Smith chart and impedance matching

In short

The Smith chart is a polar plot of the voltage reflection coefficient Γ on which circles of constant normalized resistance and reactance are overlaid, so that impedance, admittance, VSWR, return loss and the effect of a length of transmission line can all be read off one graph. Impedance matching is the design job of inserting a network — lumped L-section, quarter-wave transformer, stub or taper — between a load and a line so that the source sees Z₀ and almost no power is reflected. It is examined every year in GATE (EC) under "Smith chart, impedance matching" and in university papers on Transmission Lines, Microwave Engineering and RF Design, usually as a numerical plus a design problem. On the job it is the daily tool of RF/antenna engineers at ISRO (SAC, URSC), DRDO (LRDE, DEAL), BEL, SAMEER, Astra Microwave, Data Patterns, Tejas Networks, Qualcomm, Nokia and Ericsson India, and of anyone using a vector network analyser.

Explained

Reflection coefficient: the quantity the chart plots

• Load reflection coefficient: Γ_L = (Z_L − Z₀)/(Z_L + Z₀). Normalizing by the line impedance, z = Z/Z₀, gives Γ = (z − 1)/(z + 1) and inversely z = (1 + Γ)/(1 − Γ).
• Every passive load (R ≥ 0) maps into the unit circle |Γ| ≤ 1: the whole right half of the impedance plane is folded into a finite disc — that compression is the chart's entire reason for existing.
• Landmarks: centre Γ = 0 → z = 1 (matched); right edge Γ = +1 → z = ∞ (open); left edge Γ = −1 → z = 0 (short); top half inductive (+x), bottom half capacitive (−x).
• Moving down a lossless line: Γ(d) = Γ_L e^−j2βd — magnitude unchanged, phase retarded by 2βd. On the chart this is a pure rotation clockwise (toward the generator) on a constant-|Γ| circle; counter-clockwise is toward the load.
• One full rotation = 2βd = 2π → d = λ/2: line impedance repeats every half wavelength, and inverts (z → 1/z) every quarter wavelength.
• |Γ| > 1 (outside the chart) means negative resistance — oscillator and stability work uses a compressed Smith chart for that region.

How the chart is constructed

• Writing Γ = u + jv and z = r + jx in Γ = (z−1)/(z+1) and separating real and imaginary parts gives two families of circles — this derivation is a common 5-mark question.
• Constant-r circles: centre (r/(1+r), 0), radius 1/(1+r). All pass through Γ = +1; r = 0 is the unit circle itself, r = 1 passes through the centre, r = ∞ is the point Γ = 1.
• Constant-x circles: centre (1, 1/x), radius |1/x|. Arcs also passing through Γ = +1; x = 0 is the real (horizontal) axis.
• The two families are orthogonal everywhere they cross — hence a unique (r, x) for every point.
• Peripheral scales: angle of Γ in degrees, wavelengths toward generator (WTG) and wavelengths toward load (WTL), 0 to 0.5λ. Radial scales below the chart read |Γ|, VSWR, return loss in dB, mismatch (reflection) loss and standing-wave ratio in dB.
• Admittance (Y) chart: the same chart read with normalized y = g + jb; it is the impedance chart rotated 180°, i.e. the point diametrically opposite z is 1/z. The immittance (ZY) chart overlays both, which is what stub and L-section design actually needs — series elements move along r-circles, shunt elements along g-circles.

Reading and using the chart: standard procedures

1. Plot a load: normalize (z_L = Z_L/Z₀), find the intersection of its r- and x-circles.
2. Input impedance of a line of length l: draw the constant-|Γ| (SWR) circle through z_L and rotate clockwise through l in WTG units (or 2βl degrees); read r + jx and de-normalize by ×Z₀.
3. VSWR: the SWR circle cuts the positive real axis at r = S; S = (1+|Γ|)/(1−|Γ|).
4. Voltage maximum occurs where the rotated Γ is real positive (positive real axis, z = S); the minimum a quarter-wavelength away (z = 1/S). Distance from load to first minimum is a classic measurement-based question.
5. Load from a slotted-line measurement: given S and the distance to a voltage minimum, start at z = 1/S on the SWR circle and rotate toward the load by that distance.
6. Admittance: read the point 180° opposite, or simply use the Y-chart overlay.
7. De-embedding: a known connector or feed length is removed by rotating toward the load by its electrical length — exactly what a VNA port extension does.

