Computation of Cut off Frequencies of TE

Modes in waveguides of Arbitrary Cross

Section Using EFIE and MOM (Galerkin
Approach)

The electric field can be expressed as-
E j A e | = V

(1)
vector potential is

( ') ( , ') '
c
A J c G dc =
}

(2)
[Primed coordinate denote source region.]
And the scalar potential is:

1 1
( , ') '
c
JG dc
j
|
c e
= V
}

(3)
Where G is the two dimensional Greens function:
(2)
0
1
( , ') ( ' )
4
G H k
j
= (4)
H is the Henkel function of second kind, order zero.
R= ' is the distance between the source point
and the observation point.
(2)
0
( ') ( ) '
4
c
A J c H kR dc
j

=
}
(5)

(2)
0
1 1
( ) '
4
c
JH kR dc
j j
|
c e
= V
}

(6)
Boundary condition:


tan 0 E =
(7)
MOM IMPLEMENTATION
Expanding current in terms of expansion functions.
1

N
n n
n
J I J n
=
=

(8)
Where,
n
I = coefficient of nth basis function


n
J = nth basis function
n = unit vector in the direction of the vector basis
function.
Substituting J in vector and sacalt potential
expressions:
(2)
0
( ') ( ' ) '
4
c
A J c H k dc
j

=
}

(9)
( )
(2)
0
1
1
( ') ( ) '
4
N
n n
n
c
I J c H kR dc |
ec
=
' ' = V

}

(10)
Testing with:
( ) ( ) m
m
W c W c m =

(11)
Where m

is the unit vector in the direction of the
vector testing function.
( )
, 0 m W j A e | +V = (12)
(2)
0
1
( ) [ ( ') ( ) '
4
N
m
n n
n
c c
I W c J c H kR dc
e
=
+

} }
( )
{ }
(2)
0
1
( ) ' ]
4
n
c
J c H kR dc dc
ec
' ' V V
}

(13)

Therefore, matrix element equals
(2)
0
( ) ( ') ( ) '
4
m
mn n
c c
Z W c J c H kR dc dc
e
(
= +
(

} }

( )
{ }
(2)
0
1
( ) ( ) '
4
m
n
c c
W c J c H kR dc dc
ec
(
' ' V V
(

} }
(14)
The second term can be written as:
| | ( ) m
c
W c dc | V
}
(15)
The above integral can be evaluated using integration
by parts as:
( )
( )
0
( ) ( ) m m
c
W c W c dc | |
=
V
}
(16)
Where the first term is zero because the integral is a
closed contour integral making the upper and lower
limit same.
Thus eq () becomes-

( )
( )
1
( )
4
m
c
W c dc |
ec
(
V
(

}
(17)
Substituting back | into above equation gives-
( )
( )
( )
(2)
0
1
( ) ( ) '
4
m
n
c
W c J c H kR dc dc
ec
(
' ' V V
(

}
(18)

Thus, we get:
(2)
0
( ) ( ') ( ) '
4
m n
mn
c c
Z W c J c H kR dc dc
e
=
} }


( )
( )
( )
(2)
0
1
( ) ( ) '
4
m
n
c c
W c J c H kR dc dc
ec
' ' V V
} }
(19)
Where, the first term is the contribution of vector
potential A, while the second term is the contribution
of scalar potential phi.

Triangular basis and testing functions:
Node based triangular basis function is used which is
unity at the current node and decreases linearly to
zero at all the neighbouring nodes.
Mapping of segments of different lengths to a unit
length:
Before mapping:

nbr
m
m nbr
c c
W m
c c

,
( ) ,
nbr m
c c c e

(20)

After mapping:
(1 )
m
W u m = ,
( ) 0,1 ue (21a)
Similarly,
(1 )
n
J u n ' = ,
( ) 0,1 u' e

(21b)
( )
m
m
W
W m c
c
c
V =
c
(22)
This is a one dimensional or contour divergence.

