1294 IEEE TRANSACTIONS ON MAGNETICS, VOL.29, NO.

2, MARCH 1993

A Circuit Approach to Field Computation

C.J.Carpenter
Electrical Engineering Department, University of Bnstol, U.K. BS8 1TR
A bstracf - The desk-top availability of powerful system, and A in terms of its inductive energy, instead of the
field computation packages makes it appropriate to more esoteric idea of a vector potential of a vector field. E and
direct more attention to the needs of the user.
Although user-friendly in many important respects, become auxiliary quantities denoting differentials, thus
packages require a considerable familiarity with reversing the customary roles. The change in view is
electromagnetic field theory, and this is a major appropriate to general-purpose field packages, in which @ and
hurdle for many engineers more accustomed to circuit A are computed, so that a clear understanding of their physical
concepts. Also many design objectives can be significance is helpful, and necessary. E and B are commonly
expressed more directly in terms of circuit derived by postprocessing, suggesting this as their appropriate
parameters, such as capacitance and inductance, than conceptual role.
in terms of the field vectors and E. A simple,
although radical, change is suggested to provide a
closer link between the field and circuit views. The 11. CIRCUIT VERSUS FIELD CONCEPTS
electric potential 0 and ‘magnetic vector are a Field theory rests on the properties of E and B and, in
defined, in energy terms, as the primary field particular, on the idea of flux as a field source, as in
quantities, and are visualised as measures - o f the curlE=--dB
capacitance and inductance, respectively. E_ and at
become auxiliary symbols denoting derivatives, so or, in integral form
that the customary roles of the potentials and field
vectors are reversed. Some of the oractical e = -dJ2.
UL
advantages of the change are examined, and
illustrated by example.
where y~ is the flux linkage. This predicts the force on a
charge, or emf in a winding, in terms of the remote-action
effect of something called “flux,” which is a property of empty
I. INTRODUCTION space if the device is air-cored. The action of iron is accounted
Electromagnetic field computation may now be regarded as for by introducing a new quantity
coming of age, in the sense that very powerful and robust
computation packages capable of solving a broad range of -
B=M+PoH (3)
representing a somewhat arbitrary mixture of the
problems are now widely available, together with the low-cost magnetisation vector, M, due to the electron spins, and the
hardware needed for personal use. The contrasts with the early
days of hand calculation, followed by batch programming, are “free space flux,” p&. Likewise the electric flux, of density
immense, and the progress in software design is Q, is a source of the magnetic field, due to the remote-action
correspondingly impressive to those who remember the stage effect of something called “displacement current,” representing
when the only person who could usc a program was the one the movement of charge in a dielectric, but somewhat more
who wrote it. But the release from such constraints raises mysterious in its absence.
wider questions about the ways in which the needs of the user Circuit theory, on the other hand, is based on parameters
can now best be met. such as capacitance, inductance, and resistance, which also
One aspect is the need to distinguish between data and acquire a direct significance as physical concepts in their own
information. Field computation may easily generate very large right. It is probable that most engineers “see” capacitance, for
amounts of data, but this does not necessarily illuminate the example, as an essential characteristic of metal electrodes,
operating principles of a device, or help in a design task; requiring no “explanation” in terms of some more fundamental
indeed it may have the opposite effect. It is all too easy to property of empty space in conveying electric flux. Moreover,
bury the problem in data. This is closely related to a more it is debatable whether the “explanation” is of any real validity,
fundamental difficulty faced by the general user. Although in that the fundamental characteristic of any electrostatic
expertise in programming is no longer needed, a considerable system is that it stores energy, and this capability is measured
familiarity with electromagnetic field theory is, and this directly by the capacitance. The view of force as a consequence
remains a major hurdle for many engineers with littlc expertise of the stored energy is just as useful, and valid, as the view of
in the subject. Hopes that this situation will improve are not it as the cause. Likewise inductance is a measure of the kinetic
supported by the evidence [ 11. The difficulty lies in the need to energy of the moving charges, and there are considerable
understand the properties of the field vectors, E and J3,and the practical advantages in the concept of force as a consequence,
relationships between them, and this is also a part of the data since this leaves no doubt about the problems of interrupting
problem, since the fields, as defined in this way, are often not the current flow in highly inductive circuits. Concentrating
related to the terminal, or circuit, parameters in a sufficiently attention on the fluxes sometimes leads to a failure to
direct or easily-visualised way. It is such characteristic appreciate sufficiently clearly what actually matters to the
parameters which summarise device behaviour, and are often of engineer.
most interest to the user. The important practical point, to those using field
The paper points to the advantages of a simplc, although packages, is the extent to which a knowledge of the E, B
far-reaching, change of approach. The field and circuit “mechanism” is necessary to understand the system behaviour,
quantities can be more closely related [2,3]by interpreting the with the underlying question about its “reality” as an
fields as the electric potential @, and the magnetic vector A, explanation of what is happening. The idea of fluxes as
defining @ as a measure of the capacitive energy of any remote-action field sources provides a very complete and
beautiful mathematical structure, but is it any more than that?
0018-9464/93$03.008 1993 IEEE
 1295

