Fact-checked by Grok 2 weeks ago

Scattering parameters

Scattering parameters, commonly referred to as S-parameters, are a set of dimensionless complex numbers that describe the linear response of an electrical network to incident and reflected traveling voltage waves at its ports, typically normalized to a characteristic reference impedance such as 50 Ω.^[1]^[2] For an n-port network, the S-parameters form an n × n scattering matrix S, where the outgoing waves b are related to the incoming waves a by b = S a, with each S_ij representing the ratio of the outgoing wave at port i to the incoming wave at port j when all other ports are terminated in the reference impedance.^[3] This formulation is particularly suited to high-frequency applications in radio frequency (RF) and microwave engineering, where it simplifies the analysis of power flow, impedance matching, and signal propagation in components like amplifiers, filters, antennas, and transmission lines.^[1]^[2] The origins of S-parameters trace back to mid-20th-century network theory, with foundational work on scattering matrices for multiport systems developed by Vitold Belevitch in the 1940s, as detailed in his classical treatment of passive network synthesis.^[3] The parameters gained widespread adoption in microwave engineering following Kaneyuki Kurokawa's 1965 IEEE paper, which introduced the concept of power waves and their scattering matrix representation to address challenges in high-frequency power and gain calculations.^[4] This development was complemented by practical measurement advancements, such as Hewlett-Packard's 1967 introduction of the HP 8410 vector network analyzer, which enabled efficient swept-frequency S-parameter characterization up to 12 GHz.^[4] Today, S-parameters are a cornerstone of RF design, supported by modern tools like vector network analyzers that provide automated, error-corrected measurements for multiport devices.^[2] A key advantage of S-parameters over traditional impedance (Z) or admittance (Y) parameters lies in their measurement practicality at RF and microwave frequencies, where Z and Y require open- and short-circuit terminations that introduce significant parasitic inductance and capacitance, leading to inaccuracies.^[1]^[2] In contrast, S-parameters are obtained using matched loads (e.g., 50 Ω terminations), minimizing reflections and oscillation risks while directly relating to observable quantities like the reflection coefficient (S₁₁ = b₁/a₁ |_a₂=0) and transmission coefficient (S₂₁ = b₂/a₁ |_a₂=0).^[2] For reciprocal networks, the S matrix is symmetric (S_ij = S_ji). For passive networks, |S_ii| ≤ 1 for all ports. For lossless reciprocal networks, the S matrix is unitary (S^† S = I), ensuring conservation of energy.^[1]^[3] These attributes make S-parameters indispensable for applications in electromagnetic compatibility (EMC), signal integrity analysis, and the design of integrated circuits operating above 1 GHz.^[3]

Introduction

Definition and purpose

Scattering parameters, also known as S-parameters, describe the electrical behavior of linear electrical networks by relating the reflected voltage waves \mathbf{b} at the ports to the incident voltage waves \mathbf{a} through the matrix equation \mathbf{b} = S \mathbf{a}, where S is the scattering matrix for an N-port network.^[3] This formulation is particularly applicable to multi-port devices in high-frequency applications. The incident and reflected waves at a port are defined as

a = \frac{V + Z_0 I}{2 \sqrt{Z_0}}, \quad b = \frac{V - Z_0 I}{2 \sqrt{Z_0}},

where V and I are the total voltage and current at the port, and Z_0 is the reference impedance, typically real and positive (e.g., 50 \Omega). These wave variables capture the power flow into and out of the network, providing a physically meaningful representation based on power waves. The primary purpose of S-parameters is to facilitate the analysis and design of high-frequency radio frequency (RF) and microwave circuits, where traditional impedance (Z) or admittance (Y) parameters become impractical due to their reliance on ideal open- or short-circuit conditions that introduce significant reflections and measurement challenges at elevated frequencies.^[5] Unlike Z- and Y-parameters, which assume zero reflections and can yield unbounded values, S-parameters inherently account for mismatches by normalizing to a characteristic impedance Z_0, making them suitable for characterizing transmission lines, amplifiers, and antennas in real-world environments with non-ideal terminations.^[3] Key advantages of S-parameters include their insensitivity to specific port terminations when measured under matched conditions, ensuring consistent characterization regardless of external loading as long as the reference impedance is maintained.^[5] For passive networks, the magnitudes of the S-parameter elements are bounded by |S_{ij}| \leq 1, reflecting the physical constraint that reflected or transmitted power cannot exceed the incident power. Additionally, the matrix form enables straightforward cascading of network sections through matrix multiplication, simplifying system-level simulations and predictions in complex microwave assemblies.^[5]

Historical context

The concept of scattering parameters in electrical engineering traces its roots to scattering theory in quantum mechanics and optics during the 1930s and 1940s, where the S-matrix was developed to describe wave interactions and particle scattering without reference to unobservable internal states. In quantum mechanics, John Archibald Wheeler introduced the S-matrix in 1937 to characterize nuclear reactions,^[6] and Werner Heisenberg advanced it in the 1940s as a foundational tool for avoiding infinities in quantum field theory calculations.^[7]^[8] These ideas influenced network analysis by emphasizing observable input-output relations over internal voltages and currents, particularly useful for wave-based systems like waveguides. A key milestone in engineering applications occurred in 1945 with Vitold Belevitch's doctoral thesis at the Université Libre de Bruxelles, where he first described the scattering matrix—termed the "repartition matrix"—for lumped-element networks, focusing on power distribution among ports without measuring internal parameters.^[9] Belevitch's work laid the groundwork for radio-frequency (RF) applications by enabling stable representations of multi-port networks, influencing subsequent developments in microwave circuit design during the post-World War II era. This formulation was introduced to the microwave community in 1948 through the Radiation Laboratory series volume "Principles of Microwave Circuits," which adapted scattering concepts for high-frequency transmission lines and waveguides.^[10] The modern formulation of S-parameters emerged in 1965 through Kaneyuki Kurokawa's influential paper "Power Waves and the Scattering Matrix,"^[11] which defined power-wave variables to handle nonlinear and active devices, making S-parameters practical for microwave amplifiers and oscillators where traditional impedance parameters fail due to instability. Their adoption accelerated in microwave engineering during the 1970s, coinciding with the commercialization of vector network analyzers (VNAs); Hewlett-Packard's 1967 HP 8410 model enabled swept-frequency S-parameter measurements up to 12 GHz, and by the late 1970s, computer integration in instruments like the HP 8542A introduced error correction and automation, standardizing their use in circuit characterization.^[4] IEEE guidelines and measurement standards in the 1980s further formalized S-parameters for high-frequency testing, integrating them into protocols for component evaluation in radar and communication systems.^[4] In the 1990s and 2000s, S-parameters extended to mixed-mode formulations for differential signaling in high-speed digital circuits, with David E. Bockelman and William R. Eisenstadt introducing combined differential and common-mode parameters in 1995 to analyze balanced lines and suppress noise in multi-port networks.^[12] This development supported the rise of gigabit Ethernet and serializer/deserializer (SerDes) technologies, where mixed-mode S-parameters facilitate crosstalk and mode-conversion analysis in printed circuit boards.^[13]

Fundamental Principles

Power waves and formulation

In scattering parameter theory, power waves provide a generalized framework for analyzing multi-port networks, particularly those involving active devices where power gain or loss is significant. These waves, denoted as incident wave a_k and reflected wave b_k at port k, are defined in terms of the port voltage V_k and current I_k relative to a complex reference impedance Z_0 with positive real part, as follows:

a_k = \frac{V_k + Z_0 I_k}{2 \sqrt{\operatorname{Re}(Z_0)}}, \quad b_k = \frac{V_k - Z_0^* I_k}{2 \sqrt{\operatorname{Re}(Z_0)}}

