Microelectronic Circuits Sedra and Smith, 6th Edition Reading Notes (GMU ECE 333) David Stein, Summer 2015 1 Chapter 1: Signals and Amplifiers Microelectronics generally refers to the field of integrated circuits, up to and including microprocessors - the devices reviewed in this text may be utilized individually as discrete circuits, or as components of an integrated circuit 1.1: Signals A signal is a carrier of information that can be processed in various ways - a real-world measurement is first converted by a transducer to an electrical signal - the signal can be equivalently represented in two ways: Thévenin form (a voltage source vs (t) having a source resistance Rs (t) in series), or Norton form (a current source is (t) having a source resistance Rs (t) in parallel) - these forms are interchangeable (since vs (t) = Rs is (t), but Thévenin is preferred when the source resistance is low; Norton is preferred when the source resistance is high (see Exercise 1.1: when oriented in series, most of the source voltage is delivered to the load when the load resistance is much larger than the source resistance; when oriented in parallel, most of the source voltage is delivered to the load when the load resistance is much smaller than the source resistance) Information is contained in a signal as a time-varying magnitude of the current or voltage - the analysis of the signal is easiest to conceptualize if it can be represented mathematically, which can be difficult for arbitrary signals 1.2: Frequency Spectrum of Signals A signal can be represented according to its frequency spectrum, as a set of individual sinusoids of the form va (t) = Va sin(ωt + φ), where Va is the peak amplitude, ω is the angular frequency, and φ is the phase shift - sinusoid amplitude is commonly discussed as its average, represented by the root-mean-square, √ - the time-domain and rather than its peak value; the RMS value is simply peak 2 frequency-domain representations of a signal are equivalent and interchangeable The frequency spectrum is determined as a Fourier series (for periodic signals) and Fourier transforms (for both periodic and aperiodic signals, including arbitrary signals) - the Fourier series produces a set of sinusoids with harmonically related frequencies, i.e., multiples of a fundamental frequency - the Fourier transform is a continuous function over all frequencies, but the most significant components of the signal are typically in a select range; e.g., an audio signal encodes information for all frequencies, but only those within the range audible to humans (20 Hz - 20 kHz) are significant; similarly, an analog video signal 2 only needs to be sampled between 0 and 4.5 MHz 1.3: Analog and Digital Signals Most physical signals are analog, featuring a magnitude that can take on any value within a certain range, and the amplitude is a continuous function - other signals are digital, where the amplitude at specific times is within a defined set of discrete values, such that the signal can be represented as a sequence of rational numbers - an analog signal can first be sampled, by measuring it at regular time intervals, and then quantized or digitized, by mapping each measurement to a discrete value (usually through rounding) While many number systems may be used to represent digital measurements, binary number systems are the simplest and most common representation - the quantization can either use only two values (low and high, represented as binary 0 and 1, or as a low voltage such as 0 V and a high voltage such as +5 V), or a set of values that are represented each represented by a sequence of N binary values from b0 , b1 , . . . , bN −1 , where the recorded value is represented as the sum D = b0 20 + b1 21 + . . . + bN −1 2N −1 - the mapping or rounding that is inherent in sampling creates a quantization error, which can be reduced by increasing the resolution of the quantization (i.e., increasing N ) - quantization can be achieved by an analog-to-digital (A/D) converter, which inputs an analog input signal and outputs a series of binary values that together represent the digital output - the reverse process can be achieved through a digital-to-analog (D/A) converter, which converts a digital input signal into an analog output signal (though lost resolution cannot be recaptured) 1.4: Amplifiers Signal amplification: Transducers typically sample a signal with too little amplitude for effective processing, so the signal needs to be amplified to a larger amplitude - linear amplifiers scale the input value as a linear proportion, in order to avoid distortion that adds new information - thus, ideal amplifiers exhibit the relationship vo (t) = Avi (t), where A is the amplifier gain Voltage amplifiers operate by increasing the voltage of a signal, as above - power amplifiers operate by increasing the current with little or no voltage gain - an audio system may feature both a voltage amplifier (as a preamplifier), and a power amplifier (which feeds the voltage to the speakers with a higher current, thus amplifying the volume of both soft and loud music passages) The voltage gain of a linear voltage amplifier is defined as: 3 Av = vvoi The transfer characteristic of the linear amplifier is shown as a plot of vi against vo , and is a straight line with slope A for a linear voltage amplifier - similarly, a current amplifier has a current gain Ai = iioi - both types of amplifiers have a power gain Ap = Av Ai = PPLI = vvoi iiio Gain is often expressed in decibels on a logarithmic scale, such as voltage gain = 20 log(|Av |) dB; an increase of 6 dB indicates a doubling of the voltage, and an increase of 20 dB indicates a tenfold increase in voltage - similarly, current gain = 20 log(|Ai |) dB - however, power gain is expressed as 10 log(|Ap |) dB; on this scale, power doubles with an increase of 3 dB, and by a factor of 100 with a 20 dB increase Since amplifiers add power, they require connection to a power source - an example amplifier is connected to a positive power source VCC , which delivers a current ICC , and a negative power source VEE , which delivers a (negative) current IEE - both of the opposite terminals of the voltage sources (the negative terminal of VCC , and the positive terminal of VEE ) are connected to ground, along with a ground terminal of the amplifier - together, these power sources deliver DC power: Pdc = VCC ICC + VEE IEE The amplifier exhibits a power-balance equation: Pdc + PI = PL + Pdissipated A power efficiency for the amplifier can also be calculated as: L · 100 ν = PPdc Amplifier saturation: Amplifiers only generate a linear signal over a limited range of input voltages, where the power delivered by the power supplies can linearly scale to produce the proportional output - if the voltage is too high, amplifier saturation occurs, where the voltage increase is limited by a ceiling and distortion of the signal occurs Symbol conventions: iC (t): A time-varying current signal; i.e., a total instantaneous