EECS 117

Lecture 17: Magnetic Forces/Torque, Faradays Law

Prof. Niknejad
University of California, Berkeley
University of California, Berkeley EECS 117 Lecture 17 p. 1/??
Memory Aid
The following table is a useful way to remember the
equations in magnetics. We can draw a very good
analogy between the elds
a
E H J
D B V A

1

P M
a
I personally dont like this choice since to me E and B are real and so the
equations should be arranged to magnify this analogy. Unfortunately the equations
are not organized this way (partly due to choice of units) so well stick with conven-
tion
University of California, Berkeley EECS 117 Lecture 17 p. 2/??
Boundary Conditions for Mag Field
We have now established the following equations for a
static magnetic eld
H = J
B = 0
_
C
H d =
_
S
J dS = I
_
S
B dS = 0
And for linear materials, we nd that H =
1
B
University of California, Berkeley EECS 117 Lecture 17 p. 3/??
Tangential H
The appropriate boundary conditions follow
immediately from our previously established techniques

2
C
Take a small loop intersecting with the boundary and
take the limit as the loop becomes tiny
_
C
H d = (H
t1
H
t2
)d = 0
University of California, Berkeley EECS 117 Lecture 17 p. 4/??
Tangential H (cont)
So the tangential component of H is continuous
H
t1
= H
t2

1
1
B
t1
=
1
2
B
t2
Note that B is discontinuous because there is an
effective surface current due to the change in
permeability. Since B is real, it reects this change
If, in addition, a surface current is owing in between
the regions, then we need to include it in the above
calculation
University of California, Berkeley EECS 117 Lecture 17 p. 5/??
Normal B

2
S
n
Consider a pillbox cylinder enclosing the boundary
between the layers
In the limit that the pillbox becomes small, we have
_
B dS = (B
1n
B
2n
)dS = 0
And thus the normal component of B is continuous
B
1n
= B
2n
University of California, Berkeley EECS 117 Lecture 17 p. 6/??
Boundary Conditions for a Conductor
If a material is a very good conductor, then well show
that it can only support current at the surface of the
conductor.
In fact, for an ideal conductor, the current lies entirely
on the surface and its a true surface current
In such a case the current enclosed by even an
innitesimal loop is nite
_
C
H d = (H
t1
H
t2
)d = J
s
d H
t1
H
t2
= J
s
This can be expressed compactly as
n (H
1
H
2
) = J
s
But for a perfect conductor, well see that H
2
= 0, so
H
1t
= J
s
n H
1
= J
s
University of California, Berkeley EECS 117 Lecture 17 p. 7/??
Hall Effect
J
B
0
V
0
When current is traveling through a conductor, at any
instant it experiences a force given by the Lorentz
equation
F = qE +qv B
The force qE leads to conduction along the length of the
bar (due to momentum relaxation) with average speed
v
d
but the magnetic eld causes a downward deection
F = q xE
0
q yv
d
B
0
University of California, Berkeley EECS 117 Lecture 17 p. 8/??
Hall Effect: Vertical Internal Field
+ + + + + + +
+ + + + + + +
+
In steady-state, the movement of charge down (or
electrons up) creates an internal electric eld which
must balance the downward pull
Thus we expect a Hall voltage to develop across the
top and bottom faces of the conducting bar
V
H
= E
y
d = v
d
B
0
d
University of California, Berkeley EECS 117 Lecture 17 p. 9/??
Hall Effect: Density of Carriers (I)
J
B
0
V
H
J
Since v
d
= E
x
, and J
x
= E
x
, we can write v
d
= J
x
/
V
H
=
J
x

B
0
d
Recall that the conductivity of a material is given by
= qN, where q is the unit charge
Since v
d
= E
x
, and J
x
= E
x
, we can write v
d
= J
x
/
V
H
=
J
x

B
0
d
University of California, Berkeley EECS 117 Lecture 17 p. 10/??
Hall Effect: Density of Carriers (II)
Recall that the conductivity of a material is given by
= qN, where q is the unit charge, N is the density of
mobile charge carriers, and is the mobility of the
carriers
V
H
=
J
x
B
0
d
qN
N =
J
x
B
0
d
qV
H
=
IB
0
d
AqV
H
Notice that all the quantities on the RHS are either
known or easily measured. Thus the density of carriers
can be measured indirectly through measuring the Hall
Voltage
University of California, Berkeley EECS 117 Lecture 17 p. 11/??
Forces on Current Loops
F
F
1
F
2
B
B
x
y
z
a
b
I
1
I
2
F
Since the eld is not
uniform, the net force
is not zero. Note the
force on the sides
cancel out
F
1
= y
I
1
I
2
2a
d
F
2
= + y
I
1
I
2
2(a +b)
d
F = F
1
+F
2
= y
I
1
I
2
d
2
_
1
a

