value is 0.75 (
Therefore, the duty ratio span is 0.5)
Therefore, the value of the fine counter 24 must be set to 0 when the duty ratio is 0.25.

Note that the remaining pulse train excluding the two clock pulses used as the LATCH signal and the RESET signal becomes the aforementioned CKL pulse. LATCH, RESET, CK
Each signal of L is a clock pulse cp and a square wave signal evo
The pulse is distributed by the pulse distributor 20 based on the following.

Now, according to the embodiment, as shown in FIG. range) is used. Therefore, it is not affected by errors in matching the amplitude and loss between the sensor output e^, eB and the modulation signal eM, so there is no problem even if the number of divisions between rough positions is increased, and the accuracy is improved by increasing the number of divisions. Moreover, the stroke position can be detected stably.

As an example, if one division of the scale is 2-rings, the 1/4 section is the coarse position, and therefore the coarse counter 12 counts at 0.5-simple.

On the other hand, how many parts the coarse position should be divided into is determined by the divider 23, and as shown in Fig. 4, if the fine pulse is divided into 5O-1 = 49 pieces per half cycle of the PWM signal, then The precision counter 24 counts in units of 0.01g+-. In other words, in this case, 1/1 of one scale is
This makes it possible to detect up to 200. In this case, it goes without saying that if -25J is loaded as the initial value of the fine counter 24, it will count from 0 to 49.

In the example, one cycle of the sensor output is divided into four, but it is not limited to this, and it is possible to divide it into eight as long as it does not include the vicinity of the peak value of the sensor output. It is.

(Effect of the invention) In this invention, the
a pair of sensors that output a sine wave signal corresponding to position 1; means for pulse width modulating the sensor outputs using a high frequency sine wave as a modulation signal; means for dividing so as not to include the vicinity of the value; means for selecting a portion corresponding to each divided section from the pair of pulse width modulated signals; and converting the selected pulse width modulated signal into a position signal. Since the above-mentioned means is provided, position detection can be performed accurately and stably even near the peak value.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of an embodiment of this invention, Fig. 2 is a waveform diagram for explaining the operation of this embodiment, and Fig. 3 is the span (range) of the PWM signal duty ratio of this embodiment. FIG. 4 is a waveform diagram for explaining the distribution of seminal pulses in this embodiment. FIG. 5 is a waveform diagram of the conventional example, and FIG. 6 is a block diagram of the conventional example. Fig. 7 is a waveform diagram of the sensor output of the earlier application, Fig. 8 is a composite waveform diagram of the sensor output and modulation signal, Fig. 9 is a partially enlarged view of Fig. 8, and Fig. 10 is a block diagram of the earlier application. FIG. 11 is a waveform diagram for explaining the effect of the prior application, and FIG. 12 is a diagram showing the relationship between the sensor output and the PWM signal duty ratio of the prior application. 9A, 9B...Sensor, 12...Rough counter, 16
... Sine wave oscillator, 18... Pulse oscillator, 20.
...Bulse distributor, 23...Frequency divider, 24...Precision counter, 25...Register, 31.33...Comparator, 34...Direction discrimination circuit, 35.36...e
ll device, 39... Ronok circuit, 40-43...A
ND circuit, 44...OR circuit. Patent applicant: Kayabae Gyo Co., Ltd. Figure 3 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 11

Claims (1)

[Claims] a pair of sensors that are provided relatively shifted in the displacement direction and output sinusoidal signals corresponding to the position scale; means for pulse width modulating the outputs of these sensors using high-frequency sinusoidal waves as modulation signals; means for dividing one period of the output so as not to include the vicinity of the peak value of the sensor output; means for selecting a portion corresponding to each divided section from the pair of pulse width modulated signals; 1. A position detection signal processing method, comprising: means for converting a pulse width modulation signal into a position signal.