current IC : The time-invariant DC component of iC (t) ic (t): The time-varying alternating component of iC (t) Ic : The peak amplitude of ic (t) Thus, iC (t) = IC + ic (t), and ic (t) = Ic sin(ωt) 4 1.5: Circuit Models for Amplifiers Amplifier types: Voltage amplifier: Avo = vo vi Current amplifier: Ais = io is Transconductance amplifier (translates an input voltage to an output current): Gm = vioi Transresistance amplifier (translates an input current to an output voltage): Rm = viio 5 While each amplifier type may more natively fit a particular scenario, all four types are equivalent Voltage amplifier: According to this model, the amplifier is a voltage-controlled voltage source with an open-circuit voltage gain of Avo , an input resistance Ri , L , and the and an output resistance Ro - the transfer function is v0 = Avo vi RLR+R o RL voltage gain is Av = Avo RL +Ro - in an ideal amplifier, Ro = 0, and Av = Avo ; thus, Avo is the open-circuit voltage gain, when 100% of the resistance is at the load - similarly, the source resistance diminishes the amount of voltage measured i over the input: vi = vs RiR+R , so it is ideal to maximize the value of Ri - for s any combination of parameters, the actual gain is: RL i Av = vvos = Avo RiR+R s RL +Ro In some cases, the voltage is acceptable, but the source resistance is high, and raisin the input resistance to compensate would cause signal attenuation - a buffer amplifier can be used to drive the voltage with greater current Cascaded amplifiers: Sometimes multiple stages are needed to achieve a desired amplification - e.g., in the following scenario, the stage-1 amplifier reduces the input voltage from 100 k Ω to 1 k Ω; the stage-2 amplifier has a high voltage gain; and the stage-3 amplifier acts as a buffer amplifier to drive the amplified voltage with low output resistance 1.6: Frequency Response of Amplifiers Amplifier frequency response can be measured by providing a sinusoidal wave of a known amplitude and frequency, and measuring the amplitude and phase shift of the output (a linear amplifier will not change the frequency) - the transfer function, T (ω), has both a gain (|T (ω)| = VVoi ) and a phase shift (∠T (ω) = φ) the complete frequency response can be evaluated as a Bode plot, which reveals both the gain (amplitude response) and the phase shift (phase response) at each input frequency 6 The frequency response of an amplifier can be determined from the circuit model, but where capacitors and inductors are replaced by their reactances 1 (for capacitors, reactance = jωC ; for inductors, reactance = jωL ) - the trans(s) , and can be fer function can be represented as a complex function T (s) = VVoi (s) evaluated as T (jω) to determine the response to various physical frequencies Single-time-constant (STC) networks are circuits comprising only one reactive element and one resistance, which can be modeled according to one time conL stant τ = RC = R - different types of STC networks have a characteristic response to various inputs, such as sinusoidal, step, and pulse - most networks exhibit either low-pass-filter or high-pass-filter characteristics - tuned amplifiers feature both a low-pass filter and a high-pass filter, and therefore pass frequencies only within a desired range The components of the circuit also determine the shape of the responses at gain falloff points: internal capacitance can cause a loss of gain at high frequencies, while the inclusion of a coupling capacitor between amplifier stages can cause a loss of gain at low frequencies - integrated circuit amplifiers typically do not feature coupling capacitors, but are directly-coupled DC low-pass amplifiers 7 Chapter 3: Semiconductors Solid-state devices such as diodes and transistors are made from semiconducting materials- this chapter focuses on the physical properties and behavior of semiconductor materials Most semiconductors are made from silicon, through a specific fabrication process - an integrated circuit containing a vast number of transistors can be carved from a single silicon crystal, leading to a monolithic circuit - the conductivity of silicon is altered to create these devices by the controlled introduction of impurity atoms, in a process called doping 3.1: Intrinsic Semiconductors Semiconducting materials are defined as any material exhibiting conductivity between conductors (e.g., copper) and insulators (e.g., glass) - semiconducting materials may be made of one element, such as germanium or silicon, or compounds, typically (group III or V) + (group II or VI), such as gallium-arsenide Silicon is the most widely used semiconducting material in use today - silicon has four valence electrons, and can form four covalent bonds with neighboring atoms to form a regular lattice structure - at low temperatures, all covalent bonds are intact and all electrons are non-mobile, so silicon is an insulator; however, at room temperature, an electron can break free and drift apart and move according to an electric field (“thermal generation”) - the deficit of an electron from its original atom creates a positively charged “hole” that attracts an electron from a neighboring atom, so the “hole” moves in the opposite direction of the electric field - the number of free electrons and holes scales proportionally with temperature - the conductivity of a material is measured as a number of charge carriers per unit volume (cm3 ) - free electrons and holes also settle in other atoms and therefore stop acting as charge carriers (“recombination”) This process can be quantified: temperature affects the thermal generation rate, which in turn determines the number of charge carriers, which in turn determines the recombination rate: ni = n = p - also, p · n = (ni )2 , where: n = the current / average / typical number of free electrons per unit volume p = the current / average / typical number of holes per unit volume ni = the number of charge carriers of either type (holes or electrons) per unit volume For any material: 8 3 Eg ni = BT 2 e 2kT , where: B = a material-dependent parameter for the material Eg = bandgap energy k = Boltzmann’s constant (8.62 · 10−5 eV K ) For silicon: 3 B = 7.3 · 15015 k − 2 Eg = 1.12eV At room temperature, ni = 1.5 · 1010 chargecmcarriers 3 3.2: Doped Semiconductors The value of ni for silicon at room temperature is too small to be useful, and also fluctuates too much with temperature - carrier concentration can be improved by injecting a dopant, which introduces an impurity that substantially increases either holes or free electrons n-type doped silicon can be created with a dopant having five valence electrons, such as phosphorus; the extra electron becomes a donor, without creating a hole: the positive charge remains with the dopant (“bound charge”) - the degree of doping determines (ND ) determines the concentration of donor atoms, which is usually much greater than the number of intrinsic free electrons ni ; thus, the number of free electrons in n-type doped