JPS6148752B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to, for example, a signal processing circuit for various measurement signals, and particularly to an improvement in a signal processing circuit for correcting signal nonlinearity. Generally, there is no ideal proportional relationship between the measured quantity of various measurement signals such as temperature and pressure in an instrumentation system and the input signal of the signal processing circuit for converting it into an electrical signal. It has a linear error. For example, let the quantity to be measured be X, and the relationship between the input signal (generally the input signal of an operational amplifier) V _io is the first
As shown in Figure a, an ideal proportional relationship cannot be obtained;
It has a non-linear error as shown in figure b. Such non-linear errors in the relationship between the measured quantity X and the input signal V _io must be corrected in fields where highly accurate detection is required. Conventionally, two methods have been proposed for correcting this non-linear error: a method using a broken line approximation method and a method using a polynomial approximation method. FIG. 2 shows a correction signal Y for correcting the non-linear error of the input signal V _io of FIG. 1 using the broken line approximation method.
FIG. 3 shows a correction signal Y similarly generated by the polynomial approximation method. First, when obtaining the correction signal Y by the near-line approximation method as shown in FIG. 2 using an operational amplifier, a large number of break points are required to obtain a good non-linear error correction result. In this case, by sequentially determining the slope of each broken line starting from the 0% side on the left side of the figure, the slope coefficient of each broken line can be set by repeating a relatively simple procedure. However, a broken line function generation circuit (a circuit using an operational amplifier) that takes into account the slope of the broken line is required for each broken line, which complicates the circuit configuration for obtaining the correction signal Y and increases the number of parts. Particularly when the differential coefficient of the input signal V _io does not increase monotonically, extra adjustments such as inverting the output of the polygonal function generating circuit are required, making the configuration even more complicated. In FIG. 3, the dotted curve is an example of a correction signal generated by the polynomial approximation method, and the square term (≒X ² ) and cube term (≒X ³ ) of the input signal V _io are used, for example, by an analog multiplier. This correction signal component is generated by adding and subtracting this correction signal component using a coefficient setter. This method requires a small number of coefficient setters, the circuit itself is relatively simple, and if the settings are appropriate, it has a highly accurate correction ability. However, this method uses complicated calculations such as the method of least squares based on a large number of measured values.
After dividing the input signal V _io into components such as X, X ² , X ³ , etc.,
It is necessary to set the necessary coefficients to a constant value by dialing the opposite sign of the coefficient of each term. Therefore, unless the components of each term of V _io are known in advance, measurements and calculations are complicated and the coefficient setting work is difficult. An object of the present invention is to provide a signal processing circuit for correcting non-linear errors in an input signal, which has a small number of coefficient setters and can easily set coefficients in short steps without requiring complicated calculations. It is in. Before explaining embodiments of the present invention, the basic idea will be explained. In general, the input signal V _io containing a nonlinearity error can be expressed as a power series of the measured quantity X as shown in the following equation. V _io ( _X ) = a ₀ ″ + a ₁ ″ ^X + a ₂ ″X ² + a ₃ ″X ³ + a 4 ″ It is possible to define
In addition, the above-mentioned object is achieved by re-expressing the function as a sum of polynomials that can be easily realized in circuit terms. Assuming that the zero and sensitivity of the input signal V _io are adjusted in advance, V _io (0)=0V _io (1)=1 holds true. However, the measured quantity is 0≦X≦1. Accordingly
Equation (1) is V _io (X) = a ₁ ′X + a ₂ ′X ² + a ₃ ′X ³ + a ₄ ′X ⁴ + …(2)

