With this magic tool, choosing the right components is a piece of cake!

For multi-rail applications where efficiency is critical, the switch mode power supply (SMPS) has become the de facto standard. This is especially true in battery-powered and portable applications where long battery life is required. There are many ways to design the power chain. Buck converters, boost converters, buck/boost converters, and several other topologies can be used. What these have in common is that they require well-performing external active and passive components to make the system work optimally.

Some power IC solutions may require only three external components, such as the ADP2108 buck regulator. Because it has an internal power switch, this switch-mode regulator requires only three external components: an input capacitor, an output capacitor, and an inductor. The upper limit of external components is virtually unlimited, depending on the topology and power requirements. Faced with cost, performance, and system reliability issues in a design, designers must know which parameters are most important in order to select the appropriate components.

Everyone is familiar with resistors, and their impact on SMPS is fairly limited. However, where they are used in feedback, compensation, and current sensing, their potential impact must be understood.

When using an adjustable regulator, an external resistor divider network is typically used to divide the output voltage to provide feedback to the regulator. Here, both the resistor tolerance and the resistor temperature coefficient come into play. Newer FPGAs and processors have lower core voltages, which places tighter requirements on supply voltage tolerances. For an FPGA with a 1 V core voltage, a 5% tolerance is only 50 mV.

Figure 1 shows how resistor tolerance and resistor temperature coefficient can have a significant impact on the final design.

figure 1.

A DP2301 step-down regulator has a 0.8 V reference. The output voltage is:

If the gain of the circuit is defined as

Design the output voltage to 1 V, choose R2 = 10 kΩ, and calculate R1 = 2.5 kΩ. The circuit gain is:

If 5% tolerance resistors are used and the worst case is considered, the gain is:

This equates to a ±2% tolerance on the output voltage, which consumes a large portion of the error budget in a system that requires a 5% supply voltage tolerance.

The same design using 1% tolerance resistors would only have an error of ±0.4%.

The resistor temperature coefficient will also cause system errors. If R1 is specified to have a temperature coefficient of +100 ppm/°C and R2 is –100 ppm/°C, a 100°C rise in temperature will cause an additional 0.4% error. For these reasons, it is recommended to use resistors with a 1% tolerance or better. Resistors with temperature coefficients as low as 10 ppm/°C are readily available, but will increase system cost.

Capacitors have many roles in SMPS design: energy storage, filtering, compensation, soft-start programming, etc. Like all real devices, capacitors have parasitics that designers must be aware of. In terms of SMPS energy storage and filtering, the two most important parasitics are the effective series resistance (ESR) and the effective series inductance (ESL). Figure 2 shows a simplified diagram of a real capacitor.

figure 2.

An ideal capacitor has an impedance that decreases monotonically as frequency increases. Figure 3 shows the impedance vs. frequency for two different 100μF capacitors. One is an aluminum electrolytic type and the other is a multilayer ceramic capacitor. At lower frequencies, the impedance decreases monotonically as frequency increases, as expected. However, due to the presence of ESR, at a certain frequency this impedance reaches a minimum. As the frequency continues to increase, the capacitor begins to behave like an inductor and the impedance increases. This curve of impedance vs. frequency is called the "bathtub" curve, and all real capacitors have similar behavior.

image 3.

Figure 4 illustrates the function of capacitors in a buck converter design. The input capacitor sees a large discontinuous ripple current. This capacitor needs to be able to withstand the high ripple current (low ESR) and have low inductance (ESL). If the input capacitor ESR is too high, I*R power dissipation will occur in the capacitor. This will reduce the converter efficiency and may overheat the capacitor. The discontinuous nature of the input current will also interact with the ESL to cause voltage spikes on the input. This will introduce unwanted noise into the system. The output capacitor in a buck converter sees a continuous ripple current, which is generally low. For best efficiency and load transient response, the ESR should be kept low.

Figure 4.

Figure 5 illustrates the function of the decoupling capacitors in a boost converter. The input capacitor sees a continuous ripple current. Low ESR capacitors should be chosen to minimize the voltage ripple on the input. The output capacitor sees a larger discontinuous ripple current. Capacitors with low ESR and low ESL are needed here.

Figure 5.

In a buck-boost converter, both the input and output capacitors see discontinuous ripple current. This topology requires the use of capacitors with low ESR and low ESL.

It may be wise to connect multiple capacitors in parallel to get a higher capacitance. The capacitance increases in parallel, while the ESR and ESL decrease. By connecting two or more capacitors in parallel, you can get a higher capacitance and lower inductance and resistance. Many times, this is the only way to get the required high capacitance value and low ESR to meet the design requirements.

There are many types of capacitors to choose from. Aluminum electrolytic capacitors, tantalum capacitors, and multilayer ceramic capacitors are three of the most common types. Like most design decisions, choosing the right type involves a series of trade-offs.

