The role of power management in today’s electronic devices
In earlier days of electronic board design – say in the Seventies, when a Motorola M6800 or Intel 8080 might have been the processor of choice – power distribution was a relatively simple issue. The M6800 operated from a single -0.3 to +7VDC supply, and dissipated 0.5W typically . Related TTL logic could operate from the same voltage. Routing DC power across the board wasn’t too complex, and could be done without excessive I2R losses. In any case, energy efficiency wasn’t a critical issue at that time.
By contrast, an Intel Core i7 can have a thermal design power up to 150W, while operating on voltages reduced to a little over 1V – and lower voltages mean higher currents and potential I2R losses for a given power level. Additionally, a PCB may be designed to accommodate more than one processor type, meaning that the requirement for different DC supply voltages on the board starts to build. This trend can also be driven by other components such as FPGAs, DSPs and memory devices with their own voltage requirements and power levels ranging from microwatts to hundreds of watts.
Yet, while these factors are driving demand for more voltage levels and higher power, designers are also being challenged to pack ever more components into dwindling areas to satisfy the continuous demand for smaller yet more highly functional devices. It all adds up to a need for more voltage rails transporting higher power levels within much less space. Routing, managing and regulating power, while minimising I2R losses, space and cost, have accordingly become critical issues. Excessive I2R losses will not only damage or destroy the device, but also increase running costs, reduce battery life, and raise concerns about the device’s impact on the environment.
This article looks at the key components of power management systems, and shows how they can tackle the issues raised above. It shows how energy efficiency can be improved, while reducing cost and space, through moving from traditional, linear transformer-based designs to switch-mode solutions. The ability to achieve further improvements through various levels of IC integration are also discussed.
We complete the discussion with a look at so-called transformerless power supplies, and how they can be a valuable tool for engineers seeking to design microcontroller-based or other low power circuits with few application components.
Power system components
At a high level, we can identify two categories for power supply components:
- The AC – DC converter, which converts the wall socket AC supply into regulated DC. It comprises a rectifier, which performs the AC – DC conversion, followed by a regulator, which processes the rectifier output to produce a regulated, noise-free DC voltage for the load. It performs load regulation to maintain the output voltage level if the load current changes, and line regulation to maintain it if the input voltage changes.
- DC – DC converters are either found at the back end of the AC – DC stages, or distributed around the PCB near the points of load. These may be buck converters, which produce an output DC voltage lower than the applied input, or boost converters, whose DC output is higher than their applied input.
We can dive in to both of these categories to learn more about what’s happening within them.
AC – DC converters
AC – DC converters are either transformer-based types that use linear regulators, as shown in Fig.1, or based on one of several types of switched-mode regulators. Fig.2 shows a fly-back converter – an example of switched mode regulation.
Linear vs. switched mode operation
ROHM’s Tech Web site explains how both linear and switched-mode AC – DC converters operate, as well as their relative merits:
Linear: In Fig.1’s linear circuit, the transformer steps down (transforms) the 100VAC input into an AC voltage from which a desired DC voltage can be produced. The transformed level (the stepped down voltage generated on the secondary side of the transformer) depends on the winding ratio between the transformer’s primary and secondary.
The stepped-down AC voltage is converted to DC with a diode bridge rectifier, and smoothed out by a capacitor for final conversion to a DC voltage with small ripples. The rectified DC voltage is equal to the AC peak voltage (AC×√2) minus the forward voltage of the diode. A voltage regulator stage can be added if a more regulated output is required.
Two linear regulator types are available ; One is Standard, which has a higher voltage drop across it. For example, a TI LM78L05 will give a regulated output of 5V @100mA, but Vin must be > 7V. The other type, called Low Drop Out (LDO), operates with a lower dropout voltage. The input voltage typically needs to be only 0.6V higher than the output voltage.
Linear regulators are low cost, easy to use and small, but they are very inefficient, operate on low current only, and have limited Vin and Vout ranges.
Switching: Whereas the linear circuit transforms the AC input voltage to a lower level, Fig. 2’s fly-back switching circuit rectifies it directly for smoothing by a high-voltage capacitor. It is then chopped at a high frequency and stepped down by a transformer before being rectified and capacitor-smoothed.
