JPH07245810A - Control system using one-bit digital signal processing - Google Patents

Abstract

PURPOSE:To improve signal processing accuracy and simplify a control mechanism by controlling a control target by driving a one-bit digital signal drive type actuator by A/D converted one-bit digital arithmetic output. CONSTITUTION:An analog output signal S11 from a sensor 32 is converted to a one-bit digital output signal S12 by a one-bit A/D converter 33. This one-bit digital output signal S12 is input to a one-bit digital arithmetic unit 34, arithmetic operation is performed by the one-bit digital signal S12 at this one-bit digital arithmetic unit 34, and then one-bit digital arithmetic signal S12 is sent out. Then, the one-bit digital signal drive type actuator 35 is driven by the one-bit digital arithmetic output signal S13 for controlling the control target 31. In this way, the processing can be performed by one-bit digital signal, signal processing accuracy can be improved, and the control mechanism can be simplified.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a control system using 1-bit digital signal processing for controlling a mechanical device such as a magnetic levitation device.

[0002]

2. Description of the Related Art Conventionally, analog circuits and multi-bit digital circuits have been used for signal processing in control systems. FIG. 17 shows such a conventional multi-bit (n
(Bit) digital control system block diagram.

As shown in FIG. 17, an analog signal S ₁ from the sensor 2 is A / D converted by an A / D converter 3 into a multi-bit digital signal S ₂ having 2 ⁿ discrete values, Signal processing is performed using the digital computer 4. As a result, the output multi-bit digital signal S ₃ is converted into an analog signal S ₄ by the D / A converter 5, the analog signal is transmitted to the actuator 6, and the control target 1 is controlled. .

[0004]

However, the quantization error (rounding error) that occurs during A / D / D / A conversion in the above-mentioned conventional multi-bit (n-bit) digital control system is normal to the control system. It becomes a factor to hinder proper operation. Here, as a measure for measuring the degree of quantization error,
The ratio between the signal power S when the full-scale sine wave is input and the quantization noise power N _Q is taken and displayed in decibels. It is called the S / N ratio. In this case, n bits of A /
The S / N of the D converter is obtained from the following equation (1).

S / N _dB = 6.02n + 1.76 (1) Therefore, if the number of bits is increased, the S / N is improved and the quantization error is reduced. However, in order to make a high-bit A / D converter, sophisticated circuit technology such as laser trimming is required, and strict element characteristics are also required. In addition, in the multi-bit A / D / D / A conversion, linearity (linearity) cannot be essentially compensated, and there is a drawback that zero cross distortion occurs when switching the most significant bit.

Further, when the object to be controlled has a fast pole or when quick response is required, an A / D / D / A converter operating at a high sampling frequency may be used, but it is very expensive. There is also a cost problem. In order to solve the above problems, the present invention uses 1-bit digital signal processing, uses 1-bit digital signal processing that can easily transmit signals, improve accuracy, and simplify the control mechanism. The purpose is to provide a control system.

[0007]

In order to achieve the above object, the present invention performs calculation based on an output signal from a sensor, drives an actuator based on the calculated data, and feedback-controls a controlled object. In a control system, a 1-bit A / D converter for converting an analog output signal from a sensor into a 1-bit digital output signal and a 1-bit digital output signal from the 1-bit A / D converter A 1-bit digital arithmetic unit for sending an arithmetic output signal and a 1-bit digital signal drive type actuator driven by receiving the 1-bit digital arithmetic output signal from the 1-bit digital arithmetic unit are provided. .

[0008]

According to the present invention, as described above, the analog output signal from the sensor is converted to 1 by the 1-bit A / D converter.
Convert to bit digital output signal. This 1-bit digital output signal is input to the 1-bit digital arithmetic unit, which is operated by the 1-bit digital signal to output the 1-bit digital arithmetic output signal. The 1-bit digital signal output type actuator drives the 1-bit digital signal drive type actuator to control the controlled object.

Therefore, it is possible to obtain a control system using 1-bit digital signal processing, which can perform 1-bit digital signal processing, can easily transmit signals, can improve accuracy, and can simplify the control mechanism. Specifically, (A) oversampling and noise shaping significantly reduce the quantization error due to 1-bit A / D conversion, resulting in accuracy equal to or higher than that of the multi-bit A / D converter. Obtainable.

