Lecture 3:Microwave Network Analysis, Sparameters and impedance matching networks When do you need to carry out microwave network analysis? Objective: move from the requirement to solve for all the fields and waves of a structure to an equivalent circuit that is amenable to all the tools of the circuit analysis. Reasons to use network analysis over Maxwell’s equations: a) A field analysis using Maxwell’s equations is normally difficult and provide much more information than we need. b) We are only interested in the signal flow and the voltage and current at a set of terminals. c) Many RF/Microwave components/devices have more than 1 port and present cumbersome problems for complete field analysis (multiple interfaces). ELEC518, Kevin Chen, HKUST 1 • Circuit dimensions << wavelength – Lumped passive and active components. – Negligible phase change throughout the circuit. – Circuit theory --- Kirchhoff’s laws and Ohm’s law. • Circuit dimensions ~ wavelength – Distributed passive and active components. – Phase depends on position. Components are characterized by their dimensions, propagation constants and characteristics impedances. – Microwave network theory. ELEC518, Kevin Chen, HKUST 2 Impedance Matrix Impedance and Admittance Matrices • Two-terminal pair --> port • V and I ---> equivalent V and I. – Reference planes are defined to provide a phase reference for the equivalent V and I phasors. – At the nth reference plane, the total voltage and current are Vn = Vn+ + Vn− I n = I n+ − I n− • The impedance matrix that relates these voltages and currents: [V ] = [Z ][I ] V Z ij = i Ij I k = 0 for k ≠ j Zii: input impedance Zij: transfer impedance between ports i and j, (i ≠ j) An arbitrary N-port network. ELEC518, Kevin Chen, HKUST 3 ELEC518, Kevin Chen, HKUST 4 ELEC518, Kevin Chen, HKUST 5 Some Characteristics of Impedance and Admittance Matrices 6 The Scattering Matrix (S-parameter Matrix) • Reciprocal Networks (no active devices, ferrites, or plasmas -- no electrical or magnetic sources): defined as having identical transmission characteristics from port one to port two or from port two to port one --- circuit behavior independent of directions of waves and currents. Both Z and Y matrices are symmetric. Yij = Y ji Z ij = Z ji • Lossless Networks All the Z and Y elements are imaginary. * However, to determine Z and Y elements, both voltage and current values need to be measured. This is difficult at microwave frequencies. Furthermore, open and short circuits can easily result in oscillations in circuits. ELEC518, Kevin Chen, HKUST ELEC518, Kevin Chen, HKUST 7 • The exact value of voltage and current is difficult to define for non-TEM lines. • Difficult to deal with voltage and current in high frequency measurement. • It is more convenient to deal with the ratio of voltages or currents reflected or transmitted. • The scattering matrix of N-port networks with the same characteristic impedance at all ports is defined as V1− S11 S12 L S1N V1+ − M V2+ V2 = S 21 M M M − S NN VN+ VN S N 1 L ELEC518, Kevin Chen, HKUST 8 Generalized Scattering Matrix Conversion of impedance to admittance (Z to Y) Rule: simply rotating the Γ vector of the impedance by 180 degree. The scattering matrix of N-port networks with characteristic impedance Z0n at nth port is defined as b1 S11 b S 2 = 21 M M bN S N 1 S12 L S1N a1 M a2 M L S NN a N an = Vn+ Z0n Vn− bn = Z0n Z 0 n : Z 0 of the nth port • Sij is found by driving port j with V j+ , and measuring the reflected wave amplitude,Vi − , coming out of port i. The incident waves on all ports except the jth port are set to zero, which means that all ports are connected to matched loads. • The matched load has advantages in terms of its insensitivity to the transmission line length. ELEC518, Kevin Chen, HKUST 9 Frequency response of networks in Smith chart Impedances tend to move clockwise with frequency for passive networks ELEC518, Kevin Chen, HKUST 10 The Scattering Matrix (S-parameter Matrix) Tank Circuit: loop Why do we need this at all? small loops indicating selfresonances of the inductor and capacitor. • It is not practical to measure voltages and currents at the ports at microwave