Loaded Waveguide Junctions

In this section, scattering relations will be derived for the general case of N waveguides meeting at a load. When a load is present, the scattering is no longer lossless, unless the load itself is lossless. (i.e., its impedance has a zero real part). For $N>2$ , $v^{+}_i$ will denote a velocity wave traveling into the junction, and will be called an ``incoming'' velocity wave as opposed to ``right-going.''^C.9

Figure C.29: Four ideal strings intersecting at a point to which a lumped impedance is attached. This is a series junction for transverse waves.

Consider first the series junction of $N$ waveguides containing transverse force and velocity waves. At a series junction, there is a common velocity while the forces sum. For definiteness, we may think of $N$ ideal strings intersecting at a single point, and the intersection point can be attached to a lumped load impedance $R_J(s)$ , as depicted in Fig.C.29 for $N=4$ . The presence of the lumped load means we need to look at the wave variables in the frequency domain, i.e., $V(s) = {\cal L}\{v\}$ for velocity waves and $F(s) = {\cal L}\{f\}$ for force waves, where ${\cal L}\{\cdot\}$ denotes the Laplace transform. In the discrete-time case, we use the $z$ transform instead, but otherwise the story is identical. The physical constraints at the junction are

$$V_1(s) = V_2(s) = \cdots = V_N(s) \isdef V_J(s)$$ (C.90)

$$F_1(s) + F_2(s) + \cdots + F_N(s) = V_J(s) R_J(s) \isdef F_J(s)$$ (C.91)

where the reference direction for the load force $F_J$ is taken to be opposite that for the $F_i$ . (It can be considered the ``equal and opposite reaction'' force at the junction.) For a wave traveling into the junction, force is positive pulling up, acting toward the junction. When the load impedance $R_J(s)$ is zero, giving a free intersection point, the junction reduces to the unloaded case, and signal scattering will be energy preserving. In general, the loaded junction is lossless if and only if re$\left\{R_J(j\omega)\right\}\equiv0$ , and it is memoryless if and only if im$\left\{R_J(j\omega)\right\}\equiv0$ .

The parallel junction is characterized by

$$F_1(s) = F_2(s) = \cdots = F_N(s) \isdef F_J(s)$$ (C.92)

$$V_1(s) + V_2(s) + \cdots + V_N(s) = F_J(s)/R_J(s) \isdef V_J(s)$$ (C.93)

For example, $F_i(s)$ could be pressure in an acoustic tube and $V_i(s)$ the corresponding volume velocity. In the parallel case, the junction reduces to the unloaded case when the load impedance $R_J(s)$ goes to infinity.

The scattering relations for the series junction are derived as follows, dropping the common argument `$(s)$ ' for simplicity:

$$R_J V_J = \sum_{i=1}^N F_i = \sum_{i=1}^N (F^+_i + F^-_i)$$ (C.94)

$$= \sum_{i=1}^N (R_i V^+_i - R_i \underbrace{V^-_i}_{V_J-V^+_i})$$ (C.95)

$$= \sum_{i=1}^N (2 R_i V^+_i - R_i V_J)$$ (C.96)

where $R_i$ is the wave impedance in the $i$ th waveguide, a real, positive constant. Bringing all terms containing $V_J$ to the left-hand side, and solving for the junction velocity gives

$$V_J = 2\left(R_J + \sum_{i=1}^N R_i\right)^{-1} \sum_{i=1}^N R_i V^+_i$$ (C.97)

$$\isdef \sum_{i=1}^N{\cal A}_i(s) V^+_i(s) \protect$$ (C.98)

(written to be valid also in the multivariable case involving square impedance matrices $R_i$ [437]), where

$${\cal A}_i(s) \isdef \frac{2R_i}{R_J(s) + R_1 + \cdots + R_N}$$ (C.99)

Finally, from the basic relation $V_J = V_i = V^+_i + V^-_i$ , the outgoing velocity waves can be computed from the junction velocity and incoming velocity waves as

$$V^-_i(s) = V_J(s) - V^+_i(s)$$ (C.100)

Similarly, the scattering relations for the loaded parallel junction are given by

$$F_J(s) = \sum_{i=1}^N{\cal A}_i(s) F^+_i(s), \quad {\cal A}_i(s) \isdef \frac{2\Gamma _i}{\Gamma _J(s) + \Gamma _1 + \cdots + \Gamma _N}$$

$$F^-_i(s) = F_J(s) - F^+_i(s)$$ (C.101)

where $F_J(s)$ is the Laplace transform of the force across all elements at the junction, $\Gamma _J(s)$ is the load admittance, and $\Gamma _i=1/R_i$ are the branch admittances.

It is interesting to note that the junction load is equivalent to an $N+1$ st waveguide having a (generalized) wave impedance given by the load impedance. This makes sense when one recalls that a transmission line can be ``perfectly terminated'' (i.e., suppressing all reflections from the termination) using a lumped resistor equal in value to the wave impedance of the transmission line. Thus, as far as a traveling wave is concerned, there is no difference between a wave impedance and a lumped impedance of the same value.