This formulation ensures that the net power delivered to the port is given by P_k = |a_k|^2 - |b_k|^2, where the magnitudes squared represent available and delivered power quantities, respectively.^[14] The power-wave approach, introduced by Kurokawa in 1965, extends traditional scattering parameters to handle arbitrary reference impedances and active networks by satisfying the power relation |a|^2 - |b|^2 = power delivered to the port, which facilitates accurate analysis of devices with gain, such as amplifiers, where conventional traveling waves may not conserve power properly.^[14] Unlike voltage or current waves, which are normalized solely by the characteristic impedance magnitude and primarily describe transmission line propagation, power waves incorporate the real part of the reference impedance in their normalization, ensuring that the wave amplitudes directly correspond to power levels and enabling better handling of power conservation in non-passive systems.^[14] This preference for power waves arises because traditional traveling-wave definitions, based on a = (V + Z_0 I)/ (2 \sqrt{|Z_0|}) and similar for b, assume lossless, real Z_0 and fail to account for maximum power transfer in active or mismatched scenarios. For an N-port network, the scattering matrix \mathbf{S} relates the reflected waves to the incident waves via \mathbf{b} = \mathbf{S} \mathbf{a}, where \mathbf{S} is an N \times N complex matrix, and each element S_{ij} is defined as the ratio S_{ij} = b_i / a_j with all other incident waves a_k = 0 for k \neq j.^[14] This matrix formulation allows for the characterization of the network's linear behavior under small-signal conditions, independent of the specific excitation at other ports. While the reference impedance Z_0 can be arbitrary for generality, particularly in non-standard systems, the 50 \Omega real impedance is the conventional choice in radio-frequency (RF) engineering to align with common transmission line standards and measurement equipment.^[15]

Reciprocity conditions

In reciprocal networks, the scattering matrix S exhibits symmetry, satisfying S = S^T, which implies that the elements are equal such that S_{ij} = S_{ji} for all ports i and j.^[14] This property arises from the Lorentz reciprocity theorem applied to the underlying electromagnetic fields, ensuring that the transmission response is identical regardless of the direction of signal propagation between ports.^[16] Reciprocity holds under specific conditions: the network must be linear, time-invariant, and free of non-reciprocal elements or materials, such as isolators, gyrators, or magnetically biased ferrites that break time-reversal symmetry.^[17] These conditions ensure that the network's response to interchanged excitation and observation ports remains unchanged, a direct consequence of the symmetry in Maxwell's equations for isotropic media.^[14] The implications of reciprocity are significant in microwave engineering: the equality of transmission coefficients S_{ij} = S_{ji} means that power transmitted from port i to port j equals that from j to i, under matched conditions, which simplifies network design, cascade analysis, and measurement calibration by reducing the number of independent parameters.^[16] This symmetry also facilitates verification of network models during simulation or testing, as deviations indicate potential non-reciprocal behavior or measurement errors.^[17] In non-reciprocal cases, such as active devices (e.g., amplifiers) or ferrite-based components under magnetic bias, the scattering matrix is asymmetric, with S \neq S^T and S_{ij} \neq S_{ji}.^[18] A classic example is the ideal three-port circulator, a non-reciprocal device that routes signals unidirectionally; its scattering matrix is given by

S = \begin{pmatrix} 0 & 0 & 1 \\ 1 & 0 & 0 \\ 0 & 1 & 0 \end{pmatrix},

demonstrating the lack of symmetry while maintaining perfect matching and isolation in the forward direction.^[19] Such devices are essential for applications requiring signal isolation but violate reciprocity due to the external magnetic field or active gain mechanisms.^[20]

Behavior in lossless networks

In lossless networks, where no power is dissipated, the scattering matrix S is unitary, satisfying the condition S S^\dagger = I, with S^\dagger denoting the Hermitian transpose (conjugate transpose) and I the identity matrix. This property arises from the conservation of power, ensuring that the total power of the outgoing waves equals the total power of the incoming waves at all frequencies. The formulation in terms of power waves, as introduced by Kurokawa, underpins this unitarity for networks with purely reactive elements or ideal transmission structures. The unitarity of S imposes key constraints on the scattering parameters. Each element satisfies |S_{ij}| \leq 1, reflecting that no single output wave can exceed the incident power. Furthermore, for an N-port network, power conservation at each port i requires \sum_{j=1}^N |S_{ji}|^2 = 1, meaning the sum of the squared magnitudes of the parameters representing waves leaving port i equals unity. These relations stem directly from the rows (or columns) of the unitary matrix, guaranteeing no net power loss.^[21] For a two-port lossless network, the unitarity condition simplifies to |S_{11}|^2 + |S_{21}|^2 = 1 and |S_{22}|^2 + |S_{12}|^2 = 1, ensuring that the reflected power at each port plus the transmitted power sums to the incident power. The full unitarity also enforces phase relationships, such as S_{11}^* S_{12} + S_{21}^* S_{22} = 0 and |S_{11} S_{22} - S_{12} S_{21}| = 1, which maintain overall power balance. In reciprocal lossless two-ports, these combine with symmetry S_{12} = S_{21}.^[21] A special case is the matched lossless two-port, where reflections vanish (S_{11} = S_{22} = 0), and the transmission parameters satisfy |S_{12}| = |S_{21}| = 1. This occurs in networks perfectly impedance-matched to the reference ports, with all incident power transmitted without reflection or loss. Representative examples include the ideal transformer (with appropriate turns ratio for matching) and a lossless delay line. For a uniform lossless transmission line of electrical length \beta l, the scattering matrix is

S = \begin{pmatrix} 0 & e^{-j \beta l} \\ e^{-j \beta l} & 0 \end{pmatrix},

which is unitary and illustrates pure transmission with phase delay.^[21]

Behavior in lossy networks

In lossy networks, the scattering matrix S deviates from the unitary property observed in lossless counterparts, where S S^H = I ensures conservation of power. Instead, for lossy networks, S S^H \neq I, and the net power absorbed by the network is positive, quantified as \sum_i |a_i|^2 - |b_i|^2 > 0, where a_i and b_i are the incident and reflected power waves at port i, respectively.^[14] This absorption arises from dissipative elements such as resistors, leading to a non-conservative power balance where the total incident power exceeds the sum of reflected and transmitted powers. For a two-port lossy network, insertion loss manifests as a reduction in available power delivered to the load, directly tied to the magnitude of the transmission coefficient satisfying |S_{21}| < 1. This parameter |S_{21}| represents the fraction of incident power transmitted through the network, with losses converting the difference into heat or other forms of dissipation. In passive lossy networks, a fundamental bound ensures no amplification occurs: the spectral norm \|S\|_2 \leq 1, meaning the largest singular value of S is at most unity, which aligns with the positive semidefiniteness of I - S S^H and prevents the reflected or transmitted power from exceeding the incident power.^[22] A representative example is a matched attenuator, a passive two-port device designed to introduce controlled loss without reflections. Its scattering matrix takes the form

S = \begin{pmatrix} 0 & 10^{-\alpha/20} \\ 10^{-\alpha/20} & 0 \end{pmatrix},

where |S_{11}| = |S_{22}| = 0 indicates perfect matching at both ports, and |S_{21}| = |S_{12}| = 10^{-\alpha/20} with \alpha denoting the attenuation in decibels. For instance, a 3 dB attenuator has |S_{21}| \approx 0.707, dissipating half the incident power while transmitting the other half.^[23] Such devices are commonly used in microwave systems to protect sensitive components or balance signal levels. While active networks can exhibit deviations like |S_{21}| > 1 to achieve gain, the behavior in passive lossy networks remains constrained by the aforementioned absorption and norm conditions, prioritizing dissipation over amplification.^[14]

