Advanced Transmission Line Modeling Techniques
Advanced Transmission Line Modeling Techniques
In Section 4.0, two simple models were derived for a closed end and an open
end transmission line. In both of these models the driver is located at one end of the
transmission line and the cross-sectional area is monotonically varied based on a taper
ratio m and the exponential area equation found in Section 3.0. For a straight-line
geometry m = 0 and this is equivalent to the test line geometry shown in Figure 2.1 with
an open or a closed end boundary condition.
In most transmission line designs, the driver is not all the way at one end of the
line and the cross-sectional area is much more complicated with changes in area
occurring at several locations along the line’s length. To model these more complicated
enclosures, a second set of MathCad worksheets were formulated and are presented in
this section.
To address the offset driver geometry, the MathCad model “TL Open End.MCD”
needed to be rewritten to reflect the situation shown in Figure 6.1. When a driver is
installed offset from the closed end of the transmission line, the line is split into two
separate sections. Above the driver is typically a short transmission line with a closed
terminus. Below the driver is a longer transmission line with an open terminus. Notice
that these two transmission lines are in parallel as seen from the back of the driver.
The equivalent acoustic circuit and equivalent electrical circuit for the
transmission line with an offset driver are shown in Figures 6.2 and 6.3 respectively.
Notice that in the equivalent acoustic circuit, the acoustic impedance of the closed line
and the acoustic impedance of the open line are in parallel. Also notice the relationship
between the volume velocities at the back of the driver.
Ud = Uc + Uo
In Figure 6.1, recognize that as the driver moves into the cabinet, the air volume
displaced moves into the closed ended transmission line and into the open ended
transmission line. The volume split will depend on the relative values of the two acoustic
impedances Zac and Zao. Also notice that the volume velocity of the air at the terminus is
no longer a function of the volume velocity of the driver. The volume velocity of the air at
the terminus is a function of the volume velocity of the air entering the open ended
transmission line.
UL ε Uo
=
SL So
For the first quarter wavelength mode, offsetting the driver has no significant
impact. The most obvious impact of offsetting the driver in a transmission line is a dip or
null in the terminus response. This shows up in both stuffed and unstuffed terminus
responses. In the frequencies above the first quarter wavelength resonance, the
dramatic attenuation of the terminus output can locally reduce if not eliminate the ripple
typically seen in most transmission line system SPL responses.
Page 1 of 12
Section 6.0 : Advanced Transmission Line Modeling Techniques
By Martin J King, 07/05/02
Copyright 2002 by Martin J. King. All Rights Reserved.
The reason for the severe dip or null in the terminus response can be found in
Figure 4.2 for an unstuffed closed end transmission line. In this figure the first half
wavelength mode is seen as a spike at 171 Hz. But at the frequency corresponding to a
quarter wavelength standing wave, ~86 Hz, a deep null can be seen in the pressure
response. Since the acoustic impedance of a transmission line is the ratio of the
pressure to the volume velocity, a deep null must also exist in the acoustic impedance of
the closed end line. A deep null in this acoustic impedance will tend to short the parallel
impedances in Figure 6.2 and all of the driver’s volume velocity will be directed into the
closed end. The means that the output from the terminus will be strongly attenuated.
For a stuffed transmission line, the same phenomenon can be observed. The
damping from the fibrous tangle will tend to reduce the sharpness of all the resonant
peaks and at the same time reduce the depth of the first null resulting from the closed
end portion of the transmission line. By reducing the sharpness of the peaks and nulls,
the frequency content is also locally broadening for the volume velocities into the open
and closed ends of the transmission line. The resulting shallower dip will not attenuate
the terminus response as much as the unstuffed line but it will still have a significantly
wider effect on the total SPL system response.
In summary, offsetting the driver from the end of a transmission line will
dramatically reduce the terminus SPL response at the quarter wavelength frequencies of
the closed end section of the transmission line. By shifting the position of the driver, the
position of the dips in the terminus response can be moved higher or lower in frequency.
