Magnetics Design for Switching Power Supplies

LloydH. Dixon

Section 1

Introduction
Experienced SwitchMode Power Supply design-
ers know that SMPS success or failure depends heav-
ily on the proper design and implementation of the
magnetic components. Parasitic elements inherent in
high frequency transformers or inductors cause a va-
riety of circuit problems including: high losses, high
voltage spikes necessitating snubbers or clamps, poor
cross regulation between multiple outputs, noise cou-
pling to input or output, restricted duty cycle range,
Figure 1-1 Transformer Equivalent Circuit
etc. Figure I represents a simplified equivalent circuit
of a two-output forward converter power transformer, optimized design, (3) Collaborate effectively with
showing leakage inductances, core characteristics experts in magnetics design, and possibly (4) Become
including mutual inductance, dc hysteresis and satu- a "magnetics expert" in his own right.
ration, core eddy current loss resistance, and winding
Obstacles to learning magnetics design
distributed capacitance, all of which affect SMPS
In addition to the lack of instruction in practical
performance.
magnetics mentioned above, there are several other
With rare exception, schools of engineering pro-
problems that make it difficult for the SMPS designer
vide very little instruction in practical magnetics rele-
to feel "at home" in the magnetics realm:
vant to switching power supply applications. As a
.Archaic concepts and practices. Our great-
result, magnetic component design is usually dele-
grandparents probably had a better understanding
gated to a self-taught expert in this "black art". There
of practical magnetics than we do today. Unfor-
are many aspects in the design of practical, manu-
tunately, in an era when computation was diffi-
facturable, low cost magnetic devices that unques-
cult, our ancestors developed concepts intended
tionably benefit from years of experience in this field.
to minimize computation (such as "circular
However, the magnetics expert is unlikely to be suffi-
mils") and other shortcuts, still in use, which
ciently aware of the SMPS circuit problems caused
make the subject of magnetics more complex and
by the various parasitic elements and the impact of
confusing to us today. Ancient design equations
the specific circuit locations of these elements. This
intended for sinusoidal waveforms (and not
often results in poor decisions in the magnetic com-
clearly defined as such) are often incorrectly ap-
ponent design.
plied to rectangular SMPS waveforms.
This collection of topics on magnetics is intended
.The CGS system of units. Magnetics design rela-
to give the SMPS designer the confidence and the
tionships expressed in the modem SI system of
ability to: (I) Develop a reasonably accurate electri-
units (rationalized MKS) are much simpler than
cal circuit model of any magnetic device, to enable
in the archaic CGS system. SI equations are much
prediction of circuit performance, (2) Relate the
easier to understand and remember. Unfortu-
electrical circuit model to the magnetic device struc-
nately, in the U.S., core and magnet wire data is
ture, thus providing the insight needed to achieve an