Two-Port S-Parameters

Matrix definition

The two-port scattering matrix, commonly denoted as the S-matrix, relates the reflected (outgoing) power waves to the incident (incoming) power waves at the two ports of a linear network. It is expressed in matrix form as

\begin{pmatrix} b_1 \\ b_2 \end{pmatrix} = \begin{pmatrix} S_{11} & S_{12} \\ S_{21} & S_{22} \end{pmatrix} \begin{pmatrix} a_1 \\ a_2 \end{pmatrix},

where a_1 and a_2 are the complex amplitudes of the incident power waves at ports 1 and 2, respectively, and b_1 and b_2 are the corresponding reflected power waves.^[14] Each element of the S-matrix carries a specific interpretation tied to the network's response under matched conditions. The diagonal element S_{11} is the input reflection coefficient, defined as the ratio of the reflected wave to the incident wave at port 1 (b_1 / a_1) when port 2 is terminated in the reference impedance Z_0, ensuring a_2 = 0. Similarly, S_{22} is the output reflection coefficient (b_2 / a_2) with port 1 terminated in Z_0 (a_1 = 0). The off-diagonal elements describe transmission: S_{21} is the forward transmission coefficient (b_2 / a_1) with port 2 matched (a_2 = 0), quantifying how much of the input at port 1 appears as output at port 2, while S_{12} is the reverse transmission coefficient (b_1 / a_2) with port 1 matched (a_1 = 0). These definitions stem from the power wave formulation, which normalizes waves to account for the reference impedance.^[14] The measurement of each S_{ij} requires terminating all other ports (here, the unused port) in the reference impedance Z_0, typically 50 \Omega in microwave systems, to prevent extraneous reflections and isolate the desired response. This termination simulates an infinite transmission line matched to Z_0, absorbing all incident energy without reflection.^[24] At low frequencies, where the physical size of the network is negligible compared to the signal wavelength, the S-matrix relates directly to other classical two-port parameters, such as the impedance (Z) matrix or the ABCD (transmission) matrix, via deterministic conversion formulas that incorporate Z_0. These conversions facilitate analysis in lumped-element approximations but become more complex at higher frequencies due to distributed effects. As an illustrative example, consider a simple two-port network formed by a shunt impedance Z connected between the signal conductor and ground at the junction of two matched transmission lines, each with characteristic impedance Z_0. Due to the network's reciprocity and symmetry, S_{11} = S_{22} and S_{12} = S_{21}. The elements are given by

S_{11} = S_{22} = -\frac{Z_0}{2Z + Z_0}, \quad S_{21} = S_{12} = \frac{2Z}{2Z + Z_0}.

This form arises from computing the equivalent parallel impedance Z \parallel Z_0 seen at the input (with port 2 terminated), yielding S_{11} as the reflection coefficient for that load, and S_{21} as the ratio of the transmitted wave amplitude to the incident wave, accounting for voltage division at the junction. For instance, if Z \to \infty (no shunt), the network behaves as a through connection, with S_{11} = 0 and S_{21} = 1; if Z = 0 (short to ground), S_{11} = -1 and S_{21} = 0.^[1]

Wave variables and normalization

In two-port networks, scattering parameters describe the relationships between incident and reflected waves at the input and output ports. The incident waves are denoted as a_1 at port 1 and a_2 at port 2, while the reflected waves are b_1 and b_2, respectively. These waves satisfy the relation \begin{pmatrix} b_1 \\ b_2 \end{pmatrix} = \begin{pmatrix} S_{11} & S_{12} \\ S_{21} & S_{22} \end{pmatrix} \begin{pmatrix} a_1 \\ a_2 \end{pmatrix}, where the S-matrix elements characterize reflection and transmission behaviors.^[3] The wave variables are typically defined as power waves to ensure that the magnitudes relate directly to incident and reflected power, particularly for non-50 Ω transmission lines. For a port with real reference impedance Z_0, the power waves are given by

a = \frac{V + Z_0 I}{2 \sqrt{Z_0}}, \quad b = \frac{V - Z_0 I}{2 \sqrt{Z_0}},

where V and I are the total voltage and current phasors at the port. The inverse relations are V = \sqrt{Z_0} (a + b) and I = \frac{1}{\sqrt{Z_0}} (a - b). For the two-port case, at port 1 with characteristic impedance Z_{01},

a_1 = \frac{V_1 + Z_{01} I_1}{2 \sqrt{Z_{01}}}, \quad b_1 = \frac{V_1 - Z_{01} I_1}{2 \sqrt{Z_{01}}},

and similarly at port 2 with Z_{02},

a_2 = \frac{V_2 + Z_{02} I_2}{2 \sqrt{Z_{02}}}, \quad b_2 = \frac{V_2 - Z_{02} I_2}{2 \sqrt{Z_{02}}}.

This formulation ensures that the incident power is |a|^2 and reflected power is |b|^2, with net power delivered to the port as |a|^2 - |b|^2. The power wave definition, introduced by Kurokawa, is preferred over voltage waves for maintaining bounded S-parameters in passive networks, as voltage waves can lead to |S| > 1 in lossy or non-50 Ω systems.^[14]^[23] Normalization of the waves depends on the choice of reference impedance Z_0, which directly impacts the S-parameters, as [S = (Z](/page/S/Z) - Z_0 I)(Z + Z_0 I)^{-1} in terms of the impedance matrix [Z](/page/Z). Changing Z_0 scales the wave amplitudes and alters the matrix elements; for instance, the magnitude |S_{11}| for a given device increases if Z_0 is mismatched to the device's input impedance. In practice, 50 Ω is the standard for coaxial RF systems due to its balance of low attenuation and power-handling capability, while 75 Ω is common in video and broadcast applications for optimal power transfer over longer cables with lower loss. Measurements or simulations referenced to one Z_0 must be renormalized to another using transformation formulas, such as S' = T^{-1} S T, where T accounts for the impedance difference. For unequal port impedances (Z_{01} \neq Z_{02}), pseudo-waves are employed, normalizing each port independently with its own Z_0, which preserves the power interpretation but results in non-symmetric S-matrices even for reciprocal networks, as S_{12} \neq S_{21} in general.^[3]^[25] When a two-port network is terminated at the output with a load impedance Z_L \neq Z_{02}, mismatch introduces multiple internal reflections, altering the effective S-parameters observed at the input. The load reflection coefficient is \Gamma_L = \frac{Z_L - Z_{02}}{Z_L + Z_{02}}, and the effective input reflection coefficient becomes

\Gamma_{\text{in}} = S_{11} + \frac{S_{12} S_{21} \Gamma_L}{1 - S_{22} \Gamma_L}.

This expression accounts for the feedback from the mismatched termination, increasing reflections and reducing available power gain compared to matched conditions (\Gamma_L = 0). Such effects are critical in cascaded systems, where unaccounted mismatches can degrade overall performance by up to several dB in return loss or insertion loss.^[23]

Properties of Two-Port Networks

Gain definitions and calculations

In two-port networks characterized by scattering parameters, several types of power gain are defined to quantify the transfer of power from the source to the load, accounting for mismatches at the input and output ports. These gains are derived from the S-parameter matrix elements S_{11}, S_{12}, S_{21}, and S_{22}, as well as the source reflection coefficient \Gamma_S and load reflection coefficient \Gamma_L. The definitions distinguish between the actual power delivered under specific terminations (transducer gain), the maximum possible under ideal matching (available gain), and the power into the network under operating conditions (operating gain).^[24] The transducer power gain G_T, also known as the overall power gain, is the ratio of the power delivered to the load P_L to the power available from the source P_{avs}. It fully accounts for bilateral effects through the reverse transmission parameter S_{12}. The formula is

G_T = \frac{|S_{21}|^2 (1 - |\Gamma_S|^2)(1 - |\Gamma_L|^2)}{|(1 - S_{11} \Gamma_S)(1 - S_{22} \Gamma_L) - S_{12} S_{21} \Gamma_S \Gamma_L|^2}.

This expression, originally developed using generalized power wave analysis, highlights how reverse transmission (S_{12} S_{21}) modifies the denominator, reducing gain in reciprocal networks.^[24]^[26] The available power gain G_A represents the ratio of the power available from the network output (under conjugate match at the output port) to the power available from the source. Under the common assumption of matched source and load terminations (\Gamma_S = 0, \Gamma_L = 0) and unilateral approximation (S_{12} = 0), it simplifies to G_A = \frac{|S_{21}|^2}{1 - |S_{11}|^2}, emphasizing the impact of input reflection on maximum achievable gain.^[24] In the general case with arbitrary \Gamma_S but conjugate output matching, G_A = |S_{21}|^2 \frac{1 - |\Gamma_S|^2}{|1 - S_{11} \Gamma_S|^2}.^[24] The operating power gain G_P, or power gain, is the ratio of power delivered to the load to the power incident on the network input. Assuming matched source (\Gamma_S = 0) and unilateral behavior, it becomes G_P = \frac{|S_{21}|^2}{ |1 - S_{22} \Gamma_L|^2 }, but the full expression under operating terminations is G_P = \frac{|S_{21}|^2 (1 - |\Gamma_L|^2)}{|(1 - S_{11} \Gamma_S)(1 - S_{22} \Gamma_L)|^2} for the unilateral case. This metric is particularly useful for evaluating performance with fixed source and load impedances.^[24] Under perfectly matched conditions (\Gamma_S = \Gamma_L = 0), the insertion power gain, or simply the forward power gain, reduces to the scalar linear form |S_{21}|^2, representing the fraction of incident power transmitted through the network without reflections. The reverse gain is analogously defined as |S_{12}|^2, quantifying power transfer from output to input.^[24] These linear gains are often expressed in logarithmic form for practical analysis, where the forward gain in decibels is $10 \log_{10} |S_{21}|^2 = 20 \log_{10} |S_{21}|, with similar distinctions applied to available and operating gains (e.g., available gain in dB = $10 \log_{10} G_A). The reverse gain in dB follows as $10 \log_{10} |S_{12}|^2. This logarithmic scale facilitates comparison across frequencies and devices in RF design.^[24]