Although in some respects this is more a matter of philosophy also see that there is no mutual inductance between wires at
than engineering, it has important practical implications to all right-angles to each other. Although the practical advantages of
,who have to apply field theory, and is particularly relevant to (4) are, of course, very well known as a method of
engineers forced to delve further by the requirements of computation, especially for circuits in air, it is normally
numerical packages. In considering the possibilities of an treated as a mere mathematical equivalent, with no physical
alternative view, we observe that there is also a third. significance, and (5). or (l), is invoked as the underlying
Quantum mechanics, which is one of the most secure parts of “cause.” Yet these give much less insight, and very few
modem physics, replaces the E, fields by photons as the programs calculate L in this way.
underlying “reality”. Moreover, the appearance of @ and A, not The underlying point is that the current, not the flux,
E and B, in the Schrodinger equation supports the @, A view becomes the source of A in accordance with
of the macroscopic field, and this has led physicists such as -
V2A=-hJ (6)
Feynman, Leighton and Sands [4] to propose the at low frequency, where the current density,
corresponding change in interpretation of what is meant by the L=lc+Jf (7)
“field.” The increasing role of quantum mechanics in includes contributions from the conduction current, Lc. Lf
engineering suggests that a consistent view of the charge denotes the equivalent surface and volume currents in the iron,
interactions is very desirable. and we may likewise add any convection currents. Expressed as
a volume integral
111. INDUCTANCE
The treatment of @ as measure of capacitance, or electrical A = h l Ldv (8)
energy, is likely to cause little difficulty in electrostatics,
beyond the usual objection that the @ field cannot be regarded and combining this with (4) gives the inductance in the
as a physical entitity because the choice of the @ = 0 datum is Neumann form of an interaction between the currents, or
arbitrary. This point and others, are answered in the companion moving charges, instead of attributing it to the flux; more
paper [l], and have been discussed in more detail elsewhere [ 5 ] . specifically, the flux in empty space.
It is the view of the magnetic field as the A vector, defined in The result is in accordance with the energy density J.A/2.
inductive terms as the kinetic energy of the moving charges, Although the source currents usually flow in closed paths, this
which is less familiar, together with the definition of the is not a necessary restriction, as in circuits carrying pulses, or
electric field as @, instead of E, under non-static conditions. in antennas. Lack of closure further complicates the description
An example of the problem of using a magnetic field of the field of a current element in terms of E and B, since
computation to support a flux-based interpretation of both the flux and the current linkages are difficult to define,
inductance is that of calculating leakage reactance. Most end- and the charges which appear at the ends if the current is
windings and the like consist of a multiplicity of wires interrupted combine with B as a source of E. The exact role of
carrying different currents connected into arrays, and increases the displacement current is in doubt [ 6 ] . The problems
in operating frequency may make it necessary to calculate, and
disappear in the 9, A description, if we assume the Lorentz
minimise, the relevant reactances. Since the various wires
carry different currents, it is convenient to add the inductance gauge, since the charges are the source of @, and the A field
contributions from each, with due regard to the connection depends only on the current, so that the consequence of any
arrangements. But we cannot aSsess flux linkage with a part interruption of the current flow at the two ends is put into its
circuit, or even understand the significance of the term in this proper perspective as a matter of capacitance, not inductance.
context. What is meant by the inductance of a piece of wire ? The finite propagation velocity is modelled directly, by
Clearly energy provides one answer, since the “circuit” view of retarding the field, instead of deriving it from the interaction
magnetic energy density, L&2, distributes the total stored between the electric and magnetic displacement currents.
energy, iL/2, of the system amongst the various circuit parts, Obviously the separation between the property of
and hence identifies the inductance, L of each part. But it is capacitance, due to @ [ 11, and inductance, due to A, depends on
difficult to relate this result to flux, or to flux linkages. The a suitable definition of what is meant by A. This is provided
usual H.B/2 view of energy helps only in giving the total by the assumption that the total system energy is distributed
amount, and offers less guidance in the task of minimising the with a density (p@+ J.&/’2, instead of @.Q + H.B)/2, where p
array inductance. is the charge density. Unique value of @ and A can be obtained
Defining inductance in terms of A i.e. by choosing the Lorentz’gauge and taking the datum at a
sufficient remote point, provided that all of the net charges and
currents, including the electron spins in ferromagnetic
(4) materials, are taken into account as @ or A sources. The whole
instead of of the system must be included, so that it may not be
sufficient to define boundary conditions for A over a closed
L =Y
. (9 surface, thus replacing an exterior part of the system by an
1
provides an obvious answer. It shows that A has the physical equivalent source.
significance of inductance per unit length, times the current, i, A physical insight into what is meant by A is given by the
and this replaces the difficult idea of a magnetic vector force which it causes when it changes in time. If, for example,
potential. Computing A provides the inductance of any wire, we replace the secondary winding in a transformer by a small
group of charge, of density p, and change the primary current,
since there is no need to close the path of integration in (4). the A field is described by the force density on the group
and the effect of iron is included in most programs. We see, by
examining the way in which the various wires contribute to A (9)
that to minimise reactance we keep similar currents as far
apart, and opposite currents as close together, as possible. We in place of the corresponding force due to the field
2%

