Appendix D
Glossary
The terms and abbreviations of the course, each with the chapter that introduces it.
1 2 A B C D E F G H I J K L M N O P Q R S T U V W Z
1
- 1 dB compression point (P1dB)
- The input power (IP1dB) or output power (OP1dB) at which the gain of a block has fallen 1 dB below its small-signal value. Ch. 4 1/f³ corner
- The offset frequency below which the phase noise falls at 30 dB/decade, because flicker noise is up-converted by the DC part of the effective impulse sensitivity function; above it thermal noise gives 20 dB/decade. Ch. 11
2
- 25 % duty-cycle LO
- Four non-overlapping LO phases, each high for a quarter period, so that only one switch of an I/Q mixer conducts at a time; it removes the I/Q crosstalk of a 50 % LO and raises the current-mode gain by 3 dB. Ch. 10
A
- ABCD parameters
- The transmission (chain) matrix that relates the voltage and current at the input of a two-port to those at its output; the ABCD matrix of a cascade is the product of the matrices of its stages. Ch. 3 Additive jitter (of a divider)
- The timing noise a divider adds to the edges of its output, on top of the jitter of its input; its white part sets the divider’s phase-noise floor. Ch. 13 Adjacent-channel power ratio (ACPR, ACLR)
- The power a transmitter leaks into the neighbouring channel relative to the power in its own channel, in dBc; spectral regrowth from PA nonlinearity raises it. Ch. 16 Alexander phase detector
- A bang-bang CDR phase detector that samples the data at the centre of each bit and at the edge between bits; on a transition the edge sample shows whether the clock is early or late. Ch. 15 All-digital PLL (ADPLL)
- A PLL whose phase detector, loop filter and oscillator control are digital: a TDC (or bang-bang detector) measures the phase error as a number, a digital filter processes it, and a digitally controlled oscillator is tuned by a code. It needs no charge pump or loop-filter capacitor. Ch. 15 AM-AM and AM-PM conversion
- The change of a PA’s gain (AM-AM) and of its insertion phase (AM-PM) with the input amplitude; together they describe a memoryless nonlinear amplifier and set the distortion of an envelope-varying signal. Ch. 16 AM-to-PM conversion
- The conversion of amplitude fluctuations into phase or frequency fluctuations by a nonlinearity whose reactance depends on the amplitude, such as a varactor driven by a large swing; it is one route by which low-frequency noise reaches an oscillator’s phase. Ch. 11 π and T networks
- Three-element matching networks made of two L-sections back to back through a virtual resistance below (π) or above (T) both terminations, which lets the designer choose the network’s Q. Ch. 8 Automatic frequency calibration (AFC)
- The open-loop search, run before a PLL locks, that picks a VCO’s capacitor-bank code for the wanted channel: the VCO’s frequency is counted against the reference and compared with the target, one bank bit at a time, so that the tuning voltage settles inside its intended range. Ch. 11 Available power
- The largest power a source can deliver, obtained when the load is the complex conjugate of the source impedance; \(V_s^2/(8R_s)\) for a sinusoidal source of peak amplitude \(V_s\) and resistance \(R_s\). Ch. 1 Available power gain (GA)
- The power available at the output of a two-port divided by the power available from the source; independent of the load impedance. Ch. 3 Average power tracking (APT)
- Lowering a PA’s supply voltage with a DC-DC converter when its output power is lowered, so that the PA stays near saturation and keeps more of its efficiency; the slow, per-slot form of envelope tracking. Ch. 19
B
- Back-off
- The amount, in dB, by which a PA’s average output (or input) power is kept below its saturated or compressed power so that the peaks of an envelope-varying signal are not clipped; output back-off of at least the PAPR is typical without linearization. Ch. 16 Balun
- A balanced-to-unbalanced transformer: it converts a single-ended signal referred to ground into a differential one, or back, for example between an antenna and a differential LNA or PA. Ch. 7 Bang-bang phase detector (BBPD)
- A one-bit phase detector, usually a single flip-flop, that says only whether one edge is early or late. Its effective gain is √(2/π)/σ for Gaussian jitter of rms σ, so it depends on the jitter. Ch. 15 Barkhausen criterion
- The condition for a steady oscillation of a feedback loop: a loop gain of exactly one, in magnitude and phase (0° or a multiple of 360°), at the oscillation frequency. Start-up needs a loop gain above one; the amplifier’s nonlinearity then reduces it to one. Ch. 11 BiCMOS
- A process that offers bipolar transistors (in SG13G2, SiGe HBTs) and CMOS transistors on the same chip. Ch. 1 Block budget (link budget)
- The allocation of gain, noise figure, linearity, I/Q accuracy and phase noise to each block of a radio so that the whole chain meets the system specifications; the course radio’s is Table 5.3. Ch. 19 Blocker
- A strong unwanted signal near the receive band, from another radio or another channel, which the receiver must tolerate while it receives a weak wanted signal. Ch. 4 Blocking, out-of-band blocking
- A strong unwanted signal, usually unmodulated in the test, outside the wanted channel or outside the band, that the receiver must tolerate while still receiving a weak wanted signal; it acts through compression, reciprocal mixing and spurious responses. Ch. 19 Bode–Fano limit
- The bound on the integral of ln(1/|Γ|) over frequency for any lossless match of a reactive load; it caps the bandwidth over which a given return loss can be reached. Ch. 8 Bounce diagram
- A time–position (lattice) diagram that tracks a step as it reflects back and forth between a mismatched source and load, adding the waves to give the voltage at each end. Ch. 3 BVCEO, BVCBO
- Breakdown voltages of a bipolar transistor: collector to emitter with the base open, and collector to base with the emitter open. Avalanche current that flows out of an open base is multiplied by β, so BVCEO is several times smaller than BVCBO. Ch. 6
C
- Calibration (foreground, background)
- Measuring and correcting an analog error with on-chip circuits and digital logic: in the foreground with the radio taken out of service (for example at power-up or in production test), in the background while it operates. Ch. 19 Carrier frequency offset
- The difference between the carrier of a received signal and the receiver’s LO, from the two crystals’ tolerances; Bluetooth Low Energy allows ±150 kHz, which the channel filter must pass and the demodulator must estimate and remove. Ch. 19 Cascode
- A common-gate transistor stacked on the drain of the input transistor. It isolates the input from the output (small \(S_{12}\)), reduces the Miller effect of \(C_{gd}\), and raises the output resistance. Ch. 9 CCDF
- Complementary cumulative distribution function: the probability that the instantaneous power exceeds its average by more than a given number of decibels. Ch. 5 Channel-noise factor (\(\gamma\))
- The ratio of a MOSFET’s white drain-current noise to \(4kTg_m\); 2/3 for a long-channel device in saturation, about 0.9 for the course’s short-channel nMOS. Ch. 4 Characteristic current density
- A current per unit width (MOSFET, mA/µm) or per unit emitter area (HBT, mA/µm²) at which a figure of merit peaks: fT, fmax or NFmin. Biasing at a chosen current density keeps a design’s speed and noise when its size changes. Ch. 6 Characteristic impedance (\(Z_0\))
- The ratio of voltage to current of a wave travelling along a transmission line; RF systems use \(Z_0\) = 50 Ω for cables, connectors, filters and test equipment. Ch. 1 Charge-pump PLL
- The usual PLL of radios and clock generators: a phase-frequency detector switches two current sources that push charge into a passive loop filter, whose voltage tunes the VCO. Its linear model is due to Gardner. Ch. 12 Class A, AB, B and C
- PA classes defined by the conduction angle, set by the bias: class A conducts all the time (maximum efficiency 50 %), class B half the time (78.5 %), class AB in between, class C less than half the time (efficiency towards 100 %, output power towards zero). Ch. 16 Class D
- A switching PA in which transistors switch the output node between the supply and ground and a series-resonant filter passes only the fundamental to the load; ideally 100 % efficient, but every switching event dissipates the energy stored in the node capacitance. Ch. 16 Class E