Reflection and return loss metrics

In two-port networks characterized by scattering parameters, the input reflection coefficient \Gamma_{in} quantifies the fraction of the incident wave at port 1 that is reflected back, accounting for the influence of the load at port 2. It is expressed as

\Gamma_{in} = S_{11} + \frac{S_{12} S_{21} \Gamma_L}{1 - S_{22} \Gamma_L},

where S_{11}, S_{12}, S_{21}, and S_{22} are the scattering parameters, and \Gamma_L is the load reflection coefficient at port 2.^[27] This formula arises from the wave relations in the network, enabling analysis of how mismatches at the output affect input reflections. When the output port is terminated in a matched load (\Gamma_L = 0), \Gamma_{in} simplifies to the isolated input reflection S_{11}, which directly measures the network's inherent mismatch at port 1 under matched conditions.^[22] The return loss at the input, RL_{in}, provides a logarithmic measure of this mismatch in decibels, defined as

RL_{in} = -20 \log_{10} |\Gamma_{in}|,

indicating the extent to which power is returned to the source rather than delivered to the network; higher values (e.g., >10 dB) signify better matching and less reflection.^[27] Similarly, the output reflection coefficient \Gamma_{out} at port 2 is

\Gamma_{out} = S_{22} + \frac{S_{12} S_{21} \Gamma_S}{1 - S_{11} \Gamma_S},

with \Gamma_S as the source reflection coefficient at port 1, and the corresponding output return loss RL_{out} = -20 \log_{10} |\Gamma_{out}|.^[27] For an isolated output port (\Gamma_S = 0), \Gamma_{out} = S_{22}.^[22] Mismatch loss quantifies the power dissipated due to reflections when |\Gamma| \neq 0, representing the fraction of incident power not transmitted forward. It is given by

\text{Mismatch loss} = -10 \log_{10} (1 - |\Gamma|^2),

where \Gamma is the relevant reflection coefficient (e.g., \Gamma_{in} or \Gamma_{out}); for instance, a |\Gamma| = 0.1 yields approximately 0.04 dB loss, highlighting the impact of imperfect matching on efficiency.^[27] These metrics are essential for evaluating port matching in microwave systems, as mismatches can degrade overall performance, including power gain.^[22]

Stability and isolation parameters

In two-port networks characterized by scattering parameters, the Voltage Standing Wave Ratio (VSWR) at the input port assesses impedance matching when the output is terminated in the reference impedance. The input reflection coefficient is \Gamma_\text{in} = S_{11}, and VSWR is defined as

\text{VSWR} = \frac{1 + |\Gamma_\text{in}|}{1 - |\Gamma_\text{in}|}.

This parameter indicates the extent of standing waves due to reflections, with a value of 1 representing perfect matching and values greater than 1 signaling mismatch that can lead to power loss and signal distortion.^[28]^[29] Reverse isolation quantifies the network's capacity to block signal propagation from the output port back to the input port, which is vital for minimizing unwanted feedback. It is expressed as

\text{TI} = -20 \log_{10} |S_{12}|

in decibels, where larger TI values denote superior isolation and enhanced performance in applications like amplifiers.^[30] For evaluating unconditional stability—stability under any passive source and load terminations—several factors derived from the two-port S-parameters are employed. The S-matrix determinant \Delta = S_{11} S_{22} - S_{12} S_{21} is fundamental, with |\Delta| < 1 required to preclude oscillation risks across all terminations.^[31]^[32] The Rollett stability factor K, originally formulated by J. S. Rollett, is

K = \frac{1 - |S_{11}|^2 - |S_{22}|^2 + |\Delta|^2}{2 |S_{12} S_{21}|}.

Unconditional stability holds if K > 1 and |\Delta| < 1, ensuring no regions of instability in the Smith chart.^[33] The \mu factor, introduced by M. L. Edwards and J. H. Sinsky, offers a complementary geometric perspective and is defined as

\mu = \frac{1 - |S_{11}|^2}{|S_{22} - \Delta S_{11}^{*}| + |S_{12} S_{21}|},

where * denotes the complex conjugate. A complementary factor \mu' is obtained by interchanging ports 1 and 2. Satisfaction of \mu > 1 and \mu' > 1 alongside |\Delta| < 1 confirms unconditional stability, with the factor's value also pinpointing whether input or output port conditions contribute to potential instability.^[32] A condition of |\Delta| > 1 signals potential oscillation under specific terminations, highlighting the need for careful design to maintain |\Delta| < 1.^[31]^[32]

Multi-Port Extensions

One-port S-parameters

In scattering parameter theory, the one-port case simplifies to a single scalar parameter, denoted as S_{11}, which quantifies the reflection at the port. Defined as the ratio of the outgoing (reflected) wave amplitude b_1 to the incoming (incident) wave amplitude a_1 when no other ports are present, S_{11} = \frac{b_1}{a_1}. This formulation originates from the power wave approach, ensuring normalization to characteristic impedance for consistent power measurements across frequencies.^[34]^[14] The magnitude of S_{11}, denoted |S_{11}|, provides insight into energy behavior at the port. For lossless reactive terminations, such as an ideal capacitor or inductor, |S_{11}| = 1, indicating complete reflection without dissipation, as all incident power is returned with a phase shift determined by the reactance. In contrast, for terminations with resistive losses, |S_{11}| < 1, reflecting partial absorption of the incident power.^[34]^[1] As the input reflection coefficient \Gamma, S_{11} directly relates the port's impedance Z to the reference impedance Z_0 (typically 50 Ω) via the formula:

\Gamma = S_{11} = \frac{Z - Z_0}{Z + Z_0}.

This allows full inversion to obtain impedance from measured S_{11}:

Z = Z_0 \frac{1 + S_{11}}{1 - S_{11}}.

Such transformations enable characterization of unknown loads by converting reflection data to impedance values, essential for verifying terminations in microwave circuits.^[1]^[34] One-port S_{11} finds primary applications in assessing antenna input matching, where low |S_{11}| (e.g., below -10 dB) indicates efficient power transfer to the radiator with minimal reflection back to the feed line. It also supports load characterization, such as evaluating termination networks or filters for impedance compliance in high-frequency systems.^[35]^[34] Measurements of one-port S_{11} involve direct connection of the device under test to a single port of a vector network analyzer (VNA), following calibration with open, short, and load standards to account for systematic errors like directivity and source match. This setup isolates reflection without transmission paths, yielding magnitude and phase data across frequencies for analysis.^[36]^[34]

Four-port S-parameters

Four-port scattering parameters describe the behavior of networks with four access points, such as directional couplers and hybrid junctions, through a 4×4 scattering matrix \mathbf{S} whose elements S_{ij} relate the outgoing wave at port i to the incoming wave at port j.^[34] In these networks, the matrix elements quantify transmission, reflection, coupling, and isolation between ports, assuming normalized wave variables and characteristic impedances typically set to 50 Ω.^[37] For an ideal matched four-port network, the diagonal elements S_{ii} = 0, indicating no reflection at any port.^[38] Common configurations employing four-port S-parameters include directional couplers and hybrid circuits. In a directional coupler, power incident on port 1 (input) primarily transmits to port 2 (through port) with coefficient S_{21} = \alpha, where \alpha = \sqrt{1 - c^2} and c is the coupling factor, while a portion couples to port 3 (S_{31} = j c) and port 4 remains isolated (S_{41} = 0).^[38] The ideal S-matrix for such a forward-wave directional coupler is:

\mathbf{S} = \begin{pmatrix} 0 & \alpha & j c & 0 \\ \alpha & 0 & 0 & j c \\ j c & 0 & 0 & \alpha \\ 0 & j c & \alpha & 0 \end{pmatrix},

ensuring directionality where signals propagate preferentially in one direction.^[39] Hybrid couplers, a special case with c = 1/\sqrt{2} for equal 3 dB splitting, further introduce phase shifts; the 90° (quadrature) hybrid provides a 90° phase difference between the through and coupled outputs.^[40] Its ideal S-matrix, with ports numbered such that port 1 is input, port 2 isolated, ports 3 and 4 outputs, is:

\mathbf{S} = \frac{1}{\sqrt{2}} \begin{pmatrix} 0 & 0 & -j & 1 \\ 0 & 0 & 1 & -j \\ -j & 1 & 0 & 0 \\ 1 & -j & 0 & 0 \end{pmatrix},

highlighting the quadrature phase (-j vs. 1).^[41] The magic-T, a waveguide-based 180° hybrid, features a sum port (H-plane) and difference port (E-plane), with collinear arms providing in-phase or out-of-phase outputs depending on excitation.^[42] Reciprocal four-port networks, typical of passive microwave devices without non-reciprocal elements like isolators, exhibit symmetry in the S-matrix such that \mathbf{S} = \mathbf{S}^T, meaning S_{ij} = S_{ji}.^[37] For lossless networks, power conservation requires the S-matrix to be unitary, with each row and column having unit magnitude and being orthogonal to others, satisfying \mathbf{S}^\dagger \mathbf{S} = \mathbf{I}.^[34] This ensures that the total outgoing power equals the incoming power across all ports. Isolation is a key metric in four-port devices, defined by the transmission coefficient to the isolated port; for an ideal forward directional coupler, S_{41} = 0, implying infinite isolation and no power transfer from input to the reverse-coupled port.^[38] In practice, finite isolation arises from imperfections, quantified as $10 \log_{10} (1/|S_{41}|^2) in dB.^[40] A practical example is the Wilkinson power divider, often analyzed in multi-port contexts despite its standard three-port form (input and two isolated outputs); its ideal S-matrix at the design frequency, incorporating a quarter-wave transformer and isolation resistor, is:

\mathbf{S} = \begin{pmatrix} 0 & -j/\sqrt{2} & -j/\sqrt{2} \\ -j/\sqrt{2} & 0 & 0 \\ -j/\sqrt{2} & 0 & 0 \end{pmatrix},

providing equal power split, matching at all ports, and isolation between outputs via the resistor.^[43] Extensions to four outputs require cascaded structures but retain the core principles of the basic design.^[44]

Higher-order S-parameter matrices

Higher-order scattering parameters extend the two-port and four-port formulations to networks with an arbitrary number of ports N > 4, forming the basis for analyzing complex multiport systems such as integrated circuits (ICs) and antenna arrays. The N-port scattering matrix S is a complex N×N matrix that relates the vector of outgoing (reflected or transmitted) waves b to the vector of incoming waves a through the equation b = S a, where each element S_ij quantifies the amplitude and phase of the wave emerging from port i due to an excitation at port j, with all other ports terminated in matched loads.^[23]^[3] In practice, for uncoupled or weakly coupled systems, S exhibits sparsity, where many off-diagonal elements are zero or negligible, reducing the effective density; for instance, a fully uncoupled N-port network has a diagonal S matrix representing independent one-port reflections.^[45] Key properties of the N-port S matrix mirror those of lower-order cases but scale with system complexity. For reciprocal networks—those invariant under port interchange—S is symmetric, satisfying S = S^T, which implies S_ij = S_ji and reflects the underlying linearity and passivity without nonreciprocal elements like isolators.^[23] Passivity, ensuring no internal power generation, requires the spectral norm ||S|| ≤ 1, where the largest singular value bounds the amplification to unity or less, preventing energy gain beyond input levels.^[23] For lossless networks, S is unitary, obeying S^† S = I, which conserves power such that the sum of squared magnitudes of each row (or column) equals 1, as in

\sum_{k=1}^N |S_{ik}|^2 = 1 \quad \forall i,

guaranteeing total reflected and transmitted power equals incident power.^[23] These properties facilitate validation of measured or simulated data, with violations indicating losses, active elements, or measurement errors. In large N-port systems, computational efficiency is paramount due to the O(N²) elements in a dense S matrix, which can exceed practical storage limits for N in the hundreds or thousands. Sparsity is exploited by storing only nonzero elements using formats like compressed sparse row (CSR), drastically reducing memory usage—for example, in antenna arrays where coupling decays rapidly with distance, nonzero entries may occupy less than 5% of the matrix.^[45] Simulations often employ decompositions such as singular value decomposition (SVD) or model-order reduction techniques to approximate S for time-domain analysis or cascading networks, enabling faster circuit optimization without full matrix inversion.^[46] In multi-antenna multiple-input multiple-output (MIMO) systems, the S matrix captures mutual coupling between elements, where off-diagonal terms S_ij (i ≠ j) represent near-field interactions that degrade beamforming gain and channel capacity if not accounted for; for instance, coupling levels below -20 dB are targeted to maintain diversity in 5G arrays.^[47] Characterizing higher-order S matrices poses significant challenges, particularly for N > 4, as standard two-port vector network analyzers (VNAs) must be extended to multiport configurations with multiple receivers and switches. Accurate measurement demands de-embedding algorithms to isolate device effects from test fixtures and cables, often involving iterative transformations between S, impedance Z, and admittance Y matrices to remove parasitic influences; for large N, this multistep process amplifies errors if standards are imperfect.^[48] Specialized multiport VNAs, supporting up to 48 ports or more, are essential but increase setup complexity and calibration time, with reciprocity and unitarity checks used to verify data quality amid noise and crosstalk.^[48] These hurdles are critical in applications like IC packaging, where N can reach dozens, necessitating hybrid simulation-measurement workflows for full characterization.^[49]

Mixed-mode transformations

Mixed-mode scattering parameters provide a framework for analyzing differential signaling systems by redefining the scattering parameters in terms of differential and common-mode signals, typically for a four-port network representing two balanced pairs. This approach transforms the conventional single-ended S-parameter matrix S_{se} into a mixed-mode matrix S_{mm} using the relation S_{mm} = T^{-1} S_{se} T, where T is the transformation matrix that maps single-ended wave variables to their differential and common-mode equivalents.^[12] The transformation matrix for two differential port pairs is given by

T = \frac{1}{\sqrt{2}} \begin{bmatrix} I & -I \\ I & I \end{bmatrix},

where I denotes the 2×2 identity matrix; the upper block row corresponds to differential modes (odd symmetry, signals of equal magnitude but opposite phase), while the lower block row corresponds to common modes (even symmetry, signals of equal magnitude and phase). The resulting S_{mm} consists of 2×2 submatrices: S_{dd} for differential-to-differential transmission and reflection, S_{cc} for common-to-common, S_{dc} for differential-to-common mode conversion, and S_{cd} for common-to-differential mode conversion. These submatrices enable direct quantification of mode-specific behaviors without needing to simulate or measure individual signal combinations.^[12] This mixed-mode representation simplifies the analysis of balanced transmission lines and differential circuits by isolating pure-mode responses and conversions, facilitating metrics such as the power supply rejection ratio defined as \text{PSRR} = |S_{cc} / S_{dd}|, which assesses the suppression of common-mode noise relative to differential signal integrity. For instance, in an ideal differential amplifier, the differential-mode input reflection coefficient S_{dd11} = 0 indicates perfect matching, and S_{dc} = 0 signifies complete absence of unwanted mode conversion.