Table 1 shows the comparative merits of linear and switched mode designs.
Table 1: Linear vs. switched mode AC – DC conversion comparison
The size, weight and efficiency savings of switching designs are significant; accordingly, switch-mode AC – DC converters and power supplies dominate the market today.
Switched mode technologies
Having established that switch mode technology is now the mainstream approach to AC – DC rectification and regulation, we can review the different types of switch mode technology currently available. Information is taken from XP Power’s detailed ‘The Essential Guide to Power Supplies ’.
Fly-back: Fly-back converters, as shown in Fig.2 above, are sometimes called isolated fly-back types because the high frequency transformer provides isolation as well as energy storage. They are typically rated up to 150W.
This topology provides a low cost means of converting AC to DC power due to its simplicity and low component count. The power level is restricted by the high levels of ripple current in the output capacitor and the need to store high levels of energy in the coupled inductor in a restricted volume. Fly-back converters commonly utilize valley or transition mode controllers to reduce switching losses and green-mode controllers to minimize no-load power consumption.
The fly-back converter is used in DC/DC converters but only at low power (<50 W)="" due="" to="" the="" low="" input="" voltage="" and="" high="" ripple="" currents.<=""/>
Forward: Forward converters as in Fig.3 are typically used in power supplies which operate in the range 100-300 W. This topology uses two major magnetic components; a transformer and an output inductor. Energy transfer to the secondary and the load occurs during the switching element on-time. Forward converters are used in both AC power supplies and DC/DC converters.
There is no energy stored in the transformer; energy is stored in the output stage of the converter in the inductor and capacitor. The output inductor reduces the ripple currents in the output capacitor and the volume of the transformer is dependent on switching frequency and power dissipation.
Two Transistor Forward: At the higher end of the power spectrum, two transistor forward converters can be employed, as in Fig.4. The two switching elements operate simultaneously, halving the voltage on each switching element and allowing the use of a device with a higher current rating.
As the power rating increases, it is desirable to utilize the transformer core more efficiently by driving it through two quadrants of its available area of operation, rather than the one utilized in forward converters. This is achieved in half bridge or full bridge converters.
Half Bridge and Full Bridge: Half bridge converters are utilized in power supplies in the 150-1000W power range. This topology also uses two major magnetic components - a transformer and an output inductor - but in this case the transformer core is better utilized than in a forward converter. The switching elements operate independently, with a dead time in between, switching the transformer primary both positive and negative with respect to the centre point.
Energy is transferred to the secondary and the load during each switching element on-time by utilising a split secondary winding. This has the added benefit of doubling the switching frequency seen by the secondary, helping to reduce the volume of the output inductor and capacitor required and halving the voltage seen by each switching element.
For higher-power requirements, a full bridge converter – as shown in Fig. 6 – can be used instead.
This topology will provide double the output power for the same primary switching current, but increases the complexity of switching element drive circuits, compared to the half bridge. Half bridge and full bridge converters are used in AC input power supplies. There is also a trend to utilize this topology in low voltage bus converters.
Push-Pull: In DC/DC converters employ a similar topology to the half bridge. This is called a push-pull Converter, and is shown in Fig.7.
As the voltage applied to the switching element is typically low, this arrangement is designed to halve the primary switching current in each switching element, otherwise operation is similar to a half bridge.
LLC Half Bridge: LLC half bridge converters are popular in power ranges from 100 W to 500 W. This resonant topology utilizes Zero Voltage Switching (ZVS) to minimum switching losses and maximize efficiency. Frequency modulation is employed to regulate the output over the load range. Power transferred to the secondary, and the load, increases as the switching frequency nears that of the resonant network and reduces as the frequency moves further away. The resonant inductor (Lr) is often combined with the power transformer by controlling the leakage inductance. The LLC converter is exclusively used with a pre-regulator usually in the form of a PFC boost converter as it has limited ability to compensate for changes in input voltage.