Why match, and lumped (L-section) matching

• Reasons: maximum power transfer to the load; protection of the source/PA from reflected power; minimum ripple and loss on the feed; stable gain and noise figure (an LNA's optimum noise match differs from its power match); avoidance of standing-wave voltage peaks that puncture dielectrics.
• Conjugate match (Z_in = Z_S*) maximizes delivered power; a Z₀ match (Z_in = Z₀) maximizes flatness and minimizes reflection on the line. They coincide when Z_S = Z₀, which is the usual 50 Ω assumption.
• L-section: two reactances. If R_L > Z₀, put the shunt element across the load first, then the series element; if R_L < Z₀, series element first, then shunt. The shunt element pulls the point onto the g = 1 circle (or r = 1 circle), the series element walks it to the centre.
• Design formulas (real R_L): Q = √(R_high/R_low − 1); shunt susceptance B = Q/R_high; series reactance X = Q·R_low. Four combinations (L–C, C–L, etc.) exist; choose by DC-blocking/biasing needs and component realizability.
• Pi and T networks add a third element so that the loaded Q — hence the bandwidth — can be chosen independently; constant-Q contours drawn on the Smith chart show how far a point may be moved before bandwidth collapses.
• Lumped matching is practical up to roughly 2–3 GHz with 0402/0201 SMD parts; above that parasitics and self-resonance dominate and distributed matching takes over.

Transmission-line matching: quarter-wave transformer and stubs

• Quarter-wave transformer: for a real load, a λ/4 section of Z₁ = √(Z₀R_L) gives a perfect match at the design frequency, because Z_in = Z₁²/R_L. Narrowband; a complex load must first be rotated to a real value (to a voltage max or min) or absorbed by a series line.
• Single-stub matching: a short- or open-circuited stub of adjustable length in shunt (common, easy in microstrip) or in series (easy in coax/slabline), placed at a distance d from the load.
  • Shunt procedure: rotate from y_L toward the generator until the g = 1 circle is reached → that is d (two solutions per half wavelength); the residual susceptance +jb is cancelled by a stub of susceptance −jb.
  • Stub susceptance: short-circuited stub y = −j cot βl, open-circuited stub y = +j tan βl. The condition for reaching g = 1 reduces analytically to cos(∠Γ_L − 2βd) = −|Γ_L|.
  • Choice: shorted stubs radiate less and are preferred in coax/waveguide; open stubs avoid a via and are preferred in microstrip. Pick the shorter stub and shorter d for wider bandwidth.
• Double-stub tuner: two stubs of adjustable length at a fixed spacing d (typically λ/8 or 3λ/8) — matches without moving the stub position, so it suits lab tuners and measurement benches. First stub brings the point onto the g = 1 circle rotated by d; a forbidden region exists — loads with g_L greater than about 1/sin²(βd) cannot be matched (g_L > 2 for d = λ/8). A triple-stub tuner removes the forbidden region.
• Stub matching is inherently narrowband because both d and l are electrical lengths that scale with frequency.

