Control law design is a crucial step in servo system design. After the servo components are selected, the system's performance is ultimately determined by the control law. This paper, using the development of a radar servo system as an example, introduces the design and implementation process of the servo control law, providing a new approach for electromechanical servo system design. The proposed variable structure control (VSC) solves the problem of nonlinearity identification caused by various frictional effects and unbalanced torques. VSC theory was developed in Russia in the late 1950s and attracted the attention of Chinese scholars in the 1980s. Due to its invariance to changes in internal system parameters and external disturbances, it is of great significance for nonlinear control systems. Working principle of electromechanical servo system [align=center] Figure 1 Working principle block diagram of servo system[/align] Figure 1 is a typical working block diagram of airborne radar servo system: The digital processing unit generates digital control signals according to the working state and servo performance, which are converted into electrical power signals that can drive the motor through the d/a converter and power amplifier module. The motor drives the load to run, and the rotary transformer synchronized with the motor senses the rotation of the motor into an analog electrical signal, which is then calculated into a digital signal by the rd chip of the servo digital unit. It forms an error control signal with the ideal digital command signal, thereby forming the closed-loop control of the system. Controller design The mathematical model of the electromechanical servo system in Figure 1 includes two parts: linear and nonlinear. Equation (1) is the typical transfer function of the linear part: (1) Where t and k are parameters related to the rotational inertia, torque sensitivity, resistance and power amplifier gain of the system, and ωn and ζ are the natural resonant frequency and damping coefficient of the oscillation element, respectively, which are related to the mechanical structure stiffness of the system. Theoretical VSC Design The nonlinearity of the electromechanical servo system is mainly caused by the dry friction, backlash, and unbalanced torque of the mechanical structure, which manifests as control zero offset and dead zone. At the same time, due to the influence of components, the system contains saturation characteristics. In response to these characteristics, this paper proposes a sliding mode variable structure control method, which seeks variable structure control by establishing the sliding switching surface function s(x). For the system mode, e represents the error signal and y represents the system output. Selecting the state variable, x1=e, the state equation of the sliding mode variable structure control system can be expressed as: Simplified as: Take the sliding hyperplane as s=mx=0, where m=[m1,m2,…,mn], and the value of the row vector m can be determined by the following formula [4]: ​​i＝1,2,3,4,n is the system order. In this system, n＝4, and λ is the desired bandwidth of the system. According to the pid control law, λ＝8×2π is taken. The sliding surface equation is: s＝mx＝[λ3,3λ2,3λ,1]x. A quasi-sliding mode control with an exponential reaching law is adopted, and the designed control law is as follows: u=-（mb）-1（k1max+k2s+dsgn（s））. When the system sampling time is 1ms, k1＝0.0015, k2＝0.5, σ＝100, and the system nonlinearity is set to control zero bias, dead zone, and saturation characteristics. The step response characteristics of the system with and without nonlinearity are measured as shown in Figure 2: [align=center] a) No nonlinearity b) With nonlinearity Figure 2 Step response diagram of theoretical VSC and PID control[/align] Actual VSC design Since sliding mode variable structure control involves the system's state vector x, and the state variables x2, x3, and x4 correspond to velocity, acceleration, and jerk, respectively, the information of these three variables is usually not directly available in general electromechanical servo systems. A state observer is considered for state reconstruction, but this system is a fourth-order system, making pole placement difficult when designing the state observer. Analysis of the system transfer function shows that the oscillating element mainly affects the system's bandwidth. Therefore, under the condition of ensuring bandwidth, this element can be considered to have a relatively small impact on system performance, thus simplifying the system model. With pole configuration p = [p0 p1] and feedback coefficient f = [f0 f1]t, the observer's state equation and measurement equation are chosen as p0 = -90 and p1 = -100. The sliding surface equation is s = c0x = [λ,1]x, x = [x1 x2], x1 = e. The control law remains unchanged, and the adjustment parameters are k1 = 0.005, k2 = 0.05, and σ = 0.001. The step response characteristics of PID control and improved sliding mode variable structure control with and without nonlinearity are shown in Figure 3. [align=center] a) No nonlinearity b) With nonlinearity Figure 3 Step response diagrams of actual VSC and PID control[/align] As can be seen from the response diagrams in Figure 3, the VSC design can fully meet the system requirements. Furthermore, since the observer is easy to implement, it facilitates the engineering application of sliding mode variable structure control. The system dynamic response characteristics corresponding to each controller are shown in the attached table. The appendix compares the performance of various controllers. As can be seen from the appendix, when using VSC control, compared with classic PID control, the system's dynamic performance is more stable, and it can effectively solve the system's nonlinearity problem. Implementation of Servo Control The ultimate function of the servo system is to drive the load at a certain speed within a certain angular range in space. The system has no speed measuring device; speed control relies on angle measurements and the sampling period. Assuming the sampling period is ts, the rotational speed is ω, and the initial system position is θ0, then the position command at time k is θi(k) = θi(k-1) + ωts, and the error e(k) = θi(k) - θr(k). Where θi(0) = θ0, and θr(k) is the actual system position at time k. The system implementation involves discretizing the VSC and observer state equations. A simple discretization method is given here. Assuming the standard forms of the system state equations and measurement equations are as shown, they can be directly discretized using functions in MATLAB software. The basic discretization statement is as follows: Let a = ao - fco, b = [bo f], c = co, d = 0, and the standard model above can be obtained. The software programming language is C. From the software framework perspective, the system is divided into two main lines: one mainly uses bus interrupts for communication with the master control device; the other mainly uses system clock interrupts to complete the discretization of the controller, the formation of control commands and control states. The software framework flowchart is shown in Figure 5. [align=center] Figure 5 Software Framework Diagram[/align] Conclusion This paper describes the entire design process of a radar electromechanical servo system, focusing on the design of the control law and the implementation of the servo system. Regarding control law design, this paper introduces classical PID control and sliding mode variable structure control, and compares the two control laws. The conclusion is that sliding mode variable structure control has better control performance than classical PID control. In terms of implementation, a method for speed control using position feedback is presented, and a simple implementation method for discretizing the system state equation is provided. Finally, the software framework flow is given.