<< Chapter < Page | Chapter >> Page > |
Having learned something about how we generate signals with bipolar and field effect transistors, wenow turn our attention to the problem of getting those signals from one place to the next. Ever since Samuel Morse (and thefounder of my alma mater , Ezra Cornell) demonstrated the first working telegraph, engineers andscientists have been working on the problem of describing and predicting how electrical signals behave as they travel downspecific structures called transmission lines .
Any electrical structure which carries a signal from one point to another can be considered a transmissionline. Be it a long-haul coaxial cable used in the Internet, a twisted pair in a building as part of a local-area network, acable connecting a PC to a printer, a bus layout on a motherboard, or a metallization layer on a integrated circuit,the fundamental behavior of all of these structures are described by the same basic equations. As computer switchingspeeds run into the 100s of MHz, into the GHz range, considerations of transmission line behavior are ever morecritical, and become a more dominant force in the performance limitations of any system.
For our initial purposes, we will introduce a "generic" transmission line , which will incorporate most (but not all) features of real transmissionlines. We will then make some rather broad simplifications, which, while rendering our results less applicable to real-lifesituations, nevertheless greatly simplify the solutions, and lead us to insights that we can indeed applyto a broad range of situations.
The generic line consists of two conductors. We will suppose a potential difference $V(x)$ exists between the two conductors, and that a current $I(x)$ flows down one conductor, and returns via the other. For the time being, we will let the transmission line be"semi-infinite", which means we have access to the line at some point $x$ , but the line then extends out in the $x$ direction to infinity. (Such lines are a bit difficult to handle in the lab!)In order to be able to describe how $V(x)$ and $I(x)$ behave on this line, we have to make some kind of model of the electrical characteristics of the line itself. We can not just make up any model we want however;we have to base the model on physical realities.
Let's start out by just considering one of the conductors and the physical effects of current flowing though thatconductor. We know from freshman physics that a current flowing in a wire gives rise to a magnetic field, $H$ ( ). Multiply $H$ by $$ and you get $B$ , the magnetic flux density, and then integrate $B$ over a plane parallel to the wires and you get $$ , the magnetic flux "linking" the circuit. This is shown in for at least part of the surface. The definition of $L$ , the inductance of a circuit element, is just
We are now ready to build our model. What we want to do is to come up with some arrangement of inductors andcapacitors which will represent electrically, the properties of the distributed capacitance and inductance we discussedabove. As a length of line gets longer, its capacitance increases, so we had better put the distributed capacitances inparallel with one another, since that is the way capacitors add up. Also, as the line gets longer, its total inductanceincreases, so we had better put the distributed inductances in series with one another, for that is the way inductances addup. is a representation of the distributed inductance and capacitance of the generic transmission line.
We break the line up into sections $(x)$ long, each one with an inductance $L(x)$ and a capacitance $C(x)$ . If we halve $(x)$ , we would halve the inductance and capacitance of each section, but we'd have twice as many of them per unitlength. Duh! The point is no matter how fine we make $C(x)$ , we still have Ls and Cs arranged like we see in , with the two kinds of components intermixed.We could make a more realistic model and realize that all real wires have seriesresistance associated with them and that whatever we use to keep the two conductors separated will have some leakage conductanceassociated it. To account for this we would introduce a series resistance $R$ (ohms/unit length) and a series conductance $G$ (ohms/unit length). One section of our line model then looks like .
Although this is a more realistic model, it leads to much more complicated math. We will start out anyway,ignoring the series resistance $R$ and the shunt conductance $G$ . This "approximation" turns out to be pretty good as long as eitherthe line is not too long, or the frequencies of the signals we are sending down the line do not get too high. Without theseries resistance or parallel conductance we have what is called an ideal lossless transmission line .Notification Switch
Would you like to follow the 'Communications b : filters and transmission lines' conversation and receive update notifications?