silicon nn ≈ ND ; the number of holes, pn , is initially unchanged, but diminishes due to recombination with free electrons - accordingly, in n-type material, electrons are the “majority charge carriers” and holes are the “minority charge carriers”) - when the rates of thermal generation and recombination are equal (“thermal equilibrium”), n2i pn nn = n2i , where pn ≈ ND p-type doped silicon can be created with a dopant having three valence electrons, such as boron - as a consequence, boron accepts an extra electron from a neighboring silicon atom, creating a bound negative charge at the boron atom (“acceptor” atom) and a hole in the neighboring atom that can act as a charge carrier - the degree of doping with acceptor atoms (NA ) creates a number of holes (pp = NA ) that is much greater than the number of intrinsic charge charge carriers - as above, at thermal equilibrium, pp np = n2i , where np ≈ n2i NA Note that both n-type and p-type silicon are electrically neutral; the type of material just determines whether an electric current results in a propagation of electrons or a propagation of holes 9 3.3: Current Flow in Semiconductors Drift current: When an electric field is applied to a semiconductor crystal, holes move in the direction of the field, and free electrons move in the opposite direction The velocity of holes is: vp−drif t = µp E cm s , where: V E = strength of electric field, measured in cm 2 cm2 µp = hole mobility, measured in cm V·s - for intrinsic silicon, µp = 480 V·s , Similarly, the velocity of free electrons is: vn−drif t = −µn E cm s , where: E = strength of electric field 2 µn = electron mobility - for intrinsic silicon, µn = 1, 350 cm V·s The hole drift current density (i.e., current of holes per cross-sectional unit area) is: I Jp = Ap = qpvp−drif t For a particular section of material, the hole drift current is: Ip = Aqpvp−drif t = Aqpµp E = Jp A, where: A = cross-sectional area q = charge magnitude (for a single electron, q = 1.6 · 10−19 coulombs) p = concentration of holes Similarly, the electron drift current density (i.e., current of electrons per crosssectional unit area) is: Jn = − IAn = −qpvn−drif t And the electron drift current is: In = −Aqnvn−drif t = −Aqnµn E = Jn A The total drift current density is: J = Jp + Jn = q(pµp + nµn )E, or J = σE, where: σ = the conductivity of the material measured by σ = q(pµp + nµn ) Alternatively, total drift current can be expressed in terms of resistivity, where: J = Eρ Conversely, resistivity can be defined as: ρ= E J , where ρ = resistivity measured in terms of Ω · cm Diffusion current: When the concentration of charge carriers is higher in one part of a material than another, the charge carriers tend to diffuse, thus creating 10 a diffusion current from the high-concentration area to the low concentration area For an area of material with an abundance of holes: Jp = −qDp dp(x) dx , where Dp = the diffusivity or diffusion constant of holes, i.e., the rate at which holes can diffuse through the material dpx dx = rate of change of concentration of holes (note: this is negative, since diffusion causes concentration to fall) Similarly, for an area of material with an abundance of free electrons: Jn = qDn dn(x) dx , where Dn is the diffusivity of electrons dpx dx = rate of change of concentration of free electrons Thermal voltage: For any material: Dp n VT = D µn = µp , or VT = kT q This is known as the thermal voltage; at room temperature, T ≈ 300 and VT = 25.9 mV 3.4: The pn Junction with Open-Circuit Terminals A pn junction can be created by creating a region of p-type silicon adjacent to a region of n-type silicon, with metal contacts at each end (the metal contact at the p-type silicon end forms the anode, and the metal contact at the n-type silicon end forms the cathode): In an open circuit (when the terminals are disconnected), an interaction occurs at the interface - the edge of the p-type material, featuring an excess of holes, binds extra electrons from the edge of the n-type material - both charges become bound, creating a “depletion region” - the depletion region exhibits a diffusion current ID from the p-type region to the n-type region, which is the sum of the hole diffusion current from the p-type region to the n-type region, and the free electron diffusion current from the n-type region to the p-type region - the 11 depletion of free electrons in the n-type region causes some of the bound positive charges to remain “uncovered,” and the depletion of holes in the p-type region creates uncovered bound negative charges The creation of uncovered bound charges in the depletion region creates a voltage difference across the border, and a static electric field from the n-type material to the p-type material - this electric field serves creates a “barrier voltage” V0 , which is measurable over the depletion region (from the n-type region to the p-type region) - this voltage prevents the further migration of holes and electrons, and reduces the diffusion current Drift current: In addition to blocking diffusion of majority carriers, the electric field promotes the transport of minority carriers - e.g., a thermally generated hole that appears in the n-type region may randomly migrate toward the edge of the barrier region, and the electric field sweeps it over the border into the ptype region - this transport of minority charge carriers creates a drift current IS in the opposite direction of the diffusion current (i.e., from the n-type material to the p-type material) - the magnitude of this current is dependent upon the availability of charge carriers (and hence temperature), but independent of the actual voltage Equilibrium: In an open circuit, when ID > IS , the depletion region widens until the diffusion current falls; similarly, when ID < IS , the depletion region narrows until diffusion rises - thus, an equilibrium is established where ID = IS , and is maintained by the barrier voltage V0 - with no external voltage applied, equilibrium is established with a “junction built-in voltage” across the pn junction: D V0 = VT ln( NAn·N ) 2 i For silicon at room temperature, 0.6V < V0 < 0.9V This barrier voltage is strictly internal to the pn material, and is not observable between the cathode and anode, because contact voltages are created that balance the junction built-in voltage 12 Depletion region width: When equilibrium is reached, the width of each half of the depletion region is proportional to the concentration of majority charge carriers on the opposite side - the ratio of the widths is defined as: NA xn xp = ND , where xn = the width of the n-type portion of the junction (adjacent to the n-type region but positively charged) xp = the width of the p-type portion of the junction (adjacent to the o-type region but negatively charged) The total width q of the depletion region is: W = xn + xp = 2qs ( N1A + N1D )V0 , where F s = electrical permittivity of material (for silicon, s = 1.04 · 10−12 cm ) The width of each side can be