- A single-transistor switching PA (Sokal and Sokal, 1975) whose load network makes the switch voltage fall to zero with zero slope just before the switch closes, so the shunt capacitance is never discharged through the switch; its peak switch voltage is about 3.56 VDD. Ch. 16 Class F
- A PA whose output network presents an open circuit at the odd harmonics and a short circuit at the even harmonics, which squares the collector voltage while the current is a half sine; with the third harmonic alone the efficiency rises from 78.5 % to 88.4 %. Ch. 16 Clock and data recovery (CDR)
- A loop that extracts a sampling clock from the transitions of a received data stream and samples the data with it. Ch. 15 CML latch
- A level-sensitive storage element in CML: a track pair copies the input while the clock is high, and a cross-coupled pair holds the value while the clock is low. Two latches clocked in antiphase make a master-slave flip-flop or a divide-by-two. Ch. 13 Coexistence filter
- A band-pass filter (often an acoustic BAW filter) in front of a receiver that shares a device with transmitters in neighbouring bands, such as Bluetooth and Wi-Fi next to LTE bands 40 and 41 in a phone. Ch. 19 Coherent sampling
- Choosing the record length of a discrete Fourier transform to hold an exact integer number of periods of every tone, so that each tone falls into one bin without leakage. Ch. 2 Collector (drain) efficiency (η)
- The ratio of the RF output power at the fundamental to the DC power drawn from the supply, η = Pout/PDC. It ignores the input power. Ch. 16 Colpitts oscillator
- A single-transistor LC oscillator in which a capacitive divider (C₁, C₂) across the tank feeds part of the output back to the input; its negative resistance is −g_m/(ω²C₁C₂). Ch. 11 Common-gate LNA
- An LNA whose input is the source of a MOSFET: its input impedance is about \(1/(g_m + g_{mb})\), so it matches 50 Ω over a wide band, but its noise factor at match cannot fall below about \(1 + \gamma/\alpha\). Ch. 9 Common-mode feedback (CMFB)
- A loop in a fully differential amplifier that senses the average of its two outputs, compares it with a reference and adjusts a bias current, because the differential feedback around the amplifier does not set the output common-mode level. Ch. 10 Complex envelope (I/Q)
- The baseband signal \(I(t) + jQ(t)\) that describes a narrowband RF signal \(\mathrm{Re}\{[I(t) + jQ(t)]\,e^{j\omega_ct}\}\): its in-phase and quadrature components. Ch. 5 Complex (polyphase) filter
- A filter that acts on I + jQ as one complex signal, with a response that differs at positive and negative frequencies; a low-pass prototype shifted to +fIF selects the wanted channel and rejects the image. Ch. 5 Conduction angle (α)
- The part of each RF period, in degrees, during which the transistor of a PA conducts: 360° in class A, between 180° and 360° in class AB, 180° in class B and less than 180° in class C. Ch. 16 Conjugate match
- Terminating a source of impedance ZS in ZS*, which delivers the source’s whole available power to the load (no reflection). Ch. 8 Constellation
- The set of points in the I/Q plane that a digital modulation uses as its symbols: 4 for QPSK, 16 for 16-QAM. Ch. 5 Conversion gain
- The ratio of the IF output of a mixer to its RF input, as a voltage, current or power ratio; for a block in a 50 Ω budget, the IF voltage over the available source voltage. Ch. 10 Coupling coefficient (k)
- The mutual inductance of two coupled inductors normalized to their self-inductances, k = M/√(L1L2); 0 for no coupling and 1 for an ideal transformer. Stacked on-chip windings reach about 0.7–0.8. Ch. 7 Cross-coupled pair
- Two transistors whose gates (or bases) are connected to each other’s drains (collectors). Its differential input resistance is −2/g_m, the usual negative resistance of integrated LC oscillators; a complementary version uses an NMOS and a PMOS pair. Ch. 11 Cross-modulation
- The transfer of the amplitude modulation of a strong blocker onto the wanted signal through the third-order nonlinearity. Ch. 4 Current-commutating mixer
- A passive mixer driven by an RF current, whose switches steer that current into the inputs of a transimpedance amplifier in turn. Ch. 10 Current hogging
- The tendency of the warmest of several parallel bipolar transistors to take more current, because its base–emitter voltage falls with temperature; ballast resistors and thermal spacing counter it. Ch. 18 Current-limited and voltage-limited regimes
- The two operating regions of a tail-biased LC oscillator. In the current-limited regime the amplitude is set by the tail current, about (4/π)I R_p for a complementary pair; in the voltage-limited regime it is clipped by the supply and the transistors’ headroom, and more current only adds noise. Ch. 11 Current-mode logic (CML)
- Differential logic in which a constant (or clock-shaped) current is steered between two resistive loads by a differential pair. The output swing is small (a few hundred millivolts), so the gates are fast and draw a constant supply current. Ch. 13 Cyclostationary noise
- Noise whose statistics repeat periodically, such as the thermal noise of a switch whose conductance follows the LO; its components at frequencies one LO harmonic apart are correlated. Ch. 10
D
- dBm
- Power in decibels relative to 1 mW: \(P_{dBm} = 10\log_{10}(P/1\ \mathrm{mW})\). 0 dBm is 1 mW, −90 dBm is 1 pW. Ch. 1 DCO quantization noise
- Phase noise caused by the finite frequency step of a DCO: the rounding error of the tuning word, integrated into phase, falls at 20 dB per decade. Delta-sigma dithering of the finest bank shapes it to high offsets. Ch. 15 De-embedding
- Removing the effect of the pads and interconnects between the calibrated probe plane and the device from measured S-parameters, using dummy structures such as an open and a short. Ch. 3 Delay-locked loop (DLL)
- A loop that adjusts a voltage-controlled delay line until its output lines up with its input one period later. It has no oscillator, so its jitter does not accumulate, but it cannot synthesize a new frequency. Ch. 15 Delta-sigma (ΔΣ) modulator (in a synthesizer)
- A digital circuit that turns a fine fractional word into a stream of small integers with the right average, and pushes the error between the two (the quantization noise) to high frequencies, where the loop filter removes it. Ch. 14 Desensitization
- The loss of small-signal gain (and so of sensitivity) caused by a strong blocker that drives a stage into compression. Ch. 4 Desensitization (desense)
- The rise of a receiver’s sensitivity level, in dB, caused by an interferer or by the radio’s own transmitter, through compression of the wanted signal or noise added in the channel. Ch. 19 Design rule check (DRC)
- The automatic check of a layout against the foundry’s geometric rules (widths, spaces, enclosures, densities); a layout must pass it before it can be manufactured. Ch. 18 Digitally controlled oscillator (DCO)
- An oscillator tuned by a digital code; in an RF ADPLL, an LC oscillator whose tank capacitance is a sum of switched units in banks of decreasing step size (PVT, acquisition, tracking). Ch. 15 Digital predistortion (DPD)
- Distorting the baseband signal digitally with the inverse of the PA’s AM-AM and AM-PM characteristics, so that the cascade of predistorter and PA is linear; it lets a PA run closer to compression. Ch. 16 Digital-to-time converter (DTC)
- A programmable delay that shifts a clock edge by a digitally set time. In a fractional-N PLL it delays the divider output by the modulator’s predicted timing error, so the phase detector sees only the residue. Ch. 14 Distributed element
- A structure whose size is a sizeable fraction of the wavelength (a transmission line, a long wire), so that the voltage and current vary along it. Ch. 1 Dithering (of a DCO)
- Toggling the least significant capacitor unit of a DCO at a high rate, under control of a delta-sigma modulator, so that its average gives a finer frequency resolution and its quantization noise is moved to high frequencies. Ch. 15 Doherty amplifier
- Two amplifiers, a main one biased in class AB and an auxiliary one in class C, combined through a quarter-wave impedance inverter; the auxiliary amplifier turns on at high power and modulates the main amplifier’s load, keeping the efficiency high over the top 6 dB of output power. Ch. 16 DSB noise figure
- The noise figure of a mixer for a signal that occupies both sidebands, as in a zero-IF receiver. Ch. 10 Dual-modulus prescaler
- A prescaler that divides by P or P + 1 according to a modulus-control input; with two counters it makes a programmable divider whose fast part never changes its structure. Ch. 13
E
- \(E_b/N_0\)