(B) The circuit used in 1-bit digital signal processing has a very simple structure, and therefore can be easily manufactured without using an advanced circuit technique such as laser trimming. , Mass production is possible. Further, the accuracy can be easily improved. Furthermore, since there is no bit weighting, linearity is essentially compensated and zero cross distortion does not occur.

(C) It is not so difficult to raise the sampling frequency, so that stable control can be performed even when the controlled object has a fast pole or when the controlled object requiring quick response is handled. . (D) Since most of the system is processed by a 1-bit digital signal, the configuration of the entire system is greatly simplified. Further, by incorporating the 1-bit A / D converter in the sensor, the configuration can be further simplified.

(E) When performing remote control or the like, it is necessary to transmit a signal to a distant place, but since a 1-bit signal can be transmitted with a single cable, and signal deterioration is small, Signals can be transmitted accurately.

[0013]

Embodiments of the present invention will now be described in detail with reference to the drawings. Here, the 1-bit digital signal is
It is a discretized signal expressed only by binary values of 0 and 1. If the analog signal is converted as it is by 1-bit A / D, the quantization error naturally becomes very large. That is, when the S / N is calculated by the equation (1), the S / N is only about 7.8 [dB]. However, by increasing the sampling frequency from several times to several hundred times the basic sampling frequency (sampling frequency in multi-bit) and using noise shaping techniques such as ΣΔ modulation, the quantization noise in the signal band can be reduced. Drastically reduced.

For example, the basic circuit of a 1-bit A / D converter with secondary ΣΔ modulation noise shaping is as shown in FIG. In the figure, 11 is a first integrator to which an analog input signal is input.
The second integrator 12 is connected to 1 and is input to the second-order ΣΔ modulation noise shaping circuit 10. In the second-order ΣΔ modulation noise shaping circuit 10, the signal input from the second integrator 12 passes through the adder 13 and
It is sent to the 1-bit quantizer 14. 1-bit quantizer 1
The output signal from 4 is sent to the third integrator 18, and is fed back to the adder 13 via the 1-sample delay unit 15, the fourth integrator 16 and the fifth integrator 17.

On the other hand, the output signal of the third integrator 18 is output from the second-order ΣΔ modulation noise shaping circuit 10 via the sixth integrator 19, and is output from the first differentiator 20 and the second differentiator. A 1-bit digital output signal can be obtained via the converter 21. With this configuration, the quantization noise having a uniform distribution in the entire frequency band is normally about 15 [dB / oct] toward the high frequency region by the noise shaping circuit 10 of the second-order ΣΔ modulation. Shaped into a sloped distribution.

Further, multiplying the sampling frequency by M = 256,
Performs oversampling and sets the basic sampling frequency to 1
By cutting the band of / 2 or more with an LPF (low pass filter), from the following equation (2), about 109 [d
B] (corresponding to 17.8 bits) is obtained. S / N _dB = 10 log [[15 / 2π ⁴ ] (2 ¹ −1) · M ⁵ ] ... (2) That is, the quantization error due to the 1-bit A / D conversion due to oversampling and noise shaping is It is significantly reduced, and as a result, an accuracy equal to or higher than that of the multi-bit A / D converter can be obtained.

Radio Technology 1989 February pp. 88-9
Refer to 7 and make a 1-bit A / D converter,
An experiment of bit A / D conversion was performed. Frequency 100 [H
FIG. 3 shows a 1-bit digital signal and LPF output when a sine wave of z] is input. That is, FIG.
00Hz input sine wave, FIG. 3 (b) is a 1-bit digital output signal, FIG. 3 (c) is an LPF output signal, and FIG. 3 (d) is a 1-bit digital signal of 9.9 ms to 10.1 ms in FIG. 3 (b). Output signals are shown respectively.

That is, the black portion in FIG. 3 (b) shows a signal changing from approximately 5V in FIG. 3 (d) to approximately 0V, and the white portion in FIG. 3 (b) is approximately 5V. The signal held in the vicinity is shown. Since this 1-bit A / D converter has the third order of noise shaping, the positive / negative of the output signal is inverted. The sampling frequency is 3.072 [MHz] and the LPF cutoff frequency is 8.0 [kHz].

