Power Supply Design: Topologies and Theory
Key Takeaways
- The different converter families available for power design as well as some of their applications.
- An overview of some introductory power design topologies.
- Common-use power supply designs that cannot be categorized as non-isolative or inductive.
Power supply design is a consideration at every point on the grid, from generation to end product.
For PCBs, power comes in many forms, reflecting the diverse needs of circuits in different applications. Translating between different currents (type and value), voltages, frequencies, and other essential waveform characteristics is a necessary step for all electronics. Power animates the circuit, but any designer who has been burned or shocked by a board during development knows full well that mistakes made in designing power systems can have consequences. What may be a painful inconvenience can quickly evolve into a serious or fatal injury as current and voltages rise with the increased power demands of a board. Even disregarding worst-case scenarios, power will influence every circuit and component on a board, and overlooking this influence during design is likely to result in a board that requires significant revisions at a later stage.
While power supply design is an exceedingly broad point of conversation, the logical starting point would be the circuits themselves if form follows function. By understanding the different methodologies influencing power circuitry, designers can coax the best performance out of their design given a variety of different constraints.
Current Flow and Interconversion
Arguably the most defining aspect of power supply design revolves around what type of converter will be necessary to build around the source or onboard power. Converters can accept and outputs provide AC or DC. The different functionalities of the three styles revolve around how the converter translates between the input and output:
- DC-DC converter - These converters do not involve any transformation of the current, but rather adjust the input voltage to the necessary output levels. In today’s electronic world, there may be several instances of step-up or step-down to change the source voltage (battery, charger, etc.) to the various voltage levels required by distinct power nets, e.g., 3.3V, 5V, etc. The increase of voltage (known as boosting) and the decrease (known as step-down) utilize magnetic or electronic field storage by inductor or capacitor, respectively, to achieve the desired output voltage and any additional filtering characteristics.
- AC-AC converter - Generally, power supply types that fall into this domain see use in large-scale electronic systems such as three-phase induction motors and industrial settings. Topologies are a bit more limited than DC-DC due to their specialized applications. AC-AC converters work by acting on the waveform itself instead of the voltage levels of DC-DC that are commonly associated with digital logic levels. In addition to changing the amplitude of the waveform, AC-AC can also change the frequency of a signal, similar in effect to pulse width modulation of a digital line.
- Hybrid - Switching between AC and DC is a crucial function for integration with large-scale electronic circuits, most notably grids. Grid-level transmission requires AC (high voltage DC is making significant gains with new technology) to minimize energy losses for power traveling significant distances over cable, while many consumer electronics operate primarily on DC. Rectification and integration are required to transform from AC-DC or vice versa, respectively.
- Rectifier - Converts from AC to DC. Usually accomplished via (in increasing order of efficiency) half-wave, full-wave, and three-phase rectification. The resulting DC signal requires significant filtering to smooth out the signal and eliminate harmonics. Depending on the power, current, and voltage requirements for a particular circuit, this can be accomplished with anything from resistors and capacitors to elaborate filter networks and active devices such as voltage regulators.
- Inverter - Converts from DC to AC. While there’s a near limitless amount of periodic signals that can be created from DC, designers usually hone in on square waves (the most straightforward analog of a DC signal and therefore the best-suited for devices that operate on a binary on-off schema) or some form of a sine wave for more sensitive components.
The Basic Systems: Linear Regulators and Switched-Mode Power Supply Design
The simplest power design is that of the linear regulator. The linear regulator has an operation that is immediately intuitive to anyone with a basic understanding of electronics theory: by functioning as a resistive load, the regulator can provide a lower output voltage from a higher input through the dissipation of power in the form of heat. In effect, some power is used to heat the element to achieve the desired voltage at the load.
Depending on the change in voltage over the load as well as the supply voltage, linear regulators offer a simple and extremely stable voltage waveform as an end product with minimal EMI issues that may present themselves in fast-switching power supplies. Provided the board does not have any concerns over the heat generation in the vicinity of the regulator, whether that’s because of the overall small amount of heat produced (such as in linear dropout regulators) or because the board design can effectively sink heat, regulators are an effective implementation. However, linear regulators are poor at energy efficiency and can only drop the voltage down from the input level, thus requiring a sufficiently high source for operation. For these reasons, regulator usage is confined to boards where priorities of efficiency, heat, and nonstandard voltage levels are minimal.