- The energy per bit divided by the noise power spectral density, the SNR measure of digital communications; \(\mathrm{SNR} = (E_b/N_0)(R_b/B)\) for a bit rate \(R_b\) and noise bandwidth \(B\). Ch. 4 Eddy currents (in the substrate)
- Currents induced in the conductive silicon under an inductor by its alternating magnetic field. They flow opposite to the spiral’s current, lower its inductance slightly and dissipate power, which lowers Q at high frequencies. Ch. 7 Effective permittivity
- The relative permittivity of a uniform medium in which a wave would travel at the same speed as on a given transmission line; for a microstrip it lies between that of the dielectric and that of air. Ch. 2 Electrical length
- The length of a line expressed as the phase shift a wave accumulates along it, \(\theta = 360°\,\ell/\lambda\); a line longer than about λ/10 (36°) must be treated as a transmission line. Ch. 1 Emitter-coupled logic (ECL)
- The bipolar form of current-mode logic, with emitter-coupled pairs (and usually emitter followers between stages); in SiGe it is the fastest static logic available. Ch. 13 Envelope tracking (ET)
- Supplying a linear PA from a fast DC-DC converter whose output follows the envelope of the signal, so that the collector always swings close to the supply and the efficiency stays high in back-off. Ch. 16 Equivalent noise factor (oscillator)
- The factor F that Leeson’s model needs to reproduce an oscillator’s simulated or measured phase noise: the total phase noise in the 1/f² region divided by that of the tank’s own losses. Ch. 11 Equivalent noise resistance (\(R_n\))
- The resistance whose thermal noise, \(4kTR_n\), equals a device’s input-referred noise voltage: \(R_g + \gamma/g_m\) for a MOSFET, \(r_b + r_e + 1/(2g_m)\) for a bipolar transistor. Ch. 4 Error-vector magnitude (EVM)
- The rms distance between the received and the ideal constellation points, relative to the rms symbol magnitude; about \(1/\sqrt{\mathrm{SNR}}\) for white noise. Ch. 5 ESD protection
- Diodes (or clamps) at every pad that conduct the current of an electrostatic discharge to the supply rails; at an RF input their capacitance (tens of femtofarads) is part of the matching network. Ch. 9
F
- μ-factor
- Edwards and Sinsky’s single stability test: μ = (1 − |S11|²)/(|S22 − ΔS11*| + |S12S21|) is the distance from the chart centre to the nearest unstable load; μ > 1 means unconditional stability. Ch. 3 False lock (of a DLL)
- Locking at a delay of two or more periods, or getting stuck at the end of the delay range, because the phase detector cannot tell these states from the correct one-period lock. Ch. 15 Finger (of a transistor)
- One gate stripe of a multi-finger MOSFET layout. A device of width W is drawn as ng fingers of width Wf = W/ng that share source and drain diffusions, which cuts the gate resistance and the drain and source capacitances. Ch. 6 Finite-difference time-domain (FDTD) method
- An electromagnetic solver that discretizes Maxwell’s curl equations on a staggered (Yee) grid and steps the fields forward in time; one run with a broadband pulse gives the response over a wide frequency range. Ch. 2 Finite-element method (FEM)
- An electromagnetic solver that divides the structure into small (usually tetrahedral) elements and solves Maxwell’s equations at each frequency as a sparse linear system; it follows curved and very thin geometry closely. Ch. 2 Finite-element order
- The polynomial degree of the basis functions in a finite-element solver; first-order elements are cheap but need a finer mesh than second-order elements for the same accuracy. Ch. 18 Flash (delay-line) TDC
- A TDC in which one edge runs down a chain of inverters and the other edge samples all taps at once with flip-flops; the resulting thermometer code locates the first edge to within one inverter delay. Ch. 15 Flicker corner
- The frequency at which the flicker noise of a device equals its white (thermal or shot) noise; about 12 MHz for the course’s nMOS and a few hundred hertz for its SiGe HBT. Ch. 4 Flicker noise (1/f noise)
- Low-frequency noise whose power spectral density falls in proportion to 1/f, from carrier trapping and mobility fluctuations; strong in MOSFETs, weak in bipolar transistors. Ch. 4 Forward transit time (τF)
- The time the minority carriers of a bipolar transistor take to cross the base and the collector depletion region at a given current; it sets the diffusion capacitance gmτF and the high-current limit of fT, 1/(2πτF). Ch. 6 Fractional-N synthesizer
- A PLL whose feedback divider changes its integer ratio from one reference cycle to the next so that the average ratio has a fractional part. The channel step becomes a fraction of the reference frequency, so the reference, and with it the loop bandwidth, can be much larger than the channel spacing. Ch. 14 Fractional spur
- A spurious tone at a multiple of the fractional part of N times the reference frequency, caused by a periodic pattern in the divider’s ratio sequence and the phase error it leaves at the phase detector. Ch. 14 Frequency control word (FCW)
- The ratio fDCO/fref given to a phase-domain ADPLL, possibly fractional; the reference phase accumulator adds it once per reference cycle. Ch. 15 Frequency deviation (Δf1avg, Δf2avg)
- The modulation accuracy measures of Bluetooth Low Energy’s transmitter test: the average frequency deviation for a 11110000 pattern (225–275 kHz) and for a 1010 pattern, whose ratio to the first must be at least 0.8. Ch. 19 Frequency divider
- A circuit whose output frequency is its input frequency divided by an integer N; it divides the phase, and so the phase deviations, by N as well. In a PLL it sits in the feedback path between the VCO and the phase detector. Ch. 13 Frequency pushing
- The change of an oscillator’s frequency with its supply voltage or bias current, in Hz/V or Hz/A; through it, low-frequency noise on the supply or the tail becomes frequency noise. Ch. 11 Friis formula
- The noise factor of a cascade, \(F = F_1 + (F_2-1)/G_1 + (F_3-1)/(G_1G_2) + \dots\), with the available gains \(G_k\) and each \(F_k\) taken for the source impedance that stage actually sees. Ch. 4 Frozen-operating-point noise analysis
- A quasi-static way to find cyclostationary noise: at each instant of the large-signal cycle, a noise analysis of each device held at that instant’s terminal voltages gives its noise density there. Ch. 11
G
- Gain expansion
- A rise of gain with input amplitude, before compression sets in; it happens when the third-order coefficient has the same sign as the gain, as in a bipolar transistor’s exponential characteristic. Ch. 4 Gate resistance (Rg)
- The resistance in series with the gate of a MOSFET, from the distributed resistance of the gate electrode, the contacts and the gate wiring. It limits fmax and adds thermal noise; laying the transistor out as many narrow fingers contacted at both ends reduces it. Ch. 6 Gds2palace
- IHP’s open-source Python flow that turns a GDSII layout and an XML description of the SG13G2 stack-up into a meshed model and configuration for the Palace finite-element solver, with the ports drawn as shapes on special layers. Ch. 7 Gear shifting
- Switching a digital PLL’s loop gains from wide (fast acquisition) to narrow (low noise) once it is locked, keeping the filter’s state so that the switch causes no transient. Ch. 15 GFSK
- Gaussian frequency-shift keying: binary FSK whose frequency pulses are smoothed by a Gaussian filter of bandwidth-time product BT; constant envelope. Bluetooth Low Energy uses BT = 0.5 and a modulation index of 0.5. Ch. 5 Gilbert cell
- A double-balanced active mixer: a differential transconductor whose output currents are commutated by a quad of switching transistors into a resistive load. Ch. 10 Ground plane (on chip)
- A wide sheet of a low metal, such as Metal1, under lines and passives that carries the return current, sets the line impedance and shields the lossy substrate; the design rules require slits in it. Ch. 18 Ground–signal–ground (GSG) pad
- Three probe pads in a row, a signal pad between two ground pads, that match the tips of a coplanar RF probe; the ground pads tie the probe’s ground to the chip’s ground plane. Ch. 18 GSG pads
- Ground–signal–ground probe pads: the on-chip landing for a coplanar microwave probe, a signal pad between two ground pads. Ch. 3
H
- Harmonic balance (HB)
- A steady-state analysis that writes every node voltage as a sum of harmonics of the input frequencies, solves the linear part of the circuit in the frequency domain and the nonlinear devices in the time domain, and adjusts the harmonics with Newton’s method until the currents balance at every frequency. Ch. 2 Harmonic distortion (HD2, HD3)