Aluminum electrolytic capacitors have high capacitance and low cost, and have the best cost/F ratio of all the options. The main disadvantage of aluminum electrolytic capacitors is their high ESR, which can reach several ohms. Be sure to use switched-type capacitors because their ESR and ESL are lower than general-purpose types. Aluminum electrolytic capacitors also rely on electrolytes, which will gradually dry out, resulting in a shorter capacitor life.

Tantalum capacitors use tantalum powder as the dielectric. Compared to equivalent aluminum capacitors, tantalum capacitors offer higher capacitance in a smaller package, but at a higher cost. ESR is typically in the 100 mΩ range, which is lower than aluminum capacitors. Tantalum capacitors do not use a liquid electrolyte, so they have a longer life than aluminum electrolytic types. For this reason, tantalum capacitors are popular in high-reliability applications. Tantalum capacitors are sensitive to surge currents and sometimes require a series resistor to limit the surge current. Be sure not to exceed the manufacturer's recommended surge current rating and voltage rating. When a tantalum capacitor fails, it may burn and emit smoke.

Multilayer ceramic capacitors (MLCCs) offer very low ESR (<10 mΩ) and ESL (<1 nH) in small surface-mount packages. MLCCs are available in values ​​up to 100 μF, although the physical size and cost increase for values ​​above 10 μF. Pay attention to the voltage rating of the MLCC and the dielectric used in its construction. The actual capacitance will vary with the applied voltage, which is called the voltage coefficient. Depending on the dielectric chosen, this variation can be quite large. Figure 6 shows the capacitance vs. applied voltage for three different capacitors. X7R type dielectrics provide the best performance and are highly recommended. Ceramic capacitors are sensitive to PCB vibrations due to the piezoelectric effect of the dielectric, and the resulting voltage noise can disrupt sensitive analog circuits such as PLLs. Tantalum capacitors, which are not affected by vibration, may be a better choice in such sensitive applications.

Figure 6.

An inductor is a magnetic energy storage element, usually a coil of wire wound around a magnetic core. When current flows through an inductor, it induces a magnetic field in the core. This magnetic field is the energy storage mechanism. Since the current in an inductor cannot change instantly, when a voltage is applied to the inductor, the current ramps up. Figure 7 shows the current waveform in the inductor.

Figure 7.

When the switch is closed, the full voltage (V) appears across the inductor. The current in the inductor ramps up at a rate of V/L. When the switch is open, the current ramps down at the same rate, the magnetic field collapses, and a large voltage is generated. This magnetic field is the energy storage mechanism. Figure 8 shows a simplified model of the inductor.

Figure 8.

In addition to the inductor, there is also series resistance (DCR) and shunt capacitance. DCR is mainly caused by the coil resistance and is important for calculating the power loss of the inductor. The shunt capacitance together with the inductor may cause the inductor to self-resonate. The self-resonant frequency can be calculated by the following formula:

A good rule of thumb is to always keep the switching frequency 10 times lower than the self-resonant frequency of the inductor. In most designs, this is not a problem.

The power loss in an inductor causes the inductor to heat up and reduce efficiency. There are two main types of power loss in an inductor, and the designer needs to be aware of both. Winding resistance (DCR) losses are simply the I ² × R losses in the wire, also known as copper losses. Another factor in inductor power loss is the so-called core loss. Core loss is the combined effect of hysteresis and eddy currents in the core. Core loss is much more difficult to calculate and may not even be provided in the data sheet, but it causes power dissipation in the core and temperature rise. Analog Devices has obtained core loss information from inductor manufacturers and included it in the online design tool ADIsimPower. This allows designers to obtain accurate core loss information and its impact on the overall SMPS design.

Figure 9 shows the function of the inductor in both buck and boost power supply designs. The primary purpose of the inductor is energy storage, but it can also be used as a filter. When selecting the inductor value, first determine the maximum ripple current expected. A good starting point is to use 30% of the DC load current for a buck converter and 30% of the DC input current for a boost converter. This allows the inductor value to be calculated using the formula in Figure 9.

Figure 9.

The tolerance of off-the-shelf inductors can be as high as ±30%, which must be taken into account when calculating. In addition, the inductor should be selected according to the following formula:

Where Isat is the saturation current of the inductor. Saturation current is the current that flows when the inductance value is reduced by a certain percentage. This percentage varies from manufacturer to manufacturer and can range from 10% to 30%. When selecting an inductor, it is important to note how the saturation current varies with temperature, as the inductor will likely operate at high temperatures. A 10% reduction in inductance value is generally acceptable in the worst case. Using an inductor that is larger than necessary takes up more PCB area and is generally more expensive. Higher switching frequencies allow the use of lower value inductors.

There are two main core materials for inductors used in SMPS: iron powder core and ferrite. There are air gaps between the materials of the iron powder core, resulting in a flatter saturation curve. Therefore, inductors using this core material are more suitable for applications that require large instantaneous currents.

Ferrite core inductors saturate more quickly but have lower cost and core losses.

Choosing the right inductor value for your circuit is not a simple calculation, but most designs can accommodate a fairly wide range of inductor values.