fB = PbXB) (10) resistivity of the rails. This is the “Maxwell approach,” in the
Equation (IO) requires a cross-product, relating three sense that it was (12) which Maxwell proposed as the first of
lifferent vector directions, whereas the force given by (9) is in his “general equations” [7], expressed in terms of the
he direction of A (or in general, of its change). Thus the “electromotive force”, and using the symbol to denote VxA.
:oncept of the field due to a moving group of charge (and It forms the 4, A equivalent of the Lorentz force
ience a current element) is simplified in in two respects. p(E + uxB) + fo = 0 (13)
nstead of rings of flux, described by another cross-product, the
‘magnetic field is restricted to a single vector direction, given The principle of operation is easily described in circuit
’Y terms. Both the force on M, and the induced emf, are given by
-
A = @IC’ (11) the rate-of-change of inductance, and this can be expressed as a
where u is the velocity vector of the moving group, and @ function of the flux linkages. But (2) gives no explanation of
lefines what is here referred to as its “electric field” at the same how the change in linkage occurs, or whereabouts in the
joint. The force which is caused by a change in is likewise circuit the emf is induced. Nor is this information easily
n the direction of the change. obtained from (I), because of the relative motion between the
The use of A, instead of B, as the primary vector goes back parts of the circuit. Comparison with other forms of machine
o Maxwell, as well as Neumann, since what are now suggests that the induction is in the moving part, M, i.e. in
iniversally known as the “Maxwell” equations are not, in fact, the part on which the magnetic force is exerted, and can be
hose which Maxwell actually proposed. He founded his accounted for by “flux-cutting,” particularly since the emf
‘general equations of the electromagnetic field” [7] on the depends linearly on the velocity of M when the current is
:lectromotive force, expressed in terms of the vector A, which constant. But this does not conform with the A calculation,
le derived in Lagrangian terms from the electrokinetic and the nature of the “flux cutting” is obscure, since M carries
)roperties of the “electrical fluids” in wires. He used this emf a part of the field with it. The idea of a flux exerting the
:quation in place of (l), and defined as the curl of A, in acceleration force on M likewise raises the question of whether
the rails which are the source of the force carry the reaction to
iiccordance with the circuit approach suggested here. However, it, or not; that is, do they tend to crumple ? The IxB forces
.he concept of electric displacement current, dDJdt, was central
(10) can act only at right-angles to the current vector, 1, so
.o Maxwell’s treatment, and this, as well as dB/d[, is
that they cannot cause any axial force on the rails, and Allen
innecessary if we assume that @ and A are retarded in time, at
and Jones [9] have shown that the reaction is accounted for by
1 velocity, c, which observed empirically, instead of deducing
the field momentum, DxB. Are we to conclude from this that
it., The source equations [including (1 l)] then become valid at
the reaction to one of the largest possible acceleration forces is
311 frequencies by using retarded quantities in the integral
carried by empty space ? The system as a whole experiences no
Forms, and adding the wave term to Poisson’s equation, in
90th @ [ 11 and A. The delaying effects of polarisable materials
net force, because of the opposite effect at the far end, but the
two ends do not interact directly, if the gun is sufficiently
:an be examined in accordance with the way in which the long, and we are left with the question of how the force is
additional sources contribute to both fields [3]. conveyed from one end to the other.
Interpreting the magnetic field as A reduces it to Ax and Ay
Iv. ELECTROMAGNETIC RAILGUN components, and divides the inductance of the rectangle into
The consequences of the difference in interpretation of two parts, consisting of parallel pairs of conductors at right-
nagnetic field calculations can best be illustrated by example. angles. These parts are independent in the sense that there is no
3ne which encapsulates a number of points is the mutual inductance between them. There is some contribution
Aectromagnetic railgun, designed to produce extremely large to Ay from the small transverse component of current in the
icceleration forces on a missile, M , by passing a high current
rails, but this can be ignored for the present purpose as a
through it from a DC source (fig.1) along parallel rails. The
second-order effect, and, when described in terms of (4)-i.e.
surrent penetration, and other effects, can be studied by
dL/dt) - the induced emf is due to the rate-of-change of the
somputing A and @ in the region near to the electrical contact
self-inductance of the rails, since the movement of M does not
between M and the rails [8]. When the rails are sufficiently