Broadband matching, bandwidth limits and Indian practice

• Multisection transformers: several λ/4 sections with gradually changing impedance. Binomial (maximally flat) gives the flattest in-band response; Chebyshev gives equal ripple and the widest band for a given ripple level; Kuroda's identities and Richards' transformation convert lumped prototypes into line sections.
• Tapered lines: exponential, triangular or Klopfenstein tapers — continuous impedance variation, a smooth wideband match used in horn feeds, baluns and TRM chains.
• Bode–Fano limit: for any load with reactive storage there is a hard trade-off between achievable |Γ| and bandwidth — for a parallel RC load, ∫ln(1/|Γ|)dω ≤ π/(RC). This is why a tiny antenna cannot be matched broadband, no matter how clever the network.
• Loss: real lines and lossy components spiral the trajectory inward toward the centre; a "good looking" match on a long lossy cable may just be attenuation, not a real match. Series resistance in a matching network dissipates power and raises the noise figure.
• Typical Indian specifications and practice: antenna and RF module datasheets call for return loss ≥ 10 dB (VSWR ≤ 1.92) as a minimum and ≥ 14 dB (VSWR ≤ 1.5) for good designs; 50 Ω is standard for RF/telecom (SMA, N-type), 75 Ω for cable TV and video. Telecom equipment sold in India needs MTCTE certification from TEC (Telecommunication Engineering Centre, DoT) under the Indian Telegraph (Amendment) Rules; radio transmitters need WPC/SACFA clearance, and spectrum/emission conditions come under TRAI and DoT. EMI/EMC testing follows CISPR/IEC-based IS standards — exact clause numbers and scope are revised periodically, so verify against the current TEC/WPC notification.
• Tools: VNA measurement (SOLT/TRL calibration) displays S₁₁ directly on a Smith chart; design is done in Keysight ADS, AWR, Ansys HFSS, CST, or free tools like QUCS and smith-chart calculators. Student labs at Indian universities still use a slotted line or a trainer kit plus a paper Smith chart — both the graphical and the analytical answer are expected in exams.

Key points

• Smith chart - polar plot of Γ with constant-r and constant-x circles superimposed; valid for lossless or lossy lines.
• Normalization - always divide by Z₀ before plotting and multiply after reading; the chart is Z₀-independent.
• Γ = (z−1)/(z+1) - the mapping; z = (1+Γ)/(1−Γ) is the inverse.
• Clockwise = toward generator - rotation angle 2βl; one full turn = λ/2.
• λ/4 inverts - z → 1/z, so a short looks like an open and an inductive load looks capacitive.
• SWR circle - constant |Γ| circle; its crossing of the positive real axis equals S.
• Short / open / match - Γ = −1 (left edge), +1 (right edge), 0 (centre).
• Upper half inductive, lower half capacitive; the real axis is purely resistive.
• Admittance reading - the diametrically opposite point gives 1/z; series elements follow r-circles, shunt elements follow g-circles.
• Return loss - RL = −20 log|Γ| dB; mismatch loss = −10 log(1−|Γ|²) dB.
• Quarter-wave transformer - Z₁ = √(Z₀R_L), real loads only, narrowband.
• Single stub - move to the g = 1 circle, then cancel jb with a shorted or open stub; two solutions exist.
• Double stub - fixed spacing, adjustable lengths, has a forbidden region g_L > 1/sin²βd.
• L-section - Q = √(R_high/R_low − 1) fixes the bandwidth; no independent control without a third element.
• Conjugate match ≠ Z₀ match unless the source impedance equals Z₀.
• Loss spirals inward - attenuation can masquerade as a good match.
• Bode–Fano - reactive loads impose an absolute bandwidth-versus-match trade-off.
• Shorted stubs for coax/waveguide, open stubs for microstrip - radiation versus via convenience.

Formulas and examples

Example 1 — Γ, VSWR, return loss and voltage maximum. Z₀ = 50 Ω, Z_L = 100 + j50 Ω.

• z_L = 2 + j1 → Γ = (1 + j1)/(3 + j1) = (1+j)(3−j)/10 = (4 + j2)/10 = 0.4 + j0.2
• |Γ| = √(0.16 + 0.04) = 0.4472, ∠Γ = arctan(0.2/0.4) = 26.57°
• S = 1.4472/0.5528 = 2.618; Return loss = −20 log(0.4472) = 7.0 dB
• Mismatch loss = −10 log(1 − 0.2) = 0.97 dB, i.e. 80 % of incident power reaches the load
• First voltage maximum: rotate 26.57° toward generator → d_max = 26.57/720 λ = 0.0369λ; first minimum at 0.0369 + 0.25 = 0.2869λ. At the maximum z = 2.618 (real).

Example 2 — input impedance after a line length. Z₀ = 50 Ω, Z_L = 25 + j25 Ω, line length 0.2λ.

