# Varying transmission line parameters

Discussion in 'Electronic Design' started by bazhob, May 25, 2006.

1. ### bazhobGuest

I read that the electrical properties of transmission lines can be
described using their RLCG parameters (line resistance, inductance,
capacitance and dielectric conductance per unit length).

How can one elegantly introduce "noise" to transmission lines, so that
their properties vary within a given interval, ie +-100 Ohms, +-100uH,
+-100pF ... ? Would one use a combination of controlled resistors (ie
JFETs), inductors and varactors or is there a simpler solution?

It would also be sufficient if one could adjust those parameters
specifically (ie. add 10 Ohms, subtract 10pF)

Thank you!
Tobias

2. ### Tom BruhnsGuest

Did you want to build a lumped model of a line, with parameters you can
adjust at will, or a simulation model, or a real physical line? How
rapidly did you want to adjust the parameters, and with what accuracy?
Are you looking to do the same thing all along the line so it remains
uniform, or did you want to simulate what happens to a line if, for
example, the center conductor becomes un-centered periodically along
the length of the line?

In short, exactly what are you trying to accomplish?

I'd note that 10 ohms is a huge variation for most practical lines...

Cheers,
Tom

3. ### colinGuest

With a balanced line you could arrange it so you could vary the distance
between the 2 wires, that should vary the parameters.

Colin =^.^=

4. ### bazhobGuest

I want to build a real physical (PCB) layout. The adjustment would vary
only every 5 seconds or so (this is why I would prefer actually setting
parameters instead of using a noise source).

I am experimenting with the use of "evolutionary algorithms" in circuit
design. If there is no variation, the circuit will over-adapt to one
specific board. The tests are done on real hardware, simulation is not
an option.

I sort of want to simulate *in real* that the length of the
transmission line between two components has been extended, ie. by
physical wires, or that a 60-Ohm closed switch in the line has been
removed. I am aware that this can only be solved by approximation.

Thank you,
Toby

5. ### Tom BruhnsGuest

Hi Toby,

Well, in theory you can cascade a bunch of sections of series L and R
and shunt C and G, and with enough sections it will be a decent
approximation of a TEM transmission line. The problems I see are that
it takes a lot of sections, and in a real line R and G are typically
pretty small, so it may not be easy to find parts that will be
controllable and cover the right range of values. In addition, a real
line has a frequency dependence that will be very difficult to emulate
with lumped components. You can use varactors for variable
capacitance, as long as the signal levels you are transmitting aren't
very large, and you can similarly use something like saturable reactors
for the series inductance. All in all it doesn't sound like a very
practical thing to try to do. If the bandwidth of the signal you're
transmitting isn't too broad and the line is "long" enough, I'd say it
might well be easier to digitize the input signal, apply signal
processing to emulate the desired line, and convert back to analog at
the other end. That takes care of the transmission characteristics in
one direction at least, but not the load and source impedances without

Good luck.

Cheers,
Tom

6. ### Marc PopekGuest

you are describing more of a monte carlo analysis then adding noise. try
seeking the monte carlo analysis.
marc