(1 )
( 1)
( )
T
u
u
c
=
c A

(23)
1 m c = for first neighbour node of testing node
1 = + for rest.
(24)

1
m
T
W V =
A

(25a)
Similarly,
1
n
B
J V =
A
(25b)
Mapping equations are:
For basis functions:
B
dc du ' ' = A (26a)
For testing functions:
T
dc du = A
(26b)
Where,
B
A = basis segment length

T
A = testing segment length

( ) ( )
2 2
B nbr n nbr n
x x y y A = +

(27a)
( ) ( )
2 2
T nbr m nbr m
x x y y A = +

(27b)


cxm cym
m u x u y = +
(28a)


cxn cyn
n u x u y = +
(28b)

cxm xm
u u = for first neighbour of testing node.

xm
u = +

for other neighbours. (29a)

cym ym
u u = for first neighbour of testing node.

ym
u = +

for other neighbours. (29b)

cxn xn
u u = for first neighbour of basis node.

xn
u = + for other neighbours. (29c)

cyn yn
u u = for first neighbour of basis node.

yn
u = + for other neighbours. (29d)
nbr m
xm
B
x x
u

=
A

(30a)
nbr m
ym
B
y y
u

=
A
(30b)
nbr n
xn
T
x x
u

=
A

(30c)
nbr n
yn
T
y y
u

=
A
(30d)
m
x : x coordinate of basis node.
m
y : y coordinate of basis node.
n
x : x coordinate of testing node.
n
y : y coordinate of testing node.
nbr
x : x coordinate of neighbouring node of
basis/testing node.
nbr
y : y coordinate of neighbouring node of
basis/testing node.

The matrix element is:
1 1
(2)
0
0 0
(1 )(1 )( ) ( )
4
mn B T
Z u u m n H kR du du
e
' ' = A A
} }
1 1
(2)
0
0 0
1 1 1
( )( ) ( )
4
T B
T B
H kR du du
ec
| || |
' A A
| |
A A
\ .\ .
} }
(31)
Simplifying further:
Let
1
4
e
o = and ;
2
1
4
o
ec
=
Now, matrix element can be written as:
1 1
(2)
1 0
0 0
(1 )(1 )( ) ( )
mn B T
Z u u m n H kR du du o ' ' = A A
} }
1 1
(2)
2 0
0 0
( ) H kR du du o '
} }
(32)
The sign of the second term (due to scalar potential)
is decided as follows:
Negative sign : when first neighbour node of either
the testing node or the basis node is involved.
Positive sign: when both testing and basis nodes
first neighbours are involved and rest of the cases.


The dot product is found as:

( ) ( )
cxm xm cym ym
m n u u u u = +
(32)

( ) ( )
2
2
m B xm n T xn m B ym n T yn
R x u u x u u y u u y u u ' ' = + A A + + A A
(33)


Singularity of the Greens function:

The Hankel function of second kind is the linear
combination of Bessel function of first kind ( ) J x
v

and second kind ( ) N x
v
:
(2)
( ) ( ) ( ) H x J x jN x
v v v
=

( )
(2)
0 0 0
0 0
lim ( ) lim ( ) ( )
x x
H x J x jN x

=

2
1 ln
2
x
j

t
| |
=
|
\ .
(34)

Where 1.781 =

The singular integral will arise in the self term i.e.,
mm
Z and in adjacent terms also.
Since logarithmic singularity is integrable, it can be
handled by calculating the singular integrals
analytically or by the help of the Gaussian
logarithmic quadrature.

Self-Term:

m n =
m n
m n
x x
y y
=
=

1 m n =
T B
A = A
xm xn
u u =
ym yn
u u =
Substituting all above in equation (33):-
( ) ( )
2
2
T xn T xn T yn T yn
R u u u u u u u u ' ' = A A + A A

( ) ( )
2 2
T xn T yn
u u u u u u ' ' ( = A + A (



( )
2
2 2
T xn yn
u u u u ' ( = A + (


2 2
1
xn yn
u u + =

we get:
T
R u u' = A (35)
1 1
2
1
0 0
2
( ) (1 )(1 ) 1 ln
2
mm T
kR
Z u u j du du

o
t
| | | |
' ' = A
| |
\ . \ .
} }

1 1
2
0 0
2
1 ln
2
kR
j du du

o
t
| | | |
'
| |
\ . \ .
} }
(36)
The first term (the one due to vector potential) can be
simplified as follows:

