JPH0216852B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an improvement in a low frequency excitation type electromagnetic flowmeter.

Generally, an electromagnetic flowmeter is configured to apply a magnetic field perpendicular to the direction of fluid flow, simultaneously detect changes in electrical signals in the fluid flow path, and measure the fluid flow rate based on this. . Many modern electromagnetic flowmeters use low-frequency excitation methods, such as trapezoidal wave excitation and square wave excitation, which have superior zero point stability compared to AC excitation and DC excitation methods. There is. In a low-frequency excitation type electromagnetic flowmeter, the current supplied to the excitation coil is periodically switched between two steady-state values, and when the excitation current becomes constant, the induced voltage generated between the electrodes is sampled once each. After that, by taking the difference between adjacent sampling signals, the effects of electrochemical DC voltage and circuit-based offset voltage are removed, and a signal corresponding to the fluid flow rate is obtained. Even in such a low frequency excitation type electromagnetic flowmeter,
If sampling is not performed after a sufficient period of time has passed after the excitation current reaches a certain value, the zero point will drift. This is due to the induced voltage generated between the electrodes, a signal component proportional to the fluid flow rate, an electrochemical DC voltage, an offset voltage caused by the circuit, and an electromagnetic component generated in the loop between the electrode and electrode lead when the excitation current is switched. Noise components accompanying the switching of the excitation current are superimposed, including coupled noise and eddy current noise generated by a first-order lag circuit in which the eddy current flowing in the fluid is formed by the liquid resistance and the interfacial electric double layer capacitance of the electrode. This is because there is. The polarity of electromagnetic coupling noise and eddy current noise is reversed each time the excitation current is switched, so
This is because the difference between adjacent sampling signals cannot be eliminated, and while electromagnetic coupling noise becomes zero in a short time, eddy current noise does not become zero until a sufficient amount of time has elapsed. Therefore, in terms of zero point stability, the lower the excitation frequency is, the more advantageous it is, and some electromagnetic flowmeters in practical use have a frequency of 1/32 of the commercial power supply frequency. However, if the excitation frequency is too low, the response becomes slow and hunting occurs when a control loop is constructed.

Therefore, we supply an excitation current whose steady value repeats in the order of zero, positive, zero, and negative to the excitation coil of the electromagnetic flowmeter transmitter. If you find the difference between the signal voltage when it is positive and the signal voltage when it is positive, or the difference between the signal voltage when it is zero before negative and the signal voltage when it is negative, you can determine the influence of noise components associated with switching the excitation current. can be made smaller. However, even in this method, there is a difference in the magnitude of the noise component when the steady-state value of the excitation current is positive or negative and the magnitude of the noise component when the steady-state value is zero, so it is affected by this.

In the present invention, the eddy current generated when switching the excitation current is divided by a resistor and a capacitor formed between the electrode and the measurement fluid, and the differential noise generated between the electrodes fluctuates. Focusing on the difference between the magnitude of the noise component when the steady-state value is positive or negative and the magnitude of the noise component when the steady-state value is zero, we remove changes in the zero point due to these factors. The purpose is to

FIG. 1 is a connection diagram showing an embodiment of the electromagnetic flowmeter of the present invention. In the figure, 1 is an excitation circuit that includes a DC constant current source 11 and a constant flow rate I _s from the constant current source 11.
It has switches 12a and 12b for switching. 2 is an electromagnetic flowmeter transmitter, which includes an exciting coil 21,
Pipe 22 through which fluid flows and electrodes 22a, 23
It is equipped with b. 3 is a signal processing circuit, which includes an AC amplifier 31 that amplifies the voltage e _a induced between the electrodes 23a and 23b of the electromagnetic flowmeter transmitter 2, a switch 32 that samples the output e b of the amplifier 31, and a switch 32 that samples the output e _b of the amplifier 31; an A/D converter 33 that converts the amplifier output e _b into a digital signal;
A microprocessor 34 that performs desired digital calculations based on the digital signal from the A/D converter 33, a D/A converter 35 that converts the output of the microprocessor 34 into an analog signal, and a D/A converter 35 that converts the output of the microprocessor 34 into an analog signal.
Sample and hold the output of the converter 35 and calculate the output voltage
It has a sample hold circuit 36 that generates e _p . The microprocessor 34 performs digital calculations and also controls the switch 12 of the excitation circuit 1.
Pulses P _1a and P _1b that drive signals a and 12b, sampling switch 32 and sample hold circuit 3
generates pulses P ₂ and P ₃ that control 6.

