4.

3 Development or the Potential Function 71

To simplify writing. 1/41trE is frequently replaced by a potential

coefficient Pr

cp(r) = Pr Q
(4-26)

Evaluation of the potential becomes more demanding when the

charge's position does not coincide with the origin of the
coordinate system. Nevertheless, one can visualize again a
spherical coordinate system, whose origin coincides with the
position of the charge, and calculate the field of the point
charge with respect to a radius originating from this new origin.
This radius can be represented as the magnitude of the
difference of the two position vectors rand rq, Figure 4.3.

lP (r) = lP (x,Y,Z)

Y
Ir -rql =Y(X_x q)2+(y_y q)2+(Z_Zq)2'
x

Figure 4.3: Calculating the potential function of a point charge Q at a

field point P(rq).

Thus, we obtain the potential function of a point charge at P(rq)

cp(r) = Q or cp(r) = Prrq Q

41tE Ir - rq I
(4-27)
72 4 Gradient, Potential, Potential Function

Expressing both position vectors rq and r by their components

in the respective coordinate system gives the potential function
a more famiUar analytical form. Vectors are known to be added
or subtracted. resp .. by adding or subtracting their components.
For instance. in a Cartesian coordinate system one obtains with

Q
<p(x.y.z) = .
41tE V(X-Xq) 2 + (y-yq) 2 + (Z-Zq) 2
~------------------------------. (4-28)

The coordinate system is always chosen so as to yield the

simplest possible expressions (see A 3).

For several charges at different positions. the potential function

of each point charge may be separately calculated irrespective
of the existence of other charges. Then. the elementary po-
tential functions are superimposed to give a global potential
function integrating all sources in the region considered
(superposition principle)

<per) = Q1 + Q2 + ... + _......;::;Q:.l.l,"_ _ (4-29)

41tE'r- rq l ' 41tE' r-rQ2' 41tE' r-rQn '

<per) = <p l(r) + <P2(r) + ... + <Pn(r) ."

(4-30)

4.3.1 Potential Function of a Line Charge

Let PL be a line charge of finite length with uniform charge
density.
4.3 Development or the Potential Function 73

. ~Q dQ
PL = lim -=- (4-31)
~L dL
~o

as depicted in Figure 4.4.

z
lfJ (r)=lfJ(x,y,zl

Figure 4.4: Evaluation of the potential function of a line charge with

charge density PL(rq); the position vector rq is the
integration variable.

Interpreting an infmitesimally short line element dL as a point

charge, its potential function is given in a similar manner as in
the previous ~hapter

_...;;d.;;.;:Q___ = PL(rq)dL
d<p = (4-32)
41t£ I r-rq I 41t£ I r-rq I

Summing up the potential functions of all infinitesimal point

charges via an integratlon yields the potentlal function of a line
charge

= 1
41t£
J PL(rq) dL
I r-rq I
(4-33)
74 4 Gradient, Potential, Potential Function

For the' solution of this integral one has to specify a particular

coordinate system. Consider Cartesian coordinates and let us
place the line charge in the z-axis symmetrically about the
origin. Figure 4.5 a.

i(x-OJ 2+(y-OJ r
Z
Z

=::/~'I[Irl= I ',y,z I
I
I
'P
'P. (r,,!,,z)
L
I Y
Y

Z1
a) b)

Figure 4.5: Examples of line charges pdz)=const. for Cartesian

coordinates (a) and cylindrical coordinates (b).

Denotlng the integration variable on the z-axis as Zq (xq=O, yq=

0), the potential function is given by

1
cpdx.y.z) = - dzq , (4-34)
41t£

or upon evaluatlng the integral,

cpd x.y.z) = PL In z2 - z +, x 2 + y2 + (Z-Z2) 2

41t£ Zl-Z~x2+y2 + (z-zIl2
. (4-35)
4.3 Development or the Potential Function 7S

Usually. line charges are specified in cylindrical coordinates.

Hence. for the more general arrangement depicted in Figure
4.5b. one obtains

PL [ z2 - Z + -Jr2 + (z - Z2) 2 ]
cpdr. cp.z) = - In
41t£ Zl - z + -Jr2 + (z - Zl)2
(4-36)

Similar equations exist for circular line charges referred to as

ring charges. Mathematically they are more demanding because
of the elliptic integrals involved.