Circuits used in 1-bit digital signal processing, including the 1-bit A / D converter in FIG. 2, have a very simple structure. Therefore, advanced circuit technology such as laser trimming is used. It can be easily made without using it and can be mass-produced. Further, the accuracy can be easily improved. Furthermore, since there is no bit weighting, linearity is essentially compensated and zero cross distortion does not occur.

Next, application of the above-mentioned 1-bit digital signal processing to a control system will be described. As described above, the 1-bit digital signal processing is superior to the multi-bit signal processing currently used for control in terms of accuracy, and the non-linearity error and zero Problems such as cross distortion are also solved. Therefore, control can be performed using a signal with a small error. Also, raising the sampling frequency is not so difficult, so when the controlled object has a fast pole or when dealing with a controlled object that requires quick response,
Stable control can be performed.

Since the circuit for processing the 1-bit digital signal has a very simple structure, the portion (digital compensation element) of the conventional digital computer shown in FIG. The whole control system can be simplified by substituting the circuit for processing. Further, if the switch of the actuator is turned on / off by a 1-bit digital signal, the actuator can be driven by omitting the operation of converting it into an analog signal using the D / A converter.

FIG. 1 is a block diagram of a control system by 1-bit digital signal processing showing a first embodiment of the present invention. In this figure, the analog output signal S ₁₁ from the sensor 32 is input to the 1-bit A / D converter 33. The 1-bit digital output signal S ₁₂ from the 1-bit A / D converter 33 is a 1-bit digital calculator 34.
Sent to.

Then, the 1-bit digital arithmetic unit 3
4, the 1-bit digital operation output signal S ₁₃ from the 1-bit digital operation unit 34 is
It is sent to the 1-bit digital signal drive type actuator 35, and the controlled object 31 is controlled by driving the 1-bit digital signal drive type actuator 35. With such a configuration, most of the system is processed by a 1-bit digital signal, so that the configuration of the entire system is greatly simplified. In addition, 1 in the sensor
By incorporating the bit A / D converter, the configuration can be further simplified.

Further, when performing remote control, etc., it is necessary to transmit the signal to a remote place. However, the transmission of the 1-bit signal can be performed by one cable, and the deterioration of the signal is small. There is also an advantage that signals can be transmitted well. Next, a 1-bit digital arithmetic unit will be described. Of particular importance in this control system is a 1-bit digital calculator that constitutes a digital compensation element. In the case of multi-bit digital signal processing, the program of the computer is changed to perform arithmetic operations such as addition and multiplication, and processing such as differentiation and integration.

On the other hand, in the 1-bit digital signal processing of the present invention, the simplicity of its structure is the greatest feature, and therefore the arithmetic operations required for control do not require a computer or the like and all 1-bit digital signal processing is performed. It is important to do it alone. In the present invention, an algorithm is used in which addition / subtraction and signal amplification are performed only by 1-bit digital signal processing. Then, a computer simulation was performed based on this algorithm. Regarding the adder circuit, the circuit was actually manufactured and an experiment was conducted.

Each calculation will be described below. In addition,
In the following calculation, the 1-bit digital signal is 1 and-
It has a binary value of 1. (1) 1-bit adder / subtractor The following shows 2-input 1-bit digital addition using a counter. Regarding subtraction, one of the digital inputs 1 and -1 is inverted and the other input is added.

(A) Algorithm Counter value ≧ 0 ... Addition output is 1 (sum of two input values−1) is added to the counter. Counter value <0 ... Addition output adds −1 (sum of two input values +1) to the counter.

(2) Computer Simulation Here, two sine waves having an amplitude of 0.5 and a frequency of 1.6 [kHz] and different phases were added while changing the phase difference.
First, the input sine wave is sampled at a sampling frequency of about 3.28 [MH
z] for sampling, second-order ΣΔ modulation 1-bit A / D
The converter converts into a 1-bit digital signal. Addition is performed on the two 1-bit signals based on the above algorithm. The addition output 1-bit signal is used as a third-order IIR Butterworth type LPF with a cutoff frequency of 15.9 [kHz].
(Acts as a D / A converter in multi-bit)
Through analog addition output.