The operation of the present invention configured in this way will be explained below with reference to the waveform diagram of FIG. 2. First, switch 1
2a and 12b are controlled by drive pulses _{P 1a} and _{P 1b} as shown in _FIG . , during the period _T4 when P _1b is on, the current _Is from the constant current source 11 is switched in the opposite direction and flows through the excitation coil 21, and during the period _T1 when both P _1a and P _1b are off. , _T3 , no current is applied to the excitation coil 21.
Therefore _, as shown in _FIG .
An excitation current I _w is supplied with T ₂ and a negative excitation period T ₄ . Note that the excitation current _Iw is controlled by the switch 12a,
12b, it actually rises due to the time constant due to the inductance and resistance of the excitation coil 21, and after a delay in the falling part, it reaches a steady value, but this is not shown in the figure. An induced voltage _e a corresponding to the exciting current I _w is generated between the electrodes 23 a and 23 b of the electromagnetic flowmeter transmitter 2, as shown in FIG. 2D. The induced voltage e _a includes, in addition to a signal component V _s proportional to the flow rate F of the fluid flowing through the pipe 22, a noise component V _o accompanying the switching of the excitation current, and an offset voltage component due to the electrochemical DC potential and circuit. V _f are superimposed. The noise component V _o is the electromagnetic coupling noise generated in the loop between the electrode and the electrode lead when switching the excitation current, and the primary eddy current flowing in the fluid formed by the liquid resistance R and the interfacial electric double layer capacitance C of the electrode. Contains eddy current noise caused by delay circuits.
As a result, the induced voltage e _a
The voltages e _a1 ′, e a1 , _{e a2} , when sampled twice in the idle period and once each in the excitation period are
e _a3 ′, e _a3 , and e _a4 are given by the following equations.

e _a1 ′=V _o1 ′+V _f e _a1 =V _o1 +V _f e _a2 =V _s1 +V _o2 +V _f e _a3 ′=−V _o1 ′+V _f e _a3 =−V _o1 +V _f e _a4 =−V _s2 − V _o2 +V _f (1) And the noise component associated with switching the excitation current
V _o is constant at each switching, and decreases almost exponentially when the excitation current is constant.

Sampling interval for each period (t ₂ − t ₁ ), (t ₅ − _{t 4} )
is Δt, the damping constant K is K=exp(-Δt/CR),
Using the timing of V _o1 ^- as a reference, the effect on V _o1 ^- when changing from negative excitation to zero excitation Δt before this timing is that the noise component V _o accompanying switching of the excitation current is Focusing on the point that is constant at a certain point in time, the following is obtained.

V _o1 ^- (1) = V _o K (2) The effect of the change from zero excitation to negative excitation 3Δt ago _{is V o1 -} ⁽ 2) = _{V o} ⁽ -K ³ ) (3 ) The effect of the change from positive excitation to zero excitation before 5Δt is V _o1 ^- (3) is V _o1 ^- (3)=V _o (−K ⁵ ) (4) The effect of the change from zero excitation to positive excitation before 7Δt The effect of the change from ^negative _{excitation to zero excitation before 9Δt is V o1} ^- _{( 5} ⁾ is V _{o1 -} ⁽ 5 ⁾ )=V _o K ⁹ (6) The effect of the change from zero excitation to negative excitation before 11Δt is V _o1 ^- (6) is V _o1 ^- (6)=V _o (−K ¹¹ ) (7) before 13Δt. The effect of the change from positive excitation to zero excitation is V _o1 ^- (7) is V _o1 ^- (7) = V _o (−K ¹³ ) (8) The effect of the change from zero excitation to positive excitation before 15Δt is V _o1 ^- (8) becomes V _o1 ^- (8)=V _o K ¹⁵ (9), and so on. And these equations (2) to (9),
The sum of ... becomes V _o1 ^- . Therefore, V _o1 ^- is V _o1 ^- = ΣV _o1 ^- (m), m = 1 ~ ∞ = V _o (K - K ³ - K ⁵ + K ⁷ + K ⁹ - K ¹¹ - K ¹³ + K ¹⁵ +...) = V _o (K−K ³ −K ⁵ +K ⁷ )x (1+K ⁸ +K ¹⁶ +…) =V _o K(1−K ² )(1−K ⁴ )x [1/(1−K ⁸ )] = V _o K(1-K ² )/(1+K ⁴ )...(10).