• z_L = 0.5 + j0.5 → Γ_L = (−0.5 + j0.5)/(1.5 + j0.5) = 0.4472 ∠116.57°
• Rotate clockwise 2βl = 2(0.2)(360°) = 144° → Γ_in = 0.4472 ∠(116.57 − 144) = 0.4472 ∠−27.43° = 0.397 − j0.206
• z_in = (1 + Γ)/(1 − Γ) = (1.397 − j0.206)/(0.603 + j0.206) = 1.97 − j1.01
• Z_in = 98.5 − j50.7 Ω; S is still 2.618 (lossless line cannot change |Γ|). The load was inductive-low-R; 0.2λ of line has turned it into high-R capacitive.

Example 3 — shunt single-stub match. Z₀ = 50 Ω, Z_L = 60 − j80 Ω, short-circuited stub.

• z_L = 1.2 − j1.6 → Γ_L = (0.2 − j1.6)/(2.2 − j1.6) = 0.593 ∠−46.85°, S = 1.593/0.407 = 3.91
• Match condition cos(∠Γ_L − 2βd) = −|Γ_L| = −0.593 → ∠Γ_L − 2βd = ±126.36°
• Solution 1: 2βd = −46.85 + 126.36 = 79.51° → d = 0.110λ, where y = 1 + j1.472
• Stub must give b = −1.472: −cot βl = −1.472 → tan βl = 0.679 → βl = 34.2° → l = 0.095λ
• Solution 2: 2βd = −173.21° (+360°) = 186.79° → d = 0.259λ, y = 1 − j1.472, l = 0.405λ
• Pick solution 1 — shorter stub and shorter line give the wider bandwidth and lower loss.

Example 4 — L-section lumped match at 1 GHz. Match R_L = 200 Ω to Z₀ = 50 Ω.

• R_L > Z₀ → shunt element at the load, series element toward the source. Q = √(200/50 − 1) = √3 = 1.732
• B = Q/R_L = 1.732/200 = 8.66 mS → C = B/ω = 8.66×10⁻³/(2π×10⁹) = 1.38 pF (shunt)
• X = Q·Z₀ = 1.732 × 50 = 86.6 Ω → L = X/ω = 86.6/(6.283×10⁹) = 13.8 nH (series)
• Check: Y = 1/200 + j8.66 mS = (5 + j8.66) mS → Z = 1/Y = 50 − j86.6 Ω; adding +j86.6 Ω gives exactly 50 Ω. Loaded Q = 1.73 → usable bandwidth roughly 50–60 % of f₀.

Example 5 — quarter-wave transformer, physical length and bandwidth. Match a 100 Ω load to 50 Ω at 2.4 GHz on a substrate with ε_eff = 2.2.

• Z₁ = √(50 × 100) = 70.71 Ω
• λ = c/(f√ε_eff) = 3×10⁸/(2.4×10⁹ × 1.483) = 0.0843 m → λ/4 = 21.1 mm
• Bandwidth for Γ_m = 0.1 (RL 20 dB): 2√(Z₀R_L)/|R_L − Z₀| = 141.42/50 = 2.828; Γ_m/√(1−Γ_m²) = 0.1005; product = 0.2843 → arccos = 73.48° = 1.2825 rad
• Δf/f₀ = 2 − (4/π)(1.2825) = 2 − 1.633 = 0.367 → 36.7 %, i.e. about 2.0–2.8 GHz.

Where it is used

• Antenna and RF module design: matching patch, PIFA and chip antennas in handsets, Wi-Fi/BLE modules and IoT nodes; return loss and VSWR are the acceptance criteria in TEC/MTCTE and customer test reports.
• Defence and space electronics: DRDO (LRDE, DEAL), BEL, Astra Microwave, Data Patterns and ISRO (SAC) match transmit–receive modules, LNAs, phase shifters and feed networks in radar and satellite payloads.
• Telecom infrastructure: base-station PAs, duplexers, tower-top amplifiers and feeder runs at Tejas Networks, Nokia, Ericsson, Jio and Airtel sites; field engineers sweep feeders with a VNA/site-master and read S₁₁ on a Smith chart.
• RFIC and SoC design roles: matching and input/output networks for LNAs, mixers and PAs (Qualcomm, Broadcom, MediaTek, Signalchip, Saankhya Labs); noise-circle and gain-circle design is done on the same chart.
• Test, measurement and EMC labs: SAMEER, ERTL and NABL-accredited labs use calibration standards, tuners and Smith-chart de-embedding; also industrial RF — plasma, induction heating and medical diathermy use matching networks.
• Job roles: RF design engineer, antenna engineer, microwave/radar engineer, EMC test engineer, RF field/validation engineer — Smith-chart questions are standard in their interviews.