DC – DC converters
DC – DC converters can sometimes be part of the power supply solution, or they can be distributed around the PCB, as separate components, near the points of load (PoLs). As devices become smaller, they operate at lower voltages and higher speeds. Although the current consumption per transistor decreases, overall consumption tends to increase since each device has so many more transistors. In response, power-supply converter designs are moving toward lower voltage, higher current, and quicker response to transient load variations. For example, to cut resistive power losses, the trend in large-scale server and telecom applications is to employ an intermediate voltage—say, 12 or 24 V—combined with multiple point-of-load (PoL) dc-dc converters.
As their name suggests, DC – DC converters convert a voltage input to either a higher-level (Boost) or lower-level (Buck) voltage output.
Fig. 8 is a representation of a simple buck converter. This diagram, together with the following discussion of buck and boost converters, is taken from Texas Instruments’ ‘Introduction to power management’.
In this circuit, inductor current ripple is proportional to the ratio between the switch’s ON and OFF times – its duty cycle.
The relation between the buck converter’s input and output voltage is given by:
Fig.9 (upper) shows this concept implemented as a standard buck converter, while Fig.9 (lower) shows a synchronous buck variant. Standard buck circuits are step down only, and are most popular for point of load (PoL) applications.
A synchronous buck circuit replaces or complements the diode with an extra switch and a low side gate driver to improve efficiency. However, efficiency is not high if the converter is operated under discontinuous conduction mode (DCM).
A synchronous buck circuit with diode emulation can be used. This is similar to synchronous buck but solves the DCM efficiency performance issue. However, it uses a more expensive IC.
If the requirement is to step up the voltage rather than reduce it, then a boost converter circuit as represented in Fig. 10 can be used. Output voltage Vout relates to input voltage Vin according to the equation:
These buck circuit can only step voltages up, and is best used with current loop control rather than voltage mode if operated in continuous conduction mode. The major drawback is that the current cannot be limited, as the switch cannot be turned off.
Alternatively, an inverting buck-boost circuit can be used to provide both step up and step down capability. This is common in battery operated devices, where both buck and boost are periodically needed, depending on battery charge. Vout always has a reverse polarity to Vin, so the circuit is popular when a PCB with a positive voltage input needs a negative voltage for one or more PoLs. The circuit is electrically very noisy.
Fly-back converters are transformer-isolated versions of inverting buck-boost circuits. They can deliver either positive or negative output voltages, depending on their transformer windings. They can buck down from much higher input voltage rails, and can have multiple output voltage of different polarities (e.g. +- 12V) by using more than one secondary winding - but only one voltage rail can be controlled. The circuits are very electrically noisy, but low-cost.
SEPICs (Single Ended Primary Inductor Converters), as shown in Fig. 10, can step up and down like buck-boost, but do not invert the polarity. They are common in battery operated devices, where either buck or boost may be needed, depending on the battery charge. Unlike Boost, they can be shut-down. Their transfer function is complex, so they are typically used when a fast transient response is not required.
TI’s WEBENCH design environment can be used for a stable design. WEBENCH evaluation boards are available from Farnell - see Fig.11.
Control methods: Various methods are available for controlling buck and boost converters. Each has their own benefits and drawbacks. More information on this is available in the Texas Instruments’ ‘Introduction to Power Management’ . Guide.
Choosing the right switching frequency: Switching frequency directly impacts the power supply’s size. The higher the frequency, the smaller the inductor and capacitor; this is why small hand held device PSUs use such high switching frequency.
However, higher switching frequency also means lower efficiency, due to switching losses as well as core losses in the inductor.
Ultimately, therefore, switching frequency choice becomes a trade-off between size and efficiency, as well as cost.
So far, we have seen how most power supply applications have migrated from transformer-based to switching implementations to save energy, space and cost. Further improvements can be achieved by using IC integration for some of the power management functionality. Below are some examples showing various possible levels of integration.
For DC-DC controller implementations, Farnell offers range of integration levels:
Switching modules like LMZ14203: These are complete solutions with internal switch and inductor inside the package. The devices have a very small footprint, with all components optimised, and some come with EMC compliance. However, they are relatively expensive, more limited as the inductor value is fixed, and have a maximum current delivery of 20A.