It has the property of [Formula]. Here, V _io (X) = Y, Y (Y-1) ... (3) Y (Y-1/2) (Y-1) ... (4) Y (Y-1/4 (Y-1/ 2) (Y-1) ...(5) Similarly, if you calculate the polynomials below and calculate the sum of these polynomials, equation (2) becomes as follows: V _io (X) = a ₁ X + a ₂ Y (Y-1) +a ₃ Y (Y-2/1) (Y-1) +a ₄ Y (Y-1/4) (Y-1/2) (Y-1) +... ...(6) In general, the non-linear error included in the input signal V _io is as small as a few percent, and there is almost no sinusoidal meandering many times, so the higher the order, the smaller the value.The main terms in the input signal V _io are: Since the coefficient is “a ₁・X” which is almost “1”, for example,
The ^X4 term is small enough so it can be ignored. Therefore, in equation (6), when Y is expanded again by a polynomial ^of The term is a ₄ ′(1−
a ₁ ′). Therefore, it can be considered that almost no approximation error occurs even if the input signal V _io is approximated by up to the cubic term a ₃ of Y in equation (6). Also, since V _io (1)=1, in equation (6), Y
=1 and X=1, then a ₁ =1. Therefore, equation (6) becomes as follows. V _io (X) = X + Y (Y-1) {a ₂ + a ₃ (Y-1/2)} ...(7) As is clear from this equation (7), the Y term corresponds to the nonlinear error. Therefore, in order to obtain the correct signal V _O that corrects the nonlinear error of the input signal V _io , V _O (X) = V _io (X) + Y (Y-1) {k + m (Y-1/2)} …(8) is sufficient. This equation (8) is the correct signal V _O that corrects the nonlinear error of the input signal V _io realized by this invention.
It is. Of course, by substituting equation (7) into equation (8), we get V _O (X)=X+Y(Y-1) {(k+a ₂ )+(m+a ₃ )(Y-1/2)}...(9) Become. As shown in this equation (9), V _O (0)=0, V _O (1)
It is clear that =1 holds true. Next, when realizing equation (8), it is necessary to set k and m using a coefficient setter. This will be explained below. First, add the measured quantity X such that Y=1/2. 1/4 (k+a ₂ )=X ₁ −V _O (X ₁ ) …(10) Since X ₁ −V _O (X ₁ ) is the deviation from the ideal value, use the coefficient setter to make the deviation 0. Adjust k. That is, k=-a ₂ (11) holds true. Next, add the point where (m+a ₃ ) is large, for example, the measured quantity where Y=1/4. X ₂ −V _O (X ₂ )=3/64(m+a ₃ ) …(12) Here, since X ₂ −V _O (X ₂ ) is the deviation from the ideal value, set it to 0. Set m using the coefficient setter, m=-a ₃ ...(13). At this time, V _O (X)=X (14) holds true for any X satisfying 0≦X≦1. The present invention as explained above achieves the intended purpose by realizing equation (8). (8)
As shown in the formula, the correction signal for the input signal V _io is a function of the component {Y(Y-1)} that becomes zero near the lower limit value and upper limit value of the input signal, and the lower limit value, upper limit value, and Component that becomes zero near the intermediate value {Y
It is sufficient to generate two sets of signals having functions of (Y-1/2) and (Y-1)}. An embodiment of the present invention, which is a specific implementation example, is shown in FIG. 4 below. FIG. 4 is a diagram showing a circuit configuration realized by analog means, and FIG. 5 is a diagram showing the characteristics of each part thereof. In FIGS. 4 and 5, an adder including resistors R1 and R2, a constant voltage source V1, and an operational amplifier A1 obtains the term (V _io -1/2) as shown in the characteristic shown in FIG. 5a. Next, the analog multiplier M1 (V _io
-1/2) ² terms are created, and the output Y ₂ has the b characteristic shown in Figure 5. The next stage adder is operational amplifier A2
, resistors R3 and R4, and a constant voltage source V2, and output Y ₂ =V _io (V _io -1) of A2 is obtained. (Fig. 5C) Multiplier M2 in section B multiplies (V _io -1/2) and output V _io (V _io -1) of section A, and the output Y ₃ is as shown in Fig. 5d. V _io (V _io -1/2) (V _io -1) is obtained. In order to facilitate coefficient setting, an operational amplifier with a gain of -1 is interposed, and ±Y ₂ and ±Y ₃ are applied to both ends of coefficient setters VR1 and VR2, respectively. By doing so, the coefficients k and m in equation (8) can be set continuously from -1 to 0 to 1. An adder consisting of resistors R5, R6, R7 and an operational amplifier A5 obtains V ₀ =V _io +V _io (V _io -1)k+V _io (V _io -1/2) (V _io -1)m. Therefore, A is a function generating section where the input signal V _io becomes zero when the lower limit is 0 and the upper limit is 1, and B is a function generating section where the input signal V _io becomes zero when the lower limit is 0, the upper limit is 1, and the middle 1/2. be. The adjustment work is to turn VR1 so that the deviation of V _O from the ideal curve becomes 0 at V io = 1/2, and then turn VR2 so _that the deviation of V _O from the ideal curve becomes 0 at V _io = 1/4. Just set the . In actual work, X = 1/2 and X = 1/4 points.
Similar good results can be obtained even if VR 1 and VR 2 are set. According to this embodiment, it is possible to achieve linearity after correction that is as excellent as that of the polynomial approximation method with the extremely simple adjustment work of reducing the deviation to 0 as described above. That is, compared to the easy-to-adjust polygonal line approximation method, the adjustment can be made at a fraction of the cost, and the number of parts is equal to or less than that of the polynomial approximation method. The invention arises from the fact that the polynomial set by each coefficient setter has common factors with all of the lower order polynomials. Therefore, the higher fourth order,
It is also effective for quintic functions, etc., and the object of the present invention can be achieved regardless of whether the correction means is analog or digital. As described above, according to the present invention, it is possible to obtain a signal processing circuit that corrects an input signal having a nonlinear error with a simple circuit configuration and an easy procedure for setting coefficients.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing an example of an input signal to which the present invention is applied, FIGS. 2 and 3 are diagrams for explaining a conventional method for correcting non-linear errors of input signals, and FIG.
The figure shows an embodiment of the present invention, and FIG. 5 shows the characteristics of each part thereof. A1 to A5...Arithmetic amplifier, R1 to R7...Resistor, VR1, VR2...Coefficient setter, M1, M2...
...multiplier.

Claims (1)

[Claims] 1 In a device that corrects a nonlinear input signal with a correction signal to obtain a linear output signal,
a first function generating unit that outputs a quadratic function that becomes zero near the lower limit value and the upper limit value of the input signal, and outputs a cubic function that becomes zero near the lower limit value, the upper limit value, and the intermediate value of the input signal. a second function generating section, and combining outputs of both function generating sections to obtain the correction signal.