I-I
 usually published in the old-fashioned CGS sys-
tem, with dimensions often in inches, requiring
data conversion to apply to the SI system.
.Energy/time vs. power. Circuit designers are
comfortable in the electrical realm of Volts and
Amperes. On a Volt/Ampere plot, area represents
power. Time is not directly involved.
But on a plot of magnetic flux density, B, vs.
field intensity, H, area represents energy. Time is
always required to change flux density, because
an energy change must take place. A change in
flux density, dE, within a winding equates to the
integral volt-seconds per turn across the winding
(Faraday's Law). Time is definitely involved.
This concept takes some getting used to.
An appeal to suppliers of core materials and
wire: Old-time magnetics designers in the U.S. are
Figure. 1-2 Field Around Conductor Pair
acclimated to the CGS system, and may prefer mag-
netics data expressed in Gauss and Oersteds. But in opposite directions, i.e. a pair of wires connecting
newcomers to magnetics design, as well as experi- an electrical source to a load.
enced designers outside the U.S. prefer the interna- The solid lines represent magnetic flux, while the
tionally accepted SI system -Tesla and Ampere- dash lines represent an edge view of the magnetic
Turns. Also, cores and wire dimensional data should field equipotential surfaces. Each wire has an indi-
definitely be provided in metric units. vidual field, symmetrical and radial, with field inten-
Core losses are usually characterized as a func- sity diminishing in inverse proportion to the distance
tion of frequency using sinusoidal current-driven from the conductor. These two fields are of equal
waveforms. This is erroneous and misleading for high magnitude but opposite in polarity because the cur-
frequency SMPS applications. High frequency core rents that generate the fields are in opposite direc-
losses are primarily caused by eddy currents, which tions. As shown in Fig. 1-2, the two fields, by super-
depend on rate of flux change, not on frequency per position, reinforce each other in the region between
se. (At a fixed switching frequency, higher VIN with the two wires, but elsewhere they tend to cancel, es-
short duty cycle results in higher loss.) It would be pecially at a distance from the wires where the op-
most helpful if materials intended primarily for posing field intensities become nearly equal.
SMPS applications were characterized using rectan- Figure 1-3 shows the field associated with a sim-
gular voltage-driven waveforms, with examples ple air cored winding. The individual fields from each
shown of core loss and minor hysteresis loops under wire combine to produce a highly concentrated and
these conditions. fairly linear field within the winding. Outside the
winding, the field diverges and weakens. Although
Magnetic Field Relationships
Understanding the rules that govern the magnetic the stored energy density is high within the winding,
field is extremely valuable in many aspects of considerable energy is stored in the weaker field out-
side the winding because the volume extends to in-
switching power supply design, especially in mini-
mizing parasitics in circuit wiring, as well as in mag- finity.
A magnetic field cannot be blocked by "insulat-
netic device design.
ing" it from its surroundings -magnetic "insulation"
Figure 1-2 shows the field surrounding two par-
allel conductors, each carrying the same current but does not exist. However, the magnetic field can be
short-circuited -by placing the coil of Fig 1-3 inside

1-2
 .Flux lines are always nonnal to the magnetic
field equipotential surfaces-
.At any location, flux density is always propor-
tional to field intensity: B = .uH
Conservation of energy: At any moment of
time, the magnetic field and the current flow distrib-
ute themselves so as to minimize the energy taken
from the source. If alternative current paths exist, cur-
rent will initially flow in the path(s) resulting in
minimum stored energy, but in the long term, the cur-
rent flow redistributes so as to minimize YR loss.
Reference (R2) has more on this subject.

Transformation of Axes
SMPS circuit designers are obviously interested
in the electrical characteristics of the magnetic device
as seen at the device terminals.

Figure. 1-3 Air-Core Solenoid

a box of high penneability magnetic material, which

provides an easy path for the return flux, shielding
the coil from the environment external to the box

Important Magnetic Field Principles

.The total magnetic field integrated around any
closed path equals the total ampere-turns en-
closed by that path (Ampere's Law)
.Magnetic field equipotentials are surfaces, not
lines. (Alternatively, field intensity can be repre- Figure. 1-4 Transformation of Axes
sented as a vector nonnal to the surface.)
.Magnetic field equipotential surfaces are Figure 1-4 shows how the horizontal and vertical
bounded and terminated by the current which axes scale factors can be altered so that the B-H char-
generates the field. They are not closed surfaces, acteristic (defining a core material) is translated into
as with electric field equipotentials. a 1\1vs. :F (mmf) characteristic (defining a specific
.Flux is in the fonn of lines, not surfaces. (Flux core with magnetic area Ae and path length Re).
can also be represented as a vector.) Transforming the axes once again using Faraday's
.Flux lines are always closed loops -they never Law and Ampere's Law, the same curve now repre-
begin or end. In any arbitrary volume, the number sents the equivalent electrical characteristics of that
of flux lines entering must equal the number core when wound with N turns, fEdt vs. I.
leaving, regardless of the contents of that volume. Note that the slope of the B-H characteristic is
permeability, the slope of the 1\1vs. :F characteristic is