1
b
_
University of California, Berkeley EECS 117 Lecture 17 p. 12/??
Torques on Current Loops (I)
B
0
B
0
free rotation
about this axis
force down
force up
In a uniform eld, the
net force on the cur-
rent loop is zero. But
the net torque is not
zero. Thus the loop
will tend to rotate.
T = r F
F
1
= I
1
B
0
d x
F
2
= +I
1
B
0
d x
F
1
+F
2
= 0
University of California, Berkeley EECS 117 Lecture 17 p. 13/??
Torques on Current Loops (II)
F
1
F
2
I
I
B
0

x
y
z
T
1
= (b/2)I
1
B
0
d sin z
T
2
= (b/2)I
1
B
0
d sin z
T = T
1
+T
2
= zB
0
moment
..
I b d
. .
Area of loop
sin
In general the torque can be expressed as
T = mB
where the moment is dened as m = I Area
University of California, Berkeley EECS 117 Lecture 17 p. 14/??
Electric Motors
I
B
0
A DC electric mo-
tor operates on this
principle. A uniform
strong magnetic eld
cuts across a current
loop causing it to ro-
tate.
When the loop is || to the eld, the torque drops to zero
but the rotational inertia of the loop keeps it rotating.
Simultaneously, the direction of the current is reversed
as the loop ips around and cuts into the eld. This
generates a new torque that favors the continuous
rotation.
University of California, Berkeley EECS 117 Lecture 17 p. 15/??
Faradays Big Discovery
In electrostatics we learned that
_
E d = 0
Lets use the analogy between B and D (and E and H).
Since q = Cv and = Li, and i = q = C v, should we not
expect that

= L

i = v?
energy of field
converted to heat!
B
t
R
In fact, this is true!
Faraday was able to
show this experimentally
_
C
Ed =
d
dt
=
d
dt
_
S
BdS
The force is no longer
conservative, E =
University of California, Berkeley EECS 117 Lecture 17 p. 16/??
Faradays Law in Differential Form
Using Stokes Theorem
_
C
E d =
_
S
E dS =
d
dt
_
S
B dS =
_
S
B
t
dS
Since this is true for any arbitrary curve C, this implies
that
E =
B
t
Faradays law is true for any region of space, including
free space.
In particular, if C is bounded by an actual loop of wire,
then the ux cutting this loop will induce a voltage
around the loop.
University of California, Berkeley EECS 117 Lecture 17 p. 17/??
Example: Transformers
V
1
I
1
V
2
I
2
+

L
1
L
2
M
B
+
V
1

+
V
2

In a transformer, by denition the ux in the primary

side is given by
1
= L
1
I
1
Likewise, the ux crossing the secondary is given by

2
= M
21
I
1
= M
12
I
1
= MI
1
(assuming I
2
= 0)
Thus if the current in the primary changes, a voltage is
induced in the secondary
V
2
=

2
= M

I
1
University of California, Berkeley EECS 117 Lecture 17 p. 18/??
Generating Sparks!
+
+
V
2

I
1
I
1
V
2
t
t
large voltage!
large slope
Since the voltage at the secondary is proportional to the
rate of change of current in loop 1, we can generate
very large voltages at the secondary by interrupting the
current with a switch
University of California, Berkeley EECS 117 Lecture 17 p. 19/??
Vector Potential
Since B 0, we can write B = A. Thus
E =
B
t
=
A
t
=
A
t
If we group terms we have

_
E+
A
t
_
= 0
So, as we saw in electrostatics, we can likewise write
E +
A
t
=
University of California, Berkeley EECS 117 Lecture 17 p. 20/??
More on Vector Potential
We choose a negative sign for to be consistent with
electrostatics. Since if

t
= 0, this equation breaks
down to the electrostatic case and then we identify as
the scalar potential.
This gives us some insight into the electromagnetic
response as
E =
A
t
E =
. .
electric response

A
t
. .
magnetic response
In reality the EM elds are linked so this viewpoint is not
entirely correct.
University of California, Berkeley EECS 117 Lecture 17 p. 21/??
Is the vector potential real?
We can now re-derive Faradays law as follows
V =
_
C
E d =
_
C
d

t
_
C
A d
The line integral involving is zero by denition so we
have the induced emf equal to the line integral of A
around the loop in question
V =

t
_
C
A d
We also found that equivalently
V =

t
_
S
B dS
University of California, Berkeley EECS 117 Lecture 17 p. 22/??
The Reality of the Vector Potential
V =

t
_
C
A d
This equation is somewhat more satisfying that
Faradays law in terms of the ux. Although its
mathematically equivalent, it explicitly shows us the
shape of the loops role in determining the induced ux.
The ux equation, though, depends on a surface
bounding the loop, in fact any surface. Sometimes its
even difcult to imagine the shape of such a surface
(e.g. a coil)
University of California, Berkeley EECS 117 Lecture 17 p. 23/??
Solenoid Transformer
+
V
2

