Application of Phase Detection in Weak Signal Measurement

In the industrial production process, there are many rotating objects whose internal temperatures need to be measured and controlled. For example, the drying cylinders in the textile industry, the rolling mill rolls, the rotors of the motors, and the drafting heating rollers in the chemical fiber industry. Since the measured object rotates at high speed, the temperature must be measured in a non-contact manner. The platinum thermal resistor and the passive RC network form a sensor that can rotate with the measured object. The required excitation signal is sensed into the sensor non-contactly by magnetic coupling, and becomes the input signal Vi of the passive RC network (see Figure 1). The sensor has a frequency selection function, and its output signal Vo is sensed non-contactly by magnetic coupling to become the feedback signal V′o. The subsequent display controller analyzes and processes the signal, converts it into the corresponding temperature, and compares it with the given value, thereby controlling the surface temperature of the measured object.

and 2.1V′i and V′o are the carriers of the signal ωn——natural frequency. The RC passive network using the Wien bridge is shown in Figure 2. The network has a frequency selection characteristic. Take R1=R2=RT ( platinum thermal resistor), and its natural frequency ωn is a function of temperature T℃. Obviously, temperature T and natural frequency ωn are in a one-to-one correspondence. Take Fn=ωn/2π, Table 1 describes the corresponding relationship between temperature T and frequency Fn (partial data). And T has nothing to do with the size of V′I and V′o. In the information transmission of the entire network, ωn is the information we need, and V′i and V′o are just carriers of information. 2.2 Quantitative relationship and constraints is a sinusoidal excitation signal, ω is its angular frequency, let =Vimsinωt, then the feedback signal . Suppose the temperature of the object under test at a certain moment is T, the corresponding resistance value of the platinum thermal resistor is RT, and the natural frequency ωn of the RC network is 1RTC. Take Vim=8V, when the excitation signal angular frequency ω＜ωn or ω＞ωn, the measured feedback signal Vom is about 600mV; when the excitation signal angular frequency , the theoretical value of Vom is 0, and the actual value is about 30mV. See Figure 3.

[page]

3.1 Obtain ωn and calculate the measured temperature T.
Because we do not know the measured temperature T in advance, but the ωn of the RC network exists objectively and changes with the change of T. The function of the controller shown in Figure 1 is to obtain the ωn of the network at this moment, so as to display the measured temperature T at this moment according to Table 1.
3.2 V′i is a sweep frequency signal V′i that changes from small to large
as the excitation signal. The feedback signal is obtained through the RC network frequency selection. After rectification and AD conversion, it is converted into a digital quantity V(n). If the ω of the excitation signal increases by one step, V(n+1) can be obtained, and if ω continues to increase, V(n+2) can be obtained. The single-chip computer system continuously makes judgments. If the sizes of V(n), V(n+1) and V(n+2) are similar, it means that the ω of the excitation signal is not ωn. The excitation signal ω of the display controller continues to increase, and the system continues to judge the new V(n), V(n+1) and V(n+2). If V(n+1) is much smaller than V(n) and V(n+2) at a certain moment, then the frequency ω of the excitation signal at the moment n+1 must be stored, because this frequency ω may be the natural frequency ωn of the RC passive network, which contains the temperature information of the object being measured.
3.3 Using phase detection to lock ωn
Because the measurement and control system is located in the factory site, it is always subject to various interferences. The existence of these interferences will often produce an instantaneous minimum value V(n+1) and meet the conditions of V(n) V(n+1) and V(n+2) V(n+1) (see Figure 3). This minimum value V(n+1) is false, and its corresponding ω is not the ωn we want, so it must be eliminated. How to distinguish whether the minimum value V(n+1) is true or false?
Further analysis of the RC passive network of the Wien bridge shows that its phase-frequency characteristics (theoretical value) are shown in Figure 4. At ω=ωn, the phase of the feedback signal V′o jumps. Actual measurements of this detection system show that its jump is much larger than the theoretical value. We can use the phase jump characteristics to determine the authenticity of the minimum value V(n+1).
Every time the excitation signal ω increases by one step, as long as the step size is appropriate

, the value of φn+2 minus φn+1 is judged. If →0, the phases of V(n+1) and V(n+2) have not jumped. Although the value of V(n+1) is very small, it is false and should be eliminated. If it is very large, the phases of V(n), V(n+1) and V(n+2) have jumped, and V(n+1) is the true minimum value. At this moment, the ω of V(n+1) is what we want to lock. See Figure 4.
3.4 Summary
Once the MCU system detects that V(n) V(n+1) and V(n+2) V(n+1); and their corresponding phases do not satisfy: φn≈φn+1≈φn+2, then the ω of V(n+1) at this moment is locked and stored, and the corresponding temperature is displayed by looking up the table.

Obtaining a specific phase value or a specific phase difference is not the same as determining whether the phase difference is zero (i.e., phase difference zero detection). Our system only needs to determine whether the phase has jumped, which can be achieved using a phase difference zero detection system. The phase difference zero detection system can use analog methods, digital methods, or a combination of them, which will not be described here.