FIG. 4 shows the result of 1-bit addition simulation. That is, FIG. 4A shows a case where the phase difference between the two sine waves is 0 °, FIG. 4B shows a case where the phase difference between the two sine waves is 90 °, and FIG. Each case is shown in °. (3) Circuit fabrication and arithmetic operation experiment A 1-bit addition circuit as shown in FIG. 5 was designed and manufactured based on the addition algorithm.

In FIG. 5, 41 is a first 4-bit up / down binary counter (74ALS169), 42.
Is a delay element (74LS31), 43 is a dual 4-1 data selector (74ACT153), 44 is a clock delay circuit (74HCT00), 45 is a second 4-bit up / down binary counter (74ALS169), 46
Is a third 4-bit up / down binary counter (74
ALS169), 47 is a switch for clearing the counters 45 and 46, 48 is a delay flip-flop (7
4HCT74).

FIG. 6 shows the operation when an appropriate 1-bit input signal is applied to this 1-bit adder circuit. That is, FIG.
(A) is 4 times the sampling frequency (4fs) 12.288
[MHz] (signal S ₂₃ ), FIG. 6 (b) is the input signal A (signal S ₂₁ ), FIG. 6 (c) is the input signal B (signal S ₂₂ ), FIG.
(D) is the U / D input signal (S ₂₄ ) of the second and third 4-bit up / down binary counters (74ALS169) 45 and 46, and FIG. 6E is the input signal to the clock terminals of the counters 45 and 46. (S _25), the output signal (S ₂₆₎ shown in FIG. 6 (f) the counter value, shown in FIG. 6 (g) the third 4-bit up-down binary counter (74ALS169) 46, FIG. 6 (h) the counter ( 74HCT74) clock signal (fs) 3.072 [MHz] (signal S ₂₇ ), FIG. 6 (i) shows the counter (74HCT74) Q
Output (signal S ₂₈ ), Fig. 6 (j) shows counter (74HCT)
Q 74) (-) indicates output (signal S _29), respectively.

As described above, two 4-bit up / down counters are cascade-connected as the counter. Internal operation is 4 times sampling frequency (4fs) 1
2.288 [MHz] is used as the clock input,
(I) Add the input signal A to the counter. (Ii) Add the input signal B to the counter. (Iii) The inverted output of the most significant bit, which indicates whether the counter is positive or negative, is added to the counter. That 3
One operation is performed during one basic sampling.

Next, the 1-bit amplifier will be described.
Similar to the addition, the 1-bit signal is also amplified by using the counter. An operation for multiplying the input signal by p / q (p and q: both positive integers) was considered. (1) Algorithm The input signal is 1 ... P is added to the counter. Input signal is -1 ... Subtract p from the counter.

The counter value ≧ q ... 1 is output and q is subtracted from the counter. The counter value ≤ -q ...- 1 is output and q is output to the counter.
Add. | Counter value | <q ... Output is 0 (more precisely, 0 is 1
Converted to bit digital signal) (2) Computer simulation Here, assuming that a sine wave with a frequency of 1.6 [kHz] and an amplitude value of 0.8 is amplified, the amplification factor (gain) is changed and multiplication is performed.・ Division simulation was performed. 1-bit A / D
The same converter and LPF as in the addition simulation were used.

FIG. 7 shows the result of 1-bit amplification simulation. That is, FIG. 7A shows the case where the gain is 0.5, FIG. 7B shows the case where the gain is 0.1, and FIG. 7C shows the case where the gain is 1.2. Next, a control system of a magnetic levitation apparatus will be described as an example of an applied control system using the 1-bit digital signal processing of the present invention.

Here, a case will be described in which the 1-degree-of-freedom levitation control of a simple magnetic levitation device is performed by 1-bit digital signal processing. FIG. 8 is a configuration diagram of a control system of a magnetic levitation device showing a second embodiment of the present invention. In this figure, 51 is a magnetic levitation device as a control target, 52 is a sensor provided in the magnetic levitation device 51, 53 is a compensation element, and 54 is an amplifier as an actuator.

A simulation was carried out on the assumption that the compensating element 53 is processed by a 1-bit digital signal. That is, the displacement signal X and the velocity signal v output from the sensor 52 have K _p = 3.7 and K _d = 2.
Regarding the operation of adding the gains of 07 and sending the output of which the positive and negative are inverted to the amplifier (actuator) 54, there are a case of performing multi-bit digital signal processing and a case of performing 1-bit digital signal processing. Both were simulated using a computer.