1-3
permeance, while the slope of the fEdt vs. I charac- Ampere's Law
teristic is inductance. SI:
Systems of Units F = ~Hd£ =NI ,.. H£ A-T
The internationally accepted SI system of units
(Systeme International d'Unites) is a rationalized HzNI/f A-T/m(l)
system, in which permeability, ~ = ~o~ (~ is the
absolute permeability of free space or nonmagnetic
material = 47t°10-7;~ is the relative permeability of a
F = f Hdl! = .47rNI "' HI! Gilberts
magnetic material). In the unrationalized CGS sys-
tem, ~ = 1, therefore ~ is omitted from CGS equa-
H "" .47rNI / f Oersteds(la)
tions so that ~ = ~. But the rationalization constant
~ doesn't just disappear in the CGS system -in-
Amperes Law states that the total magnetic force,
stead, portions of this constant show up in all the
.r, along a closed path is proportional to the ampere-
CGS equations, complicating them and making them
turns in a winding or windings linked to that path,
more difficult to intuitively grasp. In the SI system,
i.e., that the path passes through. In the SI system, the
all of the "garbage" is gathered into ~, thereby sim-
units of magnetic force are expressed in ampere-
plifying the SI equations.
turns. When the field intensity H varies along the
The equations below are given in both systems -
path, H must be integrated along the path length.
SI and CGS. It is suggested that beginners in mag-
Fortunately. the simplified form shown in Eq. 1 and
netics design stick to the SI equations and ignore the
1a can be used in most situations.
CGS system until completely comfortable with the
principles involved. Then, it may be helpful to use the Faraday's Law
CGS system when working with magnetics data ex- SI:
pressed in CGS units, rather than convert the units.
dcp E
~I/>= Edt Weber(2)
dt -N : N
Table I
Magnetic Parameters and Conversion Factors
SI CGS CGS to
SI
FLUX DENSITY B Tesla Gauss 10-4
FIELD INTENSITY H A-T/m Oersted 1000/4" d<t> E 108

PERMEABILrrY 1!0 4,,-10-7 1 4,,-10-7

=--XlO8 d<{>= Edt Maxwell (2a)
dl N N
(space)
PERMEABILITY IJ,
(relative) 2
AREA (Core Window) A.,Aw m cm 10-4 Faraday's Law equates the flux rate of change
LENGTH (Core, Gap) R"ig m cm 10-2
through a winding to the volts/turn applied to the
TOT AL FLUX = JBdA oil Weber Maxwell 10-8

TOTAL FIELD ~ jHdf :1',mmf A-T Gilbert 10/41[

winding. Thus, the flux change is proportional to the
integral volt-seconds per turn (directly equal in the SI
RELUCTANCE = :1'/oil ~ A-T/Wb Gb/Mx IO9/47t
system). Faraday's Law operates bilaterally -that is,
PERMEANCE= 1/~ p 47t°lO-9

INDUCTANCE=poN2 L Heruy {Henry)

if 2.5 volts/turn is applied to winding A, the flux
(SI) through A will change by 2.5 Webers/second. If a
ENERGY w Joule Erg 10-7
second winding, B, is linked to all of the flux pro-
Ampere's Law and Faraday's Law jointly gov- duced by winding A, then 2.5 Volts/turn will be in-
ern the important relationship between the magnetic duced in B.
elements and the equivalent electrical circuit as seen Faraday's Law makes it clear that flux cannot
change instantaneously. Any flux change requires
across the windings.
time and usually a change in energy. Time is required
to move along the <1>or B axis, which is more obvious

1-4
considering the electrical equivalent scale dimensions netic system, as shown by the shaded area between
are Volt-seconds. A-B and the vertical axis. From B to C, magnetically
Note that all flux lines follow a closed loop path. stored energy is returned to the electrical circuit. The
flux lines have no beginning or end. difference between the energy put in and taken out is
hysteresis loss, the area between the two curves. At
Energy
C, the magnetically stored energy is zero.
SI:
From C to D, energy is put into the system. From
D back to A energy is returned to the electrical cir-
cuit. The area between the curves is loss. At A, the
remaining stored energy is zero.
I.Edt Joules (3) A positive energy sign indicates energy put in; a
negative sign indicates energy returning to the exter-
CGS:
nal circuit. From A to B, voltage and current are both
w = fVoloHdB = fIoEdt010-' Ergs (3a) positive, so the energy sign is positive. Although the
integrated Volt-seconds are negative at A, upward
movement indicates positive voltage. From B to C,
Energy put into and removed from the magnetic
current is positive but voltage is negative (downward
system can be determined by integrating the area
movement). Therefore the energy sign is negative.
between the characteristic and the vertical axis (B, t/J,
From C to D, current and voltage are both negative,
fEdt) on the energy plane. Energy must be integrated
hence positive energy. From D to A, negative current
over time, which is a factor on the vertical axis, not
with positive voltage indicates returning energy.
the horizontal.
It is much easier to understand this process by Permeability
using the electrical equivalent axes, Volt-seconds and SI:
Amperes. Referring to Fig. 1-5, from point A to B,
.u = .UO.Ur = B / H Tesla/A-T/m
energy from the external circuit is put into the mag-
CGS:
.u = .Ur = B / H Gauss/Oersted