.
affect its own inductance (Le. it does not move through its
long the magnetic effect of the left-hand end of the circuit can own A field). Thus the source of both the emf, and the force
be ignored, and the most convenient reference frame is one on M, is the A, field of the rails, and Ar. due to M, makes no
moving with the missile. The equation solved is derived from
the charge equilibrium condition, which can be expressed in contribution. The description of motional emf needs close
the form examination, but Ax shows clearly its source, and the nature
dA of the interaction.
p (- E@- -==+ u x g x a + fo = 0 We see that, having computed A, any “explanation” in
at
where fo denotes the non-electromagnetic force, due here to the terms of flux merely confuses an essentially simple
interaction. The computer calculation shows how the
penetration of the current and the A field, into the new
material, describes the acceleration forces on the additional
volume of conduction electrons, as the length changes. This
acceleration process requires a “potential hill,” due to $, which
is caused by the appropriate distribution of excess charge,
controlled by the source. Thus @ defines what is meant by the
tY
- X
“applied voltage,” and A the “back emf,” and we observe that
these are separate field quantities, with different sources, in
contrast with the usual E, B view (13), in which E is defined
 129

than mathematical artifices. Some authors hold sufficiently choice of a negative sign for the electron is pa
strongly to this view to object to the terms “applied voltage” important (and unfortunate) because of the way in
and “back emf,” yet, since @ is a measure of capacitance, and A circuit parameter A concentrates attention on th
of inductance, it would be a matter of great concern if the two interactions,but we can equally treat the c
properties could not be clearly separated at low frequency, even the motion of positive charge, which
though the distinction may include an arbitrary element as the mental picture and illustrates the physical
frequency is raised and retardation effects become significant. in the electromagnetic properties.
As is easily shown by other examples (the Faraday disc The fundamental requirement impos
generator is one), the two major sources of difficulty in many charges are in equilibrium, as is illustrated, for
problems of electromagnetic induction are the concepts of the need for a @ (or Hall) field in the missile M
“flux-cutting” and “flux linkage” associated with €3,together oppose the transverse momentum forc
with a failure to appreciate sufficiently clearly the roles of the conduction electrons. The electrons in the rails must also be i~
+ and A components of E. There are, of course, various ways both transverse and axial equilibrium, showing the nature o
of re-defining the gauge @e. div & for computational the reaction to the force on M which is caused by thc
convenience, and this may absorb @ as a component of A, but movement of the rail electrons. The details of the interactior
the currents and the excess charges are always separate and require more attention than is possible here, but are not needa
identifiable sources of the field, described by A and 4, to illustrate the general point that the @ and A fields mus
respectively, in the Lorentz gauge. It is these which describe always give a complete account of the forces on the charges
the circuit properties of inductance and capacitance, and it is and that no reference is needed to the usual field equations
the substitution of (13) for Maxwell’s equation (12) which is specifically equ.la, either to obtain (12), or to explain wha
the source of much unnecessary confusion and happening.
misunderstanding in describing the operation of simple
electromagnetic devices. One consequence is the not VI.FIELDVERSUS CIRCUIT PROPERTIES
uncommon belief that the “electrostatic” forces due to @ are of One of the most important of the computed parameters i
no importance in magnetic devices, whereas they are, of the system energy, which can be expressed in terms of either 1
course, essential to nearly all current flows. and B, or @ and A, both commonly offered as options withii
the same software package. The latter
V. ELECTROMAGNETIC
MOMENTUM model, in the sense that the energy
Taking the computed parameter, A, as a definition of the confined to the charges and currents,
“magnetic field” suggests that the capacitive, or @, field, may customary interpretation as properties
be appropriately referred to as “electric,” rather than the versions of some of the various prope
somewhat misleading “electrostatic”. Note that this involves a flow, are given in table 1 (note that p$ an
fundamental change in what is meant by “electric” and energy densities, and the factor 1/2 is
“magnetic”, since E is not, in general, the same as -grad 4, but energy).
includes the aAJ3t component (12, 13). A measures the kinetic Other properties fo
effect of the moving charges (in any given reference frame), as transformation appears in its E, B form in on
is shown directly by (1 l), and is described by the force which terms of the 4-vector 4, A in the other
is caused when it changes in time (9), so that the quantity PA commonly stated that what is here referred t
is, in this respect, the electromagnetic analogue of the view of energy is limited to low frequencies,
momentum mu. It is the momentum, or electromagnetic no more than an assumption, and is incon
inertia, associated with the movement of charge which well-known property of the Lorentz gauge in
characterises the property of inductance, and is directly radiation into two parts, one described in terms o
demonstrated by the arc which forms when a switch is opened. other in terms of A. These corr
We also observe that, as the flow of current through the components [5],and form an
missile, M, in the railgun carries the conduction electrons has been shown elsewhere
through a changing Ax field, their momentum changes, frequencies provided that all
causing a force of density defined in a self-consis