frequencies. • It is natural to deal with power in incident and reflected waves for microwave transmission lines. • Active devices may not be stable with short or open terminations due to oscillation. • The Scattering matrix relates the voltage waves incident on the ports to those reflected from the ports. • Most importantly, scattering matrix elements can be measured without open or short in the load, just matching loads. There is no reflected wave regardless of the length of the transmission lines used --- practical to implement. ELEC518, Kevin Chen, HKUST 11 ELEC518, Kevin Chen, HKUST 12 Example 4.4 on page 198 of Pozar. S parameters of the 3 dB attenuator circuit S-parameters for Reciprocal Networks and Lossless Networks S11 S12 0 0.707 S21 S22 = 0.707 0 • Reciprocal Networks – [S] is symmetric. For a 2x2 [S], S12=S21. • Lossless Networks – [S] is a unitary matrix. [S ]t [S ]* = [U ] N ∑S ki N * ∑ S ki S ki = 1 k =1 N ∑ S ki S kj* = 0 k =1 S = δ ij * kj k =1 ELEC518, Kevin Chen, HKUST 13 where [U] is the unit matrix. for i=j for i≠ j ELEC518, Kevin Chen, HKUST 14 Shift in Reference Plane Features on S-parameters • The reflection coefficient looking into port n is not equal to Snn, unless all other ports are connected to matched load. • The transmission coefficient from port m to port n is not equal to Snm, unless all other ports are connected to matched load. • The S parameters are properties of the network itself, and are defined under the condition that all ports are connected to matched loads. Changing the terminations or excitations of a network does not change its S parameters, but may change the reflection and transmission coefficients. ELEC518, Kevin Chen, HKUST 15 Open occurs in practical measurement setup. S11' = S11e − j 2φ1 ' S 21 = S 21e − j (φ1 +φ2 ) ELEC518, Kevin Chen, HKUST 16 The Transmission (ABCD) Matrix • Used for a cascade connection of two or more twoport connection. • Defined as V1 A B V2 V1 = A V2 + B I2 I = C D I I1 = C V2 + D I2 2 1 A two-port network: • For the cascade connection of two-port networks, we have V1 A1 B1 A2 B2 V3 I = C D C D I 1 1 1 2 2 3 ABCD Example: Quarter- and Half-Wave Transmission I2 Lines l V1 Zo ZL The ABCD matrix of a length of transmission line l of Zo and ß is A cascade connection: ELEC518, Kevin Chen, HKUST 17 A = cos ßl B = j Zo sin ßl C = j Yo sin ßl D = cos ßl ELEC518, Kevin Chen, HKUST 18 ELEC518, Kevin Chen, HKUST 20 • Note that, for cos ßl = 0 (that is, for ßl=π/2, a quarter wavelength or odd multiple) the ABCD matrix becomes simply A=0 C = j Yo B = j Zo D=0 which implies that I2 = V1/j Zo independent of V2 or I1. • Similarly, if sin ßl = 0 (that is, for ßl=π, a half wavelength or multiple) the ABCD matrix becomes simply A = -1 C=0 B=0 D = -1 which implies that V2 = -V1 and I2 = -I1 independent of the terminating impedance at end 2. ELEC518, Kevin Chen, HKUST 19 Two-port network with certain network parameters can lead to various equivalent circuit formation. T equivalent π equivalent ELEC518, Kevin Chen, HKUST 21 Impedance Matching • Why impedance matching? – Maximum power is delivered when the load is matched to the line. – Impedance matching sensitive receiver components (antenna, LNA, etc.) improves the signal-to-noise ratio of the system. – Impedance matching in a power distribution network (such as antenna array feed network) will reduce amplitude and phase errors. – Impedance matching uniquely removes the requirement for a specific reference plane. – Provide reliable and predictable interconnections between components in a system. ELEC518, Kevin Chen, HKUST 22 Methods of Impedance Matching • • • • • Matching stubs (shunts or series, single or multiple) Quarter-wavelength transformers (single or multiple) Lumped elements Tapered transmission lines Combination of the above A lossless network matching an arbitrary load impedance to a transmission line. Z0 Load Matching network Two-port matching network One-port matching ZL Multiple solutions ELEC518, Kevin Chen, HKUST 23 Design issues of the matching networks • Complexity --- Simplest design that satisfies the required specification is generally the most preferable. Cheaper, more reliable, less