Designing a power supply can be termed more of an art than a science. The transients and real-world interactions are simply too ornate to capture with any single model of a power supply system. Generally, these models construct some sort of transfer function that behaves as closely as possible with the plant or, in digital power designs, the power stage. In order to measure how close the real system behaves to the model and the control loop created to control that model, the power supply designer must measure the frequency response of the system. This data is then plotted on a Bode plot and analyzed to determine the gain and phase margin of the power supply controller design. In many cases, due to the inconsistencies between the model and the actual plant, this process is repeated multiple times during the power supply design process. This tuning of the control loop is where the artistic portion of power supply design comes into play.
Typically, frequency response measurement of power converters is done with help of an external frequency response analyzer. This requires external connections and modifications to the board, including breaking the control loop path to perform analysis.
This method poses several problems for the power supply designer:
1) Inserting the resistor changes the characteristics of the system. No longer are you measuring the response of the pure control loop because you have broken the control loop path and inserted that small resistor.
2) The process of taking these measurements is highly tedious and time-consuming.
3) In most engineering labs, there are only a few network analyzers available and rarely are they just lying around. Vying for time to use this piece of equipment can be a full-time job in and of itself.
So, what other options are there to measure the frequency response of a digitally controlled power supply system? Why, I’m glad you ask. One alternative is to use a software-based algorithm to inject the frequency into the control loop and measure the response of the system using the on-chip analog to digital converter (ADC), which is already connected in the control loop and measuring the output of the power stage. This process provides the plant frequency response characteristics and the open loop frequency response of the closed loop system. Software-based frequency response analysis removes all of the barriers mentioned above:
1) The control loop remains intact and unchanged
2) The process is automated in software already embedded on the device in the form of a library
3) No external measurement equipment is required.
The process of implementing a software-based frequency response algorithm, as Texas Instruments has done with its latest software frequency response analyzer (SFRA) library is described in the following diagram.
Image used with permission from The MathWorks, Inc.
TI’s SFRA library is designed to enable frequency response analysis on digitally controlled power converters based on C2000™ microcontrollers using software only and without the need for an external frequency response analyzer. The optimized library can be used in high-frequency power conversion applications to identify the plant and the open loop characteristics of a closed loop power.