A relatively new member of the inductor family is the multilayer chip inductor. The physical size of this chip inductor is very small (0805), which allows for ultra-small designs. Inductance values ​​are currently available up to 4.7 μH, so they are generally suitable for designs with higher switching frequencies. The small size also limits their current handling capability (about 1.5 A), so they cannot be used in higher power designs. Compared with standard wirewound inductors, chip inductors have lower cost, smaller size, and lower DCR, so designers can use them at their discretion.

If you compare shielded and unshielded inductors, shielded inductors are more expensive and have lower saturation current (for the same physical size and inductance value), but they can significantly reduce EMI. To help eliminate EMI issues in your design, it is worth using a shielded inductor. This is especially true at higher switching frequencies.

Asynchronous switching power supply designs use a passive switch. This switch is usually a diode. However, due to the forward voltage drop of the diode, the output of the asynchronous design is generally less than 3 A, otherwise the efficiency will drop significantly.

For all but the highest voltage designs, asynchronous regulators recommend the use of Schottky diodes, which have breakdown voltages up to around 100 V. Schottky diodes have a lower forward voltage drop than silicon diodes, resulting in significantly lower power dissipation.

In addition, its reverse recovery time is 0, which also eliminates the switching losses of the diode.

Schottky diodes also offer ultra-low forward voltage drop versions. However, their breakdown voltage is only about 40 V at most, and the cost is slightly higher, but it can further reduce the power consumption of the diode.

When selecting a diode, you must consider the forward voltage drop, breakdown voltage, average forward current, and maximum power dissipation. You should choose a device with the lowest possible forward voltage drop, but be sure to use the forward voltage drop value from the data sheet that is relevant to the design current. Many times, the forward voltage drop increases significantly as the forward current increases. The higher the forward voltage drop, the more power dissipated in the device. This in turn reduces converter efficiency and can potentially overheat the diode.

The forward voltage temperature coefficient of a diode is negative. This is a double-edged sword. On the one hand, as the temperature of the diode increases, the forward voltage drop decreases, so the power dissipation of the device decreases. However, due to this effect, the use of parallel diodes to divide current is not advisable, because one diode will tend to dominate and take all the current in the parallel system.

The breakdown voltage rating of the diode should be higher than the system voltage. The forward current rating should be greater than the RMS inductor current designed into the circuit. Of course, the diode must be able to dissipate enough power to avoid overheating. The maximum power dissipation rating of the selected device should be greater than the design requirements. ADIsimPower, Analog Devices’ online power design tool, has a large diode database dedicated to helping you choose the best device for your specific application.

The "switch" in a switching power supply is generally a MOSFET. Very high voltage and current designs may use an IGBT type transistor.

MOSFET is mainly divided into two categories: N-channel and P-channel , each has its own advantages.

N-channel enhancement mode devices require a positive gate-source voltage to turn on, have lower on-resistance than P-channel devices of the same size, and are less expensive.

P-channel devices require a negative gate-source voltage to turn on, have a larger on-resistance, and are slightly more expensive.

N-channel devices tend to be more difficult to drive due to the requirement for a positive gate-source voltage, as the gate may need to be driven above the system main supply voltage. This is usually handled by a simple bootstrap circuit, but adds cost and complexity to the system. The latest IC regulators include bootstrap diodes to reduce cost and component count.

P-channel devices are easy to drive without additional circuitry. The disadvantages of using P-channel MOSFETs are higher cost and higher on-resistance.

When selecting a MOSFET, one must pay attention to some key performance parameters: Rds, Vds, Vgs, Cdss, Cgs, Cgd, and Pmax (in no particular order).

MOSFET devices have maximum current ratings and maximum power dissipation ratings that must not be exceeded. Internal power dissipation comes primarily from two sources: I ² × Rds and switching losses.

When the MOSFET (switch) is on, the power dissipation comes from only one source, the I ² × Rds losses. When the switch is off, the device does not dissipate power. However, during the transition period, the device does dissipate power. The power dissipation during the transition period is called switching losses.

Figure 10 shows the switching loss curve. It is mainly caused by the capacitance on the gate, including gate-source capacitance and gate-drain capacitance. To turn the MOSFET on and off, these capacitances must be charged and discharged. Note the voltage and current waveforms in Figure 10. During the on-time, there is not only voltage across the device, but also current flowing through it. This results in V × I power dissipation in the device. The higher the frequency, the greater the switching loss. This is one of the many trade-offs in SMPS design. The lower the frequency, the larger the inductance and capacitance, and the higher the efficiency. The higher the frequency, the smaller the inductance and capacitance, but the loss is greater.

Figure 10.

When designing an SMPS, the selection of auxiliary components often takes a back seat to the controller or regulator IC, but the choice of active and passive components has a huge impact on the overall performance of the power supply. Efficiency, heat generated, physical size, output power, and cost all depend to some extent on the external components selected. In order to make the best choice, the performance requirements need to be carefully analyzed. Using integrated design tools such as ADIsimPower from Analog Devices can simplify this process.