Also, like V _{o1 -} ^, when V o1 is calculated using the timing of V _o1 as a reference, it becomes as follows.

The effect of the change from negative excitation to zero excitation before 2Δt is V _o1 (1) is V _o1 (1)=V _o K ² (11) The effect of the change from zero excitation to negative excitation before 4Δt is V _o1 ( 2) is V _o1 (2) = V _o (−K ⁴ ) (12) The effect of the change from positive excitation to zero excitation 6Δt ago is V _o1 (3) is V _o1 (3) = V _o (−K ⁶ ) (13) The effect of the change from zero excitation to positive excitation before 8Δt is V _o1 (4) is V _o1 (4) = V _o K ⁸ (14) The effect of the change from negative excitation to zero excitation before 10Δt is The influence is V _o1 (5) is V _o1 (5) = V _o K ¹⁰ (15) The influence of the change from zero excitation to negative excitation 12Δt ago is V _o1 (6) is V _o1 (6) = V _o ( −K ¹² ) (16) The effect of the change from positive excitation to zero excitation before 14Δt _is V _o1 (7) = V _o (−K ¹⁴ ) (17) The effect of changing from zero excitation to zero excitation before 16Δt is The effect of changes in excitation is V _{o1 (} 8) = V _o K ¹⁶ (18), and so on. And these (11) to (18)
The sum of the equations... becomes V _o1 . Therefore, V _o1 is V _o1 =ΣV _o1 (m), m=1~∞ =V _o (K ² −K ⁴ −K ⁶ +K ⁸ +K ¹⁰ −K ¹² −K ¹⁴ +K ¹⁶ +…) =V _o ^{( K 2 −K 4} _−K ^{6 + K 8 )} V _o K ² (1−K ² )/(1+K ⁴ ) (19).

Furthermore, when V _o2 is calculated using the timing of V _o2 as a reference in the same way as V _o1 ^- , the result is as follows.

The effect of the change from zero excitation to positive excitation before 2Δt is V _o2 (1) is V _o2 (1)=V _o K ² (20) The effect of the change from negative excitation to zero excitation before 4Δt is V _o2 ( 2) is V _o2 (2) = V _o (+K ⁴ ) (21) The effect of the change from zero excitation to negative excitation 6Δt ago is V _o2 (3) is V _o2 (3) = V _o (−K ⁶ ) (22) The effect of the change from positive excitation to zero excitation before 8Δt is V _o2 (4) is V _o2 (4)=V _o (−K ⁸ ) (23) The effect of the change from zero excitation to positive excitation before 10Δt is The effect of the change is V _o2 (5) is V _o2 (5) = V _o K ¹⁰ (24) The effect of the change from negative excitation to zero excitation 12Δt ago is V _o2 (6) is V _o2 (6) = V _o (+K ¹² ) (25) The effect of the change from zero excitation to negative excitation before 14Δt is V _o2 (7) is V _o2 (7)=V _o (−K ¹⁴ ) (26) From positive excitation before 16Δt The effect of the change to zero excitation is V _o2 (8) = V _o (-K ¹⁶ ₎ (27), and so on. And these (20)(11)～
The sum of equation (27)... becomes V _o2 . Therefore, V _o2 is V _o2 = ΣV _o2 (m), m = 1 ~ ∞ = V _o (K ² + K ⁴ − K ⁶ − K ⁸ + K ¹⁰ + K ¹² − K ¹⁴ − K ¹⁶ +…) = V _o ^{( K 2 +} _K ^{4 −K 6 −K 8 )} _{_} K ² (1+K ² )/(1+K ⁴ )...(28).