- The power of the second or third harmonic of a single-tone output relative to the fundamental, in dBc. Ch. 4 Harmonic trap
- A series LC branch from a PA’s output node to ground, resonant at a harmonic (usually 2f0), that shorts that harmonic; below resonance it looks like a capacitor and can be part of the matching network. Ch. 16 Hartley image-reject receiver
- Quadrature down-conversion followed by a 90° shift in one path and a sum, which adds the wanted signal and cancels the image. Ch. 5 Heterodyne receiver
- A receiver that converts the RF signal to a non-zero IF, selects the channel there and rejects the image with a filter before the mixer. Ch. 5 Heterojunction bipolar transistor (HBT)
- A bipolar transistor whose emitter-base junction joins two different semiconductors. In a SiGe HBT the base is a thin, heavily doped SiGe layer whose graded germanium content speeds up the electrons crossing it. Ch. 6 Hogge phase detector
- A linear CDR phase detector built from flip-flops and XOR gates: a pulse whose width follows the phase error is compared with a fixed half-bit reference pulse. Ch. 15
I
- IIC-OSIC-TOOLS
- A Docker image from JKU Linz with the open-source IC design tools (Xschem, ngspice, Xyce, KLayout, openEMS, Palace and others) and the open PDKs preinstalled; the course’s reference environment. Ch. 1 Image frequency
- The input frequency on the other side of the LO that a mixer converts to the same IF as the wanted signal: \(f_{RF} \mp 2f_{IF}\). Ch. 5 Image-rejection ratio (IRR)
- The gain to the wanted signal divided by the gain with which the image reaches the wanted channel; limited by the gain and phase mismatch of the I and Q paths. Ch. 5 Impedance transparency
- The property of a passive mixer that the baseband impedance appears, frequency-translated to around the LO, at its RF port. Ch. 10 Impulse sensitivity function (ISF)
- The phase shift that a small charge impulse injected at a node causes, as a function of the phase of the oscillation at which it arrives, normalized to the maximum charge swing; it turns an oscillator’s noise sources into its phase noise (Hajimiri and Lee). Ch. 11 In-band phase noise
- The flat part of a PLL’s output phase noise inside the loop bandwidth, set by the reference, charge pump and divider noise multiplied by N (20 log N in decibels). Ch. 12 Induced gate noise
- The noise current that the fluctuating channel charge couples onto the gate through \(C_{gs}\); it grows with frequency, is partly correlated with the channel noise, and sets an optimum Q for the input network of an inductively degenerated LNA. Ch. 9 Inductive source degeneration
- An inductor \(L_s\) in the source of the input transistor. Together with \(C_{gs}\) it makes the input impedance have a real part \(\omega_T L_s\) that is free of thermal noise; a gate inductor resonates the rest, so the input matches at one frequency. Ch. 9 Injection-locked frequency divider (ILFD)
- An oscillator running near f/N that is locked by a signal at f injected into it (for example into the tail of a cross-coupled pair). It works at millimetre-wave frequencies with little power, but only over a narrow lock range. Ch. 13 Injection pulling
- The disturbance of an oscillator by an injected signal close to its frequency, e.g. the PA output coupling into the VCO; the oscillator locks within Adler’s lock range. Ch. 5 Injection pulling
- The disturbance of an oscillator’s frequency or phase by a signal coupled into its tank near its frequency (or a harmonic of it); within Adler’s lock range the oscillator locks to the injected signal, just outside it it is pulled. Ch. 19 Input-network quality factor (\(Q_L\))
- \(Q_L = \omega_0(L_g + L_s)/R_S = 1/(\omega_0 R_S C_{gs})\) of an inductively degenerated LNA at resonance; the gate-source voltage is about \(Q_L/2\) times the source voltage. Ch. 9 Input sensitivity curve
- The minimum input amplitude (or power) at which a divider still divides correctly, plotted against the input frequency. Its minimum lies near twice the self-oscillation frequency; its right edge is the maximum toggle frequency. Ch. 13 Insertion loss (of a matching network)
- The ratio, in dB, of the power accepted by a network to the power it delivers; for an L-section about 10 log(1 + Q/QL + Q/QC). Ch. 8 Integral nonlinearity (INL)
- The deviation of a converter’s transfer characteristic from a straight line, as a function of its code. In a DTC it is a code-dependent delay error; applied to a fractional-N modulator’s large timing error, it folds the shaped noise into the loop band and, where it has steps, creates fractional spurs. Ch. 14 Integrated jitter (rms phase error)
- The square root of twice the integral of the phase noise L(f) over an offset band, in radians (or degrees); divided by 2π times the carrier frequency it is the rms timing jitter. Ch. 12 Intermediate frequency (IF)
- The frequency to which a receiver converts the wanted channel before (or instead of) baseband. Ch. 5 Intermodulation (IM2, IM3)
- Output tones at sums and differences of multiples of two input frequencies; IM3 at \(2f_1 - f_2\) and \(2f_2 - f_1\) fall next to the wanted signal, IM2 at \(f_2 - f_1\) at baseband. Ch. 4 I/Q imbalance
- Gain error ε and phase error θ between the in-phase and quadrature paths; makes \(y = K_1z + K_2z^*\) and limits the IRR to about \(4/(\varepsilon^2 + \theta^2)\). Ch. 5 ISM band
- A frequency band set aside for industrial, scientific and medical use, in which licence-free radios (Wi-Fi, Bluetooth) may also operate under power limits; the course radio uses the 2.400–2.4835 GHz band. Ch. 1
J
- Jitter tolerance
- The largest sinusoidal jitter, as a function of its frequency, that a CDR or receiver tolerates without exceeding a given bit-error rate. Ch. 15 Jitter transfer
- The transfer function from jitter at a loop’s input to jitter at its output; low-pass for a PLL or CDR. Ch. 15
K
- Kirk effect
- Base push-out in a bipolar transistor: when the electron density in the collector depletion region approaches the collector doping, the field collapses, the base widens into the collector and fT falls. It sets the current density of peak fT. Ch. 6 Knee voltage (Vknee)
- The lowest collector (drain) voltage at which the transistor still behaves as a current source; the collector cannot swing below it without saturating, which reduces the usable voltage swing of a PA to VCC − Vknee. Ch. 16
L
- Leakage and magnetizing inductance
- In the model of a transformer with coupling k, the series inductance (1 − k²)L1 of the flux that does not link the secondary, and the shunt inductance k²L1 of the flux that does. Ch. 8 Leeson’s model
- An empirical expression for the phase noise of a feedback oscillator in terms of the resonator’s loaded Q, the signal power, a noise factor and the flicker corner: a −20 dB/decade region inside the resonator’s half-bandwidth, a −30 dB/decade region below the flicker corner and a floor far out. Ch. 11 Limit cycle
- A self-sustained periodic oscillation of a nonlinear loop; in a bang-bang PLL without noise, the phase error circulates around zero with a size set by the step size and the loop latency. Ch. 15 Link budget
- The allocation of gain, noise figure, linearity and other specifications to the blocks of a radio so that the chain meets its sensitivity and interference requirements. Ch. 5 Load-line match
- An output match that presents the collector (drain) with the resistance set by the supply, the knee voltage and the output power, Ropt = (VCC − Vknee)²/2Pout, with the device’s output capacitance tuned out; it differs from a conjugate match. Ch. 17 Load-pull
- Measuring or simulating a PA’s output power and efficiency while the load impedance at the fundamental (and sometimes at the harmonics) is varied over the Smith chart; the results are drawn as contours and show the optimum load. Ch. 16 Lock range (of an injection-locked divider)
- The span of input frequencies over which an injection-locked divider stays locked; it grows with the injection strength and falls with the tank quality factor. Ch. 13 Loop bandwidth (fu)
- The frequency at which the open-loop gain of a PLL falls to one. Below it the output follows the reference (and the noise of the blocks before the VCO); above it the output follows the free-running VCO. Ch. 12 LO self-mixing
- LO power leaking to the mixer’s RF input (or radiating and returning) and mixing with the LO itself, producing a DC offset in a zero-IF receiver. Ch. 5 Low-IF receiver
- A quadrature receiver with an IF of the order of the channel bandwidth; keeps the signal away from DC and rejects the image with I/Q processing such as a complex filter. Ch. 5 Low-noise amplifier (LNA)