Specialized Applications

Amplifier design and stability analysis

In the design of microwave amplifiers, scattering parameters enable the assessment of stability by analyzing how source and load reflections affect potential oscillations. Unconditional stability occurs when the amplifier operates without oscillation for any passive source and load terminations, satisfying |Γ_S| ≤ 1 and |Γ_L| ≤ 1. This condition is met if the Rollett stability factor K exceeds 1 and the magnitude of the determinant |Δ| is less than 1, where Δ = S_{11}S_{22} - S_{12}S_{21} and K = \frac{1 - |S_{11}|^2 - |S_{22}|^2 + |\Delta|^2}{2 |S_{12} S_{21}|}. Under these criteria, the source reflection coefficient Γ_S and load reflection coefficient Γ_L must remain within the unit circle on the Smith chart to prevent |Γ_{in}| > 1 or |Γ_{out}| > 1, ensuring no amplification of reflections leads to instability.^[27] Stability circles provide a graphical tool to visualize boundaries where instability may occur, plotted in the Γ_L or Γ_S planes using the device's S-parameters. The output stability circle represents the locus of Γ_L values for which |Γ_{in}| = 1, given by \left| \Gamma_L - C_o \right| = r_o, where the center C_o = \frac{S_{22}^* - \Delta S_{11}^}{1 - |\Delta|^2} and the radius r_o = \frac{|S_{12} S_{21}|}{1 - |\Delta|^2}. Similarly, the input stability circle delineates the locus of Γ_S for which |Γ_{out}| = 1, with center C_i = \frac{S_{11}^ - \Delta S_{22}^*}{1 - |\Delta|^2} and radius r_i = \frac{|S_{12} S_{21}|}{1 - |\Delta|^2}. These circles, when plotted on the Smith chart, identify forbidden regions outside the unit disk; if the circles lie entirely outside the unit circle, the amplifier is unconditionally stable.^[27] For loaded conditions, stability requires that Γ_S and Γ_L lie in the stable regions relative to the circles—typically inside the unit circle and outside any intersecting stability circle portions—to avoid oscillation. In practice, the μ factor offers a single-parameter metric for design, defined as \mu = \frac{1 - |S_{11}|^2}{|S_{22} - S_{11}^* \Delta| + |S_{12} S_{21}|}, where μ > 1 confirms unconditional stability from the input perspective, and a similar μ' > 1 applies from the output; values greater than 1.2 are often targeted for margin in broadband designs. These factors guide the selection of terminations, ensuring reflections do not exceed unity magnitude at either port. In amplifier design, matching networks are synthesized to position Γ_S and Γ_L in stable regions while achieving desired gain and bandwidth.^[27]

Scattering transfer parameters

Scattering transfer parameters, also known as T-parameters, offer an alternative formulation to S-parameters for describing the behavior of two-port networks, with a particular emphasis on facilitating the analysis of systems composed of cascaded components. Unlike S-parameters, which relate outgoing waves to incoming waves at all ports, T-parameters relate the waves at the input port to those at the output port in a chain-like manner, making them analogous to ABCD or chain parameters but expressed in terms of normalized traveling waves. This representation is especially valuable in microwave engineering for modeling multi-stage devices such as filters, amplifiers, and transmission lines where components are connected in series.^[50] The T-matrix is defined by the equation

\begin{pmatrix} a_1 \\ b_1 \end{pmatrix} = \begin{pmatrix} T_{11} & T_{12} \\ T_{21} & T_{22} \end{pmatrix} \begin{pmatrix} b_2 \\ a_2 \end{pmatrix},

where a_1 and b_1 are the incident and reflected waves at the input (port 1), and b_2 and a_2 are the reflected and incident waves at the output (port 2), respectively. The elements of the T-matrix are derived from the corresponding S-parameter matrix as follows:

T_{11} = \frac{1}{S_{21}},

T_{12} = -\frac{S_{22}}{S_{21}},

T_{21} = \frac{S_{11}}{S_{21}},

T_{22} = -\frac{\det(S)}{S_{21}},

with \det(S) = S_{11}S_{22} - S_{12}S_{21}. These relations assume the standard reference impedances for the ports and are valid for linear passive or active networks. For reciprocal networks (S_{12} = S_{21}), the determinant of the T-matrix equals unity, preserving key physical properties during conversions. A key advantage of T-parameters lies in their suitability for cascading multiple two-port networks. When two networks with T-matrices T_1 and T_2 are connected in series (output of the first to input of the second), the overall T-matrix is simply the product T = T_1 T_2, enabling straightforward computation of the combined response without solving coupled equations. This matrix multiplication property simplifies the design and simulation of complex systems, such as distributed filter structures or amplifier chains, by allowing modular assembly of individual component matrices.^[50] However, T-parameters have limitations: they are inherently oriented toward series (cascaded) configurations and do not lend themselves easily to modeling parallel or shunt connections, where S-parameters or Y-parameters are more appropriate. Additionally, the T-matrix form is asymmetric even for reciprocal devices, as T_{12} \neq T_{21} in general, which can complicate interpretations compared to the symmetric nature of S-matrices for reciprocal systems. These drawbacks make T-parameters less versatile for general multi-port analysis but ideal for linear chain topologies.^[50] As an illustrative example, consider the cascade of two identical lossless reciprocal sections, each characterized by the same T-matrix T. The total transfer matrix becomes T_{\text{total}} = [T^2](/page/T+2), obtained via standard matrix squaring. For instance, if each section is a matched transmission line segment with S_{11} = S_{22} = 0 and S_{21} = S_{12} = e^{-j[\theta](/page/Theta)} (where [\theta](/page/Theta) is the electrical length), the individual T-matrix elements are T_{11} = e^{j[\theta](/page/Theta)}, T_{12} = 0, T_{21} = 0, and T_{22} = e^{-j[\theta](/page/Theta)}. Squaring yields a total transmission coefficient corresponding to a doubled phase shift $2[\theta](/page/Theta), confirming the additive behavior in cascaded lossless systems without reflections. This demonstrates how T-parameters efficiently capture the cumulative effects in such configurations.^[50]

Measurement Methods

Two-port measurement techniques

The primary instrument for measuring two-port scattering parameters is the vector network analyzer (VNA), which operates by generating a swept-frequency signal and quantifying the incident and reflected waves at the ports.^[51] In a typical setup, the VNA sources a continuous-wave RF signal from port 1, sweeping across a defined frequency range, while directional couplers at each port separate the incident waves (a1, a2) from the reflected or transmitted waves (b1, b2), enabling computation of the S-parameter matrix elements such as S11, S21, S12, and S22.^[51]^[52] This approach provides magnitude and phase information, essential for characterizing linear network behavior under small-signal conditions.^[51] Accurate measurements require calibration to remove systematic errors introduced by the VNA, cables, and connectors, such as directivity, source match, load match, reflection tracking, transmission tracking, and isolation.^[53] Common methods include the short-open-load-thru (SOLT) calibration, which uses known standards to solve for these errors, and the thru-reflect-line (TRL) calibration, which employs a transmission line of known length for broadband accuracy, particularly useful at higher frequencies where precise load standards are challenging.^[54]^[55] These techniques de-embed fixture and adapter effects, yielding error-corrected S-parameters that closely represent the device under test (DUT).^[56] The underlying error model for two-port VNA measurements is typically a 12-term error flow graph, accounting for imperfections in both forward and reverse measurement directions.^[53] This model uses signal flow graphs to propagate errors through the system, with forward sweeps measuring parameters like S11 and S21, and reverse sweeps capturing S22 and S12 after switching the source and receiver paths.^[55] Calibration solves a set of equations derived from the standards to determine the error coefficients, which are then subtracted from raw measurements.^[54] Beyond frequency-domain results, VNAs support time-domain analysis by applying the inverse fast Fourier transform (IFFT) to the frequency-dependent S-parameters, transforming them into time-domain responses such as step or impulse waveforms.^[57] This enables visualization of pulse propagation, identification of discontinuities, and fault location in transmission lines or interconnects by analyzing reflections in the time trace.^[58]^[59] Standard VNAs cover frequencies from kHz to tens of GHz, but measurements up to millimeter-wave bands (e.g., 110 GHz or higher) are achieved using frequency extenders, which convert the VNA's baseband output to higher frequencies via harmonic mixing while maintaining phase coherence.^[60]^[61] These extenders, often paired with specialized probes or waveguides, require adapted calibration kits to ensure accuracy in the extended range.^[60]

