Wednesday, June 30, 2010

Power Supply Filters

A power supply filter is a passive filter for linear or switching power supplies to smooth ripples or pulsations in the raw DC output. A line filter, as shown in Fig. 1, suppresses RF interference (RFI) induced into or transmitted on the AC power line or induced into or conducted from within the host product. These filters are required in products powered by switching power supplies, such as personal computers, that must comply with Federal Communications Commission (FCC) regulations limiting EMI/RFI above 10 kHz.

Figure 1 Line filter for a power supply.


The characteristic curves of the four basic types of filters are shown in Fig. 1. The frequency values on the horizontal axes are typical operating frequencies for the filters shown, and the positions on the curves labeled fC are the cutoff frequencies.

Figure 1 Filter characteristics:

(a) low-pass filter, (b) high-pass filter,
(c) bandpass filter, and (d) band-reject filter.

Passive Filters

A passive filter is a network of resistors, capacitors, and inductors configured to pass specific frequency bands while suppressing others. The upper and lower limits of the band are called cutoff frequencies. Filters are designed so that their input and output impedances match their source and load impedances. Roll-off or attenuation at the cutoff frequency is measured in decibels.A filter with high attenuation has a steep roll-off curve that is nearly a vertical slope.

Filters are configured by connecting capacitors and inductors in networks, and their schematics suggest letters or other familiar symbols. The four most common configurations are the L, T, pi, and ladder. The positions of the elements are determined by the desired function of the filter (e.g., low pass or high pass). The L filter schematic is shaped like an inverted letter L, and the T filter is shaped like the letter T. The pi filter schematic looks like the Greek letter π, as shown in Fig. 1, and the ladder filter looks like a ladder.

All capacitors can pass AC, and high frequencies pass with less opposition than low frequencies. (Capacitive reactance is inversely proportional to frequency.) But because a capacitor has conductive plates separated by an insulating dielectric, DC is completely blocked. By contrast, inductors, basically coils of wire, easily pass DC and very low frequency AC, but their ability to oppose AC is directly proportional to frequency because inductive reactance is proportional to frequency. Thus, passive filters exploit the frequency-response characteristics of capacitors and inductors.

Figure 1 Pi filter for a power supply.

Friday, June 25, 2010


  • The constant-k filter is so named because the product of its series and parallel impedances remains a constant designated k at all frequencies. These impedances can be inductive or capacitive reactances. A constant-k filter can be configured as any of the basic filter types.
  • The m-derived filter is a modified form of a constant-k filter based on a constant called m, the ratio of the cutoff frequency to the infinite attenuation frequency. An m-derived filter exhibits a sharper attenuation or roll-off curve than a constant-k filter because it has more poles. It can also be configured as any of the basic filter types.
  • The Butterworth filter exhibits an essentially flat ripple response in the passband and a sharp attenuation or roll-off curve at its cutoff frequency. It has a wide operating frequency range that extends from DC into RF. These filters can be configured as low-pass, high-pass, and bandpass. Their transient responses are much better than those of Chebyshev filters.
Filters can be identified by one or more of the following classifications:

  • The Chebyshev filter has characteristics that are similar to those of the Butterworth filter, but it trades off higher amplitude ripple response to obtain an even sharper frequency roll-off curve at its cutoff frequency. Because these are constant-k filters, they can be configured as low-pass, high-pass, and band-reject.
  • The Bessel filter is named for the mathematical functions used to design it. Its frequency cutoff characteristics are not as sharp as those of the Butterworth filter.
  • The elliptical filter is similar to a Chebyshev filter, but its passband contains even higher amplitude ripple response.
  • A filter can be further characterized by its number of poles, as determined by the number of reactive components (inductors or capacitors) within the filter. (Resistors do not count as poles because they are not reactive.) The steepness of the attenuation curve or roll-off is determined by the number of poles. For example, a six-pole filter has a steeper attenuation curve than a two-pole filter.