As shown in FIG. 9, the input signals are both sine waves having a frequency of 500 Hz, and the input displacement signal displacement X
Is 0.2, the amplitude of the input speed signal v is 0.3, and it is further assumed that the input displacement signal displacement X and the input speed signal v are out of phase by 90 °. Regarding multi-bit digital signal processing, a 12-bit A / D converter and a 14-bit D / A converter are generally used, and the sampling frequency is set to 8 [kHz]. At this time, the S / N ratio of A / D conversion is 74 [dB] according to the equation (1). A 1-bit A / D converter that corresponds to this in accuracy is
When the second-order ΣΔ modulation noise shaping is performed, the oversampling rate M is about 50.
4, but in this simulation, M = 64
And

Therefore, the sampling frequency in the 1-bit digital signal processing is 512 [kHz]. Also, LP that functions as a 1-bit D / A converter
F uses a fourth-order one, and the cutoff frequency is 8 [kHz
z]. 1-bit digital signal of the input displacement signal X as shown in FIG. 10 (1-bit A / D of the input displacement signal X
The output waveform) after the conversion, indicating 1-bit digital signal amplified by the 1-bit amplifier K _p X (output waveform after 1-bit amplification of the input displacement signal X) in FIG. 11.

Similarly, the 1-bit digital signal of the input speed signal v (output waveform after 1-bit A / D conversion of the input speed signal v) as shown in FIG.
The output waveform [K _d v] after bit amplification is shown in FIG. Further, FIG. 14 shows an addition / inversion output waveform of the 1-bit digital signal of FIGS. 11 and 13.

Actually, the actuator is operated with this 1-bit output signal as it is, but the LPF output waveforms of K _p X and K _d v are shown in FIG.
5 and FIG. 16 shows the LPF output waveform of the final output shown in FIG. For comparison, FIG. 16 also shows the result of performing the same processing in the above-described multi-bit digital signal processing and the theoretical calculation waveform obtained by mathematically amplifying and adding.

In order to measure the distortion of the waveform in FIG. 16, one period of the waveform was divided into 1024, and the mean square of the error from the mathematically calculated result at each point was taken. This value E in multi-bit digital signal processing
_{The rms} was 0.933157 and E _rms = 0.019258 in the 1-bit digital signal processing. From this, it can be said that the 1-bit digital signal processing was able to perform the calculation with extremely high accuracy.

Further, there is a delay in both multi-bit and one bit, but in terms of time, one cycle of output wave is 2.0.
In contrast to [ms], multi-bit is 95.7 [μ
s], 1-bit has a delay of 35.2 [μs],
Topologically speaking, multi-bit is about 0.300 [r
ad] and 1 bit have a delay of about 0.111 [rad].

From the above, it can be seen that 1-bit digital signal processing is far superior in accuracy to multi-bit signal processing and has less delay. It should be noted that the present invention is not limited to the above-described embodiments, and various modifications can be made based on the spirit of the present invention, and they are not excluded from the scope of the present invention.

[0045]

As described in detail above, according to the present invention, the following effects can be achieved. (1) It is possible to obtain a control system using 1-bit digital signal processing that performs 1-bit digital signal processing, facilitates signal transmission, improves accuracy, and simplifies the control mechanism.

(2) Quantization error due to 1-bit A / D conversion is greatly reduced by oversampling and noise shaping, resulting in multi-bit A / D conversion.
It is possible to obtain an accuracy equal to or higher than that of the converter. (3) Since the circuit used in 1-bit digital signal processing has a very simple structure, it can be easily manufactured without using advanced circuit technology such as laser trimming, and mass production. Is possible.

Further, the accuracy can be easily increased. Furthermore, since there is no bit weighting, linearity is essentially compensated and zero cross distortion does not occur. (4) Since it is not so difficult to raise the sampling frequency, stable control can be performed even when the controlled object has a fast pole or when the controlled object that requires quick response is handled.

(5) Since the majority of the system is processed by the 1-bit digital signal, the configuration of the entire system is greatly simplified. In addition, 1 bit A in the sensor
By incorporating the / D converter, the configuration can be further simplified. (6) When performing remote operation, etc., it is necessary to transmit the signal to a distant place, but since the transmission of the 1-bit signal can be performed with a single cable and the signal deterioration is small,
Signals can be transmitted accurately.

