Constant pressures to save energy, size, and cost all define power-supply design for equipment such as servers, telecom infrastructure, industrial automation, and consumer products. Energy-efficient and highly integrated point-of-load (POL) voltage regulators have become extremely important in satisfying these demands.
As part of a distributed power architecture (DPA), POL converters combine power control and switching functions, helping to reduce parts count and printed-circuit board (PCB) footprint while boosting power-conversion efficiency. Variations on the conventional DPA tend to eliminate the intermediate conversion stage and power the POLs directly from a higher regulated bus voltage. This eliminates some of the conversion losses incurred in a traditional two-stage architecture.
Engineers can apply this approach in a wide range of system designs using integrated POL regulators such as IR’s SupIRBuck family. These devices combine HEXFET MOSFETs offering benchmark conduction efficiency with control circuitry in the same package.
With variants covering a wide range of input voltages, these regulators can operate directly from 12-V up to 27-V rails, for example, saving the need to convert to a lower intermediate voltage. The SupIRBuck family and similar devices also provide built-in protection features and value-added functions such as a precision voltage reference allowing accurate programming of the output voltage.
To ensure optimal power efficiency across a wide range of load conditions, the choice of regulator control scheme can be critical. The SupIRBuck family uses hysteretic constant on-time (COT) modulation to maintain optimal energy efficiency throughout the operating envelope from light load to full load.
Unlike conventional voltage-mode regulators, which modulate the duty cycle to maintain regulation, a COT controller modulates the switching frequency. To maintain constant on-time, the switching frequency increases as load current increases to supply the required energy.
Conversely, the switching frequency decreases when load current is reduced. This principle allows the COT controller to maintain high energy efficiency across a wide load range, whereas constant-frequency controllers employ cycle skipping to avoid inefficient operation at low loads.
Figure 1 illustrates the major functional blocks of a regulator featuring constant on-time control. In principle, hysteretic COT control compares the feedback voltage ripple to an internal reference voltage to maintain output regulation.
During startup, the soft-start voltage is lower than the regulator reference voltage. In this phase, the soft-start voltage acts as the internal reference to ensure a smooth output ramp-up. After the soft start reaches the regulator reference, feedback tracks the regulator reference while the soft-start voltage may continue to increase.
When the feedback voltage ripple falls below the internal reference, the comparator initiates a signal to turn on the high-side power MOSFET. The high-side MOSFET will remain on for the programmed on-time set by a resistor connected between the FF and VIN (input voltage) pins.
The current through this resistor is mirrored to control the charge rate of a capacitor, which triggers the turn-off of the high-side MOSFET and turn-on of the low-side MOSFET when the capacitor voltage reaches the on-time reference voltage. In this way, the on-time of the high-side MOSFET is inversely proportional to the input voltage, so the on-time decreases when input voltage increases and increases when the input voltage decreases.
Once the trigger signal is initiated, the low-side MOSFET will remain on for a minimum duration to charge a bootstrap capacitor. During normal operation or continuous-conduction mode (CCM), the low-side MOSFET is turned off when the high-side MOSFET is turned on. The SupIRBuck regulators implement adaptive dead-time control to prevent shoot-through.
During light-load conditions, the COT controller can operate in discontinuous-conduction mode (DCM) to minimise power losses. When the inductor current ripple is larger than the load current, inductor current can flow in the reverse direction, discharging the output capacitors.
To avoid reverse current flow and prevent undesirable power loss, the low-side MOSFET is turned off when inductor current ripple reaches zero. This is achieved using a comparator to monitor the switch-node voltage and signal when the inductor current is about to reverse direction. The inductor current will remain at zero while both the high-side and low-side MOSFETs are turned off.
The subsequent switching cycle begins when the feedback voltage is discharged below the internal reference. The switching frequency, then, is modulated according to how fast the load discharges the output capacitors. Hence, it’s reduced at light-load conditions. This effectively reduces switching and gate-drive losses, which are the dominant power-loss factors at light load.
Since COT controllers use a comparator instead of an error amplifier to compare the feedback voltage ripple and the internal reference, external compensation networks generally aren’t required. This reduces the component count and the complexity of the overall design.
Moreover, since the scheme’s response time is only limited by the constant on-time of the high-side MOSFET and the minimum on-time of the low-side MOSFET, a COT controller also can deliver superior transient response. When the load steps up, the controller reaches regulation quickly and with minimal undershoot (Fig. 2).
One advantage of this is to ensure fast recovery from low-power or standby mode to full-load steady-state operation. Conversely, regulation can be regained within as little as one switching cycle when the load is reduced, since the high-side MOSFET will not be turned on unless the feedback voltage drops below the internal reference.
Real And Virtual Design Tools
The IR347x family of SupIRBuck regulators featuring COT control cover a broad range of loads from 6 A to 15 A and integrate key system-management and protection features including over-temperature protection, thermally compensated over-current protection, over/under-voltage protection, pre-bias startup power-good output, and an enable input with voltage monitoring capability. Comprising devices housed in 4- by 5-mm or 5- by 6-mm power quad flat-pack no-lead (PQFN) packages, the family also offers scalability to higher current ratings.
A choice of six evaluation boards is available, supporting the IR347x series for 27-input applications and the IR3863 and IR3865 12-V input devices. The IRDC347x evaluation boards will enable developers to solve power-conversion challenges in applications such as light industrial and battery-based equipment operating from voltage rails up to 27 V (Fig. 3). The IRDC will be particularly useful in computing projects targeting second-generation Intel Core processors as well as consumer dc-dc applications including printers, LCD TVs, and game consoles.
Successful power design today depends not only on satisfying end user demands for greener products that tick the right eco-design boxes, but also on delivering these products at lower cost and with improved performance and reliability.
Combining the high efficiency, transient performance, and part-count reduction made possible by constant on-time control, with benchmark power MOSFETs in the same package, today’s POL converters help designers meet these goals to shorten time-to-market.