Integrated Inductive Switch-Mode Power Supply: Fondamentals on Feedback Control in Current Sliding Mode

There are many strategies for controlling the power stage of a static switching converter. The focus here is restricted to non-isolated, non-resonant converters supplied with a DC input voltage, to provide a DC output voltage. Control of the output voltage must compensate for the impact of any variations, whether on the input voltage or on the output current. The [D3283] article deals with fixed-frequency voltage regulation, in voltage mode, i.e. only information on the variation in output voltage is used in the feedback loop. This involves measuring the output voltage and comparing this measurement with a reference voltage; often the reference voltage will be the value of the voltage to be obtained at output, or a fixed fraction thereof, or even a reference voltage at zero output current in the case of the Adaptive Voltage Positioning (AVP) of voltage regulators designed for processors. The comparison, if linear (analog), will produce an error voltage. This error voltage is used to drive the power stage through a modulation operation. The simplest form of modulation is pulse-width modulation (PWM). In other words, a fixed-frequency clock signal is transformed into a signal of the same frequency, but with modulated high and low durations. The duration of the high state becomes that of the energy draw from the input voltage, while the duration of the low state defines the duration of the discharge of the energy reservoir at the output, in the load. It is also possible, by the same means, to regulate the current in the load, if the current measurement produces an equivalent voltage (e.g. across a shunt). In a voltage down converter (or series chopper), if the output voltage drops, the current level in the inductor must be increased, and vice versa. Voltage information, the image of current, is used in a similar way to the direct measurement of output voltage in the case of a voltage converter (apart from a few details, linked to a concern for closed-loop stability). This fixed-frequency switching operation is of no particular interest when compared with output voltage control. Variable-frequency operation offers a more interesting degree of freedom. In this case, the comparison between the current image in the inductor and a reference value will be non-linear. The "discrete" nature of this type of control (current or voltage) immediately differentiates it from the fixed-frequency voltage control mentioned above.

Approaches can be combined to give additional transient performance to closed-loop control. For example, a first closed loop could control the current in the inductor (discrete character), and a second voltage control loop would produce the current reference to the first loop (linear character)....