[Brief description of drawings]

FIG. 1 is a block diagram of a control system by 1-bit digital signal processing showing a first embodiment of the present invention.

FIG. 2 is a basic circuit diagram of a 1-bit A / D converter with secondary ΣΔ modulation noise shaping according to the present invention.

FIG. 3 is a diagram showing an operation of the 1-bit A / D converter according to the present invention.

FIG. 4 is a diagram showing a 1-bit addition simulation result showing the first embodiment of the present invention.

FIG. 5 is a 1-bit addition circuit diagram showing a first embodiment of the present invention.

FIG. 6 is a time chart of the 1-bit adder circuit according to the first embodiment of the present invention.

FIG. 7 is a diagram showing a 1-bit amplification simulation result showing the first embodiment of the present invention.

FIG. 8 is a configuration diagram of a control system of a magnetic levitation device showing a second embodiment of the present invention.

FIG. 9 is a diagram showing an input displacement signal X and an input velocity signal v showing a second embodiment of the present invention.

FIG. 10 is an input displacement signal X showing a second embodiment of the present invention.
FIG. 4 is an output waveform diagram after 1-bit A / D conversion of FIG.

FIG. 11 is an input displacement signal X showing a second embodiment of the present invention.
FIG. 3 is an output waveform diagram after 1-bit amplification of FIG.

FIG. 12 shows an input speed signal v showing a second embodiment of the present invention.
FIG. 4 is an output waveform diagram after 1-bit A / D conversion of FIG.

FIG. 13 is an input speed signal v showing a second embodiment of the present invention.
FIG. 3 is an output waveform diagram after 1-bit amplification of FIG.

FIG. 14 shows K _P X and K _d v showing a second embodiment of the present invention.
FIG. 7 is a diagram showing an addition and inversion output waveform of FIG.

FIG. 15 shows K _P X and K _d v showing a second embodiment of the present invention.
FIG. 7 is a LPF output waveform diagram of FIG.

FIG. 16 is a final output LP showing the second embodiment of the present invention.
It is an F output waveform chart.

FIG. 17 is a block diagram of a conventional multi-bit (n-bit) digital control system.

[Explanation of symbols]

 10 Second-order ΣΔ modulation noise shaping circuit 11 First integrator 12 Second integrator 13 Adder 14 1-bit quantizer 15 1-sample delayer 16 Fourth integrator 17 Fifth integrator 18 Third Integrator 19 Sixth integrator 20 First differentiator 21 Second differentiator 31 Control target 32,52 Sensor 33 1-bit A / D converter 34 1-bit digital calculator 35 1-bit digital signal drive type actuator 41 first 4-bit up-down binary counter 42 delay element 43 dual 4-1 data selector 44 clock delay circuit 45 second 4-bit up-down binary counter 46 third 4-bit up-down binary counter 47 switch 48 final stage Delay type flip-flop 51 Magnetic levitation device (controlled object) 53 Compensation element 54 Amplifier (actuator)

Claims (4)

[Claims] 1. A control system for performing an arithmetic operation based on an output signal from a sensor, driving an actuator based on the arithmetic data, and feedback-controlling a controlled object, wherein (a) an analog output signal from the sensor is 1 bit. 1-bit A / D converter for converting to a digital output signal, and (b) 1-bit digital operation for inputting the 1-bit digital output signal from the 1-bit A / D converter and sending 1-bit digital operation output signal And (c)
A control system using 1-bit digital signal processing, comprising: a 1-bit digital signal drive type actuator driven by receiving a 1-bit digital operation output signal from the 1-bit digital operation unit. 2. The 1-bit A / D converter according to claim 1, wherein the 1-bit A / D converter performs sampling at a sampling frequency that is several times to several hundred times the basic sampling frequency, and has a ΣΔ modulation noise shaping circuit. A control system using digital signal processing. 3. The control system using 1-bit digital signal processing according to claim 1, wherein the 1-bit digital arithmetic unit has a 1-bit arithmetic circuit. 4. The control system using 1-bit digital signal processing according to claim 1, wherein the controlled object is a magnetic levitation device.