Among these equations, we obtain V _o1 = KV _o1 ^- (29) from equations (10) and (19).

Also, from equations (19) and (29), V _o1 /V _o2 = [V _o K ² (1-K ² )/(1+K ⁴ )] / [V _o K ² (1+K ² )/(1+K ⁴ ) ]=(1- ^K2 )/(1+ ^K2 )...(30) is obtained.

Therefore, the sampling voltages e _a1 ′, e _a1 , e a3 _′ , e _a3 obtained in the signal processing circuit 3 during the idle periods T ₁ and _T 3
By calculating the following equation based on , the values e _o1 , e _o2 , k corresponding to the noise components V _o1 , V _o2 and K due to switching of the excitation current can be found, and the compensation value corresponding to V _o2 - _{V o1} can be obtained. Can calculate e _o .

Therefore, if the following equation is calculated between the compensation value e _o and the sampling voltages e _a2 and e _a4 obtained during the periods T ₂ and T ₄ during which the excitation current flows using the compensation value e _o and e a1 and e _a3 , , e _p1 = (e _a2 − e _a1 ) − e _o = V _s1 e _p2 = (e _a3 − e _a4 ) − e _o = V _s2 (32), and the offset voltage component V _f can be removed and the excitation current Noise components associated with switching
V _o1 and V _o2 can also be effectively removed, and signal components V _s1 and V _s2 proportional to the fluid flow rate can be obtained.

In the signal processing circuit 3 shown in the figure, the electromagnetic flowmeter transmitter 2
After the induced voltage _{e a} from The outputs of _the amplifier 31 corresponding to the sampling voltages e _a1 ′, e _a1 , e _a2 , e _a3 ′, e _a3 , e _a4 of e a are sequentially transferred to the A/D converter 3.
3, which is converted into a digital signal and sent to the microprocessor 34. The microprocessor 34 first performs a digital operation corresponding to equation (31) using digital signals corresponding to the sampling voltages e _a1 ′, e _a1 , e _a3 ′, and e _a3 obtained during the idle periods T ₁ and T ₃ . Calculate _the compensation value by performing
Every time the digital signals corresponding to the sampling voltages e _a2 and e _a4 obtained at T ₄ are input, a calculation corresponding to equation (32) is performed by digital calculation to calculate the offset voltage component V _f and the noise component V _o . remove,
Digital values corresponding to signal components V _s1 and V _s2 proportional only to the flow rate of the fluid are sequentially output. Furthermore, A/
The sampling voltages e _a1 , e _a1 ′, e _a2 , e _a3 ′, which are input from the D converter 33 to the microprocessor 34
The digital signals corresponding to e _a3 and e _a4 are stored in dedicated registers, and their values are held until the next cycle's signal is input. The calculated compensation value is also stored in a dedicated register, and the value is determined by inputting digital signals corresponding to the sampling voltages e _a1 ′, e _a1 , e _a3 ′, and e _a3 obtained during the pause periods T ₁ and T ₃ . Each time, the calculation corresponding to equation (4) is performed and updated. In this case, calculation accuracy can be improved by using a moving average value from the past as a compensation value. The output of the microprocessor 34 is converted into an analog signal by a D/A converter 35, and is sequentially applied to a sample and hold circuit ₃₆ by pulses P3 generated at the timing shown in FIG. As a result, an output voltage e _p corresponding to a signal component V _s proportional only to the flow rate of the fluid is obtained at the output of the sample and hold circuit 36 .

In this way, in the present invention, the difference between the noise component when the steady value of the excitation current is positive or negative and the noise component when it is zero is compensated for, and the noise component that causes the drift of the zero point is effectively removed. Therefore, the stability of the zero point can be improved without lowering the excitation frequency, that is, without sacrificing responsiveness.

