Skip to main content

Power Supply Design for Embedded Systems

Key Takeaways

  • Juxtapositioning power and energy concerns in power design.

  • Encoding that maximizes energy efficiency.

  • The four hardware circuits used to build an embedded system power supply.

Bulk electrolytic capacitor array

Power supply design for embedded systems can store charge in a power bank for back–up power.

Stopping to smell the roses in electronic design is something that is rarely done, but it allows for a greater appreciation of the immense strides the industry has taken in a short amount of time. In two decades, the computing power contained within a high-end desktop can now be successfully reproduced on a handheld PCB. Shrinking dies also equivalently reduced the corresponding power demands for the same amount of processing power: what used to require a bulky power supply unit can instead be accomplished with a much more manageable circuit consisting of a handful of discrete elements.

Power supply design for embedded systems is more complex than standard devices due to the presence of complex software that minimizes energy consumption. To best optimize these crucial circuits, software analysis is equally important.

Contrasting Power and Energy

Power supply design is among the chief concerns for all electronics – including embedded systems, especially when these embedded systems are part of a portable electronics system. Systems need to balance size, weight, and cost concerns against the availability of onboard power generation, and in the likely outcome that this accommodation cannot be made, designers must be able to pivot to a lightweight power source that can sufficiently meet all the energy demands of the unit. With the aforementioned constraints, the power supply represents a finite resource whose use should be minimized to extend field life between charges. Hardware and software need to operate synergistically to reduce moment-to-moment draw unless necessary.

An important distinction exists between power efficiency and energy efficiency, though the terms are often conflated and used interchangeably. A power-efficient design minimizes the total power needs of the circuit, reducing overall complexity and the requirement for supporting features like additional heat dissipation. On the other hand, an energy-efficient design makes the most of the power that is available through intelligent component selection, layout, and additional design and manufacturing stages. Both power and energy efficiency can be optimized for a board, one can take precedence over the other, or neither may be in effect, although poor energy efficiency can point to poor implementation of the design at the board level more than any potential constraint/tradeoff.

Power Management Develops Smart Distribution Networks

Embedded systems offer several features over less-specific system architectures. Unmistakably tied to microcontrollers and therefore often integrated into the functionality of larger overall structures, embedded systems place a premium on power performance. For any one clock cycle, the power supply has to account for instructions, memory addressing, caches, and any external interfacing peripherals. Several software implementations are utilized to optimize energy usage:

  • Dynamic power networking - As the computer system consists of many encompassing features, one of the primary methods for improving energy efficiency is preventing power from flowing to subsystems that are not in use on the next clock cycle. Power and energy are tightly related, and limiting the total number of calls to the processor minimizes the time the active elements of the circuit need to be powered on. 

  • Undervolting - Undervolting aims to lower draw and prevent unnecessary heat generation, which can circumvent the need for active energy-intensive cooling processes. In CMOS gates, undervoltage offers a particularly valuable incentive due to their proliferation and the fact that power as a function of voltage scales quadratically, while runtime scales linearly. Certain system modules may even allow for alternate low–voltage operation settings.

  • Frequency scaling - The much less heralded counterpart of overclocking, which draws down consumption by limiting the operating speed of the CPU. The speed at which gates can switch states depends on the difference between the low and high logic levels. A decrease in voltage leads to a decrease in program execution speed, which can be selectively applied to calls that are time-insensitive. Additionally, there are two complementary modes, high-duty low-clock or low-duty high-clock, that exist for constant voltage situations. The latter offers the best performance in an ideal scenario due to the concept of computational sprinting but can lead to brownout conditions in low-battery situations that render it less reliable overall than the former.

The Hardware Supporting Power Supply Design for Embedded Systems

The power supply design for embedded systems needs to be able to convert power from main into a safe and compatible waveform while also bolstering reliability with uninterruptible functionality. In order starting from main, a simple yet effective power supply topology will consist of the following circuits:

  1. A transformer to step down the voltage to manageable amplitudes.

  2. A bridge rectifier to convert AC to DC.

  3. Battery and supporting capacitors to assist in charge/discharge cycles as well as signal conditioning.

  4. A voltage regulator to set a constant voltage (likely 5V) for the embedded system.

The operation of the circuit is fairly straightforward: when powered off main, the embedded system runs directly off the source power, while some power is allocated to charging the battery until it reaches capacity. Then when power from main is interrupted, the battery supplies power until it is depleted or main power is reestablished. In terms of component selection, most embedded systems do not require a large current draw, but designers may want to accommodate moderate currents (~1 A) for increased modularity. For the output capacitor, a capacitance value should be selected based on the maximum permissible ripple.

Power Up Embedded System Design With Cadence’s Simulations

With the continued growth of embedded systems in the IoT space, power supply design for embedded systems remains a noteworthy subject of investigation. Improving power design and management is a desired attribute for new IoT products that may not have consistent access to main power during field deployments. Figuring the best power performance while weighing constraints like cost and size is a challenging task, but Cadence’s PCB Design and Analysis Software offers extensive modeling tools to optimize efficiency. Coupled with the powerful  OrCAD PCB Designer layout environment, development teams can realize the most cutting–edge embedded power solutions.

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.