How can designers increase efficiency in design? Due to their flexibility and power efficiency, switching-mode power supplies encompass a significant fraction of power designs. Though there are far more permutations available to fit the exact current, voltage, and efficiency requirements of a particular board, the basic non-isolated topologies are outlined below:
- Buck - Also known as a step-down converter, it is used to reduce a high source voltage to a lower value for power networks with reduced voltage requirements. Consists of a source, switch, inductor, capacitor, and flyback diode (this is sometimes replaced by a second transistor acting as a switch for reduced losses). When the switch first closes after an open steady state, the transistor begins to store energy in the form of a magnetic field with a large voltage drop at the initial moment the circuit closes (idealized as the source voltage) and gradually decreases to zero. After a sufficient charging period, the switch is thrown open and the inductor sources current from its stored field to continue to power the circuit. As for the rest of the components, the flyback diode prevents large voltage spikes across the inductive load when a switch is opened following a steady state closed condition and the output capacitor reduces ripple for a more steady DC voltage signal. Buck converters can operate with a second transistor in a synchronous setup that minimizes the resistive losses during the “on” period of the circuit, which is especially beneficial in low-duty cycle/slow-switching applications.
- Boost - The boost converter, also known as a step up, is constructed with the same elements as a buck converter. However, these components are arranged in such a manner that the steady state open-switch condition features an inductor with a magnetic field at maximum strength as opposed to zero in the buck design. A bias fully develops across the inductor at steady state conditions, and when the switch is thrown open, the total impedance of the current loop increases. To compensate, the inductor converts some magnetic energy to the current to minimize changes to the current in the loop. In doing so, some of the bias developed across the inductor will effectively be transferred to the load. The capacitor is used to store the charge generated by the combined source-inductor series combination and supplies the load when the switch closes while the diode prevents discharge via the contacts of the switch in the open state.
- Buck-boost - The buck-boost converter is a combination of the step-up and step-down voltage functions of the respective individual circuits. The standard design is an inverting polarity converter, that is, a negative voltage provided at the source transforms into a positive value at the load and vice versa. Alternative designs allow for a non-inversion of the source voltage in the four-switch configuration or a bidirectional power transfer between source and load in a Ćuk or split-pi setup. Buck-boost converters can be used as building blocks for more complex power circuits like the single-ended primary-inductor converter (SEPIC). As in the buck and boost converters, respectively, a closed switch state charges the inductor and allows the capacitor to source voltage for the load, while the open switch condition has the inductor source current to the load directly.
One important operating parameter for switched-mode power supplies is the period current is supplied to the load when being sourced from the inductor. Power circuits are said to operate either continuously or discontinuously when the current supplied for the full duration of the inductor-sourcing cycle is greater than zero or when the current through the inductor is zero at any point during this state, respectively.
Generally, idealized power transfer equations become more complex operating in discontinuous mode as opposed to continuous mode, and greater losses may occur in discontinuous mode as well. For example, a power transfer function operating in continuous mode may be a simple ratio of the input and output voltages, while discontinuous mode may introduce the inductance value, duty cycle, and other parameters. While it may be preferable from an efficiency and design-simplicity perspective to design to a continuous operating mode, this is partially constrained by the load itself, as less intensive loads may not need to draw power for the full timeframe of inductor-sourced power.
A visual reference for the buck, boost, and buck-boost converter theory.
Special Switched-Mode Parameters: Isolative and Capacitive
Switched-mode power supplies operating with transformers are designated isolated due to the galvanic isolation present between the windings of the transformer. For both safety and signal integrity, these power supply designs offer exceptional benefits. By breaking the conduction path between different sub-circuits, designers can communicate between separate power sources operating at different voltage levels. Beyond offering isolation, these switched-mode power supplies can also elevate or drop voltage levels based on the turn ratio present in the transformer windings. Like the non-isolated switched-mode power supplies outlined above, there are a few basic topologies that can be endlessly adjusted:
- Forward - Consists of a switch, transformer, diodes, and an inductor and capacitor to create and store energy by way of their respective fields. During the switch “on” condition, energy is transmitted through the transformer gap and provides power to the load.
- Flyback - Functionally similar in design to a buck-boost converter in the partitioning of the overall circuit between on-off phases. Energy is stored within the magnetic flux of the transformer, which performs like a less-efficient magnetic storage unit than an inductor. The on-state has the inductor building energy and the capacitor at the load sourcing voltage to power the load, while the off-state transfers the magnetic energy across the transformer, powering the load and storing an electric field within the output capacitor. As mentioned, the transformer acts as a storage unit and significantly decreases the efficiency of the circuit compared to the forward topology.
- Push-pull/half-bridge - Push-pull converters are relatively unique in that both acceptable configurations of switch positions provide power to the load, instead of the more common style of the source powering some storage device, which then powers the load during the disruption of the original circuit path. Timing is extremely important in the circuit design, as both switches (generally transformers due to their responsiveness) closing results in a short circuit, and both switches' opening can develop a significant back electromotive force.
