Anda di halaman 1dari 7

Session F4E

SOME FALLACIES IN THE CIRCUIT ANALYSIS COURSE


Artice M. Davis Department of Electrical Engineering San Jos State University San Jos California 95192-0084
Abstract - This paper examines a number of common fallacies in the typical introductory circuit analysis text and offers a more logical, rigorous, and pedagogical approach for each. The majority of these fall within that segment of introductory circuit analysis dealing with active circuits and subcircuits; more, however, can be found in each of the other major areas covered by such texts. It is hoped that this critique will spur a reexamination of the topics covered in introductory circuit analysis texts. Portions of this paper have been submitted for possible pulication in the IEEE Transactions on Education. will be referred to as taping the dependent source [1, pp 167168].

II. SUPERPOSITION
The first topic to be considered is the use of superposition for analyzing active circuits. Virtually all the widely used circuits texts exhort the reader, in one fashion or another, Never deactivate a dependent source when applying superposition to an active circuit. For a sampling, see [2, pp 7375; 3, p149; 4, p126]. The circuit shown in Figure 2 will be analyzed using superposition in

I. INTRODUCTION
There are many texts available for the introductory circuit analysis text, yet all are uniformly weak in certain areas: those pertaining to very fundamental topics, in fact. Close examination shows that many of them parrot approaches, warnings, and cautionary exhortations that seem to be merely transcribed from earlier similar texts. The topics discussed in this paper are representative, though certainly not exhaustive. Before examining these topics, a simple technique will be described that, though seemingly simple, is actually a powerful tool in analyzing active circuits. Consider the voltage controlled voltage source (VCVS) shown in Figure 1.

i vs + i =g v c m x
3 6 3 6

+ vx is

Figure 2: An active circuit example violation of this rule in order to show that the exhortation is quite unnecessary. Notice that the dependent source (a VCCS, or voltage controlled current source) is shown taped. It will be left to the reader to derive the following equations by superposition, treating the VCCS as an independent current source:

i v a. A VCVS b. Its v-i characteristic

vx

vc = vx

(1) . (2)

Figure 1: A Voltage Controlled Voltage Source (VCVS) Now think about the controlling relationship as, let us say, a label affixed to the source and labeled with that equation. Next, imagine that a small piece of masking tape is placed over this equation and labeled with the literal value vc. Notice that the v-i characteristic of a dependent source is precisely the same as that of an independent source of the same type (i.e. a voltage source in this case). Thus, one can think of the VCVS as (temporarily) being considered as an independent onebut one having the literal (and unknown) value vc. Simple changes, mutatis-mutandis, work for dependent sources of other types. The process just described

Noticeand this is highly importantthat one must compute the controlling variable by superposition as well as the variable of interest. The next step is to untape the dependent source and express its value in terms of the unknowns, in this case i, and the independent source values. This gives

.
This can be rearranged into

(3)

(4)

0-7803-6424-4/00/$10.00 2000 IEEE


th

October 18 - 21, 2000 Kansas City, MO

30 ASEE/IEEE Frontiers in Education Conference F4E-23

Session F4E

If gm = 2, we note that there is no solution for is unless one also has gmvs + 15gmis = 0in which case there are an infinite number of solutions. Considering Equation (2), the same observation also holds for i. If gm 2, on has

Consider the circuit of Section II after being partitioned into two two-terminal subcircuits as shown in Figure 3. Note

i vs + i =g v c m x
3 a 6 3 b 6

+ vx is

.
This, together with Equation (2), gives

(5)

(6)

The reader can easily verify this result applying standard techniques. This analysis has brought to light several things. One can, indeed, deactivate dependent sources when applying superpositionby taping the dependent sources as a first step, computing the controlling variables as well as the desired variables, and then untaping the dependent sources and expressing their dependent variables in terms of the unknown variables and the independent source variables. There are two fundamental ideas behind this procedure. The first is that a dependent source has the same v-i relationship as that of an independent one with an added algebraic constraint relating its value to that of a controlling variable elsewhere in the circuit. The second is that one must apply the algebraic constraint to the total variables (that is, not to the partial responses by themselves). One final comment: most texts only present analyses of active circuits whose numerically valued parameters are carefully calculated to yield a unique solution (see any of the previously cited references [2-4]). In reality, the more interesting questions about active circuits relate to the existence and uniqueness of solutions for various values of the dependent source parameters. In practice, these circuits are often the dc equivalents of active circuits having capacitance or other effects, and the lack of dc solutions is reflected in instability of the circuit when these elements are considered. In the terminology of [5, p8], the example just considered is sometimes ill-posed, though other circuits could be always ill-posed or always well-posed.