( )
1 1
2
1
0 0
1 1
2
1
0 0
2
( ) (1 )(1 ) 1 ln
2
2
( ) (1 )(1 ) ln
mm
VP
T
T
T
Z
k
u u j du du
j
u u u u du du

o
t
o
t
=
| A | | |
' ' A
| |
\ . \ .
| |
' ' ' + A
|
\ .
} }
} }
(37)
The integral in the second term of equation (36) can
be broken into three integrals as follows:
1 1
01 03 11
0 0
(1 )(1 ) ln 2 u u u u du du I I I ' ' ' = +
} }
(38)
Where,
1 1
01
0 0
ln I u u du du ' ' =
} }
(39a)
1 1
03
0 0
ln I u u u du du ' ' ' =
} }
(39b)
1 1
11
0 0
ln I uu u u du du ' ' ' =
} }
(39c)

Rewriting equation (36) we get:
( )
| |
1 1
2
1
0 0
2
1 01 03 11
2
( ) 1 ln (1 )(1 )
2
2
( ) 2
mm
VP
T
T
T
Z
k
j u u du du
j
I I I

o
t
o
t
=
| A | | |
' ' A
| |
\ . \ .
| |
+ A +
|
\ .
} }
(40)
1 1
0 0
1
(1 )(1 )
4
I u u du du ' ' = =
} }
(41)
The integrals of equation (38) when calculated using
logarithmic quadrature [appendix A] give:

01
1.5 I = (42a)
03
0.75 I = (42b)
11
0.4375 I = (42c)
Therefore, equation (36) can be rewritten as:
( )
| |
2
1
2
1
2 1
( ) 1 ln
2 4
2
( ) 0.4375
mm
VP
T
T
T
Z
k
j
j

o
t
o
t
=
| A | | | (
A
| |
(
\ . \ .
| |
+ A
|
\ .

Or
( )
( )
2
1
1 2 2
( ) 1 ln 0.4375
4 2
mm
VP
T
T
Z
k j
j

o
t t
=
A ( | |
A
| (
\ .
(43)

The second term (due to scalar potential) can be
simplified as follows:

( )
1 1
2
0 0
1 1
2
0 0
2
1 ln
2
2
ln
T
mm
SP
k
Z j du du
j
u u du du

o
t
o
t
| A | | |
' =
| |
\ . \ .
| |
' ' +
|
\ .
} }
} }
(44)

Rewriting equation (36) we get:
( ) ( )
2 01
2 2
1 ln
2
T
mm
SP
k j
Z j I

o
t t
A ( | |
=
| (
\ .

(45)
( ) ( )
2
2 2
1 ln 1.5
2
T
mm
SP
k j
Z j

o
t t
A ( | |
=
| (
\ .
(46)

Adjacent term:
1 m n =

1
1
m n
m n
x x
y y

=
=

1 m n =
T B
A = A
xm xn
u u =
ym yn
u u =
Substituting all above in equation (33):-
( ) ( )
2
2
1 1
( ) ( )
n n T xn T xn n n T yn T yn
R x x u u u u y y u u u u

' ' = + A A + + A A

( ) ( )
2
2
T xn T xn T xn T yn T yn T yn
R u u u u u u u u u u ' ' = A A A + A A A

( ) ( )
2 2
1 1
T xn T yn
u u u u u u ' ' ( = A + A (



( )
2
2 2
1
T xn yn
u u u u ' ( = A + (


2 2
1
xn yn
u u + =

we get:
1
T
R u u' = A
(47)
1 1
2
1
0 0
2
( ) (1 )(1 ) 1 ln
2
mm T
kR
Z u u j du du

o
t
| | | |
' ' = A
| |
\ . \ .
} }

1 1
2
0 0
2
1 ln
2
kR
j du du

o
t
| | | |
'
| |
\ . \ .
} }
(48)
The first term (the one due to vector potential) can be
simplified as follows:

( )
1 1
2
1
0 0
1 1
2
1
0 0
2
( ) (1 )(1 ) 1 ln
2
2
( ) (1 )(1 ) ln 1
mm
VP
T
T
T
Z
k
u u j du du
j
u u u u du du

o
t
o
t
=
| A | | |
' ' A
| |
\ . \ .
| |
' ' ' + A
|
\ .
} }
} }
(49)
The integral in the second term of equation (36) can
be broken into three integrals as follows:
1 1
21 22 31
0 0
(1 )(1 ) ln 1 2 u u u u du du I I I ' ' ' = +
} }
(50)
Where,
1 1
21
0 0
ln 1 I u u du du ' ' =
} }
(51a)
1 1
22
0 0
ln 1 I u u u du du ' ' ' =
} }
(51b)
1 1
31
0 0
ln 1 I uu u u du du ' ' ' =
} }
(51c)