- H-bridge - Commonly found in robotics or any device where power needs to be provided to allow motors to run forward and backward (such as turning a wheel on a car). The H-bridge allows power to pass through to two different terminals, provided the configuration would not cause a shoot-through condition that would short the motor.
Up to this point, the discussion has neglected purely capacitive methods of power design. Commonly known as charge pumps, circuits in this mold can generate integer and fractional multiples of an input voltage using a two-stage process similar to magnetic charging switched-mode power supplies. A particular open or closed switch configuration first charges the capacitors while the alternate condition allows the source and capacitors to supply in series, combining the values of the sources at the load. Capacitive switched-mode power design has a more narrow area of application, primarily in the realm of low-power electronics, where efficiency is more valuable than absolute throughput.
Isolation and transformers form an important part of any power design infrastructure.
Best Practices for Improving Power Efficiency
The selection of the power topology is equally as important as the enactment of best practices to coax out optimal performance for a particular configuration. Luckily, these design criteria apply broadly to any choice of power design:
- Limiting high voltage/current spikes - Signal spikes associated with switched-mode power supplies can prove dangerous to power circuitry and influence nearby signals by inductive coupling. To be clear, it is not just high voltage or current levels that need to be considered, but also high rates of change, i.e., dv/dt and di/dt. A continuous operation mode will limit spikes more than discontinuous operation, but designers can also limit these spikes by keeping current loops and return paths direct and short. Certain nodes, like switches and gates on MOSFETs, necessitate a reduction in the surface area to reduce capacitance; while this increases inductance, the overall effect is much less pronounced than reducing the loop size.
- Parasitics - The realities of circuit fabrication mean discrete, idealized components are out the window. While individual instances of their occurrence are small and easy to trivialize, these effects can become detrimental in aggregate if unaccounted for. Parasitics take the form of three essential circuit characteristics:
- Resistance - The thickness of the conductors in the direction normal to the plane of the board is linearly and inversely proportional to the resistance. Vias also provide a nominal resistance, with a larger annular ring and barrel length increasing the resistance.
- Capacitance - Wherever copper features exist in the same position in the plane but on different layers, capacitive coupling occurs. This capacitance is trivial, except for short distances between the layers, but this capacitance gets multiplied through all of the areas of the net. Hence for populous nets like power, this coupling could prove particularly vexing. Decreasing the overlapping area of the conductors or increasing the distance between the layers will reduce this value.
- Inductance - The best method to reduce inductance in a trace is to route over ground, with a shorter distance between the conductor and the ground plane reducing the size of the current loop.
- Heat - Power sources are likely to generate significant amounts of heat during operation, but designers can invest in many mitigating techniques to properly gird the board. For passive options, large copper pours and air-facing surface areas will be the primary options for dissipation. The larger the pours and the less interruption with breaks in the plane, the more effectively heat can travel through the board and away from its point of generation. Larger and more involved packages are likely to need a thermal pad of some sort; always use maximum specs for thermal pad design to maximize contact between pad and pin and to provide additional space for thermal vias. Heat sinks may also be utilized to increase the air-surface contact area for additional cooling via convection. Remember that Ohmic heating is a result of current–methods to improve current pathing and passage will also improve the heat flow as well.
In general, the layout process (placement and routing) should follow a logical flow of both size and priority. Large components or those that require a certain position on the board (e.g., edge, away from tall components for vertical connector, etc.) should be placed/routed before those that can be arranged with greater flexibility. Not only will prioritizing in this manner improve performance, but it will likely save on time that would be spent correcting the layout. As always, keep traces short and direct, and refer to the manufacturer’s datasheet.
Power supply design is a broad topic encompassing many supporting design concepts in addition to being a topic all its own. The necessity of best practices when designing power cannot be overlooked: systems with poor power delivery, grounding, or other issues are likely to experience substandard performance and disruptions during operation. Without a toolset to evaluate the various aspects of board design that can influence power performance, designers can be left in the dark as to how well a circuit is performing. Fortunately, Cadence’s PCB design and analysis software provide a thorough evaluation of work-in-progress to reduce time spent on prototyping and revisions. Coupled with OrCAD PCB Designer, users can realize even the most complex and demanding boards with a full slate of features built for ease of use and power.
Leading electronics providers rely on Cadence products to optimize power, space, and energy needs for a wide variety of market applications. To learn more about our innovative solutions, talk to our team of experts or subscribe to our YouTube channel.