In the above description, the case where the offset voltage component V _f is constant has been exemplified, but V _f changes due to fluctuations in the electrochemical DC potential. _{Sampling voltages} e _a1 ′, e _al , e _a2 , e a3 ′, e _{a3 ,} e _a4 , e _a5 ′,
e _a5 , e _a6 , e _a7 ′, e _a7 , and e _a8 are given by the following equations.

e _a1 ′=V _o1 ′+V _f0 e _a1 =V _o1 +V _f0 +V _f1 +V _f2 +V _f3 +… e _a2 =V _s1 +V _o2 +V _f0 +3V _f1 +9V _f2 +27V _f3 +… e _a3 ′=−V _o1 ′+V _f0 +4V _f1 +16V _f2 +64V _f3 +... e _a3 = -V _o1 +V _f0 +5V _f1 +25V _f2 +125V _f3 +... e _a4 = -V _s2 -V _o2 +V _f0 +7V _f1 +49V _f2 +343V _f3 +
… e _a5 ′=V _o1 ′+V _f0 +8V _f1 +64V _f2 +512V _f3 +… e _a5 =V _o1 +V _f0 +9V _f1 +81V _f2 +729V _f3 +… e _a6 =V _s3 +V _o2 +V _f0 +11V _f1 +121V _f2 +1331V _f3 +
… e _a7 ′=−V _o1 ′+V _f0 12V _f1 144V _f2 +1728V _f3 +… e _a7 =−V _o1 +V _f0 +13V _f1 +169V _f2 +2197V _f3 +… e _a8 =−V _s4 −V _o2 +V _f0 +15V _f1 +225V _f2 +3375V _f3
+… (33) Therefore, if the signal processing circuit 3 first obtains the compensation value e _o and essentially performs the calculation of the following equation, the noise component
V _o1 , V _o2 can be removed, the offset voltage component can be removed by linear approximation, and the flow rate components V _s1 ,
An output e _p related to V _s2 can be obtained.

e _p = 1/2 (−e _a4 +e _a3 +e _a2 −e _a1 )−e _o =V _s1
+V _s2 /2 (34) However, e _o =2K ² /1−K ² (e _a3 −2e _a3 ′+3e _a1 −2e _a1 ′/
2)=V _o2 −V _o1 k=e _a3 −2e _a3 ′+3e _a1 −2e _a1 ′／2e _a3 −3e _a3 ′
+2e _a1 −e _a1 ′=k Furthermore, if the signal processing circuit 3 essentially calculates the following equation, the noise components V _o1 and V _o2 can be removed, and the offset voltage component V _f can be removed by quadratic approximation. .

e _p = 1/4 (e _a6 −e _a5 −2e _a4 +2e _a3 +e _a2 −e _a1 )
−e _o = 1/4 (V _s1 +2V _s2 +V _s3 ) (35) However, e _o =2K ² /1−K ² (−e _a5 +2e _a5 ′−2e _a3 +3e _al −
2e _a1 ′/4)=V _o2 −V _o1 k=−e _a5 +2e _a5 ′−2e _a3 +3e _a1 −2e _a1 ′/−2e _a
5 +3e _a5 ′−2e _a3 ′+2e _a1 −e _a1 ′=k Furthermore, if the signal processing circuit 3 essentially calculates the following equation, the noise components V _o1 and V _o2 can be removed, and the offset voltage component can be reduced by 3 It can be removed by approximating the following equation.

e _p = 1/8 (−e _a8 +e _a7 +3e _a6 −3e _a5 −3e _a4 +3e _a3 +
e _a2 −e _a1 )−e _o = 1/8 (V _s1 +3V _s2 +3V _s3 +V _s4 ) (36) where e _o =2K ² /1−K ² (e _a7 −2e _a7 ′+2e _a5 +2e _a5 ′− 5e _a3 ＋2e _a3 ′＋3e _a
1 −2e _a1 ′/8)=V _o2 −V _o1 k=e _a7 −2e _a7 ′+2e _a5 ＋2e _a5 ′−5e _a3 +2e _a3 ′
＋3e _a1 −2e _a1 ′／2e _a7 −3e _a7 ′−2e _a5 ＋5e _a5 ′−2e _a3 −
e _a3 '+2e _a1 -e _a1 '=k Furthermore, even if the following equation is substantially calculated, the noise components V _o1 and V _o2 can be removed, and the offset voltage component can be removed by cubic approximation.