lossy. • Bandwidth --- Normally, it is desirable to match a load over a band of frequencies. Increased bandwidth usually comes with increased complexity, e.g. using multistage matching. • Frequency --- Matching networks are usually optimized for a particular frequency. ELEC518, Kevin Chen, HKUST 24 L networks consist of two reactive components (inductor and capacitor), which results in eight different configurations. • Implementation --- Choose the right type of matching networks, either tuning stub or transmission line. • Adjustability --- This maybe required for applications where a variable load impedance occurs. Matching with Lumped Elements (L Networks) Network for zL inside the 1+jx circle (smith chart). Network for zL outside the 1+jx circle (smith chart). ELEC518, Kevin Chen, HKUST 25 Analytic solution for the matching network elements ELEC518, Kevin Chen, HKUST and Case 1: load impedance inside the 1+jx circle ---> RL>Z0 For a match looking into the matching network, we have Z 0 = jX + Z 1 X L Z0 + − 0 B RL BRL Both solutions are applicable for impedance matching at a single frequency. But one solution may be preferable over the other one when other performance, e.g. bandwidth, is considered. 1 jB + 1 /( RL + jX L ) Solving for X and B from the two equations for real and imaginary parts, Case 2: load impedance outside the 1+jx circle ---> RL<Z0 Solutions are: X ± RL / Z 0 RL2 + X L2 − Z 0 RL B= L RL2 + X L2 X = ± RL ( Z 0 − RL ) − X L B=± *Note: B is always real (RL>Z0) and has two solutions. One solution is capacitive (positive) and the other one is inductive (negative). ELEC518, Kevin Chen, HKUST X= 26 27 ( Z 0 − RL ) / RL Z0 *Analytic solution is computing intensive and lack of intuition. ELEC518, Kevin Chen, HKUST 28 Smith Chart Solutions of Matching Networks Impedance effect of series and shunt connections of L and C to a complex load in the Smith Chart The effect of connecting a single reactive component (either capacitor or inductor) to a complex load zL • The addition of a reactance connected in series with a complex impedance results in motion along a constant-resistance circle in the combined Smith Chart. • A shunt connection produces motion along a constantconductance circle. A general rule of thumb for rotation in the Smith Chart • When an inductor is involved, we rotate in the direction that moves the impedance into the upper half of the Smith Chart. • In contrast, a capacitance results in the move toward the lower half. ELEC518, Kevin Chen, HKUST 29 ELEC518, Kevin Chen, HKUST 30 Solution: Example: Step 1: Compute normalized transmitter and antenna impedances. Since no characteristic impedance Z0 is given, we arbitrarily select Z0 = 75 Ω for simplicity. 1 We have zT = ZT /Z0 = 2 + j 1 zA = ZA /Z0 = 1 + j 0.2 Step 2: Taking into account the first element (the shunt capacitor) connected to the transmitter. Move down on the circle of the constant conductance. Step 3: Taking into account the next element (the series inductor) connected to the transmitter. Figure 1: Transmitter to antenna matching circuit design. ELEC518, Kevin Chen, HKUST Move up on the circle of the constant resistance. 31 ELEC518, Kevin Chen, HKUST 32 Step 4: Draws the complex conjugate of the antenna impedance in the Smith Chart for maximum power transfer. This should be the output impedance of the matching network. Design of the matching network using ZY Smith Chart zM = zA* = 1 - j 0.2 Step 5: Find the normalized impedance of the intersection of two circles. zTC = 1 - j 1.22 and the corresponding admittance of yTC = 0.4 + j 0.49. There is another path connecting zM and zT. The normalized susceptance of the shunt capacitor is jbC = yTC - yT = j 0.69 C = bC /(ωZ 0 ) = 0.73 pF What does this mean? and the normalized reactance of the inductor is jxL = zA - zTC = j 1.02 L = ( xL Z 0 ) / ω = 6.09nH ELEC518, Kevin Chen, HKUST 33 Procedures of designing impedance matching networks using Smith Chart 34 Design procedures cont’ 1 Find the normalized source and load impedances. 2 In the Smith Chart, plot ircuits of constant resistance and conductance that pass through the point denoting the source impedance. 