It can be convenient to define a reference impedance for S-paramters
which is not the characteristic impedance of all the pieces of
transmission line in the system, and keep the same reference impedance
for all your work. For example, 50 ohms is commonly used, even though
the system contains sections of microstrip of various impedances.
But that's just a convenience for system analysis and design. If you
want to try to assign even a little physical significance to
reflection coefficient as used on a piece of line, you really should
be using the line's characteristic impedance as the reference
impedance. In addition, you should realize that it's going to make
sense only in a linear, time-invariant system with steady-state
excitation, with only one source of excitation (at a time). In
addition, it is of course a function of frequency, just as the line's
characteristic impedance is.
As others have noted, the magnitude of the reflection coefficient can
be greater than unity with a passive line and load. Don't try to read
too much physical significance into that, however.
Although the classic definition of (V)SWR involves knowing a voltage
maximum and a voltage minimum on a line, I much prefer a definition in
terms of forward and reverse voltages. That allows me to think about
SWR at a point on a uniform line, and realize that it will be
different at different points (because of line attenuation). In fact,
the _definition_ I use for reflection coefficient is Er/Ef, or
equivalently the ratio of electric fields (or magnetic fields)
associated with forward and reverse waves (which then applies also to
non-TEM waveguides). From that definition, it's straightforward to
determine that rho = (Zl-Zo)/(Zl+Zo). And in keeping with the idea
that you cannot have a voltage magnitude minimum less than zero, and
because I believe it's more practical than the classic definition to
have an SWR definition I can apply to any point on a line, my working
SWR defintion is SWR = (|Ef|+|Er|)/(|(|Ef|-|Er|)|). This will align
well with the usual formula that SWR = (1+|rho|)/(1-|rho|) when
|rho|=1, but it never gives you a negative SWR. If you can accept my
definition of SWR, we can talk about SWR. If you can't, then I just
won't talk with you about SWR, and limit the discussion to reflection
coefficient which we presumably would be able to agree on.
Cheers,
Tom
(Dr. Slick) wrote in message . com...
From Pozar's Microwave Engineering (Pg. 606):
Reflection Coefficient looking into load = (Zl-Zo)/(Zl+Zo)
Where Zl is a purely real load impedance, and Zo is the
purely real characteristic impedance reference.
When you change Zo, you change the normalized center of the
Smith Chart, and therefore the Reflection Coefficient and SWR, looking
into the same load.
Check it out...
Slick