Penneability is a measure of a magnetic material

-the amount of flux which a magnetic field can push
through a unit volume of the material. Penneability is
roughly analogous to conductivity in the electrical
realm.

Figure 1-5 Energy Plane

1-5
 (a) (b) (c)

Figure 1-6- Energy Plane, Sinusoidal Voltage Drive Examples

cluding parasitic inductances, such as shown in Fig.

Permeance, Reluctance
1-1.
SI: This will be discussed in a later section
p =1/ !1{. = <1>/.'!= BAIHf!
Inductance
Webers/A-T SI:
P=I/!I{.=~~.A/1!
Henrys (4)
CGS: L=N2P=~~N2.A/£

P=l/!l(.= <I>/:F= BAIH£ CGS:

Maxwells/Gilbert Henrys (4a)

L = 47tIJ.rN2.A/£.10-9
P=l/!/{.=~.A!f!

When the material characteristic, permeability, is Inductance has the same value in the SI and cas
applied to a magnetic element of specific area and systems. In the SI system, inductance is simply the
length, the result is permeance. In the SI system, permeance times the number of turns squared.
permeance is equal to the inductance of a single turn.
At Home on the Energy Plane
Reluctance, the reciprocal of permeance, is It is, of course, possible to plot any type of elec-
analogous to resistance in an electrical circuit. (Don't trical device on the energy plane JEdt vs. I. Figure 1-6
push this analogy too far -reluctance is an energy
shows several different devices -inductors, capaci-
storage element, whereas resistance is a dissipative tors, resistors -driven by a sinusoidal voltage wave-
element.) Reluctance and permeance can be defined form. Before looking at Figure 1-7 which shows the
for the entire magnetic device as seen from the elec-
waveforms involved, try to identify the devices repre-
trical terminals, but it is most useful to define the re-
sented in Fig. 1-6.
luctance of specific elements/regions within the de-
vice. This enables the construction of a reluctance
The answers:
model -a magnetic circuit diagram -which sheds
considerable light on the performance of the device Fig.1-6 (a) is an air core inductor, ideal and
and how to improve it. From the reluctance model, lossless. Current lags the applied voltage, but as
using a duality process, a magnetic device can be shown in Fig. 1-7, the inductor current is in phase
translated into its equivalent electrical circuit, in-
with JEdt plotted on the vertical axis of the energy

1-6
plane. As the wavefonn traverses from B to C, volt-
age and current are both positive -energy is put into
the inductor. from C to D, current is positive but volt-
E
age is negative -the same energy previously stored is
IR
given back to the circuit.
Fig.1-4 (b) is an ideal, lossless capacitor. Current
leads the applied voltage and is therefore 180° out of
phase with fEdt. From A to B, voltage and current are
both positive -energy is put into the capacitor. From
B to C, the same energy previously stored is returned fEdt
to the circuit.
Fig.l-6 (c) is a resistor. Current is in phase with
applied voltage, therefore current leads fEdt by 90°. A B c D A
Since voltage and current are in phase, their signs are
always the same. The energy sign is always positive -
energy is always put into the resistor, never returned IL
to the circuit. The entire area within the ellipse repre-
sents loss.
Of course, Faraday's Law does not apply to a re-
sistor or capacitor. Therefore the vertical fEdt scale
for these devices cannot be translated into flux.
Fig.1-6 (d) is an inductor with idealized metal Ic

alloy core with low frequency hysteresis, driven into

saturation. A tape-wound Permalloy core driven at
low frequency (no eddy currents) approaches this
characteristic. The area within the characteristic is
hysteresis loss. The only energy returned to the cir-
cuit is the area of the thin wedges above and below
the characteristic. Only when the core saturates, tak-
ing on the characteristic of an air core, is any energy
stored and returned.
The dashed characteristic shows a minor hystere-
sis loop occurring with reduced drive, which does not
take the core into saturation.