calculated as: A xn = W NAN+N D D xp = W NAN+N D The charge stored in each half of the depletion region is defined as: |Q+ | = qAxn ND , and |Q− | = qAxp NA , where A = the cross-sectional area of the material Note that the charges are balanced: |Q+ | = |Q− | 3.5: The pn Junction with an Applied Voltage When a DC voltage is applied across the terminals of a pn junction, with the higher potential at the p terminal, the device is forward-biased; when the voltage is higher at the n terminal, the device is reverse-based In reverse bias, when a voltage VR is applied, the barrier voltage increases to V0 + VR , which significantly reduces ID - however, IS is not affected by the barrier voltage, as it is rate-limited by thermal generation, so a small reverse current q appears - the width of the depletion region also expands: W = 2s ( N1A + N1D )(V0 + VR ) The amount q of charge stored in the junction is: ND QJ = A 2s q( NNAA+N )(V0 + VR ) D In forward bias, the voltage VF reduces the build-in voltage V0 , which dramatically raises the diffusion current ID , and diminishes the width of the depletion region - the total observed current is I = ID − IS Carrier injection: When forward-biased, the pn junction allows many more of 13 the majority carriers to cross the barrier into the opposite region - at the border of the depletion region in the n-type region, an excess number of holes are deposited, in the amount of: V pn (xn ) = pn0 e VT , where xn refers to the boundary between the depletion region and the n-type region 2 pn0 = the equilibrium concentration of n nin This amount is the sum of the equilibrium concentration, pn0 , and the excess holes that are deposited by the voltage; the excess is measured as: V Excess = pn0 e VT − pn0 The concentration continues deeper into the diffusion region, and diminishes at an exponential rate: −x−xn V pn (x) = pn0 + pn0 e VT e Lp , where x − xn = the distance from the depletion region border Lp = the diffusion length of the material (smaller value = faster recombination of excess charge carriers with majority carriers in the region = more rapid dropoff in concentration) The diffusion current density under an applied voltage is: −x−xn V D Jp (x) = q Lpp pn0 (e VT − 1)e LP The recombination of holes with electrons in the n-type region is strongest at the border of the depletion region, and diminishes further from the border - the applied voltage causes a current that injects electrons into the n-type region to replenish those that participate in recombination - the total current in the n-type material is therefore the same as the current at the border: V D Jp (xn ) = q Lpp pn0 (e VT − 1) The same process occurs with the diffusion of electrons into the p-type region, 14 and recombination with holes - the current density created by this process is: V n VT Jn (−xp ) = q D − 1) Ln np0 (e The total current, as the sum of both currents, is determined as: V D n )(e VT − 1) I = Aqn2i ( Lp NpD + LD n NA In reverse-bias, when the voltage is negative and significantly greater than the thermal voltage, the exponential term is essentially zero, so: V I = Is (e VT − 1) ≈ −Is , where D n Is = Aqn2i ( Lp NpD + LD ) n NA However, at a large negative voltage, breakdown occurs at a negative voltage V = −VZ Zener breakdown: When VZ < 5V, breakdown occurs due to the zener effect, where the voltage is strong enough to break some covalent bonds, thus generating a significant number of mobile charge carriers - only a small change in voltage in the neighborhood of −VZ is sufficient to create a large current Avalanche breakdown: When VZ > 7V, breakdown occurs due to the avalanche effect - in this model, the minority carriers crossing the depletion zone have enough kinetic energy to break covalent bonds - the liberated charge carriers also have a large kinetic energy, and thereby break covalent bonds between other toms - the resulting “avalanche” of freed charge carrier creates the breakdown current through the junction 3.6: Capacitative Effects in the pn Junction The pn junction stores charge in two areas: the depletion region (predominant during reverse bias), and the n-type and p-type regions (predominant during forward bias) - different mechanisms are involved in each area Junction√capacitance: The reverse-biased junction stores charge according to: QJ = α q V0 + VR , where ND α = A 2s q NNAA+N D This can also be expressed as: dQ Cj = dVRj |VR =VQ = 2√Vα+V 0 r At zero reverse-bias, the capacitance of the junction is Cj0 = α √ 2 Vo The effects described above result in a pn junction that changes abruptly at the junction boundary - other devices have a graded junction with a gradient from n-type material to p-type material - the junction capacitance of these pn junctions changes as a function of the grading coefficient 15 Diffusion capacitance: In the forward-biased junction, carrier injection causes recombination with the majority carriers in each region, causing an excess of minority carriers - this excess manifests as a stored charge in each region - for the p-type region, the stored charge is determined as: Qp = Aq(pn (xn ) − pn0 )Lp = τp Ip , where τp = L2p Dp = hole lifetime in the n-type region Similarly, in the n-type region, the stored charge is: Qn = τn In The time constants τp , τn represent the average time for an injected charge carrier to recombine with a majority charge carrier in the region - the charge carriers lost to recombination at this rate are replenished by the current Ip , In - the total stored charge during forward bias is: Q = τp Ip + τn In this can be expressed as a function of the total current: Q = τT I, where τT = the mean transit time of the junction - the capacitative effect is therefore determined as: Cd = ( VτTT )I 16 Chapter 4: Diodes Some circuit devices exhibit nonlinear characteristics, but can be used in linear electronics - diodes are two-terminal devices that exhibit nonlinear i − v characteristics - diodes are useful for many purposes, such as rectification and amplification - several models can be used to represent the diode at varying levels of accuracy and complexity 4.1: The Ideal Diode Diodes typically have a positively charged terminal (an anode) and a negatively charged terminal (a cathode) - an ideal diode simply has two modes: when a positive voltage is applied over the anode and cathode (“forward-biased”), the diode behaves as a short circuit; no voltage is absorbed, and current passes through freely (the diode is “on”) - when a negative voltage is applied over the anode and cathode (“reverse-biased”), the diode absorbs all voltage, and passes no current (the diode is “off”) - the i − v characteristic of this device is piecewise linear: a horizontal zero-current section at v ≤ 0, and a vertical section at v = 0+ Rectification: Ideal diodes can be used as a rectifier circuit, which converts a bidirectional alternating signal into a unidirectional signal - the input signal vi has an average value of zero, and the output vo has a positive average value (which can be considered a “DC component” over which the output signal alternates) Diode logic gates: A combination of diodes can function as a logic gate when provided with discrete input voltages (such as 0 volts representing a “low” value and +5 volts representing a “high” value) - a logic OR and AND gates can be formed this way: 17 4.2: Terminal Characteristics of Junction Diodes Simple diodes can be formed from a pn junction, which conducts substantial current in the forward direction and almost no current in reverse - the typical pn junction exhibits a positive vertical current at a small positive voltage −VZK ; a negative vertical current at a certain large negative voltage; and a zero current at intermediate voltages (the “breakdown” region): In a more accurate representation, the zero-voltage reverse-biased region exhibits a small leakage current, and the transition points represent nonlinear, exponential growth over a small voltage range: 18 The forward-bias region exhibits current according to the equation: v i = IS (e VT − 1) IS = saturation current (specific to a particular diode, and temperature-sensitive) v = input voltage VT = thermal voltage = kT q −23 J k = Boltzmann’s constant = 8.62 · 10−5 eV K = 1.38 · 10 K T = temperature in kelvin q = electron charge = 1.6 · 10−19 C Alternatively, VT = 0.0862T mV; at room temperature, VT = 25.3 mV Saturation current scales proportionally with the cross-sectional area of the diode: if the diode cross-section doubles, IS doubles - saturation current is also very temperature-sensitive: at T + = 5◦ , IS doubles - however, IS is typically in the range of 10v−12 for silicon diodes, so for any significant amount of forward current, i ≈ IS e VT Diode voltage can also be expressed as v = VT ln IiS - a ratio of voltages and currents can be expressed as V2 − V1 = VT ln II21 In practice, forward current is typically negligible for any positive voltage up to about 0.5 V (the “cut-in voltage”), and then turns almost vertically positive therefore, when a pn junction diode is “on” with any forward voltage exceeding the cut-in voltage, the diode absorbs a constant voltage (about 0.5 V) and otherwise operates as a short - diodes typically have a current rating indicating a range of forward currents over which the diode functions as expected 19 Since IS is temperature-sensitive, the voltage drop is also temperature-sensitive: Reverse-biased region: The current in this region is largely determined by the saturation current: i ≈ IS - actual diodes typically exhibit a somewhat larger reverse current (e.g., 1 nA) due to leakage effects, which are also temperaturedependent and tend to double for T + = 10◦ Breakdown region: When the reverse voltage exceeds a threshold value ( VZK ), voltage increases steeply - the diode is not destroyed in this process, but the abrupt change in the neighborhood of VZK may cause current to change rapidly over a small fluctuation in voltage 4.3: Modeling the Diode Forward Characteristics Exponential model: The most accurate description of the forward characteristic VD is the exponential model presented above, where ID = IS e VT when VDD ≥ 0.5 V - when the diode is included in a circuit featuring a resistor, the current can also be modeled according to Kirchhoff’s current law: ID = VDDR−VD Graphical analysis: The Kirchhoff’s current law equation provides a line in the v − i plot with slope − R1 (the “load line”) - the exponential model provides a non-linear equation for the current through the diode - the intersection of these equations provides an intersection point (the “operating point”) indicating the actual voltage and current of the diode 20 Iterative analysis: The equations above can be applied iteratively until the values converge Example: For a diode that exhibits 1 mA at a voltage of 0.7 V, find the current ID and voltage VD when positioned in a circuit with VDD = 5 V and R = 1 k Ω. Answer: Start with the load line equation, ID = VDDR−VD : ID = 5−0.7 1000 = 4.3mA Next apply the exponential equation, V2 − V1 = VT ln II12 = 25mV ln II12 : V2 = 0.7 + (25mV · ln( 4.3mA 1mA )) = 0.738 V Apply the load line equation again: ID = 5−0.738 1000 = 4.262mA Apply the exponential equation again: V2 = 0.738 + (25mV · ln( 4.262mA 4.3mA )) = 0.738 V This yields convergence at ID = 4.262mA, V2 = 0.738V, which is our answer Constant-voltage-drop model: A less accurate but easier estimate of diode behavior can just presume that the diode imposes a constant voltage drop on a forward voltage: 21 Ideal-diode model: Completely neglecting the voltage drop enables very quick analysis, but exhibits significant inaccuracy; the constant-voltage-drop model is almost as conceptually easy and provides a much better approximation Small-signal model: In some cases, a small alternating signal vd (t) is imposed on a large DC voltage VD - we can find the slope of the exponential curve at the quiescent point VD , and then presume that over this small range, curve is relatively linear - the linearity is estimated according to the incremental resistance rd = VIDT = VTVD - this small-signal approximation works when |vd (t)| ≤ 0.5mV IS e V T Voltage regulation: Diodes are used to create a regulated voltage, where a circuit provides a constant DC voltage in spite of changes in the load and/or changes in the power supply voltage - because the diode exhibits a nearly constant voltage over a large range of currents, the input to a diode will exhibit a relatively steady 0.7 V over a wide range of currents - higher voltages can be achieved by connecting diodes in series; e.g., a series of three diodes provides a steady 2 V over a wide range of currents 4.4: Operation in the Reverse Breakdown Region - Zener Diodes The abrupt change in the reverse breakdown region can be useful in some contexts, such as voltage regulators, because the diode maintains its voltage with high precision over a range of currents - zener diodes are manufactured that operate specifically in this range - a zener diode can be positioned in a reversebias position and subjected to a negative voltage from anode to cathode (i.e., a positive voltage over the cathode to anode); the diode consumes a very specific voltage ( VZ ) over a range of currents ( IZ ) - the consumed voltage does change slightly, based on an incremental resistance rz , such that δV = rz δI, but this value is typically in the range of 10 Ω - in the actual “knee” region near VZK , the resistance changes abruptly, so it is best to avoid operating a zener diode 22 in this region - zener diodes are available with VZ in the range of a few volts to hundreds of volts, and can operate within a range of currents A simple model of zener diodes is reflected by the equation VZ = VZ0 + rz IZ in practice, VZ0 is very close to VZK The zener voltage is temperature-dependent, and is defined by a temperature coefficient, TC (mV / ◦ C), which can be positive or negative - each zener diode has a particular profile curve for TC at a range of voltages; many