Multi-port measurement approaches

Multi-port vector network analyzers (VNAs) extend the capabilities of standard instruments to characterize networks with N > 2 ports, using either integrated multi-port designs or external switch matrices to handle configurations up to 48 ports for applications like phased-array antennas and large-scale interconnects.^[62]^[63] These systems maintain high dynamic range and phase stability across all ports, enabling accurate S-parameter extraction for complex devices where port interactions are critical.^[64] Switched configurations, in particular, route signals from a core two-port VNA to multiple device ports via RF switches, allowing sequential access without requiring fully integrated high-port hardware.^[65] The measurement process involves exciting one port at a time with the incident wave while terminating all other ports in matched loads, typically 50 Ω, to isolate transmission and reflection responses.^[66] This sequential excitation yields the full N × N S-parameter matrix, requiring N² independent measurements—each a complete frequency sweep—to capture all elements, as the matrix is not necessarily symmetric for non-reciprocal devices.^[67] Calibration, often building on two-port standards like SOLT, is adapted for multi-port use through automated routines that define error terms for each port pair.^[68] De-embedding is essential to isolate the device under test (DUT) from fixture effects, such as probes or transitions, using techniques like multimode Through-Reflect-Line (TRL) calibration extended to four or more ports for precise reference-plane shifting.^[69] Generalized analytical methods employ embedding theorems to mathematically subtract known fixture S-parameters from the raw multi-port data, ensuring the resulting matrix reflects only the DUT behavior.^[48] These approaches are particularly vital in high-frequency setups where parasitic effects can dominate. Challenges in multi-port measurements intensify with larger N, as the quadratic increase in configurations leads to extended acquisition times and higher costs, often limiting throughput in production environments.^[65] Switch-induced losses and crosstalk further complicate accuracy, while for mixed-mode analysis in differential networks, baluns facilitate conversion from single-ended to balanced ports but constrain bandwidth and introduce phase imbalances.^[70] Advanced on-wafer probing addresses integrated circuit characterization by deploying multi-needle arrays for direct port access, minimizing parasitics in sub-millimeter wave applications.^[71] For very large port counts, probabilistic uncertainty quantification evaluates error propagation across the matrix, aiding reliability assessment without exhaustive re-measurements.^[72]

