Study on a Method to Improve the Stability of the Sensor Signal Based on PD Controller

In order to solve the high frequency oscillation generated from the video signal of the camera, the study proposes a method to improve the stability of high frequency oscillation of common sensor signal, which means the high frequency oscillation signal generated from the sensor module can increase the phase margin of the unstable signal through PD controller, thus produces a stable video signal and then through the circuit of AGC (Automatic Gain Control), the automatic compensation of the frequency can be achieved and finally the fourth-order low-pass circuit is used to meet the requirement of cutoff frequency of the sensor signal processing and the stability of the signal and the high frequency performance can be ultimately improved.


INTRODUCTION
The definition of the video image is determined by the resolution of camera and the high frequency characteristic of the signal processing circuit of the camera.With further research and development of materials technology, the current resolution of the camera has been greatly improved, but the definition of the image has not been greatly improved and the reason of that is because of the limit of the high frequency characteristic of the signal processing circuit (Bi et al., 2009).The greater high-frequency component in signal processing is, the more details of the image there are and the resolution is higher too.The difficulty in current video processing lies in the high frequency oscillation in the process of camera's circuit module design, resulting in a significant decline of the high-frequency performance in video signal processing.The situation is also occurred in other sensor signal processing, such as the signal processing of 3D laser radar that is used for robots information acquisition (Zhang, 2000).As the oscillation of high-frequency signal makes the decline in the performance of the high-frequency signal, the signal collected is not rich enough, resulting in the inaccuracy of robot path planning, so the processing of path planning can be only solved by using algorithm in the MCU (Zhang et al., 2007).
The approach of using the software algorithm to process the signal oscillation problem has been proposed to solve the problem in the domestic and foreign literature with longer space, which is still a good method, but it adds additional the problems of digital signal processing.Professor Zhang dian-fu has put forward a method to use FPGA to improve the circuit characteristics by achieving large-scale integrated circuit (Zhang and Zhao, 2009), but this increases the cost, which makes the circuit structure more complex and brings other parasitic problem as well.
We want to improve high frequency characteristics of camera by adopting the design of advance controller to prevent high frequency oscillation of camera signal processing circuit and improve the stability of the whole circuit and apply the method to other similar sensor signal processing.

OVERALL SYSTEM DESIGN
To improve the stability and solve the problem of the oscillation of the sensor's output signal waveform, the current primary task is to improve the phase margin (generally an increase of 60° is preferred).There are several ways to improve the phase margin (Nehorai and Acoustic, 1994).The design uses a phase lead theory in control theory, which designs a priori-correction device (PD controller).The controller can make the system have an additional open-loop zero, increase the phase margin and improve system's stability and smoothness and eliminate oscillation generated from analog sensor circuit (Zhang, 2000).From the Bode diagram of the controller as shown in Fig. 1, it is concluded that it has a feature that the phase margin of output signal is increased.
As it can be seen from the overall block diagram of the design in Fig. 2, although an unstable signal is output from the analog circuit of the sensor, when the signal output goes through the PD controller, the phase margin can make the signal more stable as there is a phase lead feature in PD controller.However, the PD controller will result in signal attenuation in some certain frequency (Isik, 2010), so according to the requirements of the video signal processing, the phase must not be less than -180° when the high frequency band is at 100k.In order to meet the requirement of frequency, compensation AGC circuit is used (automatic gain of the controllable voltage) and compensation is made for the attenuation part.That is to say, take the output signal of PD controller as the input of AGC (automatic controllable voltage gain), the comparison voltage of which is provided by the output of the peak value detection circuit.The cutoff frequency will be greatly improved since the signal is processed by AGC.In the design scheme in Fig. 3, in order to meet the requirements of the video signal processing, a low-pass filter circuit is added finally so that the cutoff frequency can meet the requirements and the waveform is stabilized and enlarged, so the high frequency characteristic of the sensor circuit is improved.

KEY CIRCUIT DESIGN
The same kind sensor simulation circuit: In order to find similar sensor vibration rule, we first build a sensor analog circuit to simulate such sensor circuit that generates high-frequency oscillation.The circuit has features such as narrow pass band, waveform that is easy to have oscillation, on basis of which general method of signal processing is expected to be found.The circuit is also a controlled object of the method of the research.(The circuit taken as an example in the study is mainly used for processing video signal).
The transfer function of the circuit system can be calculated from the parameters of each device as well as sensor analog circuit diagram in Fig. 4  The Bode diagram as shown in Fig. 3 can be drawn from MATLAB simulation: Through the analysis of amplitude-frequency characteristic curve and phase frequency characteristic curve, it's found that the circuit begins to decay in the range about 4.5k and the cutoff frequency is around 4.5KHZ or so.The phase margin is not great under the frequency and the phase is less than -180° in the high-frequency end.
Through the above analysis, it is found that the system is an unstable high-frequency system and the circuit is in an extremely unstable state.The signal processing oscillation existing in a lot of sensor circuit can be simulated through the circuit.

The design of circuit module of PD controller:
According to the design of circuit module of PD controller as shown in Fig. 5, the corresponding resistance can be calculated and the capacitor value is: R1 = R2 = 27kΩ, C = 276pF.When a circuit is built on breadboard for simulation debugging, it is found that the output waveform of analog sensor is significantly improved.In order to improve its effectiveness, the capacitance value is increased to 303pF and when it is tested again, the waveform displayed on an oscilloscope gets better improvement.
The waveform comparison chart after the improvement of the high-frequency part is as shown in Fig. 6: Known from the Fig. 6 and 7: the mid and high frequency phase are increased, the phase margin are increased and the system tends to be stable.The circuit design of PD controller is successful.

CONCLUSION
Through the added PD controller and AGC circuit in the test circuit module, the phase of high frequency in circuit has been raised and phase margin has been increased, thereby the system has been stable and high frequency signal performance of system has been improved, making the signals collected by the camera real and rich.The method that PD controller is creatively added in the signal processing circuit presents a solution to the treatment of high frequency oscillation of general sensor signal.

Table 2 :
The rule of variation of the voltage with the frequency after accessing to frequency compensation circuit

please see Annex VII for more details about data)
 Test results for output voltage range are shown in Table 1:  The rule of variation of the voltage with the frequency after accessing to frequency compensation circuit is as shown in Table 2:  Noise Test: Vn = 8mv