- The first active stage of a receiver: it presents a matched input to the antenna, adds as little noise as possible and gives enough gain that the noise of the following stages hardly matters. Ch. 9 L-section
- A two-element matching network: a series reactance on the low-resistance side and a shunt reactance on the high-resistance side; its Q is fixed by the transformation ratio, Q = √(Rhigh/Rlow − 1). Ch. 8 Lumped element
- A resistor, capacitor or inductor much smaller than the wavelength, so that the voltage and current are the same along it at any instant. Ch. 1 Lumped port (EM)
- An excitation in an electromagnetic simulation that connects a reference impedance (usually 50 Ω) across a small gap, for example vertically from the ground plane to a line; its own parasitics belong to the reference plane, not to the structure. Ch. 18
M
- Macromodel
- A compact circuit, such as a set of RC states and controlled sources from vector fitting, that reproduces a block’s simulated or measured S-parameters inside a circuit simulator, including transient analysis. Ch. 18 MASH modulator
- Multi-stage noise shaping: a cascade of first-order accumulators, each fed with the previous one’s residue, whose carries are recombined so that the output error is the last stage’s error shaped by (1 − z−1)L. It is unconditionally stable; an order-L MASH 1-1-…-1 adds between −(2L−1 − 1) and 2L−1 to the integer ratio. Ch. 14 Mason’s unilateral gain (U)
- The power gain of a two-port after it has been made unilateral with a lossless, reciprocal feedback network and matched at both ports. It does not change under such embeddings, which is why its unity-gain frequency fmax is a property of the device. Ch. 6 Matching network
- A (nearly) lossless two-port that transforms a load impedance into the impedance a source or device should see: an L, π or T network, a transformer or transmission-line sections. Ch. 8 Maximum available gain (MAG)
- The power gain of an unconditionally stable two-port (\(K > 1\)) with both ports conjugate-matched: \(\mathrm{MAG} = \vert S_{21}/S_{12}\vert\,(K - \sqrt{K^2-1})\). Ch. 1 Maximum oscillation frequency (\(f_{max}\))
- The frequency at which Mason’s unilateral power gain U of a transistor falls to one; above it the device cannot amplify power. Ch. 1 Maximum oscillation frequency (fmax)
- The frequency at which Mason’s unilateral gain U of a transistor falls to one: the highest frequency at which the device can amplify power, and so oscillate. It depends on the gate (base) resistance and on Cgd (Cμ) as well as on fT. Ch. 6 Maximum stable gain (MSG)
- The figure of merit \(\vert S_{21}/S_{12}\vert\) of a potentially unstable two-port (\(K < 1\)): the largest gain obtainable after the device is just made stable with added loss. Ch. 1 Medium-power amplifier (MPA)
- An amplifier stage that delivers a moderate output power (around +10 dBm at millimetre waves), used as a driver or as the output stage of a short-range transmitter. Ch. 17 Metal slit
- A narrow opening that the design rules require in metal wider than a set width (30 µm in SG13G2), to relieve mechanical stress and keep chemical–mechanical polishing uniform. Ch. 18 Microstrip line (on chip)
- A transmission line made of a signal strip over a ground plane, separated by a dielectric; on chip, for example a TopMetal2 strip over a Metal1 ground, about 16 µm wide for 50 Ω in SG13G2. The ground plane shields the lossy substrate. Ch. 7 Millimetre wave (mm-wave)
- Frequencies from 30 to 300 GHz (the ITU’s EHF band), where the free-space wavelength is between 10 mm and 1 mm. Ch. 1 Millimetre wave (mm-wave)
- Frequencies from 30 to 300 GHz, where the free-space wavelength is 10 to 1 mm; on chip, interconnects of a few hundred micrometres are already a sizeable fraction of a wavelength. Ch. 17 MIM capacitor
- A metal-insulator-metal capacitor: two metal plates separated by a thin, high-permittivity dielectric in the back end of line. In SG13G2 it sits between Metal5 and TopMetal1 and gives 1.5 fF/µm² (cap_cmim; cap_rfcmim adds the series inductance, resistance and substrate network). Ch. 7 Minimum noise figure (NFmin)
- The lowest noise figure a two-port can reach, obtained when it is driven from the optimum source reflection coefficient Γopt. With the noise resistance Rn it gives the noise figure for any other source. Ch. 6 Mismatch loss
- −10 log10(1 − |Γ|²), in dB: the drop in delivered power caused by the reflection at a mismatched port. Ch. 3 π model (of an inductor)
- A lumped equivalent circuit of a two-port spiral: a series branch (inductance with a frequency-dependent resistance) bridged by a capacitance Cp, and at each port a shunt branch of the oxide capacitance Cox in series with the substrate’s Rsi ∥ Csi. Fitted to electromagnetic simulation or measurement, it is used in circuit simulation. Ch. 7 Modulation index (h)
- In FSK, the peak-to-peak frequency deviation divided by the bit rate, \(h = 2\Delta f/R_b\); with h = 0.5 each bit advances the phase by ±90°. Ch. 5 MOM capacitor
- A metal-oxide-metal capacitor made of interdigitated fingers on several stacked metal layers, which uses the lateral (fringe) field between the fingers. It needs no extra mask; SG13G2’s cap_cmomi gives about 1.3 fF/µm² with Metal1 to Metal5. Ch. 7 Multi-project wafer (MPW)
- A shared fabrication run in which many designs occupy one reticle and share the mask cost; IHP runs open-source MPWs for designs in SG13G2. Ch. 19 Multi-time partial differential equation (MPDE)
- A formulation that gives each widely separated time scale of a circuit (for example a carrier and its envelope) its own time variable, so that a signal with fast and slow parts can be simulated efficiently. Ch. 2
N
- Negative resistance
- A one-port whose small-signal voltage and current have opposite signs, so that it delivers power instead of absorbing it. Connected across a lossy tank it cancels the tank’s loss; an oscillator starts when the negative resistance is smaller in magnitude than the tank’s parallel resistance. Ch. 11 Noise cancelling
- A wideband LNA technique in which an auxiliary amplifier senses the input node and adds the input signal to the output so that the main transistor’s noise, which appears in phase at input and output, cancels while the signal adds (Bruccoleri et al.). Ch. 9 Noise factor and noise figure (F, NF)
- The ratio of the signal-to-noise ratio at the input of a block to that at its output, with the source at \(T_0\) = 290 K; the noise figure is \(\mathrm{NF} = 10\log_{10}F\) in decibels. Ch. 4 Noise folding
- The mixing of high-frequency, noise-shaped phase error down to low offsets by a nonlinear phase detector or charge pump (for example, UP/DN current mismatch). It raises the in-band noise floor and creates fractional spurs. Ch. 14 Noise modulating function (NMF)
- The periodic factor by which the instantaneous spectral density of a cyclostationary noise source (such as a switching transistor’s drain noise) varies over the oscillation cycle; multiplied by the ISF it gives the effective ISF of that source. Ch. 11 Noise parameters
- The four numbers (\(F_{min}\), \(R_n\) and the complex \(\Gamma_{opt}\)) that give the noise factor of a two-port for any source impedance; \(F_{min}\) is reached when the source reflection coefficient is \(\Gamma_{opt}\). Ch. 4 Noise resistance (Rn)
- The noise parameter that sets how fast the noise figure of a two-port rises when the source moves away from Γopt. ngspice reports it in ohms. Ch. 6 Noise share
- The fraction of a receiver’s noise factor contributed by one block, (Fk − 1) divided by the gain before it, relative to the total; it says which block to improve first. Ch. 19 Noise temperature (\(T_e\))
- The temperature at which the source resistance would produce, referred to the input, the noise a block adds: \(T_e = (F-1)T_0\); used for very low-noise amplifiers, where NF is a fraction of a decibel. Ch. 4 Noise transfer function (of a PLL)
- The transfer from one noise source to the output phase: low-pass and scaled by N for the reference, the divider and the charge pump; band-pass for the loop-filter resistors; high-pass for the VCO. Ch. 12 Non-quasi-static (NQS) effects
- Effects of the finite time the channel charge of a MOSFET needs to follow its terminal voltages. At high frequency they appear as a resistance in series with the gate capacitance and as a delay of the transconductance. Ch. 6 N-path filter
- N switched capacitors driven by non-overlapping LO phases; the low-pass response of the capacitors is translated to a band-pass response centred on the LO frequency. Ch. 10