Figure 1-7 Sinusoidal Voltage Drive

1-7
 A,B,G,H
(a)

Figure 1-8- Rectangular Voltage Drive Examples

eddy currents) does not change shape radically when

Rectangular Voltage Drive Waveform
frequency or voltage or the waveshape are changed.
Sinusoidal waveshapes are not relevant in most
But with a resistor, unlike the inductor, the energy
SMPS applications. Figure 1-8 shows the same de-
plane plot expands in all directions as a function of
vices driven by a symmetrical rectangular voltage
voltage, the plot changes vertically inversely with
waveform (not including the capacitor, which would
frequency, and changes shape as a function of the
require an infinite current at the instantaneous voltage
driving waveshape, as seen in the difference between
transitions). Figure 1-9 shows the corresponding
Fig. 1-8 (b) and 1-6 (c).
waveforms.
And no matter how many Volt-seconds are ap-
Fig.1-8 (a) is the same lossless air-core inductor
plied to the resistor, it will never saturate like the in-
as in 1-6 (a). Although the characteristic looks the
ductor in Fig.1-8 (c).
same as with sinusoidal drive, Fig. 1-9 waveforms
Noting the striking similarity between the resistor
reveal that the characteristic dwells at its extremes
characteristic of Fig. 1-8 (b) and the dash line un-
during times of zero applied voltage.
saturated square-loop inductor characteristic of Fig.
1-8 (c), raises an interesting question: If inductance is
Fig.1-8 (b) The same resistor as in 1-6 (c) plots
defined as the slope on the plot of JEdt vs. I, then the
as a rectangle rather than an ellipse. The rectangular
resistor in (b ) is an inductor -it has infinite induc-
voltage and current waveforms exist at three distinct
tance along B to C and F to G, just like the unsatu-
levels. From B to C, the voltage and current are both
rated inductor in 1-8 (c).
at a constant positive level, while JEdt slowly rises.
But if a resistor is defined as a device that does
Thus, the current is constant while JEdt changes.
not store energy, only dissipates energy, then the in-
From C to D, current suddenly collapses to zero,
ductor of Fig. 1-8 (c) is a resistor! !
where it dwells until time E, because JEdt does not
change while the voltage is zero. At time F, the cur- Notes on the SMPS Environment
rent suddenly changes to its constant negative level, Transformer Definition: A true transformer is a
where it remains while JEdt slowly drops toward G. magnetic device with multiple windings whose pur-
Fig.1-8 (c) The same idealized metal-cored in- pose is not to store energy, but to transfer energy in-
ductor as Fig.1-6 (d) exhibits the same shape on the stantaneously from input to output(s). In addition,
energy plane although the driving waveshape is quite the windings are often electrically insulated to pro-
different. vide high voltage dc isolation between input and out-
In fact, with any practical inductor with magnetic put. The turns ratio can be adjusted to obtain opti-
core, the low-frequency hysteresis loop (excluding mum relationship between input and output voltages.