I I ,
in accordance with (9). This is the puxl3, or LxB, force given
in (12) or (14), and emerges as another aspect of the
I Property Field Circuit*
“electromotive” force, but acting in the transverse direction. It energy density
is the total force on M, rather than the force density, which is
usually of most interest, and is given by
F = i (A1 - A2) (15)
either by integrating (15), or by direct inspection. Here i
denotes the current, and the subscripts the Ax values at the two
ends of M. It is obvious that the force is directed to the right,
since this is the direction of the electromagnetic momentum of
conduction electrons in the rail which they are leaving, and the
two cross-product rules which so bedevil the customary
description in terms of are replaced by the electromagnetic
1298

observable forces on the charges, in accordance with (12). V2$ + k 2 $ = 0 (16)

The railgun example illustrates the way in which the J.N2 is satisfied by Cp, as by E and B, and imposing the zero-energy
view of the energy is more directly relevant to what is going surface condition Cp = 0 gives the usual forms of solution, but
on, since the physical parts consist of charges, not fields. If expressed in terms of a scalar, instead of the various vector
the current is constant, so also is 1.N2 in the missile, M, and components. The analogy with a vibrating membrane is
the increase in the energy of the system is caused by the particularly close. The simplest launching element can be
change in Ax, due to the changing length of the rails. Both the visualised as a small charged sphere at the end of a short
momentum and the energy changes occur in the newly- current-carryingwire, oriented in the axial direction, so that the
energised material, and it is this which is the focus of the field Cp field determines the sphere capacitance, and shows how this
computation. is modified by the charges induced on the waveguide walls. To
The most familiar aspects of the “circuit” model are the avoid evanescent modes a launching array is required, arranged
descriptions of energy and power, but they extend also to force. so that the end-wall source conditions match those in the Cp
The “field” view is the Maxwell stress, since this, like all the wave, giving a distributed capacitance. The current in the
other properties, depends only on E and B, together with D and supply wire, or wires, likewise launches a single-componentA
-
H . The “circuit” description is what may be termed the field, oriented in the direction of propagation, showing how
“Maxwell force”, in (12), since this was the first of Maxwell’s the inductance of the launching wires is modified by the
“general equations.” Like all the other “circuit” properties, it cavity. The two properties are closely linked physically, and
takes the form of a product of charges, or currents, and one of this is reflected in the similarity in the AZ and Cp fields, which
the field measures Cp and A (see also (1 7)). The version given satisfy the same equation and boundary condition. The
in (13), on the other hand, is a “hybrid” 121, in the sense that it combination defines the usual transverse-magnetic,or E-mode,
mixes charges and E, B fields. This “hybrid” nature is reflected but the five components of E and are replaced by two, and
in the way in which the “pure” field equations, including (l), the scalar Q is sufficient to illustrate both. Table 1 shows that,
are often given as the essential foundation of electromagnetic to obtain a complete description, the Cp and A fields must be
theory, without explicit reference to (1 3) (or to the version supplemented by the source charges and currents (flowing in
given in (12), on which Maxwell founded his theory). the axial direction), and this gives another two variables. But
since the wall currents are important in many practical
VII. WAVEGUIDES applications, these have also to be added to the description in
The customary definition of A as a vector potential makes terms of E and B. This helps to illustrate the difference
the choice of both the gauge and the potential datum between the two alternative views. In one the field is a self-
essentially arbitrary, and these are commonly given as grounds supporting entity created by the interaction between the E and
for rejecting Q and 151 as physical entitities. Reversing the roles B vectors, and the charges are secondary, whereas in the other
of B and A gives a very different view, since the Lorentz gauge the Q and 4 fields are no more than convenient numerical
acquires a particular significance in separating the Q and A measures of the charge interactions. These include the charges
energies, together with the corresponding radiation terms [SI, and currents on the walls, which become a part of a circuit
and the datum likewise acquires a meaningful role [I], so that whose capacitance and inductance parameters are distributed, in
