The watchword in electronic design today is miniaturization. We want to get as much functionality into as small a space as is possible. In a modern switched-mode power supply (SMPS), the inductors and capacitors required for filtering take up more space than any other part of the device. The appropriate values, and therefore the sizes of these passive components, are inversely proportional to the switching frequency employed. So, a higher switching speed is the best way to accomplish the goal of reducing component size and saving space.
A corollary advantage herein is the fact that air core inductors can be employed, rather than inductors with metal cores. This not only reduces the losses associated with metal core inductors, but also makes the device far cheaper and easier to manufacture. Additionally, because of the relatively small capacitor values called for, troublesome electrolytic capacitors are often not needed.
Losses at the Switch
The problem inherent in using a higher switching frequency is the losses within the MOSFET switch itself, which are due, in large part, to the internal capacitances that, inevitably, reside within the MOSFET semiconductor itself. These losses are often described as being the result of “stress” at the switch. Unfortunately, this effect increases with frequency; each time the switch turns on or off, there is a loss of power. Some of the effects, particularly EMI and RFI, can be ameliorated by the use of so-called snubber circuits, but overall efficiency suffers. The end result is that Pulse-Width Modulated (PWM) switchers are effectively limited in frequency to 1 MHz or so. Resonant converters, on the other hand, can operate at frequencies in the VHF range, 100 MHz or better.
Soft Switching
Resonant conversion, or soft switching, is a way to cope with the losses at the switch. The strategy calls for the MOSFET switch to transition at the point where there is as close to zero voltage and zero current (ZVZC) needing to transverse across it as possible. Doing this does require some very complex control circuitry. A simpler, easier solution to implement has been for switching to occur at the point of either zero current (ZCS) or zero voltage (ZVS). A disadvantage of ZVS is that losses do not go down proportionally with the output load. ZCS, on the other hand, is generally limited as to how high a frequency it can operate at. Not surprisingly, ZVS has gotten the most attention as the way to implement very-high-frequency resonant conversion.
The implementations are quite complex, but the concept, on a basic level, is more easily illustrated. In both diagrams, S represents the semiconductor switch. The figure on the left shows current-mode resonant switching, while the figure on the right shows voltage mode.
In current mode, the resonant action is initiated by the closing of the switch, and it ensures zero current (ZCS) at the time the switch turns on. In voltage-mode resonant switching, the resonant capacitor is connected directly in parallel with the switch, ensuring zero-voltage switching.
Rather than adjusting the duty cycle as a means to control the output voltage as today’s PWM switchers do, resonant converters often rely on burst-mode control; the duty cycle and frequency of operation are held constant. By this method, the entire device is turned on and off to maintain the desired output voltage. The switching occurs at a much lower rate than the VHF speed at which the device is operating internally. The concept is illustrated below.
When the output goes above what is called for, the device turns off, and the load is powered by Cout. When it falls sufficiently, the entire converter comes back on. The more frequently the converter is turned on and off, the more the operation’s overall efficiency suffers. A larger capacitor holds more charge, which makes for less frequent switching. That presents a real design trade off—bulky capacitor and greater efficiency vs. smaller capacitor and lesser efficiency.
Design Considerations
Boost converters are DC-to-DC converters that return a higher output voltage than the applied input voltage. Classical SMPS boost converters operating at relatively slow speeds are described in “Topologies for Power Conversion Devices.” A power section schematic of an experimental boost converter operating at 75 MHz is presented below.
In the example above, ZVS is achieved by carefully choosing Lf, Lr, Ce and Cr so that when the MOSFET switch opens, the voltage appearing across the device’s drain and source will rise and then fall back to zero by the time half a switching period passes. This design does depend on both a fixed duty cycle and a fixed switching frequency.
It is also desirable for the voltage to not only be zero at the switch when it turns off, but that it’s derivative with respect to time must also be zero when the device turns on. This is called Class E switching and it is an extremely important design goal, as the reduced stress presented to the switching semiconductor translates, again, to less spurious electromagnetic radiation.
There are many other methods under development. In one version, inductors and capacitors are added to the inverter to resonate with harmonics of the fixed switching frequency to shape the drain-to-source voltage at critical times to relieve the switching stress. In all cases, since the output voltage is controlled, essentially, by turning the device on and off many times per second, an important consideration is that any design of this sort must be able to achieve steady state output in as few cycles as possible. Stability is also enhanced by designing the rectifier portion of the converter to appear, as closely as possible, as a pure resistor to the inverter portion.
The main efficacy of VHF resonant conversion devices is not in greater efficiency than what is offered by classical SMPS’s. The advantages they present are smaller size, and far lower EMI and RFI, which can be quite critical in a wide variety of applications. At this point in time, a quick web search of this topic will yield primarily academic research articles, indicating that these types of devices have only just begun their journey towards ultimate commercialization.