Post subject: Transmission line Matching network Posted: Wed
Nov 01, 2006 1:12 am
Joined: Tue Apr 18, 2006
hi...i have couple of question on basic transmission
line matching network design:
how do you choose what Zo (characteristic
impedance) one might want for the matching series and stubs? since they
are matching networks Zo doesn't have to be 50 Ohm then how do you go
about selecting it? does higher Zo mean lower alpha (attenuation loss)?
in other words, is Zo proportional to 1/alpha?
what exactly is
dispersion and which has less dispersion: microstrip or coplanar? how
is diespersion related to attenuation and Zo. thanks
Post subject: Posted: Thu Nov 02, 2006 2:41 pm
Joined: Mon Jun 27, 2005 2:02 pm
The 50-ohm characteristic
impedance is a compromise between the need for minimum attenuation and
the ability to deliver maximal power.
In a system of Zo=77 ohm, there
is a minimal attenuation and in a system of Zo=33 ohm there is maximal
power capability. Another consideration is the dimension (Width9 of
the line depending on the substrate and frequency. So 50 ohm was chosen
as the best trade-off to all of these constraints.
is a different signal velocity versus the frequency - similar to the
group delay in filters (After all a transmission line is a filter).
The dispersion is much smaller in Coplanar and Stripline than in microstrip.
In the following line you can find more about the dispersion
of different transmission lines:
Post subject: Posted: Fri Nov 03, 2006
Joined: Fri Feb 17, 2006 12:07 pm
Location: London UK
In addition to the points IR has
made, remember that the input impedance (and therefore the inductance
or capacitance) of a short section of line depends on its intrinsic
characteristic impedance, and the terminating impedance.
= Z0 * (Zr + jZ0 * TAN(2pi*length/lambda))
(Z0 + jZr * TAN(2pi length/lambda))
Z0 is the intrinsic characteristic
impedance, Zr is the terminating impedance, and lambda is the wavelength
in the dielectric of the substrate. The transformer ratio is thus Z0/Zr
The intrinsic characteristic impedance is dependent on physical
dimensions and the substrate permittivity.
So if a matching network
calls for a particular L and C combination, then for a particular center
frequency you have to find the right length, width and junction point
for the series and shunt element traces.
Thank goodness we have
software these days to do all this. But at least we need to know what
the software is trying to do, in order to understand the answers and
judge if they make sense.
Posted: Wed Nov 29, 2006 8:12 pm
Nov 29, 2006 5:51 pm
how do you choose what
Zo (characteristic impedance) one might want for the matching series
It all depends mostly on what bandwidth you need.
In order to go from from a certain (usually) low impedance to a
50 ohm impedance you use matching techniques that will transform the
impedance from the one you have at the device to a certain intermediate
value from where you will use another element to move again. In most
cases for today's higher frequency designs they are done using microstrip
transmission lines and stubs that will rotate the impedance point in
a smith chart around the transmission line's characteristic impedance
and the degrees it rotates will be determined by the length of the transmission
There are infinity solutions to a matching network ranging
from a one step to multiple steps using the transmission lines or capacitive
stubs that you mentioned depending which way in the smith chart you
want to rotate but they all have very different broadband responses.
The multiple step approach usually allows you to go from point A
to point B keeping the broader bandwidth because the Q of the network
will be lower.
Transformations with devices like baluns allow
you to get your point B closer to the match requiring lower Q networks
since you're starting from a closer point.
As a general rule
of thumb broader bandwidth matches are better for temperature variations
and for manufacturing because they allow for variations in device impedance
with little or no degradation in performance.