EMBEDDED SYSTEM POWER Power design considerations for embedded systems

Related Vendors

Rogers Corporation

dSPACE GmbH

MERSEN

As intelligent mobile devices become ever smaller yet more functional, supplying sufficient, high integrity battery power to their embedded systems becomes increasingly challenging. This article describes some techniques to address these challenges.

An embedded system provides real time computer processing power from a compact physical format. As such, it has a dedicated function within a larger mechanical or electronic system. Traditionally, the ‘larger system’ could be something like a factory machine, or a bank ATM. With the rise of the IoT, however, it is just as likely to be a smaller and often mobile device, such as a medical instrument, digital camera or card swipe machine – or a commercial, military or recreational UAV.

Clearly, mobile systems must be battery-powered – and this immediately creates a problem for their designers. They are under constant pressure to provide more functionality and performance to maintain a competitive edge, but enhancements demand more power. This reduces the operational time available between battery charges, and detracts from the product’s appeal. Meanwhile, there is usually also pressure to make the product smaller – leaving less room for a well-sized battery.

Excessive power consumption can create other problems too. Higher power drawn inefficiently leads to heat losses, creating a threat to the embedded system, its surrounding equipment, and possibly its users. In any case, manufacturers must act diligently to minimize their carbon footprint, to meet the expectations of their customers, employees, and shareholders.

To mitigate this power challenge, designers can take several steps towards developing a power supply that is right for their embedded system. Firstly, they can review the active and passive components needed for their system design, and make choices to minimize their power consumption. Then, they can make a block diagram of the embedded design, so that the main components and their power consumption can be identified.

Once this is known, a power supply of appropriate capacity can be designed. As well as having the right capacity, embedded system power supplies should be designed in accordance with some important power integrity practices, as we shall see.

EMBEDDED SOFTWARE CHALLENGES

Challenges of Embedded Software Engineering

Choosing and using embedded system components

Use the lowest voltage level that will be tolerated by all the devices in the system, as their power consumption is directly proportional to this voltage. Digital CMOS devices’ power consumption is also directly proportional to clock frequency, so reducing this can help as well.

If just a few chips require a higher voltage, it may be worth the cost and space of using DC-DC converters to supply them, while running the rest of the system at the lower voltage. In some circumstances it may be desirable or even essential to distribute the supply voltage at a high level across the system, and convert it down for power-hungry low voltage devices by using DC-DC buck converters located close to them. This avoids inefficiency and I2R losses in the cables or PCB tracks between the power source and the points of load.

Switching DC-DC converters rather than low-dropout (LDO)/linear regulators should be used if possible, as they are far more energy efficient in conversion. Another point to remember is that analog or mixed-signal devices’ current consumption can also increase with temperature.

Ready-made modules such as RS232/RS485 communications interfaces, Wi-Fi or Bluetooth modules, cameras, sensors or other IoT devices also contribute considerably to power demand. It is important to check their idle as well as active power consumption. Selecting the right technology is also essential; Wi-Fi, ISM RF, BLE, Zigbee and other standards can all be used to transmit data, but designers have to balance their relative power consumption with the user experience they offer.

Leaving devices like communications modems always on can be wasteful. If possible, these should be powered down when not in use. However, this option, even if available, may not always be viable. A device’s inrush current on repeated power-up cycles may be unacceptable, as might be the power-up time delay until the device is fully powered and ready for use.

Some embedded systems include displays, with various technologies, performances, and user experiences – 7-segment, character LCD, monographic LCD or TFT for example. There is often a temptation to use higher-performance displays to make the embedded system more attractive, or provide better operating information. However, this must be balanced against power consumption, as the display can contribute 50 % to 60 % or more of the total power demand.

If a high-performance display is mandatory, mitigating techniques like power gating the backlight or using grayscale mode can be used, or a smaller display can be chosen. LED indicators’ consumption can also be reduced by reducing their brightness with a higher-value series resistor, or blinking the LED instead of leaving it always on. A 10 % on time provides adequate indication.

Pullup/down resistors and voltage dividers also dissipate power, so savings can be made by choosing the highest possible resistor values.

Choice of the microcontroller unit (MCU) is also important if its power consumption contributes significantly to the total power drawn. In the experience of engineering blogger Pallav Aggarwal , the Texas Instruments MSP430 MCU is an attractively power-frugal device. MCU power consumption can be further reduced by careful use of power saving/sleep and other operational modes. Using best practices for writing efficient code is important for reducing CPU idle time and hence power consumption.

Power supply design strategy

Once the embedded system has been finalized, and its actual power consumption determined by estimation and then by measurement, a suitable power supply and power delivery network can be developed. According to PCB design company Altium , any power supply design and power conditioning system should meet the following requirements:

Stable voltage/current outputs: These should have as little noise as possible. This involves suppressing ripple (for AC->DC->DC conversion), suppressing or filtering switching noise (DC->DC conversion), and designing the rest of the system to meet EMC standards.

Transient immunity: Many embedded systems carry standards that dictate the level of required immunity they should have to the transient response from a power supply. This is related to the slew rate of a component receiving a transient power signal and is typically measured in V/μs.

EMC standards: Active components in switching regulators or buck-boost converters can create noise that interferes with nearby components. Designers should familiarize themselves with design guidelines to ensure EMC compliance when designing a power supply for an embedded system.

Ensuring embedded system power integrity

Before taking an embedded system into production, simulation should be used to ensure its power integrity. This allows identification of hot spots on a PCB, and determination of the immunity of different components to switching transients. The power supply must also maintain a steady DC output when many ICs switch simultaneously.
Altium Designer is one available package that contains the design tools necessary for identifying power integrity problems in an embedded system design.

(ID:47999069)