O
- OFDM
- Orthogonal frequency-division multiplexing: data carried on many closely spaced subcarriers generated with an inverse FFT; used by Wi-Fi, LTE and 5G. Ch. 5 Operating power gain (GP)
- The power delivered to the load divided by the power that enters the two-port; independent of the source impedance. Ch. 3 Optimum load (Ropt), load line
- The load resistance that lets the collector swing between the knee voltage and its maximum while the current swings between zero and its maximum: Ropt = (VCC − Vknee)²/(2Pout). It is set by the voltage and current limits of the device, not by a conjugate match. Ch. 16 Optimum loop bandwidth
- The loop bandwidth that minimizes the integrated jitter, close to the offset where the in-band noise and the free-running VCO noise are equal. Ch. 12 Optimum source reflection coefficient (Γopt)
- The source reflection coefficient, relative to the reference impedance (50 Ω), that gives a two-port its minimum noise figure. The source impedance that achieves it is Zopt. Ch. 6 Oscillator figure of merit (FoM)
- FoM = −L(Δf) + 20 log(f₀/Δf) − 10 log(P/1 mW): the phase noise normalized to the carrier frequency, the offset and the power consumption, in dB; it compares oscillators of different frequencies and powers. Ch. 11 Oscillator tuning word (OTW)
- The digital code that sets a DCO’s frequency, the output of an ADPLL’s loop filter after normalization by the DCO gain. Ch. 15 Outphasing (Chireix, LINC)
- Splitting an envelope-varying signal into two constant-envelope signals of varying phase difference, amplifying each with an efficient saturated PA and summing the outputs; the envelope reappears in the sum. Ch. 16
P
- Pad capacitance
- The capacitance from a signal pad to ground (its shield or the substrate), parallel plate plus fringing; at millimetre waves it is part of the matching network. Ch. 18 Parametrized cell (pcell)
- A layout cell generated by a script from parameters, such as a transistor’s width and number of fingers or a capacitor’s size, supplied with the PDK so that devices match their models and rules. Ch. 18 Passive mixer
- A mixer whose switches (MOSFETs in triode) carry no DC current and only commutate the RF signal; current-mode when it drives the virtual ground of a TIA, voltage-mode when it drives a high impedance. Ch. 10 Passivity (of a model)
- The property that a model never generates power: all singular values of its S-matrix are at most one at every frequency; a non-passive macromodel can make a transient simulation oscillate. Ch. 18 Peak-to-average power ratio (PAPR)
- The ratio of the peak instantaneous power of a signal to its average, usually quoted at a probability (e.g. exceeded 0.01 % of the time); sets how far a linear PA must back off. Ch. 5 Periodic small-signal analysis
- The analysis of a small signal or of noise in a circuit whose operating point repeats every LO period; the circuit couples only the frequencies f + n f_LO, so it can be solved exactly with one matrix for all sidebands (the harmonic transfer or conversion matrix). Ch. 10 Periodic steady state (PSS)
- The waveform a periodically driven or oscillating circuit settles to once its start-up transient has died out; also the name of the analyses (shooting, harmonic balance) that find it directly. Ch. 2 Phase-frequency detector (PFD)
- Two flip-flops and a reset gate that compare the edges of the reference and the divided VCO output; the width of the UP or DN pulse is the time between the edges, and the circuit also detects a frequency difference. Ch. 12 Phase-locked loop (PLL)
- A feedback loop that makes an oscillator’s divided phase follow a reference: a phase detector compares the two, and a filtered version of the error tunes the oscillator. In lock the output frequency is exactly N times the reference. Ch. 12 Phase noise (\(\mathcal{L}(\Delta f)\))
- The single-sideband noise power of an oscillator in a 1 Hz bandwidth at an offset Δf from the carrier, relative to the carrier power, in dBc/Hz; for small phase deviations it is half the one-sided spectral density of the phase. Ch. 11 Polar transmitter
- A transmitter that sends the phase through a PLL or phase modulator and the envelope through the PA’s supply (envelope elimination and restoration). Ch. 5 Port (simulation)
- A source with a defined reference impedance (usually 50 ohm) that a simulator uses to excite a circuit and to compute its S-parameters; in ngspice a voltage source with portnum and z0, in Xyce a P device. Ch. 2 Power-added efficiency (PAE)
- The ratio of the RF power the amplifier adds to the DC power it draws, PAE = (Pout − Pin)/PDC. It approaches the collector efficiency when the power gain is high. Ch. 16 Power amplifier (PA)
- The last active stage of a transmitter: it delivers the required RF power to the antenna, and its efficiency usually dominates the power consumption of the transmitter. Ch. 16 Power combining
- Adding the outputs of several PA cells, with transformers, transmission lines or a distributed active transformer, to reach an output power that one cell cannot deliver within its breakdown and current limits. Ch. 16 Power-constrained noise optimization
- Choosing the width of an LNA’s input transistor for the lowest noise figure at a fixed bias current; for long-channel devices it gives an input-network quality factor of about 4.5 (Shaeffer and Lee). Ch. 9 Power waves
- The incident and reflected waves a = (V + Z0I)/(2√Z0) and b = (V − Z0I)/(2√Z0) at a port, normalized so that |a|²/2 and |b|²/2 are powers. Ch. 3 Prescaler
- The first, fastest divider of a synthesizer’s feedback chain, which brings the VCO frequency down to where slower, programmable logic can count. Ch. 13 Probe calibration
- The measurement of known standards (short, open, load, thru) on a calibration substrate that moves the reference plane of a vector network analyser to the probe tips. Ch. 18 Propagation constant (γ)
- γ = α + jβ = √((R′ + jωL′)(G′ + jωC′)): the attenuation α (Np per unit length) and phase constant β = 2π/λ of a wave on a line. Ch. 3 Proximity effect (current crowding)
- The redistribution of current in a conductor caused by the magnetic field of neighbouring conductors. In the inner turns of a spiral the field of the outer turns pushes the current to the inner edges, so the resistance rises faster with frequency than the skin effect alone predicts. Ch. 7 Pulse-swallow counter
- A programmable divider made of a dual-modulus prescaler P/(P + 1), a program counter NP and a swallow counter S: the prescaler divides by P + 1 for S cycles and by P for the rest, so N = P NP + S. Ch. 13 Push-push oscillator
- An oscillator whose output is taken at a common-mode node, where the two halves of a differential circuit cancel their fundamentals and add their second harmonics, giving twice the core’s frequency. Ch. 11
Q
- QPSK, QAM
- Quadrature phase-shift keying (four phases, 2 bits per symbol) and quadrature amplitude modulation (M points on a square grid, log2M bits per symbol). Ch. 5 Quadrature VCO (QVCO)
- Two LC oscillators coupled so that they lock 90° apart, giving I and Q outputs at the oscillation frequency without a divider; to do so each runs slightly off its tank’s resonance. Ch. 11 Quality factor (Q) of a passive component
- The ratio of the reactance of a component to its series resistance, |Im Z|/Re Z, or equivalently 2π times the peak energy it stores per cycle divided by the energy it loses per cycle. An inductor or capacitor with Q = 20 at 5 GHz loses 1/20 of its reactive power as heat. Ch. 7 Quarter-wave transformer
- A line a quarter wavelength long that transforms a resistance RL into Z1²/RL; with Z1 = √(Z0RL) it matches RL to Z0 at one frequency. Ch. 3
R
- Radio frequency (RF)
- Frequencies used for radio communication, from a few kilohertz to about 300 GHz; in circuit design, the frequencies at which the wavelength, parasitics and power transfer must be taken into account. Ch. 1 Raised-cosine pulse
- A Nyquist pulse, free of inter-symbol interference, whose spectrum rolls off smoothly over a fraction \(\alpha\) (the roll-off) beyond half the symbol rate; split between transmitter and receiver as two root-raised-cosine filters. Ch. 5 Reciprocal mixing
- A strong interferer mixing with the phase noise of the LO, which spreads it over the wanted channel: \(P_B + \mathcal{L}(\Delta f) + 10\log B\). Ch. 5 Reference plane
- The place at which a port’s voltage, current and waves are defined; calibration and de-embedding move it, for example from the cable to the probe tip and then to the device. Ch. 3 Reference spur