zener diodes have a current point where TC = 0, and therefore are not temperaturedependent when operated at that current - the temperature coefficient can also be reduced by connecting a reverse-biased zener diode in series with a forwardconducting diode that has an opposite temperature coefficient 4.5: Rectifier Circuits Diodes are useful as rectifiers in many context - example: in an AC-to-DC power supply, an alternating current is provided to a full-wave diode rectifier that produces a fully rectified current, then to a low-pass filter that averages each rectified cycle (but which still exhibits a time-dependent “ripple”), and finally to a voltage regulator that provides a constant current to a load (This process often begins with a step-down transformer: a first voltage, such as 120 V, runs through a first coil with N1 turns that induces a current in a second 2 coil with N2 turns, so that the voltage fed into the diode rectifier is 120 · N N1 - in addition to stepping down the voltage, this also electrically insulates the equipment from surges on the line) Half-wave rectifier: ( 0, vs < VD Using the constant-voltage-drop model, v0 = - two relevs − VD , vs ≥ VD vant factors for rectifier diodes: maximum current-handling capability, and peak inverse voltage (PIV) that diode must withstand without reaching breakdown - best to use diodes where VZK is well above the PIV - rectifier diodes also do 23 not work well for small-voltage signals; these scenarios call for a combination of a diode and an op-amp Center-tapped full-wave rectifier: A full-wave rectifier can be designed by center-tapping the step-down coil of the transformer, and using two diodes that direct current away from either end - the average power of this rectifier will be larger than for the half-rectifier - however, each diode experiences PIV = 2Vs − VD Bridge rectifier: A bridge rectifier can be designed as a Wheatstone bridge, featuring a combination of two pairs of diodes, each of which passes current over a load resistor during half of a cycle and resists the flow of current during the other half of the cycle - bridge rectifiers use twice as many diodes as a center-tapped rectifier, and also imposes a two-diode voltage drop instead of a one-diode voltage drop - however, the bridge rectifier does not require center-tapping, and therefore required a coil that is half the length of the center-tapped rectifier - also, the PIV is distributed over two diodes, as PIV = Vs − VD Peak rectifier: 24 (Note: This is a half-wave peak rectifier; a full-wave peak rectifier can be constructed by also including a diode across the load resistor) - in order to transform an AC rectified voltage into a DC voltage, a capacitor can be added in parallel with the load resistor - the capacitor rapidly charges when the voltage is near its peak, and then discharges to maintain the voltage while the rectified current falls - as the capacitor discharges, it loses a bit of voltage until the next peak of the source voltage - the result is a nearly-DC voltage with a “ripple” as the capacitor charges and discharges, which is determined as: Vo = Vp − 12 Vr , where Vp = peak voltage Vr = ripple voltage magnitude, where Vr = fILC The conduction interval δt is defined as: Vp cos(ωδt ) = Vp − Vr Peak rectifiers can be used as peak detectors, such as an AM demodulation devices Precision half-wave rectifier: A “superdiode” can be generated with the addition of an op-amp to provide a half-wave rectified output v0 that captures the full amplitude of the input signal - this occurs because the feedback loop causes the op-amp to drive an output voltage of va = vi + vD , i.e., higher than the input voltage - the actual output voltage, after the diode voltage drop, is vo = va − VD = vi 25 4.6: Limiting and Clamping Circuits Voltage limiter: Since diodes provide a (nearly) constant voltage drop in a forward-biased range, diodes can be used to impose as a voltage limit - the constant-voltage-drop model causes this to appear as a “hard limit” with a sharp inflection point, but real diodes provide a “soft limit” with a more gradual approach to the cutoff voltage - limiting can be imposed on positive or negative voltage (by reversing the orientation of the diode), and can be offset by placing a voltage source in series with the other side of the diode - two limits can be imposed with a pair of diodes Clamped capacitor: When this voltage source provides a negative voltage of −V , current flows through the diode (which has zero output) until the capacitor is fully charged when the voltage source then shifts to a positive voltage +V , the diode resists the flow of current, with a magnitude of +V − (−V ) - the result is that where the input voltage varies from −V to +V , the diode exhibits an output voltage of 0 to (+V − (−V )) - that is, this diode adds a DC component to the fluctuating AC source; the output is said to be “clamped” at a higher value - adding a resistor in parallel with the diode allows current to flow during each cycle, causing a slightly skewed output Voltage doubler: 26 The combination of a clamped capacitor and a peak rectifier causes an output that is twice the magnitude of the input 4.7: Special Diode Types Schottky-Barrier Diode (SBD): This type of diode is not formed by a pn junction, but by an n-type material in contact with metal - while operating very similarly to a diode, this diode does not exhibit capacitance in the n-type material, thus avoiding capacitance and raising the switching speed of the diode the forward voltage drop is also lower (0.3 to 0.5 V, vs. 0.6 to 0.8 V) Varactors: The reverse-bias capacitance of the depletion region is voltagedependent, and can be used in some applications (e.g., self-tuning radio receivers) - special diodes called varactors can be designed that exhibit this property with consistent voltage-dependent variance Photodiodes: Most diodes are shielded from light, because photons can break covalent bonds in the depletion and create a current - optoelectronics can make use of this effect by exposing the junction to light that increases the current of a set voltage when exposed to light - solar cells also make use of this effect to transform light into voltage Light-emitting diodes: A diode made with a particular type of material (directbandgap material) causes light to be emitted when carrier-injected minority carriers recombine with majority carriers in each region - LEDs can be made to produce light across a variety of wavelengths, including single-wavelength LEDs (“laser diodes”) Optoisolator: An LED can transform voltage into light that is then captured by a photodiode and converted back into voltage - this combination enables the transmission of voltage with electrical isolation, which reduces interference and the effects of current surges and spikes 27 Chapter 5: MOS Field-Effect Transistors Many types of three-terminal devices allow the voltage between two terminals to act as a switch that controls the flow of current