Subcircuit A Subcircuit B Figure 3: The circuit of Section II after partitioning that the letters a and b identify the terminals common to the two subcircuits. Subcircuit A is shown in Figure 4 with these 3

vs + i =g v c m x
Figure 4: Testing subcircuit A

+ vt

it

terminals symbolized by small open circles and with a test source applied. It will be left to the reader to show that

(7)

This is clearly the KVL equation for the Thvenin equivalent subcircuit shown in Figure 5. Notice that a common warning

a 3gmvx + vs + b
Figure 5: The Thvenin equivalent subcircuit has been ignored: the controlling variable of the dependent source is outside the subcircuit for which the Thvenin equivalent is to be found. Furthermore, the equivalent resistance can be found by deactivating the dependent source as well as the independent one. For a sampling of exhortations against these procedures, see [2, p 82; 3, p141; 6, p127]. (Reference [6] is an earlier edition of [4]. Apparently in [4] the authors have decided to more or less ignore Thvenin/Norton equivalents for circuits with dependent sources.) In the approach just demonstrated, there is an extra dependent source present. From a purely historical perspective, it might be argued that the equivalent is thus not a Thvenin equivalent. This, however, is merely an issue of terminology and the important issue is the derivation of an equivalent subcircuit. There is another aspect of this derivation that is worthy of note. Perhaps due to the way the Thvenin equivalent is often derived, most texts discuss only superposition for deterOctober 18 - 21, 2000 Kansas City, MO

III. EQUIVALENT SUBCIRCUITS


The next topic to be considered will be the derivation of Thvenin/Norton equivalents for two-terminal subcircuits. It is to be remarked that most introductory texts fail to distinguish between a subcircuit and a circuitand to notice that it is the only the terminal v-i characteristic of the subcircuit that affects its behavior when incorporated into a larger circuit.

0-7803-6424-4/00/$10.00 2000 IEEE

30th ASEE/IEEE Frontiers in Education Conference F4E-24

Session F4E

mination of its parameters; that is, one finds the open circuit voltage, then applies a test source to the subcircuit formed by deactivating the independent sources (only) to find the equivalent resistance. In fact, it is the terminal v-i characteristic that defines the behavior of a subcircuit, so one can apply a literal-valued test source, compute the resulting terminal voltage, and identify the open circuit voltage as the term not involving the terminal current and the equivalent resistance as the parameter multiplying the terminal current. The identification of the equivalent resistance as a multiplier is the reason the test source must have a literal value. This procedure gives both Thvenin parameters in one stepa oneshot analysis. This is particularly useful when the subcircuit contains dependent sources. Similar comments, of course, apply to derivation of the Norton equivalent. To see how the procedure just developed fits into the normal scheme of things, consider a different partition of the circuit just treatedthe one shown in Figure 6, where

,
or, simplifying

(8) (9)

If gm = 2, we see that it = -vs/15; in other words, the equivalent is simply that of a current source: an ideal Norton equivalent with infinite equivalent resistance. In that case, of course, the Thvenin equivalent does not exist. If g m 2,

(10)

This results in the Thvenin equivalent shown in Figure 8.

c
5

vs + i =g v c m x

a 6 3 b 6

+ vx

it
c

+ 3(2-gm)

vs

2-gm

+ v t
d

s
Figure 8: The new Thvenin equivalent subcircuit Notice that absorbing the controlling variable into the defined subcircuit changes the equivalent resistanceand it is this fact that is undoubtedly responsible for the exhortation against finding the equivalent resistance by deactivating the dependent sources as well as the independent ones. Again, the singular case occurring when gm = 2 is to be noted; it is such cases as this which distinguish active circuits from their passive counterparts; this argues for the study of such circuits to be done from a more general and parametric point of view.

Figure 6: A different partitioning of the example circuit a different pair of terminals, c and d, have been identified. Notice that a voltage source has been chosen as a test source. In general, one can choose the type of termination arbitrarily, though not in certain circumstances. For instance, if the subcircuit is equivalent to an ideal voltage source one cannot choose a test voltage source and if it is equivalent to an ideal current source one cannot choose a test current source. (The ideal termination is a norator, an element that can support an arbitrary current and voltage [7].) There is, however, a pedagogical reason for choosing a voltage source in this case: it emphasizes that the terminal voltage instantly determines the controlling variable of the dependent source, and this is a point overlooked in the typical introductory text. One can, of course, use the equivalent shown in Figure 5 to draw the circuit as shown in Figure 7. It will be left to