A filter is a circuit that passes certain frequencies while suppressing others. This property isuseful for eliminating unwanted frequencies and separating wide frequency bands into multiple channels. A passive filter does not require a power source, but because it dissipates input power it cannot provide either current or voltage gain. Moreover, it has a limited frequency range. Signal loss caused by filtering with a passive filter is called insertion loss.

By contrast, an active filter can perform the same functions as a passive filter, but it can perform those functions over a wider frequency range, and it can provide current or voltage gain. Although an active filter requires a power source, it does not need a bulky inductor.


There are four basic types of filter:

  1. A low-pass filter can pass all frequencies from zero to its cutoff frequency, and block all frequencies above the cutoff.
  2. A high-pass filter can block all frequencies below its cutoff frequency, and pass all frequencies above the cutoff. Its response is the inverse of the low-pass filter.
  3. A bandpass filter can pass all frequencies within a band defined by lower and upper cutoff frequencies, and block all frequencies above and below that band.
  4. A band-reject or notch filter can block all frequencies between its lower and upper cutoff frequencies, and pass all frequencies above and below that band. Its response is the inverse of the bandpass filter.


A power transformer can transform 50- to 60-Hz AC line power to voltages suitable for rectification to regulated DC. They are made in volume as standard products for the linear power supplies in such products as TV sets, VCRs, and stereos. Their laminated iron or steel cores are made from stacks of E- and I-shaped stampings assembled around toroidal bobbins. Power transformers intended for use in switching power supplies that switch at 400 Hz to 50 kHz are wound on ferrite cores because the reactance losses from laminated iron cores limit efficient operation to about 400 Hz.


An audio or voice transformer is similar to a power transformer, but it operates over a wider frequency range. These transformers can conduct DC in one or more windings, transform voltage and current levels, and act as impedance matching and coupling devices, or as filters. A limited range of voice frequencies within the 20 Hz to 20 kHz audio band can be passed by audio transformers.


A pulse transformer is a miniature transformer that generates fast-rising output pulses for timing, counting, and triggering such electronic devices as thyristors (silicon controlled rectifiers) [SCRs] and triacs) and photographic flash lamps.


A circuit-board transformer is made for circuit-board mounting. Classed in this group are miniature power, audio, and pulse transformers. Some have low profiles, as shown in Fig. 1-21, to permit circuit cards in card cages to be stacked closely together. Typically, these transformers are dipped in epoxy resin to seal them from dirt and moisture. Some windings have pin terminations for circuit-board insertion, and others have pads for surface mounting.

Figure 1-21 Transformer for circuit-board mounting.


A radio-frequency transformer is designed to function efficiently at radio frequencies. Unlike low-frequency transformers, they are wound on air-core bobbins because neither ferrite nor laminated iron cores are efficient at radio frequencies.


A toroidal transformer is wound on a ring-shaped core made by winding long thin continuous sheet metal strips around a cylindrical form. Both the primary and secondary windings are wound on the core by special machines designed to be able to pass wire through and around the open core. Toroidal transformers are more efficient and lighter than comparably rated laminated-core transformers, and they do not emit an audible chatter.


An inductor provides a known amount of inductance in an AC circuit. It is made by winding a length of copper wire around a cylinder or other form to make a coil or toroid. The value of inductance can be increased by inserting a core of high magnetic permeability material such as iron or ferrite within the coil. Factory-made standard inductors have values that range from less than 1 μH to about 10 H. Small inductors are used in tuned RF circuits, and large inductors are widely used in tuned audio circuits. However, the inductors with the largest values are used as filter chokes in linear power supplies. A perfect inductor would have only pure inductive reactance, but real inductors have a finite resistance. The inductance value of a variable inductor can be adjusted over a finite range by changing the number of turns in the coil or moving a permeable core in or out of the coil. At high UHF and microwave frequencies, short lengths of copper or aluminum wire serve as inductors.


A transformer transfers electrical energy from one or more primary circuits to one or more secondary circuits by means of electromagnetic induction. It consists of at least one primary winding and one secondary winding of insulated wire on a common core. No electrical connection exists between any primary or input circuit and any secondary or output circuit, and no change in frequency occurs between the two circuits.