This geometry can be used to reduce the midrange ripple inherent in a transmission line
design with a driver mounted at one end. It should also be recognized that shifting the
driver does not change the basic quarter wavelength frequencies that would be
calculated for the entire transmission line length. Only the amount of excitation applied to
each quarter wavelength mode is changed.
The MathCad worksheet “TL Open End.MCD” was reconfigured to model the
equivalent circuits in Figures 6.2 and 6.3. The third worksheet is called “TL Offset
Driver.MCD”. This MathCad computer model can be used to simulate an offset driver, in
a tapered, straight, or expanding fiber damped transmission line. This model will handle
a wide variety of geometries with Dacon Hollofil II stuffing densities between 0.0 lb/ft3
and 1.0 lb/ft3. The only restriction on the geometry is that the closed end cross-sectional
area Sc must be greater than zero. If the user is trying to simulate a pointed closed end,
a very small numerical value for the area can be specified for Sc.
Page 2 of 12
Section 6.0 : Advanced Transmission Line Modeling Techniques
By Martin J King, 07/05/02
Copyright 2002 by Martin J. King. All Rights Reserved.
Figure 6.1 : Offset Driver Geometry
Closed Ended Sc
Transmission
Line Section
Uc
Uo
Open Ended
Transmission
Line Section
UL
SL
where
Page 3 of 12
Section 6.0 : Advanced Transmission Line Modeling Techniques
By Martin J King, 07/05/02
Copyright 2002 by Martin J. King. All Rights Reserved.
Figure 6.2 : Acoustic Equivalent Circuit for a Transmission Line with an Offset Driver
Ud
pg Zao Zac p
Uo Uc
Where :
pg = pressure source
= (eg Bl) / (Sd Re)
Ud = Uo + Uc
then :
ε = uL / uo
Page 4 of 12
Section 6.0 : Advanced Transmission Line Modeling Techniques
By Martin J King, 07/05/02
Copyright 2002 by Martin J. King. All Rights Reserved.
Figure 6.3 : Electrical Equivalent Circuit for a Transmission Line with an Offset Driver
Rg + Re Lvc
Zec ec
Zeo eo
Where :
eg = voltage source
= 2.8284 volt
ed = Bl ud
Page 5 of 12
Section 6.0 : Advanced Transmission Line Modeling Techniques
By Martin J King, 07/05/02
Copyright 2002 by Martin J. King. All Rights Reserved.
Modeling a Transmission Line with a Change in Cross-Sectional Area :
At the Junction :
• Pressure is continuous
• Volume velocity is continuous
p1 S1 p2 S2 p3
U1 U2 U3
u(x,t) = ∂ξ(x,t)/∂ t
positive sign convention
Element 1
Element 2
pi
pt
pr
S1 = area of Element 1
S2 = area of Element 2
At the junction, the pressure and the volume velocity must be the same in both
pipes. As a pressure wave travels along pipe 1 and arrives at the junction, a part of the
wave will be transmitted into pipe 2 while a second part of the wave will be reflected
back into pipe 1. Transmission and reflection of a wave will occur at any discontinuity in
the acoustic impedance. A discontinuity can be a change in the cross-sectional area, a
Page 6 of 12
Section 6.0 : Advanced Transmission Line Modeling Techniques
By Martin J King, 07/05/02
Copyright 2002 by Martin J. King. All Rights Reserved.
sharp change in the taper rate of the cross-sectional area, or a sudden change in the
fiber stuffing density.
Z22 1
−
U1 Z21 Z21 U2
p = Z Z
Z11 p2
1 11 22
− + Z12
Z 21 Z21
Z33 1
−
U2 Z32 Z32 U3
p = Z Z
Z22 p3
2 22 33
− + Z23
Z32 Z32
Z22 1 Z33 1
− −
U1 Z21 Z21 Z32 Z32 U3
=
p Z Z
1 11 22 Z11 Z22 Z33 Z22 p3
− + Z12 − + Z23
Z 21 Z21 Z32 Z32
So the pressure and volume velocity at one end of a complicated geometry can now be
expressed in terms of the pressure and volume velocity at the far end. Simple
multiplication of 2 x 2 matrices can be performed to quickly work through 1, 5, 10, or
even 100 one-dimensional acoustic elements. After compiling all of the elements, the
pressure and volume velocity at one end of a transmission line is a related to the
pressure and volume velocity at the far end of the line by a single 2 x 2 transfer matrix.