Switching regulators like LM25011: These solutions include an internal switch, but need an external inductor, as shown in Fig. 13. The circuit has a small foot print(but not as small as with a module), and their switch drive circuitry is optimised. They are much cheaper than power modules but need an inductor, and also have some more flexibility due to external components.
Switching controllers like LM3150: As Fig.14 shows, this device has no internal switch or inductor. The IC has an internal gate driver, but this is not optimised with the switch; WEBENCH can help with switch selection. This offers the greatest flexibility, but means more development time, larger footprint, extra components and more routing.
Higher-level integration: ICs that integrate to a much higher level than converter implementations are also possible. NXP, for example, offers power management integrated circuits (PMICs) that provide optimum integration for a wide range of devices . They combine power management, system control, battery management and interfaces with scalable functional safety and other system-specific functions.
High-efficiency solutions are designed to extend battery life, reduce power dissipation and minimise EMC. They are designed for use with i.MX applications processors, networking and standard processors. Fig.15 shows a typical example.
The PMICs include switching and linear regulators, battery management functions, optimized power modes management, and one-time programmable (OTP) memory for flexible configurability. The devices are used in automotive, consumer, industrial and networking applications.
Example of a high-power system with multiple DC – DC converters
Texas Instruments’ PMP11399 is a complete PMBus power system for three ASIC/FPGA cores, DDR3 core memory, VTT termination, and auxiliary voltages commonly found on high-performance Ethernet switches. It’s a 12V/300W power supply with nine PoL buck converters supplying six different voltages: 5, 3.3, 1.5, 1.2, 1, and 0.85 V. Power levels range from 260 mW to 60 W.
Transformerless power supplies
We have seen how switchmode power supplies have become mostly preferred over linear transformer types. However, in some small, low-power applications, both types are too large and expensive; the linear type because of its transformer, and the switcher because of its inductor/MOSFET/controller components.
A low-cost alternative, known as transformerless power supplies, is available in two versions: Resistive and capacitive. These are described fully in Microchip’s Application Note AN954 , from which the extracts below are taken.
Note that both types present an electrocution hazard, as neither has a transformer for power-line isolation.
Capacitive transformerless power supply: Fig.17 shows a capacitive transformerless power supply. The load voltage remains constant as long as current out (Iout) is less than or equal to current in (Iin). Iin is limited by R1 and the reactance of C1. R1 limits inrush current, and is chosen so that it does not dissipate excessive power, yet is large enough to limit inrush current.
Resistive transformerless power supply: Fig.18 shows a basic resistive transformerless power supply. Instead of using reactance to limit current, this design simply uses resistance. As with the capacitive power supply, Vout remains constant as long as current out (Iout) is less than or equal to current in (Iin).
Comparison of capacitive and resistive power supplies: Both types have several advantages as well as drawbacks.Advantages of a capacitive power supply:
- Significantly smaller than transformer-based types
- More cost-effective than transformer or switched-mode types
- More efficient than a resistive transformerless power supply
- Not isolated from the AC line voltage, which creates safety issues
- Higher cost than a resistive power supply
Advantages of a resistive power supply:
- Significantly smaller than transformer-based types
- More cost-effective than transformer or switched-mode types
- Lower cost than a capacitive transformerless power supply
- Not isolated from the AC line voltage, which creates safety issues
- Less energy efficient than a capacitive power supply
- Dissipation in R1 creates heat losses
This article has shown how the proliferation of voltage levels, lower voltages (so higher currents), higher component densities, more power-hungry components and concerns about heat dissipation and green performance have combined to make power distribution a real challenge for hardware design engineers.
However, we have also reviewed the tools now available to designers; switched mode technology and its deployment into AC-DC circuits, regulators and DC-DC converters both as part of a power supply or for positioning near to the PoLs.
By using these topologies, together with appropriate design tools and optimal levels of integration, reliable and cost-effective solutions can be found even in the face of the challenges posed by today’s diminishing-size, higher-performance devices.
The role of power management in today’s electronic devices - Date published: 15th December 2018 by Farnell element14