3/1/2005 Filters.
doc 1/8
Filters
A RF/microwave filter is (typically) a passive, reciprocal, 2-
port linear device.
Pinc Pout
Filter
If port 2 of this device is terminated in a matched load, then
we can relate the incident and output power as:
2
Pout = S21 Pinc
We define this power transmission through a filter in terms
of the power transmission coefficient T:
Pout 2
T = S21
Pinc
Since microwave filters are typically passive, we find that:
0 ≤ Τ ≤1
in other words, Pout ≤ Pinc .
Jim Stiles The Univ. of Kansas Dept. of EECS
3/1/2005 Filters.doc 2/8
Q: What happens to the “missing” power Pinc − Pout ?
A: Two possibilities: the power is either absorbed (Pabs) by
the filter (converted to heat), or is reflected (Pr) at the input
port.
I.E.:
Pabs
Pinc Pout
Filter
Pr
Thus, by conservation of energy:
Pinc = Pr + Pabs + Pout
Now ideally, a microwave filter is lossless, therefore Pabs = 0
and:
Pinc = Pr + Pout
which alternatively can be written as:
Pinc Pr + Pout
=
Pinc Pinc
P P
1 = r + out
Pinc Pinc
Jim Stiles The Univ. of Kansas Dept. of EECS
3/1/2005 Filters.doc 3/8
Recall that Pout Pinc = Τ , and we can likewise define Pr Pinc as
the power reflection coefficient:
Pr 2
Γ = S11
Pinc
We again emphasize that the filter output port is terminated
in a matched load.
Thus, we can conclude that for a lossless filter:
1=Γ+Τ
Which is simply another way of saying for a lossless device
2 2
that 1 = S11 + S21 .
Now, here’s the important part!
For a microwave filter, the coefficients Γ and Τ are
functions of frequency! I.E.,:
Γ (ω ) and Τ (ω )
The behavior of a microwave filter is described by these
functions!
Jim Stiles The Univ. of Kansas Dept. of EECS
3/1/2005 Filters.doc 4/8
We find that for most signal frequencies ωs , these functions
will have a value equal to one of two different approximate
values.
Either:
Γ (ω = ω s ) ≈ 0 and Τ (ω = ω s ) ≈ 1
or
Γ (ω = ω s ) ≈ 1 and Τ (ω = ω s ) ≈ 0
In the first case, the signal frequency ωs is said to lie in the
pass-band of the filter. Almost all of the incident signal
power will pass through the filter.
In the second case, the signal frequency ωs is said to lie in
the stop-band of the filter. Almost all of the incident signal
power will be reflected at the input—almost no power will
appear at the filter output.
Jim Stiles The Univ. of Kansas Dept. of EECS
3/1/2005 Filters.doc 5/8
Consider then these four types of functions of Γ (ω ) and
Τ (ω ) :
1. Low-Pass Filter
Τ (ω ) Γ (ω )
1 1
ω ω
ωc ωc
Note for this filter:
⎧≈1 ω < ωc ⎧≈ 0 ω < ωc
⎪ ⎪
Τ (ω ) = ⎨ Γ (ω ) = ⎨
⎪≈ 0 ω > ωc ⎪≈ 1 ω > ωc
⎩ ⎩
This filter is a low-pass type, as it “passes” signals with
frequencies less than ωc , while “rejecting” signals at
frequencies greater than ωc .
Q: This frequency ωc seems to
be very important! What is it?
Jim Stiles The Univ. of Kansas Dept. of EECS
3/1/2005 Filters.doc 6/8
A: Frequency ωc is a filter parameter known as the cutoff
frequency; a value that approximately defines the frequency
region where the filter pass-band transitions into the filter
stop band.
According, this frequency is defined as the frequency where
the power transmission coefficient is equal to ½:
Τ (ω = ωc ) = 0.5
Note for a lossless filter, the cutoff frequency is likewise
the value where the power reflection coefficient is ½:
Γ (ω = ωc ) = 0.5
2. High-Pass Filter
Τ (ω ) Γ (ω )
1 1
ω ω
ωc ωc
Jim Stiles The Univ. of Kansas Dept. of EECS
3/1/2005 Filters.doc 7/8
Note for this filter:
⎧≈ 0 ω < ωc ⎧≈1 ω < ωc
⎪ ⎪
Τ (ω ) = ⎨ Γ (ω ) = ⎨
⎪≈ 1 ω > ωc ⎪≈ 0 ω > ωc
⎩ ⎩
This filter is a high-pass type, as it “passes” signals with
frequencies greater than ωc , while “rejecting” signals at
frequencies less than ωc .
3. Band-Pass Filter
Τ (ω ) Γ (ω )
1 1
∆ω ∆ω
ω ω
ω0 ω0
Note for this filter:
⎧ ≈ 1 ω − ω 0 < ∆ω 2 ⎧≈ 0 ω − ω 0 < ∆ω 2
⎪ ⎪
Τ (ω ) = ⎨ Γ (ω ) = ⎨
⎪ ≈ 0 ω − ω > ∆ω 2 ⎪≈ 1 ω − ω 0 < ∆ω 2
⎩ 0 ⎩
Jim Stiles The Univ. of Kansas Dept. of EECS
3/1/2005 Filters.doc 8/8
This filter is a band-pass type, as it “passes” signals within a
frequency bandwidth ∆ω , while “rejecting” signals at all
frequencies outside this bandwidth.
In addition to filter bandwidth ∆ω , a fundamental parameter
of bandpass filters is ω 0 , which defines the center frequency
of the filter bandwidth.
3. Band-Stop Filter
Τ (ω ) Γ (ω )
1 1
∆ω ∆ω
ω ω
ω0 ω0
Note for this filter:
⎧ ≈ 0 ω − ω 0 < ∆ω 2 ⎧≈ 1 ω − ω 0 < ∆ω 2
⎪ ⎪
Τ (ω ) = ⎨ Γ (ω ) = ⎨
⎪ ≈ 1 ω − ω > ∆ω 2 ⎪≈ 0 ω − ω 0 < ∆ω 2
⎩ 0 ⎩
This filter is a band-stop type, as it “rejects” signals within
a frequency bandwidth ∆ω , while “passing” signals at all
frequencies outside this bandwidth.
Jim Stiles The Univ. of Kansas Dept. of EECS