Figure 1-20 Transformer schematic symbols: (a) step-up transformer, (b) step-down transformer, and (c) multiple-wound transformer.

If an AC voltage is applied to the primary winding of a transformer, an electromagnetic field forms around the core and expands and contracts at the input frequency. This changing field cuts the wires in the secondary winding and induces a voltage in it. The voltage that appears across the secondary winding depends on the voltage at the primary winding and the ratio of turns in the primary and secondary windings. Schematic diagrams for three commonly specified transformer configurations are shown in Fig. 1-20.

A step-up transformer, as shown in Fig. 1-20a, has twice the number of turns in its secondary winding as it has in its primary winding, so the voltage across the secondary winding will be twice that of the voltage across the primary winding. Similarly, a step-down transformer, as shown in Fig. 1-20b, has half as many turns in its secondary as in its primary, so the secondary voltage will be half that of the primary voltage. A multiple-winding transformer, as shown in Fig. 1-20c, provides three separate output voltages that also depend on the ratios between primary and secondary windings.

All of these transformer configurations obey the law of conservation of energy. In transformers this can be interpreted as the equality of the products of voltage and current or power in both primary and secondary windings, except for losses. Thus, the power input at the primary winding is nearly equal to the power output at the secondary winding or the sum of the secondary windings if there are more than one.

If, for example, the voltage at the secondary terminals of the transformer is twice that
of the primary terminals, the current at the secondary terminals must be about half that at the primary terminals to keep the product of voltage and current, which is equal to power, constant. An ideal transformer would be 100 percent efficient because the power output would be equal to the power input. But, because losses reduce the efficiency of most transformers to about 90 percent, output power is about 10 percent less than input power. The total loss is the sum of ohmic resistance loss, eddy-current induction loss, and hysteresis (molecular friction) loss, all caused by the changing polarity of the applied current.

Most transformers transform voltage or current up or down, but an isolation transformer provides secondary voltage and current that are essentially the same as the primary voltage and current (except for resistive losses) because both windings have the same number of turns. These transformers prevent the transfer of unwanted electrical noise from the primary to the secondary windings, thus providing isolation.

The transformers closely associated with electronics are the power, audio, pulse, and RF transformers. They are rated according to the products of their secondary voltages and current in voltamperes (VA) or watts. The transformers specified for most electronic applications are rated for less than 100 VA or 100 W, but some switching power supplies have transformers rated to 1 kW.

Military Standard MIL-T-27 is the mandatory guide for workmanship on mil-spec transformers, but it is also widely used as a guide in the manufacture of commercial units. Commercial transformers that are connected to the AC power line are usually certified by a national organization for conformance to recognized safety guidelines because faults or failures in these transformers could cause electrocution or fires.


A variable capacitor is a capacitor whose capacitance value can be adjusted by turning a shaft or screw. Used almost exclusively in RF circuits, there are two classes: tuning and trimmer. Their dielectrics can be plastic, ceramic, glass, or air.


A tuning capacitor is a variable air-dielectric capacitor with plates that move within other plates to change the overall capacitance value. A single gang-tuning capacitor, as shown in Fig. 1-19, has a set of aluminum plates called the rotor mounted on a shaft so that the plates interleave with a matching set called the stator mounted on a rigid spacer. When the rotor shaft is turned by a knob, the rotor plates move in or out between the stator plates without touching them. A change in knob position alters the capacitance value, which is directly proportional to the area of the interleaved plates. Capacitance values can be from 1 to 500 pF. They are used to tune radio receivers, transmitters, and oscillators.

Figure 1-19 Tuning capacitor.


A trimmer capacitor is a small variable capacitor with air, ceramic, plastic, glass, or other dielectric that is used for fine-tuning RF circuits. They have capacitance values from 2 to about 100 pF. Made in many different styles, plate spacing is changed to alter the capacitance value by turning an adjustment screw.