the choice between the E, B and “circuit” models is a matter of accordance with the wavelength, but remain identifiable. Thus
practical convenience. This can be established only by the circuit description which emerges from the E, view as
examining examples, and it is appropriate to choose to one at something equivalent, but separate, becomes an integral part of
the other end of the frequency spectrum. the model in which Cp and A define both the fields, and the
What is commonly seen as the most important result of the relevant circuit parameters.
-
E, B model is the prediction of energy radiation, from antennas Another method of launching a wave is from a wire
and circuits operating at a sufficiently high frequency, and this carrying current in the transverse direction, and this transverse-
is also one of the most difficult areas for many engineers A, or, more briefly, A-mode, corresponds to the usual
wishing to understand field theory in applications such as transverse-electric, or H-mode. Since the wire is, effectively,
electromagnetic compatibility. Yet the underlying point is an short-circuited, it produces no Cp. Thus an appropriate term for
essentially simple one when expressed in terms of retardation, the alternative is the $-mode, usually referred to as the
since it is evident on inspection that the capacitances and transversemagnetic,or E-mode. The two solutions separate the
inductances acquire resistive components when the Cp and A + , A 4-vector into two parts, each with two components.
fields from one part of the circuit no longer amve at other Thus the waveguide modes are as easily separated and
parts in phase with the local charge and current sources. The derived in terms of the circuit parameters $ and A as in terms
treatment of the Cp field in this way is briefly discussed in [l], of E or B, and they provide an economical and mnenomic
and expanded elsewhere [ 5 ] . But it applies only to open-wire description. They are also easily interpreted physically. The A-
systems, not waveguides. mode transfers energy magnetically, or inductively, like a
Waveguide modes are one of the most important examples transformer operating at low frequency, for example, in which
of the use of the field vectors E and B (or H) to model the the energy transfer between the windings is completely
propagation conditions, and this traditional view is reflected in described by the transverse A. The Cp mode is comparable to a
the usual computational methods. But the modes can be capacitor operating at low frequency, in which the energy
described at least as simply in terms of Cp and A, and these transfer between the plates is due to +. We prefer A-modes in
variables are now being used for both modal and 3-dimensional waveguides for the same reasons as we prefer magnetic
solutions [ 11-14]. The results may be given in the usual 51, B machines. These two forms of energy transfer correspond to
(or H) form, but can also be interpreted directly in capacitance the well-known property of the Lorentz gauge in separating Q
and inductance terms, suggesting a useful alternative view. and A, so that they can be treated as if propagating
The Helmholtz equation independently of each other, and it is easily shown that this is
 1299

equivalent to the separate propagation of the p+/2 and 1 . u 2 The second term gives the contribution to d N d t due to the
energiesj51. When viewed in this way, the Hertz dipole, for motion, and this is the term used in (14) to calculate the force
+,
example, can be divided into sources of due to spheres at the which is caused by the change in momentum of a charge
two ends, together with a source of A due to the current in the moving through an A field. These are the forces at circuit
bridging link [1,5]. Its coupling with another dipole is limited bends, due to A. But the moving charge also “sees” a change
to the inductive term when the two are sufficiently remote, but in +, given by u.A, and this accounts for the forces between
by suitably orienting it in a waveguide either the capacitive parallel currents, commonly given in texts as the example of
(Le. +), or inductive (Le. A) component can be selectively the relativistic explanation of magnetism, although (1 1) shows
enhanced. that the whole of the A interaction is relativistic. Equation
(17) reduces (12) to its first two terms only, if we substitute
+
the and d g d t values “seen” by the moving charge (or, more
accurately, as assigned to it in the reference frame relative to
which it is in motion). The result is to describe all
electromagnetic interactions in +-like and A-like (Le.
capacitive and inductive) terms, in place of the E and B forces
in (13), and the examples given show some of the advantages
of this change.
Two further points emerge. Whereas the distinction
between the E and B entities, in (13), depends on the velocity
of the test charge by which they are observed, the distinction