References

[1]
[PDF] Scattering Parameters - University of California, Berkeley
Many active devices could oscillate under the open or short termination. • S parameters are easier to measure at high frequency. The measurement is direct and ...
[2]
[PDF] S-Parameter Techniques – HP Application Note 95-1
S-parameters are important in microwave design because they are easier to measure and work with at high frequencies than other kinds of parameters. They are ...
[3]
A Primer on Scattering Parameters, Part I: Definitions and Properties
Abstract—This primer offers a simple and comprehensive overview of the properties and usage of the scattering parameters of linear n-port elements.
[4]
Measuring S-parameters: The First 50 Years | Microwave Journal
Mar 24, 2008 · In 1965, an IEEE article entitled “Power Waves and the Scattering Matrix” by Kaneyuki Kurokawa of Bell Labs was among the first technical ...Missing: origin inventor
[5]
The purpose of this paper is as follows
That is why scattering or S-parameters were developed. S-parameters have many advantages over the previously mentioned H, Y or Z-parameters. They relate to ...
[6]
From S-matrix theory to strings: Scattering data and the commitment ...
This paper provides a historical analysis of how string theory was developed out of -matrix physics, aiming to clarify how modern string theory, as a theory ...
[7]
The S-Matrix Is the Oracle Physicists Turn To in Times of Crisis
May 23, 2024 · In the 1940s, Werner Heisenberg, a pioneer of quantum mechanics, expected that a revolutionary new theory would soon replace particle physics.Missing: 1930s | Show results with:1930s
[8]
[PDF] In Memoriam - Vitold Belevitch
In his thesis he introduced the revolutionary concept of scattering matrix, or repartition matrix, as he called it. In fact, this concept was discovered ...
[9]
http://mathscinet.ru/files/Belevitch.pdf
[10]
https://babel.hathitrust.org/cgi/pt?id=mdp.39015002027871;view=1up;seq=9;size=75
[11]
(PDF) Mixed-Mode S-Parameters and Conversion Techniques
The ratio of the differential signal to the common signal can be expressed by converting the S-parameter of single-ended mode to the S-parameter of mixed mode.
[12]
Power Waves and the Scattering Matrix
Insufficient relevant content. The provided URL (https://ieeexplore.ieee.org/document/1125964) points to a page titled "Power Waves and the Scattering Matrix," but the content snippet is empty and does not contain the definition of the one-port scattering parameter S11, its relation to the reflection coefficient, or impedance, including any formulas. No additional details or URLs can be extracted from the given content.
[13]
What Are S-parameters? - Ansys
also known as S-parameters — refer to the elements in a mathematical matrix describing the behavior of an electrical network (or ...Missing: bounded magnitudes
[14]
[PDF] Power Waves and the Scattering Matrix - Semantic Scholar
Power Waves and the Scattering Matrix · K. Kurokawa · Published 1 March 1965 · Physics · IEEE Transactions on Microwave Theory and Techniques.Missing: Kaneyuki | Show results with:Kaneyuki
[15]
[PDF] Scattering parameters
Jul 21, 2005 · ... Vitold Belevitch in 1945.[5] The name used by Belevitch was repartition matrix and limited consideration to lumped-element networks. The ...
[16]
Impedance Mismatch and S-Parameter Reciprocity | NWES Blog
Jun 1, 2025 · Reciprocity breaks when fundamental assumptions (passive, linear, time-invariant, isotropic media) are violated. Networks with non-reciprocal ...
[17]
[PDF] 5. Circulators and Isolators
Sep 29, 2006 · A circulator is a matched, lossless but non-reciprocal 3-port device, whose scattering matrix is ideally: 0 0 1. 1 0 0. 0 1 0. ⌈. ⌉. │. │. = │.
[18]
https://www.nwengineeringllc.com/article/impedance-mismatch-and-s-parameter-reciprocity.php
[19]
[PDF] 4.3 – The Scattering Matrix
Mar 4, 2009 · An analysis of the scattering matrix can tell us if a certain device is even possible to construct, and if so, what the form of the device must ...
[20]
[PDF] Microwave Engineering, 4th Edition - WordPress.com
Pozar, David M. Microwave engineering/David M. Pozar.—4th ed. p. cm. Includes bibliographical references and index.
[21]
[PDF] RF engineering basic concepts: S-parameters
Abstract. The concept of describing RF circuits in terms of waves is discussed and the. S-matrix and related matrices are defined.Missing: (V + sqrt(
[22]
[PDF] S-Parameters... circuit analysis and design (PDF) - HP Memory Project
5. "Quick Amplifier Design with Scattering Parameters," by William H. Froehner, Electronics, October 16, 1967.Missing: history inventor
[23]
Converting S-Parameters from 50Ω to 75Ω Impedance
Nov 21, 2003 · This application note presents an easy way to perform S-parameter measurements by treating the input and output impedance of the cable device as 50Ω.
[24]
[PDF] a1 b1 ΓS ΓL - Armms
[2]. Bodway G. E., “Two Port Power Flow Analysis Using Generalized Scattering Parameters”, Microwave J.,. May 1967, pp. 61-69. [3]. Edwards M. L., Sinsky J. H. ...
[25]
[PDF] S (scattering) Parameters - Sandiego
S11 is equal to the ratio of a reflected wave and an incident wave with. Zl=Zo. Thus, S11 can be plotted on a Smith chart and the input impedance of the two- ...Missing: sqrt( ieee
[26]
Voltage standing wave ratio (VSWR) - Microwave Encyclopedia
A reflection coefficient is defined as the ratio of reflected wave to incident wave at a reference plane. This value varies from -1 (for a shorted load) to +1 ( ...
[27]
S-Parameters and the Reflection Coefficient
Nov 13, 2023 · S-parameters are a valuable tool for calculating reflection coefficients and transmission gains for a two-port network's input and output sides.
[28]
Reverse Isolation - Keysight
The equivalent S-parameter is S12. Why Measure Reverse Isolation? An ideal amplifier would have infinite reverse isolation-no signal would be transmitted ...
[29]
Microwaves101 | Stability factor - Microwave Encyclopedia
The Mu factors tell you something that K-factor does not. They tell you which side of your circuit is the likely culprit when it goes unstable. However, if K is ...
[30]
2.6: Amplifier Stability - Engineering LibreTexts
May 22, 2022 · There are two suitable stability criteria commonly used, the $k$-factor and the $\mu$-factor, which will now be considered.Missing: isolation | Show results with:isolation
[31]
K-Factor Derivation - Microwave Encyclopedia
Professor Dr. SF Peik derives the conditions for stability (Rollet's stability factor, known as "K") of an active two-port network.
[32]
Microwaves101 | S-parameters - Microwave Encyclopedia
The scattering wave concept was further popularized around the time that Kaneyuke Kurokawa of Bell Labs wrote his 1965 IEEE article "Power Waves and the ...Missing: original paper
[33]
[PDF] Antenna Impedance Matching – Simplified - Abracon
Impedance matching is designing an antenna's input impedance to match the RF circuitry's output impedance, often 50 Ω, for maximum efficiency.
[34]
Select a Calibration Type - Keysight
Reflection Standards (OPEN, SHORT, LOAD) are connected to only ONE of the ports to be calibrated. The lower port number of the ports to be calibrated is ...
[35]
Microwaves101 | Basic network theory - Microwave Encyclopedia
If you are measuring a network that is known to be reciprocal, checking for symmetry about the diagonal of the S-parameter matrix is one simple check to see if ...
[36]
Microwaves101 | Directional Couplers - Microwave Encyclopedia
Directional couplers are four-port circuits where one port is isolated from the input port. Directional couplers are passive reciprocal networks.
[37]
Microwaves101 | Hybrid (3 dB) couplers - Microwave Encyclopedia
Hybrid couplers are the special case of a four-port directional coupler that is designed for a 3-dB (equal) power split.
[38]
None
### Ideal Scattering Matrix for the 90 Degree Hybrid Coupler
[39]
Microwaves101 | Magic Tees - Microwave Encyclopedia
A magic tee is a four-port, 180 degree hybrid power divider, realized in waveguide. Originally developed in World-War II, and first published by WA Tyrell in a ...
[40]
[PDF] The Wilkinson Power Divider | EMPossible
Scattering Parameters for Wilkinson Power Divider. Putting everything together, the scattering matrix for the Wilkinson Power Divider is: S = −j. 2. 0 1 1. 1 ...
[41]
Broadband Four-way and Eight-way Wilkinson Example
This design uses multi-stage Wilkinson dividers for wideband performance, dual-band WiFi, 1:4 and 1:8 configurations, and compact dual splitter assemblies.
[42]
[PDF] Updatable Closed-Form Evaluation of Arbitrarily Complex Multi-Port ...
Dec 23, 2024 · This paper proposes a method to update scattering parameter evaluations using a closed-form approach for complex multi-port network connections ...
[43]
Fast Multichannel Inverse Design through Augmented Partial ...
Jan 4, 2024 · ... scattering matrix S = CA–1B – D and then computing the entire S in a ... sparsity of the input profiles, and the high-performance MUMPS package.
[44]
[PDF] Mutual Coupling in MIMO Wireless Systems - Sites at Lafayette
Mutual coupling in MIMO systems occurs when current on one antenna induces voltage on nearby elements due to close spacing, affecting radiation and reception ...<|control11|><|separator|>
[45]
Generalized analytical formulation for de‐embedding of multiport ...
Jul 29, 2022 · Once the S-parameters are measured, the de-embedding is performed by a multistep approach involving the bidirectional transformation of the S- ...
[46]
[PDF] Sparse Matrix Mapping Draft 8 - IBIS Open Forum
Dec 15, 2009 · An index-pair specifies the row and column index in the n-port matrix mapped into which the [Network. Data] is mapped by [Sparse Matrix Mapping] ...
[47]
Transfer S-parameters - Microwave Encyclopedia
Transfer S-parameters express input quantities in function of output quantities, allowing easy cascading of blocks, unlike normal S-parameters.
[48]
What Are Vector Network Analyzers? | VNAs Explained - Tektronix
Behind each of the two VNA port connectors is a directional coupler (green boxes in Figure 7). ... Figure 14 shows an example of a swept frequency transmission ...
[49]
https://www.ibis.org/adhoc/interconnect_wip/Sparse_Matrix_Mapping_draft8_markup.pdf
[50]
[PDF] Advanced Calibration Techniques for Vector Network Analyzers
Other situations that present measurement challenges are those that are mechanically difficult, such as physically long devices, fixed test-port positions, or.
[51]
Specifying Calibration Standards and Kits for Keysight Vector ...
Solving the full 2-port twelve-term error model using the short/open/ load/ thru (SOLT) calibration method is an example of only one of the many measurement ...
[52]
[PDF] Network Analyzer Error Models and Calibration Methods
tracking errors. The TRL and TRM calibration methods need a two port measurement system in order to calibrate. The TRM calibration is the easiest to use.
[53]
[PDF] Applying Error Correction to Network Analyzer Measurements
The smoother, error-corrected trace produced by a two- port calibration subtracts the effects of systematic errors and better represents the actual performance ...
[54]
[PDF] Time-Domain Reflectometry & S-Parameter Channel Models
Increasing Impulse Response Resolution. • Could perform ifft now, but will get an impulse response with time resolution of. • To improve impulse response ...
[55]
[PDF] Conversion of Scattering Parameters to Time-Domain for Imaging ...
Abstract—This article discusses the conversion of scattering parameters from an antenna input port to the time domain for imaging applications.
[56]
Time Domain Analysis of RF Systems - Siglent
Time domain analysis can be set to low pass impulse, step, or bandpass modes. The most important characteristic is the gating function. Use the gating function ...
[57]
Keysight Introduces Two Frequency Extenders and Calibration Kit to ...
Sep 9, 2025 · The 85065A Precision Calibration Kit 0.5 mm enables precise measurements up to 250 GHz when used with the broadband VNA solution. The broadband ...
[58]
Vector Network Analyzer Extension Modules (VNAX) - Virginia Diodes
VDI's VNA Extenders deliver high performance network analyzer frequency extension into the THz range. Models cover 26 GHz to 1,500GHz with additional bands ...
[59]
Vector network analyzers - Rohde & Schwarz
Rohde & Schwarz offers a wide range of versatile, high-performance network analyzers up to 1.1 THz and standard multiport solutions up to 48 ports.
[60]
Network Analyzers | Keysight
Typical selection criteria include frequency range, number of required ports from 2 up to 24 (multiport) or even 48 in a switch matrix, output power, and ...
[61]
Vector Network Analysis with Up to 48 Ports - Microwave Journal
Mar 14, 2014 · The matrix ports can be used just as with a standard two-port or four-port VNA. Calibration is very easy using automatic multiport calibration ...
[62]
Reducing the Complexities Associated with Multiport Component ...
While a network analyzer with radio-frequency switches provides a solution, it adds complexity to test design. Switching time decreases test throughput compared ...
[63]
Multiport method for the measurement of the scattering parameters ...
The multiport method for the precise measurement of the scattering parameters of N-port devices with a two-port vector network analyzer (VNA) is presented.
[64]
Measurement of the scattering-parameters of planar multi-port devices
The measurement of the scattering parameters of multi-port devices with the help of a vector network analyzer (VNA) with two measurement ports is described.Missing: techniques | Show results with:techniques
[65]
Multiport Network S-parameter Restoration and Calibration with 2 ...
Jan 30, 2023 · In this paper, a multiport network S-parameter restoration and calibration procedure is proposed. Taking advantages of the analysis method ...Missing: techniques | Show results with:techniques
[66]
Multimode TRL technique for de-embedding of differential devices
In this paper, the practical use of the multimode TRL calibration technique for de-embedding purposes is discussed. The focus is on the four-port case, since ...
[67]
Mixed Mode Scattering Parameters: What Are They and How Do I ...
Nov 21, 2019 · Mixed mode scattering parameters are a set of network parameters used to characterize differential circuits.Missing: sqrt( | Show results with:sqrt(<|control11|><|separator|>
[68]
On-wafer multi-port circuits charaterization technique with a two-port ...
This method gives an efficient measurement of the scattering S-parameters of these devices at their ports reference planes. Furthermore, it provides cost saving ...
[69]
Uncertainties of Multiport VNA S-Parameter Measurements Applying ...
Aug 22, 2012 · Using the concept of general node equation, a generalized formula for the S-parameter deviations with respect to the error terms has been ...Missing: techniques | Show results with:techniques