S-Parameters and Smith Charts

Nick Gamroth Oct 2004

Abstract The following are my self-study notes on S-Parameters and Smith Charts. So its probably all wrong.

S-Parameters

Check out this two-port network:

2-Port Zs
a1 a2

Vs

Zl

b1

b2

Figure 1: Thats hot

How about those arrows? Those are the incident and reected voltage waves at each port, normalized to the characteristic impedance, Z0 . Heres how that looks: Vi1 a1 = Z0 Vi2 a2 = Z0 Vr1 b1 = Z0 Vr2 b2 = Z0 (1) (2) (3) (4)

Nicholas Gamroth
Where, Vi1 = Voltage wave incident on port 1, Vi2 = Voltage wave incident on port 2, Vr1 = Voltage wave reected from port 1, Vr2 = Voltage wave reected from port 2. Thats OK, but its not that great. This is though: S11 = S22 = S21 = S12 = b1 a1 b2 a2 b2 a1 b1 a2 (5)
a2 =0

(6)
a1 =0

(7)
a2 =0

(8)
a1 =0

Wow. Its not obvious to me why these are useful, but they are. S11 and S22 are input and output reection coecients, respectively. S11 is the input reection coecient with the output port terminated in a matched load, as in when ZL = Z0 , a2 > 0. S22 is the output reection coecient when the input port is terminated in a matched load. S21 and S12 are the forward and reverse transmission gains, respectively with both being terminated in matched loads. I guess.

Reection Coecients

Reection coecients are the same as in transmission lines. The have a complex and real part, and represent the quality of the impedance match between the load and the characteristic impedance of the line. = r + j i = ZL Z0 ZL + Z0 (9)

Sometimes, folks normalize ZL , so you can write: = Not too shabby. ZL 1 ZL + 1 (10)

Nicholas Gamroth

Impedance and Admittance

Remember impedance and admittance? I didnt, so I wrote it down. Its not that important for the purpose of this paper, but I think its something I ought to know.

3.1

Impedance
Z = R + jX (11)

Where Z =impedance, R =resistance, and X =reactance. For and inductor and a capacitor, X = jL X= 1 jC (12) (13)

3.2

Admittance
Y = G + jB (14)

Where Y =admittance, G =conductance, and B =susceptance. For and inductor and a capacitor, X= 1 jL (15) (16)

X = jC

Smith Charts

Now is when it gets hot. Check it out. In the Smith chart in gure 2, the blue circle segments are the circles of constant reactance, and the green circles are the circles of constant resistance. Thats the simplest explanation of a Smith chart I can nd. You can plot things like a complex impedance on a Smith chart. Lets try it. Z = 50 + j 100 First, normalize this complex impedance to 50 z = 1 + j2 (18) (17)

That was easy. The point right at the center of a Smith chart is the origin. Most charts give more detail that Figure 2, and theyll have

Nicholas Gamroth

Figure 2: YOW!
things like numbers that you can use to plot things. So, the real part of z is the resistance, and the imaginary part is the reactance. To plot this complex number on the Smith chart, just go to 1 on the circle of constant resistance, and 2 on the circle of constant reactance. Simps! I did this on Figure 3. Point O is the origin, point A is the intersection of the circles of resistance 1 and reactance 2. Hmm, i wonder what the reection coecient of this complex resistance is? Ok, its = 0.5 + j 0.5, or in polar, 0.707/45deg, on most Smith charts, theres a angle scale printed around the red circle in Figure 3. Im not sure about the magnitude, but it appears to be the distance between OA and O to the red circle. Yeah, thats it. But I dont know a way to get it by visual inspection (with eyeballs). Heres another tip: to plot something like 10 j 50, put that mess on the lower half of the chart. Thats where the X < 0 circles live. Actually, everything on the top half of a Smith chart represents inductance, and everything on the bottom half represents capacitance. As in an inductive or capacitive impedance. Oh yeah, and when you plot an impedance, itll change with frequency. So, if you have a RL circuit, its impedance on the smith chart will look like a line around one of the circles of constant resistance (because resistance doesnt change with frequency, natch). Plus, it normally moves clockwise as frequency increases. Thats about it for plotting things on Smith charts. http://www.web-ee.com/primers/les/SmithCharts/smith charts.htm.

Nicholas Gamroth

R=1

X=2 O

theta

Figure 3: Plot it

Amplier Stability

Check out Figure 4. Thats a generalized amplier setup. The S is the amplier. You can use S-Parameters to gure out if the amplier is unconditionally stable, and if its not, you can gure out when it is, and how to design source and load circuits to make it stable.
s L

Zs

Vs

Source Match

Load Match

Zl

in

out

Figure 4: Pump it up!