Interview questions

• Q: Why is the Smith chart drawn inside a unit circle? A: Because for any passive load (R ≥ 0) the reflection coefficient magnitude is ≤ 1, so the whole right half of the impedance plane maps into |Γ| ≤ 1.
• Q: What does one complete rotation around the chart correspond to? A: Half a wavelength of line (2βd = 2π), which is why line impedance repeats every λ/2.
• Q: Which direction do you rotate to move from load to source? A: Clockwise — the "wavelengths toward generator" scale — through an angle 2βl.
• Q: Why is a λ/4 shorted stub used as an RF choke or bias feed? A: A λ/4 line inverts impedance, so a short at one end looks like an open circuit at the other — it blocks RF while passing DC.
• Q: How do you read admittance on an impedance Smith chart? A: Take the point diametrically opposite through the centre on the same |Γ| circle; that is 1/z.
• Q: Difference between conjugate matching and Z₀ matching? A: Conjugate matching (Z_in = Z_S*) maximizes power delivered from a given source; Z₀ matching minimizes reflections on the line. They coincide only when the source impedance equals Z₀.
• Q: A single-stub match gives two solutions — which do you choose? A: Normally the shorter stub and shorter spacing, because they give lower frequency sensitivity, lower loss and less radiation.
• Q: Why can't a double-stub tuner match every load? A: For a fixed spacing d, loads with normalized conductance above about 1/sin²(βd) fall in a forbidden region; a third stub or a different spacing is needed.
• Q: VSWR is 1.05 at the input of a 30 m feeder but the antenna is clearly mismatched — explain. A: Cable attenuation spirals the trajectory toward the chart centre, so loss is masking the mismatch; measure at the antenna port or de-embed the cable.
• Q: What does the Bode–Fano criterion tell a designer? A: That a reactive load sets an absolute limit on how good a match can be held over a given bandwidth, so small antennas cannot be both well matched and broadband.
• Q: How does the chart help with lumped L-section design? A: Series elements move the point along constant-resistance circles and shunt elements along constant-conductance circles; you walk from the load to the centre using those two moves.
• Q: What is plotted when a VNA shows S₁₁ on a Smith chart? A: The measured input reflection coefficient referenced to 50 Ω after calibration — the same Γ, so impedance can be read directly.
• Q: Open stub or shorted stub in microstrip, and why? A: Usually open — it avoids drilling a via to ground — but the open end radiates and has fringing capacitance, so shorted stubs are preferred in coax and waveguide.

Common mistakes

• Plotting the un-normalized impedance, or forgetting to multiply back by Z₀ after reading the chart.
• Rotating the wrong way (toward load instead of generator) or using βl instead of 2βl for the rotation angle.
• Using the WTG scale when the problem moves toward the load, and vice versa — the two scales run in opposite directions.
• Applying a quarter-wave transformer directly to a complex load, instead of first rotating it to a real impedance.
• Confusing return loss with mismatch (insertion) loss, or quoting return loss as a negative number while calling it a loss.
• For a shunt stub, cancelling the residual reactance instead of the residual susceptance — stub design must be done on the admittance chart.
• Assuming VSWR changes along a lossless line, or that a matched load means zero current/voltage variation on a lossy line.
• Ignoring component parasitics and via inductance above 1–2 GHz, so a network that matches on paper is off-frequency on the board.

Related topics

Electromagnetic waves and transmission lines · RF circuit design basics · Microwave engineering basics · Antennas and wave propagation · S-parameters and network analysis · Filter design and resonant circuits