Tantalum electrolytic capacitors are made in three styles: (1) wet foil, (2) wet anode, and (3) solid anode. Tantalum capacitors typically have higher CV ratings than aluminum electrolytic capacitors with the same capacitance values. The dielectric formed, tantalum oxide (Ta2O5), has nearly twice the dielectric constant of aluminum oxide. All tantalum capacitors are inherently polarized. As a group, they offer long shelf life, stable operating characteristics, high operating temperature ranges, and higher CV ratios than aluminum electrolytic capacitors. However, they are more expensive than comparably rated aluminum capacitors and have lower voltage ratings.


A wet-foil tantalum capacitor is made by a process similar to that used in making an aluminum electrolytic capacitor. These capacitors can withstand voltages of up to 300 VDC. Packaged in tantalum cases, they are primarily specified for military/aerospace and highreliability applications.

Figure 1-16 Wet-slug tantalum electrolytic capacitor.


A wet-anode tantalum capacitor, as shown in Fig. 1-16, is made from a porous tantalum pellet that is formed by pressing finely ground tantalum powder and a binder in a mold and firing it in a vacuum furnace at about 2000°C. Heat welds or sinters the powder into a solid spongelike pellet with a large effective surface area. A thin film of tantalum oxide is grown electrochemically on the pellet and electrolyte is added. Packaged in silver or tantalum cases, their CV ratios are about 3 times those of wet-foil tantalum capacitors.

Figure 1-17 Epoxy-dipped solid-slug tantalum capacitor.


A solid-anode tantalum capacitor, as shown in Fig. 1-17, is also made from a porous pellet anode. A thin film of manganese dioxide that is chemically deposited on the tantalum oxide dielectric serves as a solid electrolyte and cathode. Then a layer of carbon and conductive paint is applied to complete the cathode connection. The most popular and lowest-cost tantalum capacitors, they are available with either radial or axial leads. They are dipped or molded in plastic resin to form protective jackets. Some are also enclosed in tantalum cases for further environmental protection. These capacitors have the longest lives and lowest leakage current of any tantalum capacitors. They can have capacitive values of 0.10 to 680 μF, capacitive tolerances of +- 10 to 20 percent, and maximum voltages of 50 V. The popular ratings are 1 to 10 μF.

Figure 1-18 Tantalum chip capacitor.


A solid-anode chip tantalum capacitor, as shown in Fig. 1-18, is made by the same methods as the radial-leaded version, but it is packaged in a leadless molded epoxy case for bonding to surface-mount cards or hybrid circuits. They can have capacitive values of 100 pF to 100 μF, capacitive tolerances of +- 5 to 20 percent, and maximum voltages of 50 V.


An aluminum electrolytic capacitor is made by sandwiching a paper separator soaked in electrolyte between two strips of etched aluminum foil, as shown in Fig. 1-15. The paper spacer prevents a short circuit between the cathode and anode foils. The layers of materials are wound in jelly-roll fashion and inserted in an aluminum case. External connections are made from the electrodes to the outside terminals of the case. Direct current is passed through the terminals of the capacitor, causing a thin dielectric layer of aluminum oxide to form on the anode. The electrolyte in contact with the metal foil is the cathode. A plus sign marks the positive terminal of an aluminum electrolytic capacitor.

These capacitors offer high CV ratios and are low in cost. but they exhibit high DC leakage and low insulation resistance. They also have limited shelf lives, and their capacitance values deteriorate with time. Standard units are available in radial- or axial-leaded cases in a wide range of sizes and values. The most commonly specified values are between 4.7 and 2200 μF with working voltages up to 50 VDC. These capacitors are polarized, and this property must be observed when connecting the capacitor in a circuit or it will be destroyed,

Figure 1-15 Aluminum electrolytic capacitor.

Nonpolarized aluminum electrolytic capacitors are available for use in AC circuits for such applications as speaker crossovers and audio filtering. Two polarized capacitors are placed in series with their cathode terminals connected. The anode terminals form the external circuit connections, and the cathode terminals are isolated from the external circuit by an insulator. These capacitors are rated from 1 to 10 μF with maximum working voltages of 50 VDC.