- A pair of sidebands at the reference frequency on either side of a PLL’s output, caused by a periodic ripple on the control voltage: charge-pump leakage, UP/DN mismatch, timing skew, or coupling of the reference. Ch. 12 Reflection coefficient (Γ)
- The ratio of the reflected to the incident voltage wave at a termination, (ZL − Z0)/(ZL + Z0); its magnitude is at most 1 for a passive load. Ch. 3 Regenerative (Miller) divider
- A divide-by-two made of a mixer and a low-pass or band-pass filter in a loop: the mixer multiplies the input by the output, and the loop sustains the f/2 product. It does not oscillate without an input. Ch. 13 Resistive shunt-feedback LNA
- A wideband LNA whose input match comes from a feedback resistor \(R_f\) around an inverting amplifier: \(R_{in} \approx (R_f + R_L)/(1 + g_m R_L)\). Ch. 9 Resistive termination
- Input matching of an amplifier by a shunt resistor equal to the source resistance; simple and broadband, but the resistor’s own noise alone sets the noise figure to at least 3 dB. Ch. 9 Retiming
- Re-sampling the output of a slow, noisy divider stage with a fast, clean clock (an earlier divider output or the VCO), so that the output edges inherit the jitter of the fast clock instead of that of the slow logic. Ch. 13 Return loss (RL)
- −20 log10|Γ|, in dB: how far below the incident power the reflected power is. A return loss of 10 dB (|S11| = −10 dB) reflects 10 % of the power. Ch. 3 RF design hexagon
- Razavi’s picture of the six quantities of an RF circuit that trade against each other: noise, linearity, gain, frequency, power and supply voltage. Ch. 1 rfmode (SG13G2)
- An instance parameter of the SG13G2 MOSFET models. rfmode = 1 selects the RF model card: gate resistance, a substrate resistance network, non-quasi-static channel charge and extra layout capacitances, needed above a few gigahertz. Ch. 6 Rollett stability factor (K)
- K = (1 − |S11|² − |S22|² + |Δ|²)/(2|S12S21|); with |Δ| < 1, K > 1 means unconditional stability. Ch. 3
S
- Sampled-loop stability limit (Gardner’s limit)
- The loop bandwidth, a sizeable fraction of the reference frequency, above which a charge-pump PLL becomes unstable because its phase detector acts only once per reference period. The continuous-time model cannot predict it. Ch. 12 Scattering parameters (S-parameters)
- The matrix that relates the power waves leaving the ports of a network to the waves arriving at them, with every port terminated in a reference impedance (usually 50 Ω). Ch. 3 Second-order intercept point (IIP2)
- The input power at which the extrapolated fundamental and second-order intermodulation (slope 2) lines meet; the key linearity number of direct-conversion receivers. Ch. 4 Self-interference
- Interference that a radio causes to itself: its PA or harmonics pulling the VCO, LO leakage self-mixing to DC, its reference and fractional spurs, or another radio in the same device transmitting next to the receiver’s band. Ch. 19 Self-oscillation frequency (of a divider)
- The frequency at which a CML divide-by-two oscillates with no input, as a two-stage ring oscillator. Driven near twice this frequency it divides with the smallest input amplitude. Ch. 13 Self-resonance frequency (SRF)
- The frequency at which the parasitic capacitance of an inductor (or the parasitic inductance of a capacitor) resonates with the component itself, so that its reactance passes through zero. Above it an inductor behaves as a capacitor and a capacitor as an inductor; a component is used well below its SRF. Ch. 7 Self-resonant frequency of a capacitor
- The frequency at which a capacitor’s series inductance cancels its capacitance; above it the capacitor behaves as an inductor. A capacitor near its self-resonance is a good RF short (decoupling) but a poor value-defined capacitor. Ch. 17 Sensitivity
- The smallest input power at which a receiver still meets its error-rate target: \(-174\ \mathrm{dBm/Hz} + 10\log_{10}B + \mathrm{NF} + \mathrm{SNR}_{min}\); Bluetooth Low Energy defines it at 0.1 % bit-error rate. Ch. 4 Shooting method
- A time-domain steady-state method that searches for the initial state from which one period of transient simulation returns to the same state. Ch. 2 Short-circuit current gain (h21)
- The ratio of the output to the input current of a two-port with its output short-circuited, y21/y11; for a transistor it falls at 20 dB/decade above a few hundred megahertz. Ch. 6 Shot noise
- The current noise of carriers crossing a potential barrier independently (a pn junction, the base-emitter junction of a bipolar transistor): \(\overline{i_n^2} = 2qI\,\Delta f\). Ch. 4 SiGe HBT
- A silicon-germanium heterojunction bipolar transistor: an npn transistor whose base contains germanium, which speeds up the electrons crossing the thin base and gives transit frequencies of hundreds of gigahertz. Ch. 1 Sign-sign LMS
- An adaptive algorithm that updates a coefficient by a fixed step in the direction given by the sign of the error times the sign of the reference signal; with a bang-bang phase detector it calibrates the gain of a DTC from one bit per reference cycle. Ch. 19 Single-stub match
- A transmission-line match: a series line rotates the load to the circle g = 1 of the admittance chart, and an open or short-circuited shunt stub cancels the remaining susceptance. Ch. 8 Skin depth (δ)
- The depth at which an alternating current in a conductor falls to 1/e of its surface value, δ = 1/√(πfμσ): 1.85 µm in SG13G2’s TopMetal2 at 2.44 GHz. When δ is smaller than the thickness of a trace, its resistance rises with frequency. Ch. 7 Smith chart
- The plane of the reflection coefficient Γ, overlaid with the circles of constant normalized resistance and reactance (or conductance and susceptance), on which impedances, lines and matching networks are drawn. Ch. 3 Spectral efficiency
- Bit rate per hertz of occupied bandwidth (b/s/Hz); \(\log_2M/(1+\alpha)\) for M-point modulation with raised-cosine pulses of roll-off \(\alpha\). Ch. 5 Spectral leakage
- The spreading of a sinusoid’s energy into neighbouring frequency bins when a discrete Fourier transform covers a non-integer number of its periods. Ch. 2 Spiral inductor
- A planar coil drawn in the thick top metals of a chip, square or octagonal, with N turns of width w and spacing s and an outer diameter of 100–300 µm; the inner end is brought out on a lower metal (an underpass) or, in a symmetric spiral, the turns cross over at the top and bottom. Ch. 7 Spurious-free dynamic range (SFDR)
- The range of input powers from the noise floor to the level at which third-order intermodulation products reach the noise floor: \(\tfrac{2}{3}(\mathrm{IIP3} - P_{floor})\). Ch. 4 SSB noise figure
- The noise figure of a mixer for a signal in one sideband, counting the source noise of both sidebands (and of the LO harmonics) at the output; 3 dB above the DSB figure of an ideal mixer. Ch. 10 Stability circle
- The circle on the Smith chart of the source (or load) terminations for which |Γout| = 1 (or |Γin| = 1); it separates the stable terminations from the unstable ones. Ch. 3 Stabilizing network
- Resistive elements added to a transistor stage, such as a resistor in parallel with a capacitor in series with the base, so that it is unconditionally stable (μ > 1) at every frequency where the device has gain, at the cost of some gain. Ch. 17 Stacked transistors (PA)
- Several transistors in series between the output and ground, each with its control terminal driven or AC-grounded so that the output voltage swing divides between them; it lets low-breakdown devices handle a larger supply and output power (the cascode is a stack of two). Ch. 16 Standing-wave oscillator
- An oscillator whose resonator is a transmission line carrying a standing wave, kept up by gain cells along the line; every point on the line oscillates in phase, with an amplitude that follows the wave. Ch. 11 Start-up margin
- The ratio of the small-signal negative conductance of an oscillator’s active circuit to the loss conductance of its tank, at the worst corner; designs aim for two to three. Ch. 11 Stub
- A short- or open-circuited length of transmission line used as a reactance, jZ0 tan βd or −jZ0 cot βd. Ch. 3 Substrate network
- The resistances of the silicon substrate under and around a transistor, which load its drain and source junction capacitances. It makes the output conductance rise with frequency and couples noise between devices. Ch. 6 Switched-capacitor bank
- A set of binary-weighted capacitors, each switched in or out of an oscillator’s tank by a MOS switch, that tunes the frequency in coarse steps (bands); a small varactor tunes finely within each band. Ch. 11 Switched capacitor (capacitor bank)