Rewriting equation (36) we get:
( )
| |
1 1
2
1
0 0
2
1 21 22 31
2
( ) 1 ln (1 )(1 )
2
2
( ) 2
mm
VP
T
T
T
Z
k
j u u du du
j
I I I

o
t
o
t
=
| A | | |
' ' A
| |
\ . \ .
| |
+ A +
|
\ .
} }
(52)
1 1
0 0
1
(1 )(1 )
4
I u u du du ' ' = =
} }
(53)
The integrals of equation (38) when calculated using
logarithmic quadrature [appendix A] give:

21
1.5 I = (54a)
22
0.75 I = (54b)
31
0.3125 I = (54c)
Therefore, equation (36) can be rewritten as:
( )
| |
2
1
2
1
2 1
( ) 1 ln
2 4
2
( ) 0.3125
mm
VP
T
T
T
Z
k
j
j

o
t
o
t
=
| A | | | (
A
| |
(
\ . \ .
| |
+ A
|
\ .

Or
( )
( )
2
1
1 2 2
( ) 1 ln 0.3125
4 2
mm
VP
T
T
Z
k j
j

o
t t
=
A ( | |
A
| (
\ .
(55)
The second term (due to scalar potential) can be
simplified as follows:

( )
1 1
2
0 0
1 1
2
0 0
2
1 ln
2
2
ln 1
mm
SP
T
Z
k
j du du
j
u u du du

o
t
o
t
=
| A | | |
'
| |
\ . \ .
| |
' ' +
|
\ .
} }
} }
(56)
Rewriting equation (36) we get:
( ) ( )
2 21
2 2
1 ln
2
T
mm
SP
k j
Z j I

o
t t
A ( | |
=
| (
\ .

(57)
( ) ( )
2
2 2
1 ln 1.5
2
T
mm
SP
k j
Z j

o
t t
A ( | |
=
| (
\ .
(58)

Result:

Case1: Rectangular Waveguide

Discretization density= 8 per wavelength.
Frequency step for plotting = 0.1 GHz.



2 2
1
2
cmn
m n
f
a b c
| | | |
= +
| |
\ . \ .


Frequency(GHz) from
expression
Frequency(GHz)
from from code

7.500 7.500
15.00 15.00
16.77 16.70
21.32 21.00

Case 2: L shaped Waveguide
Discretization density= 8 per wavelength.
Frequency step for plotting = 0.1 GHz.




Cut off k(in cm) in
reference
Cut off k(in cm)
from code

1.8917 1.927
2.9159 2.953
4.8755 4.964
5.2463 5.299

Appendix A
The abscissas and weight factors for Gaussian
integrands with logarithmic singularity [Ref] are as
follows:-
1
1
0
( ) ln ( )
n
i i
i
f x x dx w f x
=
=

}

n=2
abscissa weights
.11209 -0.718539
0.602277 -0.281461

1 1
01
0 0
ln I u u du du ' ' =
} }

(1)
1
0
0
ln I x x dx ' =
}

(2)
1
0
ln( ) ln( )
b
x
x
Ia I
x x dx x x dx ' ' ' ' = +
} }

(3)
Consider
a
I :-
y x x dy dx ' ' = =
0
ln
a
x
I y dy =
}

(4)
/ z y x =
1
0
ln( )
a
I x zx dz =
}

1
0
ln ln x x x z dz = +
}

(5)
Consider
b
I :-
y x x dy dx ' ' = =
1
0
ln
x
b
I y dy

=
}

(6)
/ z y x =
1
2
0
(1 ) ln(1 ) I x x z dz =
}

1
0
(1 ) ln(1 ) (1 ) ln x x x z dz = +
}

(7)
Substituting (7) & (5) into equation (3), we get:-

1
0
0
ln (1 ) ln(1 ) ln I x x x x xdx = + +
}

(8)
1 1
01
0 0
ln (1 ) ln(1 ) ln
Ic
I x x x x zdz dx
(
= + + (
(

} }
(9)