4.3.2 Potential Function of a General

Charge Configuration

In much the same way as in the previous section. the potential

function of a surface charge element with density Ps is given by

ps(rq)dS
dcp="';""';::;--- (4-37)
41t£ I r-rq I

and the potential function of a three-dimensional volume charge

element with charge density pv by

pv(rq)dV
dcp = - - ' - - - - (4-38)
41t£ I r-rq I

The most general case of a charge configuration consisting of

point charges as well as line. surface. and volume charge
densities. is shown in Figure 4.6.
 4 Gradient, Potential, Potential Function

Figure 4.6: Superposition of elementary potentials to a global po-

tential function cp(r); integration variable rq.

By superposition the global potential function is obtained as

(4-39)

It is to be noted that most technical problems differ from the

problem formulation above in that, usually, Qne does not deal
with isolated charges or charge density distributions in an
unbounded space, but with charges on electrodes whose surface
charge distributions are unknown. These problems are referred
to as boundary-value poblems. In this case, one needs to solve
the potential equation presented in the next section.

We have paid much attention to the potential functions of

elementary charge types because the potential equation of
technical problems frequently requires a numerical solUtion,
and because some modem computer-aided field calculation
methods, for example charge simulation and boundary-element
4.3 Development or the Potential Function 77

method ect. are based on the superposition of elementary po-

tentialjunctions of various types of image charges (see 9).

4.4 Potential Equations

Apart from the superposition of elementary potential functions

of discrete. infinitesimal point charges. a field-s potential func-
tion can be evaluated by solving the potential equation V2<p = O.
subject to the field region-s characteristic boundary conditions.
In the following sections we will derive the potential equations
for electric fields with and without space charges (Laplace and
Poisson fields).

4.4.1 Potential Equations for Fields

without Space Charges

As was shown in Section 3.2.4. the field region surrounding

charges obeys

div D =0 (4-40)

Together with the constitutive equation D = EE and the relation

E = -grad cpo we obtain

div EE =0 or div grad cp =0 (4-41)

For instance. for a scalar field cp(x.y.z) in a Cartesian coordinate

system the result of the gradient operation yields

dCP dCP dCP (4-42)

grad cp(x.y.z) = dX ax + dy ay + dZ az
78 4 Gradient, Potential, Potential Function

Evaluating this vector field's divergence according to Section

3.2.4, gives

(4-43)

Considering a va v = 1 and discarding the zero terms we obtain

()2<p ()2<p ()2<p

div grad <p = V· V<p = V2<p = A<p = ()x2 + dy2 + ()z2 = 0 . (4-44)

The operator V2= A is called Laplacian or del squared. Other

coordinate systems yield similar equations (see A 3).

The differential equation V2<p = 0 is called the Laplace equation.

Its solutions <p(x,y,z) are called harmonic potentials. They
generate so-called Laplace fields; these are charge-free source
field regions.

For a given field the solution of Laplace's equation is obtained

taking into account the field's characteristic boundary condi-
tions. One distinguishes between

- boundary-value problems of the first type (Dirichlet problem):

The solution <p(x,y,z) must attain certain given potentials on

the boundaries (electrode contours).

- boundary-value problems of the second type (Neumann

problem):

The first derivative of the solution <p(x,y,z) must attain certain

given values on the boundary.

In general, an analytic solution of Laplace's equation for given

boundary conditions is only manageable for simple problems
4.4 Potential Equations 79

with a high degree of symmetry, e.g., the electric field of a pa-

rallel-plate capacitor without fringe effects, Figure 4.7.

'P (0) =V

v ..

Figure 4.7: Parallel-plate capacitor. fringing fields neglected. e.g.

electrode of a guard-ring capacitor.

In the uniform field of a parallel-plate capacitor the field

quantities do not vary in y and z (dcp/dz = 0, dcp/dy = 0). There-
fore, Laplace's equation reduces to

(4-45)

Integrating d 2 cp/dx2 once yields

(4-46)

integrating once more

(4-47)

The integration constants are determined exploiting the known

boundary values
80 4 Gradient, Potential, Potential Function

(4-48)

Hence. we obtain as solution

v
<p(x) = - (fX+V (4-49)

Taking the gradient.

d<p
E(x) = - grad m
T
= - -dx a ~ (4-50)

we obtain for the x-component of the electric field strength

Ex = - grad <p(x) = - d<p V

dx = d = const. (4-51)

This result is not exactly surprising and could have been found
as well without direct integration of Laplace's equation. Hence.
this example serves only to demonstrate the basic procedure.
For a two-dimensional problem the solution effort rapidly in-
creases. Regarding three-dimensional real-world problems
without symmetry and various dIelectrics. for example the
interior of a high-voltage power transfonner or:three-phase gas-
insulated switch-gear. only numerical solution methods em-
ploying powerful software· and digital computers can cope.

4.4.2 Potential Equations for

Electric Fields with Space Charges

As was shown in Section 3.2.3. electric fields with space

charges obey