Fig.2
‘01
Aharanov-Bohm effect
+
between the and A fields does not. Secondly, although the
two terms in (17), like the Y+ and aA/& components in (12),
may sometimes be equal but opposite, this does not imply
that they have no physical significance. It is precisely this
separation of opposite components [16] which underlies the
VIII. OTHER ASPECTS interaction shown in fig.2, just as the behaviour of
+
As was remarked in section 2, the appearance of and A as transformers,and many other electromagneticdevices, depends
on the separation of the 41 and A forces on stationary charges.
the “field measures in the Schrodinger equation supports the
circuit view, and the Aharonov-Bohm effects I151 show the The equilibrium condition E = 0 in a resistanceless conductor
practical consequences. Packets of charge moving along the tells us little that is useful about the behaviour of the device of
axis of a closed toroidal winding, carrying direct current, are which the conductor forms a part, because it does not separate
subjected to a different change of quantum phase from those the capacitive and inductive components of the interaction.
following a path around the outside (fig.2), because they are
subjected to different A fields. Since there is no B field outside IX. CONCLUSIONS
the toroid, it is generally assumed that classical theory predicts In expanding the role of field computation one essential
no interaction, although the A-field does. We see that the requirement is that all users have a clear understanding of what
source of the apparent inconsistency is contained within the E, is being computed. It has been suggested that the use of a
B description, since the motion of the charge along the axis circuit description of the electromagnetic properties and forces
generates both an A and a B field, and induces an emf in the has much to recommend it, when using general-purpose
packages computing 4 and A, since these can be interpreted as
+
toroid. Likewise a field [1,5] predicts an interaction in the
measures of the circuit parameters of capacitance and
absence of any local E, demonstrating the underlying point
that the effects of the E, B fields are not purely local, as is so inductance, familiar to all engineers. The circuit view of A is
more easily understood than is the concept of a magnetic
often assumed. The magnetic version has been examined more vector potential. The description of the A field in terms of
closely in [16]. The two interactions are due to changes in the
forces on charges is also essentially simpler than the
relevant capacitances and inductances, described by macroscopic
energy changes, which are the Same whether described by and + corrcsponding interpretation of B, and is more easily related to
A, or E and B (table 1), and the practical implications are not the currents which cause it. Propagation can be described and
calculated as a direct consequence of retardation, instead of the
confined to the quantum-mechanical level.
more difficult idea of a Poynting vector ExH.
The example shows that, when computing I$and A, their
direct physical interpretation in circuit terms is wellfounded, Although the circuit view is very familiar in terms of the
and, together with the other examples, shows that it is also stored energy density (p++L.A)/2. in place of &.D+H.B)/2, its
convenient and helpful. There are, of course, many more practical implications seem to have attracted little attention,
consequences of the change in view than can be examined here, other than by authors such as O’Rahilly [17]. It has been
but one requires comment. shown elsewhere to provide a general view, valid at all
Replacing (13) by the Maxwell equation (12), replaces E by frequencies, provided that other properties are interpreted in a
+
the separate effects of and A, but leaves the velocity term self-consistent way, so that inductance is “explained” by the
electromagnetic inertia of the charges themselves, instead of
expressed in terms of B (Le. curl . We may, however, fluxes in the empty spaces around them. Examples have been
remove the concept of B, as well as E, by expanding
uxYx4 = Y(u.A) - (u.DA
analysed to illustrate the way in which the computation of +
(17) and A can provide a clearer insight into the behaviour of
where operates only on A, not u; that is devices operating at both power and communication
V&.U= u . M (18) frequencies, than does the E, B description obtained by post-
processing. Indeed the attempt to “explain” the behaviour in
300