e _p = 1/4 (−e _a7 +3e _a6 −2e _a5 −2e _a4 +3e _a3 −e
_a2 ) −e _o = 1/4 (3V _s3 +2V _s2 −V _s1 ) (37) where e _o =2K ² /1−K ² (e _a7 −2e _a7 ′+2e _a5 −2e _a5 ′−5e _a3 +2e _a3 ′＋3e _a
1 −2e _a1 ′/8)=V _o2 −V _o1 k=e _a7 −2e _a7 ′+2e _a5 +2e _a5 ′−5e _a3 +2e _a3 ′+3e _a
1 −2e _a1 ′／2e _a7 −3e _a7 ′＋2e _a5 −5e _a5 ′−2e _a3 −e _a3 ′
+2e _a1 -e _a1 '=k Furthermore, in the above description, the case where the output e _b of the amplifier 31 is given to the A/D converter 33 via the sampling switch 32 has been exemplified, but as shown in FIG. 3, the amplifier output e _b may be applied to the A/D converter 33 via the integrator 37. In this case, if the integration time t _s is selected to be an integral multiple of the commercial power supply cycle, the influence of power supply frequency noise can be removed. In FIG. 3, the integrator 37 includes a resistor RI, an operational amplifier OP,
Integrating capacitor CI connected to OP's feedback circuit
and a timing switch that controls the input integration time.
TS and a reset switch RS that resets the previous integral value immediately before the start of integration, and TS and RS are connected to pulses from the microprocessor 34.
Those driven by P ₄ and P ₅ are illustrated.
An example of the sampling time at this time is shown in FIG. In addition, in Figure 4, zero, positive,
Since each zero and negative period is selected to be 3Δt, V _o1
The relationship between V o2 and V _o2 is expressed by the following equation.

V _o1 /V _o2 = [V _o K ² (1-K ³ ) / (1 + K ⁶ )] / [V _o K ² (1 + K ³ ) / (1 + K ⁶ )] = (1-K ² ) / (1 + K ² ) (38) Furthermore, in the above, sampling is performed once when the excitation current is at positive and negative steady values, but if sampling is performed two or more times, the response and S/N ratio against electrical noise will be further improved. Furthermore, although the compensation value is calculated by sampling twice during each of the rest periods T ₁ and T ₃ when the steady-state value of the excitation current is zero, the compensation value may be calculated by sampling three or more times.

As explained above, in the present invention, the excitation current is switched before the differential noise accompanying switching of the excitation current disappears, and the voltage induced between the electrodes is increased twice or more during the period when the steady value of the excitation current is zero. Using the data obtained through sampling, a compensation value due to differential noise that occurs with switching of the excitation current is calculated each time the excitation current is switched, and this compensation value is used to perform algebraic addition to the intermediate flow rate and differentiation. Since the noise is compensated for, the zero point does not change even if the differential noise fluctuates, while maintaining a fast response speed. However, the zero point does not fluctuate.

[Brief explanation of drawings]

Fig. 1 is a connection diagram showing an embodiment of the electromagnetic flowmeter of the present invention, Fig. 2 is a waveform diagram for explaining its operation,
FIG. 3 is a connection diagram showing another embodiment of the electromagnetic flowmeter of the present invention, and FIG. 4 is a waveform diagram for explaining its operation. 1... Excitation circuit, 2... Electromagnetic flowmeter transmitter, 2
1... Excitation coil, 23a, 23b... Electrode, 3
... Signal processing circuit, 31 ... AC amplifier, 32 ...
...Sampling switch, 33...A/D converter, 34...Microprocessor, 35...D/
A converter, 36...Sample hold circuit, 37
...integrator.

Claims (1)

[Claims] 1. In an electromagnetic flowmeter in which an excitation current is supplied to the excitation coil of an electromagnetic flowmeter transmitter in the order of a steady value of zero, negative, zero, positive, the electromagnetic flowmeter transmitter sampling means for sampling the voltage induced between the electrodes two or more times and at least once during a period of a steady value other than zero; Compensation value calculation means for calculating a compensation value by calculating the difference between each differential noise caused by switching between when is positive or negative and when is zero; 1. An electromagnetic flowmeter comprising intermediate flow rate calculation means for calculating an intermediate flow rate using the intermediate flow rate, and flow rate calculation means for calculating a flow rate output by algebraically adding the compensation value to the intermediate flow rate.