The boundary conditions described in Section 4.0 can now be applied to calculate the
acoustic impedance and if applicable the terminus volume velocity.
The MathCad model “TL Sections.MCD is based on the transfer matrix method
described above and simulates an offset driver in a fiber filled transmission line. The
number of acoustic elements can be increased in both the closed and open ended
Page 7 of 12
Section 6.0 : Advanced Transmission Line Modeling Techniques
By Martin J King, 07/05/02
Copyright 2002 by Martin J. King. All Rights Reserved.
sections of the transmission line so that very complex geometry changes and stuffing
density schemes can be modeled.
After making the corrections and improvements to the MathCad worksheets over
the past two years, my understanding of tapered and expanding transmission lines
improved considerably. I began to see some real benefits for both tapered and
expanding transmission lines. One thing I studied was the changes that occur, in the
straight transmission line’s quarter wavelength resonant frequencies, when a tapered or
expanding geometry is introduced.
Length = c / (4 x f)
= (342 m/sec) / (4 x 40 Hz)
= 84.15 in
= 2.14 m
Table 6.1 shows the area at the driven end of the transmission line S0 and the
area at the open end (terminus) of the transmission line SL for the three different
assumed geometries. Again, all three transmission lines have the same length and
internal volume.
Figure 6.5 shows the magnitude of the air velocity at the terminus end of the
transmission line assuming a 1 m/sec velocity at the driven end. This applied 1 m/sec air
velocity is assumed to be uniform over the entire area S0. As the frequency of the driven
end increases from 1 Hz to 1000 Hz, thirteen separate resonant frequencies are excited.
The sharp peaks in the plots in Figure 6.5 define the frequency of each resonance.
Page 8 of 12
Section 6.0 : Advanced Transmission Line Modeling Techniques
By Martin J King, 07/05/02
Copyright 2002 by Martin J. King. All Rights Reserved.
Looking in Figure 6.5 at the magnitude of ε for the frequencies below 10 Hz you
can see that the value is different for each of the three transmission line geometries. The
magnitude is equal to the ratio of the driven area S0 over the terminus area SL. This
indicates that at very low frequencies, the volume of air moving into the line at x = 0 is
equal to the volume of air moving out of the line at x = L. This relationship is derived as
follows for frequencies below 10 Hz.
ε = u(L,t) / u(0,t)
= u(L,t) / (1 m/sec)
S0 x u(0,t) = SL x u(L,t)
S0 x (1 m/sec) = SL x u(L,t)
S0 / SL = u(L,t) / (1 m/sec)
ε = S0 / SL
Also notice in Figure 6.5 that the resonant peaks above 100 Hz appear to occur
at approximately the same frequencies. However it can also be seen in Figure 6.5, that
the first resonance for each of the transmission lines occurs at a different frequency.
Table 6.2 summarizes the resonant frequencies of the first five modes for each of
the three transmission lines. Also shown, in the second column of Table 6.2, are the
ideal quarter wavelength frequencies that would be calculated based on a classical
solution of the one-dimensional wave equation. The classic solution of the one-
dimensional wave equation can be found in most undergraduate physics or acoustics
textbooks. The problem being solved in these textbooks is typically a constant cross-
section pipe with a boundary condition specified at each end. The boundary conditions
used to solve the wave equation for an open ended transmission line are a sinusoidal
velocity applied at the driven end S0 and a zero pressure (or velocity maximum) defined
at the open terminus end SL.