1
0
(1 ) ln(1 )
c
I x x dx =
}

let 1-x = y ; dy = -dx

1
0
ln
c
I y ydy =
}

Substituting in (9):
1 1
01
0 0
2 ln ln I x x zdz dx
(
= +
(

} }

1
01
0
(2 1) ln I x xdx = +
}

(10)
Using logarithmic quadrature gives
0
I = -1.5


1 1
03
0 0
ln I u u u du du ' ' ' =
} }
(11)
Using (2)
1
03 0
0
I u I du =
}

Using eq (8):
1 1
03
0 0
ln (1 ) ln(1 ) ln I x x x x x zdz dx
(
= + +
(

} }
(12)
1 1
2 2
0 0
ln ( ) ln(1 ) ln
d
I
x x x x x x zdz dx
(
(
= + +
(

} }

1 1 2
2 2
0 0
ln ( ) ln(1 ) ln
2
d
I
x
x x x x x dx zdz
(
(
= + +
(

} }
(13)
Consider
d
I :-
let 1-x = y ; dy = -dx
0
2
1
(1 (1 ) ) ln( )
d
I y y y dy =
}

1
2
0
( ) ln( ) y y y dy = +
}


Substituting in (13)
1
03
0
1
ln
2
I x xdx
(
= +
(

}

(14)

Using logarithmic quadrature gives
03
I = -0.75


1 1
11
0 0
ln I uu u u du du ' ' ' =
} }
(15)
1
1
0
ln I u u u du ' ' ' =
}

(16)
1
0
ln( ) ' ln( )
f
x
x
Ie I
x x x dx x x x dx ' ' ' ' ' = +
} }

(17)
Consider
e
I :-
y x x dy dx ' ' = =
0
( ) ln
e
x
I x y y dy =
}

(18)
/ z y x =
1
0
( ) ln( )
e
I x x zx zx dz =
}

1 1
2 2
0 0
(1 ) ln ln (1 )
e
I x z z dz x x z dz = +
} }
(19)
Consider
f
I :-

y x x dy dx ' ' = =
1
0
( ) ln
x
f
I x y y dy

= +
}

(20)
/ z y x =
1
0
(1 ) ( (1 )) ln(1 )
f
I x x z x x z dz = +
}

1
0
1
0
(1 ) ln(1 ) ( (1 ))
(1 ) ( (1 )) ln
x x z x z dz
x z x z z dz
= + +
+
}
}

(21)

Substituting (19) & (21) into equation (16), we get:-

1 2 2
1
0
1
0
1
ln ln(1 ) ln
2 2
(1 2 ) ln
x x
I x x z z dz
x z z dz

= + +
+
}
}
(22)

Using (16):
1
11 1
0
I u I du =
}

(23)

Using(22)& (23):

2 2
1
1 1 11
0
0 0
1
ln ln(1 )
2 2
ln (1 2 ) ln
x x
x x
I x dx
z zdz x z zdz
(
+
(
(
=
(
+ +
(

}
} }

(24)



1 3 3
0
1 1
1 1 2 3
0 0 0 0
ln ln(1 )
2 2
ln (1 2 ) ln
2 3
g
I
x x x
x x dx
x x
z zdz z zdz
(
(

= +
(
(

| |
+ +
|
\ .
}
} }

(25)

Consider
g
I :-
let 1-x = y ; dy = -dx
0 3
1
1 (1 )
ln( )
2
g
y y
I y dy

=
}
(26)

1
2
0
( ) ln( ) y y y dy = +
}

Substituting in (25)

1 3 3
11
0
1 (1 ) 1 2
ln
2 2 2 2
x x x x x
I xdx
(
= + + +
(

}


1
3 2
11
0
3 5 1
ln
2 6 3
I x x x xdx
(
= + + +
(

}
(27)
Using logarithmic quadrature gives
11
I = -0.4375

1 1
21
0 0
ln 1 I u u du du ' ' =
} }
(28)