IV. NODAL ANALYSIS


It might be supposed that nothing new can be said about a method as classical as nodal analysis; this supposition, however, would be wrong. The circuit in Figure 9, the same one

v3+vs vs +
3

v2
6

it
+ 3gmvx 9 3 6

+
3

v1 is

+ vx

c + v t d

ic v3
6

vx

vs +

Figure 9: The example circuit prepared for nodal analysis we have treated in two other ways, will be analyzed using nodal analysis to support this contention. The circuit has already been prepared for nodal analysis by identifying the nodes (and supernode), assigning unknown node voltages, October 18 - 21, 2000 Kansas City, MO

Figure 7: Using the previous equivalent subcircuit the reader to derive the terminal v-i relationship, which is 0-7803-6424-4/00/$10.00 2000 IEEE
th

30 ASEE/IEEE Frontiers in Education Conference F4E-25

Session F4E

and applying the voltage source constraints to determine the others. The controlling relationship of the dependent source has been omitted simply for convenience of circuit labeling. The first step would have been to tape it anyway, so it will temporarily be treated as an independent one. It will be left to the reader to verify the following equations:

vs vs +
3

v2=0
6

v1=0 +
3

ic v3=0
6

vx

is

, , .

(11) (12) (13)

Figure 11: The forced equivalent circuit well known (and, indeed, easy to see) that, in matrix form, the left sides of the first terms in Equation (16) can be written in the form shown in Equation (17). The generalization to more than three nodes is obvious and well-known and can be written by inspection from the circuit. What is not so widely known, however, is the fact that the right-hand sides of Equation (16) can also be written down immediately by inspection

The interest for us here is the general form of these equations: they are all linear combinations of the unknown voltages and the source values. (In fact it is this action of the resistive portion of the circuit in producing circuit variables that are linear combinations of all the source variables that allowed the deactivation of the dependent sources when using superposition.) Thus, they have the form

(17)

, .

(14) (15)
of the forced equivalent circuit using the interpretation just given. The result is

where f is a multilinear function of the argument variables. This means that the nodal equations have the generic form The first term can be interpreted as the sum of the currents leaving the given node due to the action of the node voltages only, with the sources deactivated, and the second as the sum of the currents leaving that node due to the action of the sources only. A slight rearrangement gives

(18)

(16)
Putting these two results together using the matrix form of Equation (16) results in

This equation states that the sum of the currents leaving a node under the influence of the node voltages alone, with the sources deactivated is equal to the sum of the currents entering that node under the influence of the sources alonewith the node voltages forced (or clamped) to zero. The first of these two interpretations will be referred to as the deactivated equivalent, and the second as the forced equivalent. These two equivalent circuits are shown in Figures 10 and 11. It is

. (19)

v3

v2
3 6 3 6

+ vx

v1

Recall that the dependent source has been taped at this point, so Equation (19) is immediately extensible to the general case of an n-node circuit with independent (or taped) sources. It is naturally of interest to inquire whether or not the nodal equations are solvable. This question is not addressed satisfactorily in any of the available introductory circuits texts except for [1, pp130-131]. The only other that comes close is [8, pp850-851], in which the demonstration is deferred to an appendix, where a celebrated theorem of October 18 - 21, 2000 Kansas City, MO

Figure 10: The deactivated equivalent circuit

0-7803-6424-4/00/$10.00 2000 IEEE

30th ASEE/IEEE Frontiers in Education Conference F4E-26

Session F4E

Hadamard is invoked to establish the positive definite nature of the nodal conductance matrix (the square matrix on the left side of (19))and the celebrated theorem is quite difficult to find in the literature! In fact it is quite easy to establish these conditions in a truly elementary fashion. First note that the matrix form of the nodal equations has the form

,
x

(20)

where G is (for the general case) the n n nodal conductance matrix and is the nx1 matrix of currents into the nodes due to sources alone. Now it is generally knownand familiar to students from the elementary row reduction process for matricesthat G is nonsingular if and only if unique solution to the homogeneous equation is the

have already mentioned means that the nodal conductance matrix is nonsingular, and so (20) always has a solution if the deactived network is connected and all resistors are positive. Suppose the deactivated network is not connected. This can only arise if the original circuit is not connected or if there is at least one cutset of current sources (current sources whose replacement by open circuits separates the circuit into two separate parts). In this case, one need only define multiple references to salvage the nodal method. Look at Figure 13. Right away, a comment is essential. This circuit is, in reality, not well posed for the term independent source means that one can adjust it to any desired value1. This is not the case with the example circuit for the two current sources must always have the same valuewhich means that at least one of them must be a dependent source. However, the