Comparing the second and the fourth columns of Table 6.2, the MathCad model
consistently calculates lower resonant frequencies then the classic textbook solution.
This is due to the frequency dependent acoustic impedance, see Figure 4.6, specified at
the terminus in the MathCad model instead of the zero pressure boundary condition
assumed in the textbook solution. The acoustic impedance boundary condition makes
the pipe appear to be slightly longer which leads to lower resonant frequencies. A similar
situation occurs when sizing the length of a port in a bass reflex enclosure. The effective
length of the port, used in most design calculations, is typically longer then the actual
physical length of the port.
Page 9 of 12
Section 6.0 : Advanced Transmission Line Modeling Techniques
By Martin J King, 07/05/02
Copyright 2002 by Martin J. King. All Rights Reserved.
Also notice in Table 6.2, that the resonant frequencies above 100 Hz are
approximately the same. As stated previously, when discussing Figure 6.5, the first
mode occurs at different frequencies for each of the three transmission line geometries.
For the tapered transmission line, the lowering of the first resonant frequency would lead
to a shorter line for a 40 Hz design goal. For the expanding transmission line, the
increase in the first resonant frequency would have the opposite effect of requiring a
longer line for a 40 Hz design goal.
For most transmission lines, the cross-sectional area is usually held constant or
tapered. Looking back at the work done by Bailey(7) and then by Bradbury(8), the
sketches in Bailey’s article would indicate that they were working with test results
obtained from tapered transmission lines. Suppose that the expected quarter wavelength
resonant frequencies were calculated in the same manner as those shown in the second
column of Table 6.2 using the classic equation f = c / (4 x L). Now recognize that the
measured results were probably more typical of the resonant frequencies listed in the
third column of Table 6.2. The tapered transmission line would have exhibited a lower
resonant frequency for the first mode, when compared to a straight transmission line, but
correlated closely with the higher frequency modes of the straight transmission line.
Bradbury’s theory contends that only the low frequency sound waves couple with the
fibers, through a viscous damping coefficient, resulting in motion of the fibers and a
reduction in the speed of sound due to the added moving mass of the fibers. This
postulated reduction in the speed of sound would result in a lower resonant frequency
for the first quarter wavelength mode as observed in the test data. At higher frequencies,
the fibers are not coupled to the sound waves and do not move, so the speed of sound
is unchanged and the resonant frequencies are closer to the expected values.
If Bradbury did not include the impact of the tapered geometry on the anticipated
quarter wavelength frequencies in his analysis of Bailey’s test data, then I have to
wonder if this oversight was the impetus for the moving fiber equations. Remember, my
conclusion in Section 2.0 was that the fibers are not moving during normal operation of a
transmission line. This conclusion was verified in Section 5.0 when the damping
coefficient and speed of sound were empirically determined as functions of stuffing
density from the test data.
Page 10 of 12
Section 6.0 : Advanced Transmission Line Modeling Techniques
By Martin J King, 07/05/02
Copyright 2002 by Martin J. King. All Rights Reserved.
Figure 6.5 : Magnitude of the Terminus Velocity for a 1 m/sec Driven Excitation
100
Epsilon Magnitude
10
εr
1
0.1
1 .10
3
1 10 100
−1
r⋅ dω⋅ Hz
100
Epsilon Magnitude
10
εr
1
0.1
1 .10
3
1 10 100
−1
r⋅ dω⋅ Hz
100
Epsilon Magnitude
10
εr
1
0.1
1 .10
3
1 10 100
−1
r⋅ dω⋅ Hz
Page 11 of 12
Section 6.0 : Advanced Transmission Line Modeling Techniques
By Martin J King, 07/05/02
Copyright 2002 by Martin J. King. All Rights Reserved.
Summary :
Two more new MathCad models have been described, “TL Offset Diver.MCD”
and “TL Sections.MCD”, in this section. Advanced modeling techniques that can be
addressed using these models are offsetting the driver along the length of a
transmission line and a completely general geometry and stuffing scheme.
Page 12 of 12