1
2
0
ln 1 I x x dx ' =
}

(29)
1
21 2
0
I I du =
}
(30)
1 1
0 1
ln(1 ) ln( 1)
b
x
x
Ia I
x x dx x x dx

' ' ' ' = + +

} }

(31)
Consider
a
I :-
1 y x x dy dx ' ' = =
0
1
ln
a
x
I y dy

=
}

(32)
/1 z y x =
| |
1
0
(1 ) ln(1 ) ln
a
I x x z dz = +
}

1
0
(1 ) ln(1 ) (1 ) ln x x x z dz = +
}

(33)
Consider
b
I :-
1 y x x dy dx ' ' = + =
0
ln
x
b
I y dy =
}

(34)
/ z y x =

1
2
0
ln ln I x x x z dz = +
}
(35)
Substituting (33) & (35) into equation (31), we get:-
1
2
0
ln (1 ) ln(1 ) ln I x x x x xdx = + +
}
(36)

Which is equal to
0
I

Therefore from eq (30) & (9)

21 01
1.5 I I = = (37)

1 1
22
0 0
ln 1 I u u u du du ' ' ' =
} }
(38)

Using (29)
1
22 2
0
I u I du =
}
(39)
Using eq (36) in above:
1 1
22
0 0
ln (1 ) ln(1 ) ln I x x x x x zdz dx
(
= + +
(

} }
(40)
Which is same as
03
I

Therefore
22 03
I I = = -0.75 (41)


1 1
31
0 0
ln 1 I uu u u du du ' ' ' =
} }
(42)

1
3
0
ln 1 I u u u du ' ' ' =
}

(43)
1 1
0 1
ln(1 ) ' ln( 1)
f
x
x
Ie I
x x x dx x x x dx

' ' ' ' ' = + +

} }
(44)

Consider
e
I :-

1 y x x dy dx ' ' = =
0
1
(1 ) ln
e
x
I x y y dy

=
}

(45)
/(1 ) z y x =
1
0
(1 ) (1 (1 )) ln( (1 ))
e
I x x z x z x dz =
}
1
0
1
0
ln(1 ) (1 (1 ))
(1 )
(1 (1 )) ln
e
x x z x dz
I x
x z x zdz
(
+
(
(
=
(
(
(

}
}



1 1
2
0 0
ln(1 )
(1 ) ln ln
2
e
x
I x zdz z zdz
(

= +
(

} }

(46)
Consider
f
I :-

1 y x x dy dx ' ' = + =
0
(1 ) ln
x
f
I x y y dy = +
}

(47)
/ z y x =
1 1
0 0
1 ln ln (1 ) ln
2
f
x
I x x x z z x z zdz
(

= + +
` (
)

} }
(48)

Substituting (48) & (46) into equation (43), we get:-


1 2
2
0
1
2
3
0
1 1
2
0 0
(1 )
ln(1 ) (1 ) ln
2
(1 ) ln 1 ln
2
ln (1 ) ln
x
x x zdz
x
I x z zdz x x
x z zdz x x zdz
(

+
(
(
(
| |
( = + +
|
( \ .
(
(
+
(

}
}
} }

(49)

Using (42) & (43) :
1
31 3
0
I uI du =
}

(50)

Using(49)& (50):

1 1 1 2
2
0 0 0
1 1 1
2 2
31
0 0 0
1 1 1 1
3 2
0 0 0 0
(1 )
ln(1 ) (1 ) ln
2
(1 ) ln 1 ln
2
ln (1 ) ln
g
I
x x
x dx x x dx zdz
x
I x x dx z zdz x x
x dx z zdz x x zdz
(

+
(
(
(
(
| |
(
= + +
|
(
\ .
(
(
+
(
(
(

} } }
} } }
} } } }

(51)

Consider
g
I :-
let 1-x = y ; dy = -dx

0 2
1
(1 )
ln( )
2
g
y y
I y dy

=
}
(52)

1
3 2
0
1
( ) ln( )
2
y y y dy = +
}


Substituting in (51)

1 3 3
11
0
1 (1 ) 1 2
ln
2 2 2 2
x x x x x
I xdx
(
= + + +
(

}


1
3 2
31
0
3 1 1
ln
2 6 6
I x x x xdx
(
= + + +
(

}

(53)

Using logarithmic quadrature gives

11
I = -0.3125 (54)