People write S-Parameters in a matrix, like this: S= S11 S12 S21 S22 (19)

There are two important numbers that S-Parameters give you,

Nicholas Gamroth

that can determine the stablility of an amplier. The rst one is the determinant of the S matrix and is called . The second is K, which is given by the following formula: K= 1 + ||2 |S11 |2 |S22 |2 2 |S12 S21 | (20)

If || < 1 and K > 1, then the amplier is unconditionally stable. Otherwise, you can create a stability circle to see when the amplier is stable and when its not. For the output, the center of this circle is CL and the radius is RL CL =
) (S22 S11 |S22 |2 ||2

(21)

RL =

For the input, the center is CS and the radius is RS CS =

) (S11 S22 |S11 |2 ||2

S12 S21 |S22 |2 ||2

(22)

(23)

RS =

These circles can be plotted on a Smith chart for added fun. Take a gander at Figure 5. Those got plotted using the equations just given. Theres a couple of conditions to determining whether or not the amplier is stable. First, theres a couple of parameters: D1 = |S11 |2 ||2 D2 = |S22 |2 ||2 (25) (26)

S12 S21 |S11 |2 ||2

(24)

D1 is associated with the source, and D2 is associated with the load. Depending on the sign of the D parameter, the stable region of the amplier is either outside or inside the stability circle. For Dx > 0, the amplier is stable outside of the stability circle, and for Dx < 0, the amplier is stable inside the stability circle. If you look at Figure 5, and assume that D2 > 0, then we can say that for the load, the amplier is stable for all regions that arent to the upper right of the arc labeled Load Stability Circle. If things go well, youll have a unconditionally stable amplier. If this happens, depending on the Dx parameter, the stability circle will either be outside of the Smith chart, or completely encompass

Nicholas Gamroth

Load Stability Circle

Source Stability Circle

Figure 5: Feet shoulder width apart for better stability

it. If you see this, give yourself a pat on the back for making such a wonderful amplier. Otherwise, dont act so cocky. Since all that mess would be a hassle to calculate and draw, look at the S.J. Orfanidi reference to download a whole gang of MATLAB scripts to do it for you. Thats what I made the Smith chart graphics with, and its pretty easy to use. Mostly, you just have to put in the S-Parameters and it does the rest for you.

Unilateral Figure of Merit

U=
S S S | |S11 12 21 22

Heres a little gem called the Unilateral Figure of Merit. 1 |S11 |2 1 |S22 |2 (27)

A lot of times, you dont get a lot of reverse transmission, aka |S12 | << |S21 |. When this happens, you can call the two-port unilateral. Heres gu : 1 gu = (28) |1 U |2

Nicholas Gamroth

gu is called the relative gain ratio, and if its near 1, you can probably treat the two-port as unilateral. A lot of things simplify somewhat in power calculations because you can set S12 == 0.

Power Gains

When you make an amplier, you want it to plump up the volume as much as possible. That doesnt always happen if you just put in loads and sources all willy-nilly. You gotta think about that stu. The rst step in thinking about that stu is guring out how much gain you can get, and things like that. Theres three of these gains that tell you things like that. Theyre all sort of messes, so Im not going to write them down. I will list them though. They are: transducer power gain, GT , available power gain, Ga , and operating power gain, Gp . So, GT seems to be the most important, because it accounts for the eects of both the source and load impedances. Whats most important is that the transducer power gain is maximized when the two-port is simultaneously conjugate matched. That happens when in = S and L = out . This also makes the three gains given above equal. This gain is called the maximum available gain. Looks like K 1 for this to happen. Im mostly condensing this stu from the Sophocles Orfanidis (what an awesome name) book listed in the references, so for more detail, take a look there.

Designing Matching Networks

Now things might actually get useful, at least for ampliers. Well see. So the big deal here is guring out S and L to conjugate match the source and load. The next two equations are the long way to get this done: S12 S21 L (29) S = S11 + 1 S22 L L = S22 + S12 S21 S 1 S22 S (30)

Thats not too bad, but if you look at it close, you get excited when you see what happens in the unilateral case (S12 = 0):
S = S11 L = S22

(31) (32)

Nicholas Gamroth

What could be easier? Its possible to solve it in the bilateral case (bilateral is the opposite of unilateral, dude), but it involves some parameters I didnt mention. Theyre easy to get though, and the industrious reader will have no trouble guring out how to solve the bilateral case. (Hint, theyre simultaneous equations.)
C1
Zs

Vs

L1

Load Match

Zl

Figure 6: Its a match

OK. Enough fooling around. Lets design a matching network for a two-port. Look at gure 4. Were going to design a source matching network that looks like Figure 6. There are other ways to do impedance matching, but I like circuits, so were going to use circuits. So, rst we nd the simultaneous conjugate match by using the equations above. Next, convert to an impedance (were just doing the source side, so convert S ). Now we have ZS which we convert to Zin by taking its complex conjugate. Finally, we just plug this Zin , along with our ZS into a couple of equations, and were done. Those equations are: (Note: Z = R + jX ) Q=
2 XS RS 1+ RL RS RL

(33) (34) (35)

X1 =

OK, were not really done. The Xs from these equations are like complex impedances for the matching network, so we need to pick values for them. To pick the value for the inductor, use X1 like this: X1 (36) L= And to nd the value for the capacitor, use X2 like this: 1 C= (37) X2 Where is the frequency youre running the circuit at.

X2 = (XL RL Q)

XS RS Q RS RL 1

Nicholas Gamroth

References

Since I didnt know anything about S-Parameters or Smith charts before I wrote this, its all plagarized. Most of it was taken from the following. S.J. Orfandis. Electromagnetic Waves & Antennas, Chapter 12. Draft available at www.ece.rutgers.edu/ orfanidi/ewa http://www.web-ee.com/primers/les/SmithCharts/smith charts.htm.

10