(21) 2 2 v1 v2 2 v3

This follows from an elementary argument that goes like this. If the deactivated equivalent circuit is connected, that is, if there is a path of resistors (the only type of element in the deactivated equivalent)between an arbitrary pair of nodes, then there is a path of resistors from the reference node to any nonreference node. This is indicated schematically in Figure

4V +

is is

+ v3

vR

v2 + vR3
6 3 6

+ vR 2

v1

Figure 13: A circuit with a cutset of current sources deactivated circuit splits into two parts, and nodal analysis does not work. (The reader should check this by writing the nodal equations and noting that the conductance matrix is singular.) It can be modified to work, however, by defining two reference nodesone in each part of the deactivated circuit. This is shown in Figure 14. Again, the reader should

+ vR
1

Figure 12: Resistor paths in the deactivated equivalent circuit 12, which is a reproduction of Figure 10 with the resistor element voltages labeled and paths shown connecting the reference node to each of the three nonreference nodes. Look, now at the top path from the reference node to the node labeled v3. One can write the node voltage v3 as

4V +

2 2

is is

(22)
Figure 14: Multiple reference nodes for nodal analysis write the nodal equations and show that the conductance matrix is nonsingular. At this stage, nodal analysis has been quite fully developed for circuits with positive resistors and sources and an argument has been given to show that nodal analysis will always succeed provided care is taken in the choice of refer1. This precludes both cutsets of independent current sources and loops of independent voltage sources. October 18 - 21, 2000 Kansas City, MO

Now consider the total power P absorbed by the deactivated circuit. One can write

(23)
since there are no sources in this equivalent circuit. However, if all the resistors have positive values (3 and 6 in the example), then each of the individual resistor voltages must be zeroand this implies that each of the node voltages is also zero. Note that this statement follows quite intuitively from conservation of energy, but can be justified in a very rigorous way by invoking Tellegens Theorem (which does not rely upon nodal analysis for its validity). This, as we 0-7803-6424-4/00/$10.00 2000 IEEE

30th ASEE/IEEE Frontiers in Education Conference F4E-27

Session F4E

ences and that circuits with loops of voltage sources are rejected as being ill-posed. Now, however, one must deal with the fact that the circuit being analyzed has a dependent source. It must be untaped and the controlling variable expressed in terms of the unknown node voltages (and the source values). For the example being treated, one can rewrite Equation (19) as

In general, a solution can exist for all values of the dependent source parameters, not exist for some values, or not exist for any values. This is never even hinted at in most introductory treatments, though the advanced treatment in Reference [5] devotes much of the text to this issue.

V: SUMMARY
Though many more examples could be presented, space limitations preclude them. The interested reader can, however, refer to [1] for a more leisurely presentation of introductory circuit analysis on an elementary, yet rigorous, level.

. (24)

REFERENCES
Rearranging, one has the following equation. 1. Davis, A., Linear Circuit Analysis, PWS, 1998. 2. Kemmerly, J. and Hayt, W., Engineering Circuit Analysis, 5th Edition, McGraw-Hill, 1993.

(25)

3. Nilsson, J. and Riedel, S., Electric Circuits, 5th Edition, Addison-Wesley, 1996. 4.Johnson, D., Johnson, J., Hilburn, J., and Scott, P., Electric Circuit Analysis, 3rd Edition, Prentice-Hall, 1997.

The determinant of the nodal conductance matrix is

5. Fosseprez, M., Nonlinear Circuits: Qualitative Analysis of Non-linear, Non-reciprocal Circuits, Wiley, 1992.

(26)

6.Johnson, D., Johnson, J., Hilburn, J., and Scott, P., Electric Circuit Analysis, 2rd Edition, Prentice-Hall, 1992. 7. Bruton, L., RC-Active Circuits, Prentice-Hall, 1980. 8. Desoer, C. and Kuh, E., Basic Circuit Theory, McGrawHill, 1969.

Thus, the matrix equation (25) has no solution if gm = 2; otherwise,

, ,
and

(27) (28)

(29)

This solution requires quite a bit of algebra, but solution of the same circuit using several techniques is useful for persuading students thatthough the big machinery of nodal analysis always worksspecial solution techniques are useful for certain circuits; alternatively, it might be useful in encouraging students to learn symbolic mathematics software packages. 0-7803-6424-4/00/$10.00 2000 IEEE
th

October 18 - 21, 2000 Kansas City, MO

30 ASEE/IEEE Frontiers in Education Conference F4E-28

Session F4E

0-7803-6424-4/00/$10.00 2000 IEEE


th

October 18 - 21, 2000 Kansas City, MO

30 ASEE/IEEE Frontiers in Education Conference F4E-29

Anda mungkin juga menyukai