- A fixed capacitor connected or disconnected by a MOS switch, used in binary-weighted banks to tune an LC oscillator in coarse steps. Its on-state Q and its on/off capacitance ratio trade against each other through the RonCoff product of the switch. Ch. 7
T
- Tapped-capacitor match
- An impedance transformer of a series and a shunt capacitor that, with the load inductor, presents a higher resistance to the drain of an amplifier than the 50 Ω load. Ch. 9 TDC quantization noise
- The rounding error of a TDC, uniform over one resolution step when the measured time is busy; it gives an in-band phase-noise floor (2πΔtresfDCO)2/(12fref). When the measured time repeats, it appears as spurs instead. Ch. 15 Telegrapher’s equations
- The two first-order differential equations that relate the voltage and current along a transmission line through its series impedance R′ + jωL′ and shunt admittance G′ + jωC′ per unit length. Ch. 3 Temperature-compensated crystal oscillator (TCXO)
- A crystal oscillator whose frequency drift with temperature is corrected by a compensation network; the usual reference of a radio synthesizer, specified by its stability (ppm) and its phase noise. Ch. 12 Thermal noise
- The random voltage across any resistance at a temperature above absolute zero, caused by the thermal motion of its carriers: \(\overline{v_n^2} = 4kTR\,\Delta f\), white up to far beyond radio frequencies. Ch. 4 Third-order intercept point (IIP3, OIP3)
- The input (IIP3) or output (OIP3) power at which the extrapolated fundamental (slope 1) and third-order intermodulation (slope 3) lines of a two-tone test meet. Ch. 4 Third-order loop filter
- A charge-pump loop filter with an extra RC section (R3, C3) after the usual R1–C1 branch and shunt C2. The extra pole attenuates the ripple at the reference frequency and the high-frequency noise of a fractional-N divider, at the cost of some phase margin and the noise of R3. Ch. 12 Time-domain reflectometry (TDR)
- Launching a fast step into a line and watching the reflected voltage at its input; the delay locates a discontinuity and the waveform identifies it (resistive, capacitive or inductive). Ch. 3 Time-to-digital converter (TDC)
- A circuit that measures the time between two edges as a number. In an ADPLL it measures the fractional DCO phase at each reference edge; its resolution sets the loop’s in-band phase-noise floor. Ch. 15 Touchstone file
- The standard text format for S-, Y- or Z-parameters of an n-port against frequency (.s1p, .s2p, …), with an option line such as ‘# Hz S RI R 50’. Ch. 2 Transceiver
- A radio that both transmits and receives, usually sharing the antenna and the frequency synthesizer between its receiver and transmitter. Ch. 1 Transducer power gain (GT)
- The power delivered to the load divided by the power available from the source; the gain an amplifier actually provides between given terminations. Ch. 3 Transient noise
- Noise represented by random sources inside a time-domain simulation; in ngspice-47 only independent sources with trnoise or trrandom produce it, not resistors or transistor models. Ch. 2 Transit frequency (fT)
- The frequency at which the short-circuit current gain h21 of a transistor falls to one, about gm/2π(Cgs + Cgd) for a MOSFET. It is usually extrapolated at −20 dB/decade from a frequency where h21 is still large. Ch. 6 Transit frequency (\(f_T\))
- The frequency at which the short-circuit current gain \(\vert h_{21}\vert\) of a transistor falls to one. Ch. 1 Transmission line
- A pair of conductors whose length is a sizeable fraction of a wavelength (more than about λ/10), so that voltage and current vary along it and travel as waves; modelled by distributed R′, L′, G′ and C′ per unit length. Ch. 3 T/R switch
- A switch that connects the antenna to the receiver or to the transmitter, for radios that do not transmit and receive at the same time. Ch. 1 True single-phase-clock (TSPC) logic
- Dynamic CMOS logic, due to Yuan and Svensson, in which every clocked transistor is driven by the same clock phase, so no complementary clock is needed. A TSPC flip-flop has nine transistors; it is fast and draws power only when it switches, but it loses its state at very low clock frequencies. Ch. 13 Tuning gain (\(K_{VCO}\))
- The slope of a voltage-controlled oscillator’s frequency against its control voltage, in Hz/V (or rad/s/V). It sets the loop gain of the PLL and how much noise on the control line becomes phase noise. Ch. 11 Tuning ratio (of a varactor)
- The ratio Cmax/Cmin of the largest to the smallest capacitance over the available control voltage range. Together with the fixed capacitance of the tank it sets the tuning range of an LC oscillator. Ch. 7 Two-point modulation
- Modulating a PLL at the divider and directly at the VCO at once, so that the data reaches the output with a flat response wider than the loop bandwidth. Ch. 5 Two-point modulation
- Frequency modulation of a PLL applied at two places at once: the divider ratio (low-pass through the loop) and the VCO’s tuning input (high-pass through the loop). With matched gains the two responses add to a flat one, wider than the loop bandwidth. Ch. 14 Two-tone test
- A linearity measurement with two equal tones closely spaced in frequency (here 2.440 and 2.441 GHz), from which IM3, IM2 and the intercept points are read. Ch. 4
U
- Unconditional stability
- A two-port is unconditionally stable if no passive source or load makes the magnitude of its input or output reflection coefficient reach 1; otherwise it is potentially unstable. Ch. 3 Unilateral approximation
- Neglecting S12 so that the input and output can be matched separately; the unilateral figure of merit u says whether the error is small. Ch. 3 Unilateral power gain (U)
- Mason’s invariant gain of a two-port made unilateral with lossless feedback; it is the same for every lossless reciprocal embedding, and the frequency where U = 1 is fmax. Ch. 3 Unit interval (UI)
- One period of the clock or bit being discussed, used as the unit of phase or timing error: in an ADPLL one DCO period, in a CDR one bit period. Ch. 15
V
- Varactor
- A capacitor whose value is tuned by a DC voltage. In CMOS and BiCMOS processes it is usually an accumulation-mode MOS capacitor (an n+ gate over an n-well), such as SG13G2’s sg13_hv_svaricap, whose capacitance falls from the oxide capacitance in accumulation to a smaller value in depletion. Ch. 7 Variable-slope DTC
- A digital-to-time converter in which a gate charges a digitally switched capacitor bank and the next gate fires when the ramp crosses its threshold. It is simple, but its delay is not linear in the code, because the slope of the ramp, and with it the delay of the next gate, changes with the load. Ch. 14 Vector fitting
- A method that approximates sampled frequency responses, such as EM-simulated S-parameters, by a rational function with stable poles, from which an equivalent SPICE circuit can be built. Ch. 17 Vector network analyser (VNA)
- The instrument that measures S-parameters: it sends a wave into each port in turn and measures the amplitude and phase of the waves that come back from every port. Ch. 3 Vernier TDC
- A TDC with two delay lines of slightly different stage delays; the resolution is the difference of the two delays, finer than one gate delay. Ch. 15 Virtual resistance (Rv)
- The intermediate resistance at the junction of the two L-sections of a π or T network; it sets the network’s Q. Ch. 8 Voltage-controlled delay line (VCDL)
- A chain of delay stages whose delay is set by a control voltage or current; the controlled element of a DLL. Ch. 15 Voltage standing-wave ratio (VSWR)
- The ratio of the maximum to the minimum voltage amplitude along a mismatched line, (1 + |Γ|)/(1 − |Γ|). Ch. 3
W
- Weaver image-reject receiver
- An image-reject receiver that replaces Hartley’s 90° shift with a second quadrature mixing step. Ch. 5 Window function
- A weighting (Hann, Blackman and others) applied to a record before a discrete Fourier transform to reduce spectral leakage, at the cost of a wider main lobe and a loss of amplitude accuracy when a tone falls between bins. Ch. 2
Z
- Zero-IF (direct-conversion) receiver
- A receiver whose LO sits at the carrier, converting the channel directly to baseband with quadrature mixers; sensitive to DC offsets, flicker noise and second-order distortion. Ch. 5 Zero-voltage switching (ZVS)
- Closing a switch when the voltage across it is zero, so that no charge stored in its capacitance is dissipated; in class E it is accompanied by zero voltage slope (ZdVS) for robustness. Ch. 16