I
1
I
1
Consider the magnetic coupling between a solenoid
and a large loop surrounding the solenoid.
We found that for an ideal solenoid, B = 0 outside of the
cylinder. Certainly we can assume that B 0 outside of
this region.
University of California, Berkeley EECS 117 Lecture 17 p. 24/??
Vector Potential Outside of Solenoid
Then the voltage induced into the outer loop only
depends on the constant ux generated within the
center section coincident with the solenoid
Whats disturbing is that even though B = 0 along the
loop, there is a force pushing electrons inside the outer
metal.
The force is therefore not magnetic since B = 0.
The viewpoint with vector potential, though, does not
pose any problems since A = 0 outside of the loop.
Therefore when we integrate A outside of the loop,
there is a nonzero result.
University of California, Berkeley EECS 117 Lecture 17 p. 25/??
Circuit Application of Transformers
V
+
2

+
V
1

The transformer is a
very important circuit
element
Before switching power
supplies, transformers
were ubiquitous in
voltage/current
transformation
applications (taking wall
voltage of say 120V and
converting it to say 3V).
In fact, the name trans-
former comes from this
very application
University of California, Berkeley EECS 117 Lecture 17 p. 26/??
Voltage Transformer
V
1
= L
1
dI
1
dt
+M
dI
2
dt
V
2
= M
dI
1
dt
+L
2
dI
2
dt
If I
2
0, or for a light load on the secondary, we have
V
1
= L
1
dI
1
dt
V
2
= M
dI
1
dt
V
2
V
1
=
M
L
1
= k

L
1
L
2
L
1
= k
_
L
2
L
1
= n
University of California, Berkeley EECS 117 Lecture 17 p. 27/??
Power Transmission
Transformers are also used to boost the voltage for long
range power transmission
This follows since power loss proportional to I
2
R, so to
transmit a given power P, its best to use the largest
voltage to minimize the current I = P/V .
This is the reason we use AC power versus DC, since
transformers dont work with DC!
University of California, Berkeley EECS 117 Lecture 17 p. 28/??
Summary So Far...
Lets summarize what weve learned thus far. There are
no magnetic charges, so
B = 0
and electric elds diverge on physical charge
D =
Faradays laws tell us that
E =
B
t
and Ampres law relate magnetic elds to currents by
H = J
University of California, Berkeley EECS 117 Lecture 17 p. 29/??
Are These Equations Complete?
Are these equations complete and self-consistent? In
other words, do they over-specify the problem or are
some equations still missing? Furthermore, are they
self-consistent?
Mathematics tells us that (H) = 0, which implies
that
J = 0
But this can only hold for steady elds. In general, by
conservation of charge we know that
J =

t
University of California, Berkeley EECS 117 Lecture 17 p. 30/??
Maxwells Displacement Current
In other words, we have to add something to the RHS of
Ampres eq. to make it self-consistent!
Maxwell was the rst to make this observation. Since
D = , its natural to add a displacement current to
the Ampres eq.
H = J +
D
t
This now makes our eq. self-consistent since
H = 0 = J +
D
t
H = 0 = J +
D
t
= J +

t
University of California, Berkeley EECS 117 Lecture 17 p. 31/??
Magnetic Field of a Capacitor
I
S
D
t
R
V

S
J dS = 0
S

J dS = I
C
1
Now we can resolve a contradiction in Ampres eq. If
we consider the magnetic eld of the following circuit,
we know that there is a magnetic eld around loop C
1
since current cuts through surface S
1
But Ampres law says that any surface bounded by C
1
can be used to calculate the magnetic eld. If we use
surface S
2
, then the current cutting through this surface
is zero, which would yield a zero magnetic eld!
University of California, Berkeley EECS 117 Lecture 17 p. 32/??
Displacement Current of a Capacitor
The answer to this contradiction is displacement
current. If current is owing in this circuit, then the
electric eld between the capacitor plates must be
changing. Thus
D
t
= 0
So the displacement current cutting surface S
2
must be
the same as the conductive current cutting through
surface S
1
_
S
1
J
c
dS =
_
S
2
D
t
dS
University of California, Berkeley EECS 117 Lecture 17 p. 33/??