erms of flux may merely confuse and obscure an essentially I.Bardi and O.Biro, "An efficient finite-element formulation
,imple interaction, as in the electromagnetic railgun. without spurious modes for anisotropric waveguides', IEEE
Much attention has been directed to A, since it is the Trans. on Microwave Theory Tech., MTT-39 (7),pp. 1133-
1138, 1991
nagnetic fields which cause most difficulty, but the @ field Y.Aharonov and D.Bohm "Significance of electromagnetic
)lays an equally important role in most electromagnetic potentials in the quantum theory," Phys. Rev., ~01.115,pp.
levices, in a way which tends to be obscured by electric field 485-491, 1959
lescriptions in terms of E. The difference in view is shown by C.J.Carpenter "Electromagnetic energy exchanges due to
he equation for force on which Maxwell founded his theory, charges moving through constant, or zero, magnetic field,"
low generally used in the form of (13) to define the E and B IEE Pr0c.A vo1.138 (1). pp. 55-70, Jan. 1991
lectors, but more useful for numerical computation in the A.O'Rahilly "Electromagnetic theory: a critical examination
orm proposed by Maxwell. of fundamentals," Longmans, Green & Co. 1938;
republished by Dover, 1965
Taking the circuit view of the field makes the concept of
lux in vacuo unnecessary, and replaces the B vector in iron by
he electron spins (defining the magnetisation vector M),or by C. John Carpenter was born in 1927, and worked in
he equivalent currents. Electric flux in dielectrics is likewise industry both before, and after, graduating from the University
eplaced by the polarisation vector [ I ] . The supposed of London in 1948. This was followed by an M.Sc.(Eng.)
'mechanism" of interlinking fluxes, in vacuo, "explaining" degree obtained by part-time work on the numerical solution of
lropagation, is replaced by the assumption that @ and A are fields, using hand computation methods. His industrial
etarded, giving a much simpler view, and a more economical experience in manufacturing, research, and development led to
lescription of waveguide modes. Since the @, A interpretation a Lectureship in Electrical Machines at Imperial College in
)f what is meant by the "field" is in accordance with the 1953, where he developed an interest in both the fundamentals,
ichrodinger equation, it is, perhaps, surprising that it has not and practical application, of electromagnetic theory at power
more fully explored, and exploited. frequency. He acted as consultant for several firms on field
problems in large transformers, turbo-generators, and actuators.
REFERENCES His papers include several given prizes by the Institution of
C.J. Carpenter "Teaching electromagnetism in terms of the Electrical Engineers, including the Institution Premium, and
potentials instead of the 'Maxwell' equations" Digests of the this work was recognised by the award of the D.Sc.(Eng.)
Fifth Biennial IEEE Conf. on Electromagnetic Field degree of the University of London in 1979.
Computation, Claremont, CA, 1992, paper TOE2; Also IEEE A secondment in 1961 to the University of the West Indies,
Tram. Educ., Vol. 36, No. 2, May 1993. In press. in Trinidad, as Professor of Electrical Engineering, lasted 3
C.J. Carpenter "Electromagnetic energy and power in terms
of charges and potentials instead of fields", IEE Proc., A
years. It entailed setting up a department in a new Engineering
vo1.136(2), pp. 55-65, 1989 Faculty and seeing the first students through to graduation.
C.J.Carpenter "Comparison of the practical advantages of John finally left Imperial College in 1979, to take up other
alternative descriptions of electromagnetic momentum" IEE interests, and spent some years teaching technicians over a
Proc. A, vol. 136(3), pp. 101-113, 1989 wide range of subjects, including digital circuit techniques.
R.P.Feynman, R.B.Leighton, and MSands "The Feynman This led to a renewed interest in electromagnetism, and helped
lectures on physics" vol.11 Addison-Wesley 2nd ed., pp. 15- to clarify long-standing problems, leading to a series of papers
14, 1966. on field concepts and to the continuing development of an
C.J.Carpenter "Electromagnetic theory without electric flux" alternative approach. John is now attached to Bristol
IEE Proc. A vo1.139 (4), pp. 189-209, July 1992 University as a Visiting Fellow, having retired from full-time
A.P.French and J.R.Tessman "Displacement currents and
magnetic fields" Am.J.Phys., ~01.31,pp. 201 -204, 1963
employment in 1987.
J.C.Maxwell " A treatise on electricity and magnetism."
v01.2, Oxford University Press, 3rd ed., 1892 (reprinted
1955) chapts.7, 8 and 9
D.Rodger and P.J.Leonard "Modelling the electromagnetic
performance of moving rail-gun launchers using finite
elements" 6th Symp. on E.M.Launch Technology, Austin,
Texas, April 1992
J.E.Allen and T.V.Jones "Relativistic recoil and the railgun"
J. Appl. Phys., ~01.67,pp 18-21, Jan.1990.
V.G.Endean, Comments on ref.2 and reply, IEE Pr0c.A (in
press)
Kim, C.Y., Yu, S.D., Harrington, R.F., Ra, J.W., and Lee,
S.Y., "Computation of waveguide modes for waveguides of
arbitrary cross-section" IEE Pr0c.H vo1.137(2), pp. 145-
149, April 1990
C.F.Bryant, B.Dillon, C.R.I.Emson, J.Simkin, and
C.W.Trowbridge, "Solving high frequency problems using
the magnetic vector potential with the Lorentz gauge"
Compumag 1991 (IEEE Trans. MAG-27 1991)
C .R .I. Ems o n and C .W .T r o w br i dg e Fin i t e e 1em en t s
"

solutions to 3D waveguide problems using the magnetic

vector potential" Int. Conf. on Computation in
Electromagnetics, London, Nov.1991, IEE Conf. Pub.
no.350, pp 276-279 (and pp.280-283)