through the third terminal - there are two types of transistors: metal-oxide-semiconductor field-effect transistors (MOSFETs) and bipolar junction transistors (BJTs) - MOSFETs are widely used, particularly in integrated circuits, and can be smaller, more energy-efficient, and more simply and cheaply manufactured than BJTs, and integrated circuits often pack large numbers of MOSFETs into a small package - MOSFETs can also exhibit analog behavior, and ICs often implement “mixed-mode” designs 5.1: Device Structure and Physical Operation Basic MOSFET structure: A p-doped single-crystal silicon wafer substrate is provided, and two regions (the “source” and the “drain”) are heavily n-doped - a layer of silicon dioxide (SiO2), about 1 to 10 nm (10 to 100 angstroms), is developed on the surface to cover the region between the source and drain (the “channel” region), and a layer of metal is deposited atop the silicon dioxide to form a gate electrode (“G”) - metal contacts are formed to the source (“S”), drain (“D”), and substrate or body (“B”) - because the oxide layer physically insulates the gate electrode from the device body, very little current flows through the gate 28 This structure forms pn-junctions with the source and drain regions - these regions are ordinarily kept reverse-biased, with the drain higher than the source - a positive voltage applied to the gate terminal causes current to flow through the channel between the source and drain - the length ( L ) and width ( W ) of the channel affect the performance of the device - note that, at this basic level, the source and drain are physically identical and can be physically reversed At zero voltage, the MOSFET acts as a pair of diodes (between the p-type material and the source, and the p-type material and the drain), and impose a resistance on the order of 1012 Ω The application of a positive voltage (vGS is applied to the gate pushes free holes under the gate further downward in the substrate; the bound negative charges of the donor atoms are uncovered, creating a negatively charged region - this voltage also draws electrons into this region, right under the oxide layer, from the n-type source and drain regions - this “induced” n-type region forms a channel (“inversion layer”) - a positive voltage is then applied between source and drain (vDS ) allows current to flow through the channel - this structure therefore forms a n-channel MOSFET in a general region of p-type material Capacitative effects: The gate and channel regions form act as a parallel-plate capacitor separated by the silicon dioxide insulator, with positive charge accumulating in the gate and negative charge accumulating in the channel region the electric field created in the channel region controls the conductivity of the region and the resulting current when a voltage vDS is applied The gate has a threshold voltage Vtn , usually between 0.3 and 1 volt - when vDS = 0, the channel has zero voltage, and the voltage across the oxide is (vGS - an effective voltage or overdrive voltage is defined as vOV = vGS − Vtn - the capacitance of the channel is defined as: C = COX (W L), where W, L = the width and length of the channel COX = oxide capacitance (often expressed in square microns), and is defined as: OX , where COX = tOX OX = the permittivity of silicon dioxide = 3.90 = 3.45 · 10−11 F/m tOX = the thickness of the oxide layer The amount of charge stored in the channel is defined as |Q| = CvOV = Cox W LvOV It is sometimes desirable to express charge in terms of channel length: |Q| unit channel length = Cox W vOV The charge stored in the channel increases proportionally with vOV , and is often depicted as a “deeper” channel 29 When a small positive voltage VDS is imposed between the drain and the source while a channel is induced, the voltage creates an electric field defined as: |E| = vDS L Electron drift occurs in the direction of source to drain, the amount of: Electron drift velocity = µn |E| This drift causes a current from drain to source, in the amount: iD = µn Cox W L vOV vDS In this small-voltage scenario, the channel acts as a linear resistor, with conductance: gDS = µn Cox W L vOV The first factor, µn Cox , is determined by the material manufacturing process, and is often expressed as the process transconductance parameter: kn0 = µn Cox , in units VA2 The second factor, W L , is often called the aspect ratio, and is the easiest factor to select in order to achieve a desired conductance; manufacturing processes typically have sufficient precision to create a particular minimum channel length, and are characterized as such, e.g., a 45-nm manufacturing process The product of the process transconductance parameter and the aspect ratio is often expressed together as the MOSFET transconductance parameter: kn = kn0 W L Therefore, when vDS is small, the MOSFET operates as a voltage-controlled resistor: 1 rDS = µ C W (v n ox L GS −Vtn ) The small-vDS MOSFET therefore has the following iD − vDS characteristic: The resistance has slope (vGS − Vtn ); when vGS = Vtn , the line has a slope of zero, i.e., infinite resistance Triode region: As vDS increases while vGS is held at a constant overdrive voltage VOV > Vtn , vDS exists as a voltage drop over the length of the channel, from vGS = Vtn + VOV to vGD = vGS − vDS = Vtn + VOV − vDS : 30 The voltage difference between the gate and the source/drain regions determines the depth of the channel, so the channel is deepest at the source and shallowest at the drain - as vDS increases toward VOV , the channel resistance increases as well, until VDS = VOV = VGS − Vtn , at which point the taper of the channel at the drain equals zero The current can be expressed as a function of the area of the channel: 1 iD = kn0 W L (VOV − 2 vDS )vDS Saturation region: As vDS scales beyond VOV , the channel is no longer induced at the drain end, and in fact at VDS > VGS − Vtn , the positive voltage of the drain causes the channel to not exist even farther from the drain (“channel pinch-off’) - as the iD −vDS characteristic shows, current saturates in this region of operation, and is no longer dependent on vDS - the saturation region occurs at: VDSsat = VOV = VGS − Vtn 2 iD = 12 kn0 W L VOV Subthreshold region: It is not entirely true that no current flows at vGS < Vtn ; at values of vGS very near Vtn , a small current flows - the current exhibits exponential growth based on vGS s, much like ordinary pn junctions - some devices are making use of the subthreshold region 31 p-channel MOSFET: The same engineering process can result in the creation of a p-channel MOSFET - starting with an n-type material, a pair of p+ source and drain regions are formed - an oxide insulating layer is formed between, with a gate polysilicon layer, and metal terminals are formed for the source, drain, gate, and grounded body - the MOSFET exhibits a threshold voltage Vtp , which is negative by convention - and a channel forms when vGS < Vtp (i.e., when the negative voltage of the gate exceeds