3 Plot circles of constant resistance and conductance that pass through the point of the complex conjugate of the load impedance. 4 Identify the intersection points between the circles in steps 2 and 3. The number of intersection points determins the number of possible L-section matching networks. (cont’) ELEC518, Kevin Chen, HKUST ELEC518, Kevin Chen, HKUST 35 5 Find the values of the normalized reactances and susptances of the inductors and capacitors by tracing a path along the circles from the source impedance to the intersection point and then to the complect conjugate of the load impedance. --- there are usually multiple paths (multiple solutions). 6 Determine the actual values of inductors and capacitors for a given frequency. ELEC518, Kevin Chen, HKUST 36 Smith Chart Solution 2 (not using combined ZY Smith Chart) Example 5.1 on Page 254 of Pozar Design an L section matching network to match a series RC load with an impedance ZL = 200 - j 100 Ω, to a 100 Ω line, at a frequency of 500 MHz. Step 3: Convert back to impedance. Step Solution: Step 2 1 Therefore we have b = 0.3, x = 1.2 (check this result with the analytic solution). Then for a frequency at f = 500 MHz, we have C= b = 0.92 pF 2πfZ 0 L= xZ 0 = 38.8nH 2πf 3 37 ep St ELEC518, Kevin Chen, HKUST 4 Step 2: Move the load impedance to the impedance circle of 1+ jx (done in admittance Smith Chart) -- add j 0.3 in susceptance ep St Step 1: Convert the load impedance to admittance by drawing the SWR circle through the load, and a straight line from the load through the center of the Smith Chart. Step 4: Move to the center of the Smith Chart by adding an series inductor ELEC518, Kevin Chen, HKUST 38 There are two solutions for the matching networks. In this case, there is no substantial difference in bandwidth between the two solutions. Is there another solution? ELEC518, Kevin Chen, HKUST 39 ELEC518, Kevin Chen, HKUST 40 Impedance Matching Using a Single Shunt Stub Microstrip Matching Networks • A microstrip line can be used as a series transmission line, as an open-circuited stub, or as a short-circuited stub. Single-stub tuning • A series microstrip line together with a short- or opencircuited shunt stub can transform a 50-Ω resistor into any value of impedance. Shunt stub Tuning Procedures: • Find the proper d so that Y = Y0 + jB • Choose the stub susceptance (decided by l) to be -jB Example 5.2 on Page 259 of Pozar Tuning parameters: d and stub reactance or susceptance Series stub ELEC518, Kevin Chen, HKUST Design a single-stub shunt tuning network to match a load impedance ZL = 15 + j 10 Ω to a 50 Ω line. 41 ELEC518, Kevin Chen, HKUST 42 Solution: y2 zL Working with the Smith Chart! yL d2 y1 Solution 1 has a significantly better bandwidth than solution 2. d1 Shorter stub produces wider bandwidth. ELEC518, Kevin Chen, HKUST 43 ELEC518, Kevin Chen, HKUST 44 Microstrip Discontinuities 90O bend or corner: Microstrip lines and Propagation Velocity Fields exist partly in air and partly in dielectric. c εr Propagation velocity is between the velocity in air c and the velocity in the dielectric c . d w < v PCB < c εr Define an “effective dielectric constant” εre v PCB = c ε re ε re = εr +1 εr −1 2 + 2 (1) capacitance arises through additional charge accumulation at the corners --- particularly around the outer point of the bend where electric fields concentrate. (2) inductances arise because of current flow interruption. (3) the impedances of these additional components can be comparable to the line characteristic impedance at microwave frequencies. 1 1 + 12d / W ELEC518, Kevin Chen, HKUST 45 Matched microstrip bends: compensation techniques Using curved and mitered bends to reduce the effect of the additional capacitance --- mitered bends are as good as, or better than curved bends at frequencies up to 10 GHz. ELEC518, Kevin Chen, HKUST Step changes in width (impedance steps) T-junctions Effect: the length of the w2 section is lengthened. Typical value for the mitering fraction is 1 − b / 2 w = 0 .6 46 Effect: length correction is needed. Mitering fraction 1− b / 2 w b = 0.57 w ELEC518, Kevin Chen, HKUST 47 ELEC518, Kevin Chen, HKUST 48 ELEC518, Kevin Chen, HKUST 49 ELEC518, Kevin Chen, HKUST 50