Loop Bandwidth Optimization and Dynamic Performance Improvement of Digitally Controlled High Voltage Power Supply
Digital control has become increasingly prevalent in high voltage power supplies, offering advantages in flexibility, programmability, and advanced control algorithms. The performance of a digitally controlled supply depends critically on the control loop design. Loop bandwidth optimization enables fast response to disturbances while maintaining stability, improving the dynamic performance of the power supply.
Digital control systems use analog to digital converters to sample the output, digital processors to compute the control action, and digital to analog converters or pulse width modulators to apply the control. The digital controller implements the control algorithm in software, enabling sophisticated algorithms that would be impractical with analog circuits.
The control loop bandwidth is the frequency at which the loop gain drops to unity. Higher bandwidth enables the controller to respond to higher frequency disturbances, improving the transient response. However, higher bandwidth reduces the phase margin and can lead to instability if pushed too far. The optimal bandwidth balances response speed against stability margin.
Digital control introduces delays that limit the achievable bandwidth. The sampling delay is half the sampling period on average. The computation delay is the time for the processor to calculate the control action. The PWM delay is the time from calculation to PWM update. These delays add phase lag that reduces the phase margin and limits the crossover frequency.
The sampling rate must be high enough to capture the dynamics of interest. The Nyquist criterion requires sampling at least twice the highest frequency of interest. In practice, sampling rates of ten to twenty times the bandwidth are typical for good control performance. Higher sampling rates reduce the sampling delay but increase the computational load.
The control algorithm determines how the controller responds to errors. Proportional integral derivative control is the most common algorithm. The proportional term provides immediate response to errors. The integral term eliminates steady state error. The derivative term provides anticipatory action based on the error rate of change. The controller gains must be tuned for the specific system.
Advanced control algorithms can improve performance beyond what PID control achieves. State space control uses a model of the system to predict behavior and optimize control. Model predictive control optimizes the control action over a future horizon, accounting for constraints. Adaptive control adjusts the controller parameters based on operating conditions. These algorithms can achieve better performance but require more computational resources.
Compensation design shapes the loop response to achieve the desired bandwidth and stability margins. The plant transfer function describes the response of the power stage. The compensator transfer function is designed to provide the desired loop response when combined with the plant. Bode plot analysis guides the compensation design.
Digital implementation of the compensator uses discrete time approximations of the continuous time transfer function. The bilinear transform maps the continuous time transfer function to the discrete time domain. The implementation must account for the discrete time nature of the control and the effects of sampling.
Quantization effects in the digital implementation can affect performance. The analog to digital converter has finite resolution, introducing quantization noise. The digital to analog converter or PWM also has finite resolution. These quantization effects can cause limit cycles or degraded performance. The resolution must be sufficient to make quantization effects negligible.
Verification of the control performance requires testing under various conditions. Step response testing measures the response to sudden load or reference changes. Frequency response testing measures the loop gain and phase. Disturbance rejection testing measures the response to input voltage variations. The test results verify that the design meets the performance specifications.
Simulation enables optimization of the control design before hardware implementation. Circuit simulation models the power stage with the digital controller. The simulation can explore the parameter space to find optimal settings. The simulation must accurately model the delays and quantization effects of the digital implementation.