1-8
 from the input in its mutual inductance during one
portion of the switching period, then delivers energy
to the output during a subsequent interval.
Since the magnetic core material itself is incapa-
ble of storing significant energy, energy storage is
accomplished in a non-magnetic gap(s) in series with
the core.
Although mutual inductance is an essential ele-
ment in a flyback transformer, leakage inductances
remain undesired parasitic elements.
F G Core Material Limitations: In dc applications,
inductors are thought of as current operated devices.
Even the smallest dc voltage will ultimately saturate
C,D E,F the magnetic core, unless offset by the IR drop in the
winding.
In high frequency SMPS applications, the major
core material limitations are saturation and core
losses, both of which depend upon flux swing. In
these applications, transformer and inductor windings
are usually driven with rectangular voltage wave-
Il forms derived from low impedance sources. Since the
voltage, pulse width, and number of turns are quite
accurately known, it is easy to apply Faraday's Law
to determine the flux swing and appropriately limit it.
In a ferrite core transformer, magnetizing current is
difficult to determine accurately. It depends entirely
on the core material characteristic which varies
widely with temperature and with the flatness of the
mating surfaces of the core halves. Fortunately, the
magnetizing current in a transformer is small enough
to be of less concern than the flux swing.
In an inductor or flyback transformer, the mag-
netizing current is vitally important, because it repre-
Fig. 1-9 Rectangular Voltage Drive
sents the energy storage required by the application.
In this case, the magnetizing current can be calculated
A practical transformer does store some energy in
quite accurately using Ampere's Law, because it de-
mutual (magnetizing) inductance and leakage induc-
pends on the very predicable characteristics of the
tances, which degrade circuit performance in several
gap in series with the core, and the uncertain core
important respects. These inductances are normally
contribution to energy storage is negligible.
considered undesirable parasitics, whose minimiza-
tion is one of the important goals of transformer de- Points to Remember:
sign. .Magnetic field equipotentials are surfaces,
Inductor Definition: An inductor is a device bounded by the current generating the field-
whose purpose is to store and release energy. A filter .All flux lines form complete loops that never be-
inductor uses this capability to smooth the current gin or end, normal to field equipotentials.
through it. A flyback transformer is actually an in- .Flux change cannot occur instantaneously -time
ductor with multiple windings. It stores energy taken is required -an energy change occurs.

1-9
.Energy added and removed is quantified by inte-
grating the area between the characteristic and
the vertical axis.
.On the energy plane, upward movement in Quad-
rants I and IV or downward movement in Qll and
QIll add energy to the device. Moving downward
in QI and QIV , or upward in Qll and QIll returns
energy to the circuit.
.The purpose of an inductor is to store energy. In a
transformer, energy storage represents an unde-
sired parasitic element.

References
"R-nurnbered" references are reprinted in the
Reference Section at the back of this Manual.

(1) T.G. Wilson, Sr., "Fundamentals of Magnetic

Materials," APEC Tutorial Seminar, 1987

(R2) "Eddy Current Losses in Transformer Windings

and Circuit Wiring," Unitrode Seminar Manual SEM400,
1985

1-10
 IMPORTANT NOTICE

Texas Instruments and its subsidiaries (TI) reserve the right to make changes to their products or to discontinue
any product or service without notice, and advise customers to obtain the latest version of relevant information
to verify, before placing orders, that information being relied on is current and complete. All products are sold
subject to the terms and conditions of sale supplied at the time of order acknowledgment, including those
pertaining to warranty, patent infringement, and limitation of liability.

TI warrants performance of its products to the specifications applicable at the time of sale in accordance with
TI’s standard warranty. Testing and other quality control techniques are utilized to the extent TI deems necessary
to support this warranty. Specific testing of all parameters of each device is not necessarily performed, except
those mandated by government requirements.

Customers are responsible for their applications using TI components.

In order to minimize risks associated with the customer’s applications, adequate design and operating
safeguards must be provided by the customer to minimize inherent or procedural hazards.

TI assumes no liability for applications assistance or customer product design. TI does not warrant or represent
that any license, either express or implied, is granted under any patent right, copyright, mask work right, or other
intellectual property right of TI covering or relating to any combination, machine, or process in which such
products or services might be or are used. TI’s publication of information regarding any third party’s products
or services does not constitute TI’s approval, license, warranty or endorsement thereof.

Reproduction of information in TI data books or data sheets is permissible only if reproduction is without
alteration and is accompanied by all associated warranties, conditions, limitations and notices. Representation
or reproduction of this information with alteration voids all warranties provided for an associated TI product or
service, is an unfair and deceptive business practice, and TI is not responsible nor liable for any such use.

Resale of TI’s products or services with statements different from or beyond the parameters stated by TI for
that product or service voids all express and any implied warranties for the associated TI product or service,
is an unfair and deceptive business practice, and TI is not responsible nor liable for any such use.

Also see: Standard Terms and Conditions of Sale for Semiconductor Products. www.ti.com/sc/docs/stdterms.htm

Mailing Address:

Texas Instruments
Post Office Box 655303
Dallas, Texas 75265

Copyright  2001, Texas Instruments Incorporated