the threshold voltage) - in order to discuss this in a more familiar manner, the equations are written in terms of magnitude: |vGS | > |Vtp | - the process transconductance parameter, kp0 = µp Cox , further determines the transistor transconductance parameter kp = kp0 W L - PMOS devices used to be prevalent but are now supplanted by NMOS devices, because µn exceeds µp by a factor of 2 to 4, allowing for faster device speeds Complementary metal-oxide semiconductors (CMOS): Many scenarios require both NMOS and PMOS devices - this can be achieved by starting with one material, forming a “well” in which a device of the other type is formed, and creating a silicon dioxide isolation structure in between 5.2: Current-Voltage Characteristics An n-type MOSFET is typically shown as the following symbol, with an arrow showing the direction of current flow: As noted earlier, the MOSFET is a symmetric device, so the source and drain regions are arbitrarily selected to match the direction of current flow - in an n32 type MOSFET, the drain is always more positive than the source, and current flows from drain to source; in a p-type MOSFET, the drain is always more negative than the source, and holes (and hence current) move from source to drain The iD − vDS characteristics are mapped by setting vGS , and measuring the current over the range of vDS : The current reflects the three regions described above: • vGS < Vtn : Cutoff region (including sub-threshold region) • vDS < vOV = vGS − Vtn : Triode region (including the near-linear smallvoltage region) • vDS > vOV = vGS − Vtn : Saturation region MOSFETs can also be modeled as an iD − vGS characteristic by setting vDS and measuring the current over a range of vGS : The current of this circuit remains the same: 2 iD = 12 kn0 W L (vGS − Vtn ) This configuration is useful as an amplifier, but its nonlinearity at low voltages 33 can be problematic - however, in a large-voltage content that operates in the linear region, the vGS model can be considered equivalent to a voltage-controlled current source dependent upon vGS Saturation-mode output resistance: The VCCS model described above is somewhat idealized, by presuming that iD and vDS are unrelated in saturation mode - in reality, as vDS > vOV increases, increasing channel pinch-off causes a voltage drop due to the reduction in channel length (“channel-length modulation”), and iD increases with vDS - this resistance manifests as a small additional term in the current equation: 2 iD = 12 kn0 W L (vGS − Vt n) (1 + λvDS ), where λ = a device-specific parameter indicating the proportionality of vDS and iD due to channel-length modulation - this factor is determined by process technology, and is inversely proportional with channel length (since smaller channels are more susceptible to length reduction) - λ is sometimes expressed as VA = λ1 , which in turn is often expressed as VA = VA0 L based on process technology at a V particular length; VA0 is often in the range of 5 to 50 µm - for a particular vGS , output resistance can be represented as: δiD −1 ) = λI1D = VIDA , where ro = ( δv DS ID = drain current without channel-length modulation Characteristics of p-channel MOSFET: The p-channel MOSFET behaves in an almost identical manner, and exhibits an iD − vGS characteristic: 2 iD = 12 kp0 W L (vSG − |Vtp |) (1 + |λ|vSD ) 5.3: MOSFET Circuits at DC (Examples of circuit analysis and design problems) 5.4: Applying the MOSFET in Amplifier Design The basic MOSFET operates as a voltage-controlled current source, where VGS controls ID - although this behavior is nonlinear, configurations can be devised that provide a near-linear response Simple voltage amplifier circuit: 34 The current output of the MOSFET is passed through a load resistor, and the voltage between the resistor and the drain is the output: vDS = VDD − iD RD The voltage transfer characteristic (VTC) of this amplifier is as follows: When VGS < Vt , the MOSFET is in the cut-off mode; as VGS rises above Vt , the MOSFET begins transmitting current in the saturation mode, in a near-linear manner; and when VGS > VD S, the amplifier transitions to triode mode and the output becomes nonlinear The MOSFET can be used in a linear application by selecting a bias point, Q, where VGS and VDS are in the middle of the near-linear region - a small signal vgs (t) can be imposed on a DC voltage of VGS to obtain an output in the middle of the near-linear region, such that vDS (t) scales linearly with vGS (t) = VGS + vgs (t) The voltage gain of the amplifier at this point is Av = as: 35 dvDS dvGS , which is defined RD Av = −kn VOV RD = − 2IVDOV The output is 180 degrees out of phase with the input (hence the negative gain), and is proportional to RD , kn , VOV The gain is restricted by the magnitude of the input: DD |Avmax | = 2V VOV Determining VTC by graphical analysis: The quiescent point Q can be determined by graphing the curves for each input signal vGS (t): For a selected VDD and VGS , each selection of vgs (t) leads to a value vGS (t) that generates a curve of iD over the range of vDS , as determined by RD : iD = VRDD − R1D vDS D The load line has four important points: • A : vGS = Vt , so MOSFET is in cutoff mode and no current flows • Q : vGS = VGS , i.e., quiescent • B : the load line enters the triode region • C : vGS = VD D, when the MOSFET acts as a closed switch with resistance rDS (“closure resistance”) Choosing Q: If we presume that RDS is fixed but VGS is constant, the intended use of the amplifier may guide the selection of Q - choosing a point too close to either axis may create problems: if too close to the x-axis, a signal swing of vGS < Vt causes the MOSFET to turn off, so the signal is “clipped” due to insufficient “head room”; and if too close to the y-axis, a signal swing of vGS < vDS will cause the MOSFET to enter the triode region and produce distorted output due to insufficient “leg room” 36 5.5: Small-Signal Operation and Models For a particular amplifier circuit, the DC bias current ID and bias voltage, at the bias point Q, can be found by setting vGS = 0: 1 kn (VGS − Vt )2 = 12 kn (VOV )2 ID = r2 VDS = VDD − RD ID This provides an amplifier with the range of vgs as: VDS + Vt + VGS < vgs < VDD The current for any value of vGS can be written as: iD = 21 kn (VGS − Vt )2 + kn (VGS − Vt )vgs + 21 kn (vgs )2 The first term is the DC bias current; the second term is directly proportional to vgs ; and the third component is nonlinear distortion, which is minimized to promote linearity by keeping vgs within a small range: vgs 2VOV When calculated over a small range, the last term can be disregarded, and current can be expressed as: iD ≈ ID + id , where id = kn (VGS − Vt )vgs The linearity of this region can be expressed as the transconductance of the amplifier: δiD d = kn (VGS − Vt ) = kn VOV = δv gm = vigs GS The voltage drop can similarly be expressed as the sum of the bias voltage and the signal voltage: vDS = VD D − RD (ID + id ) vds = −id RD = −gm vgs RD The amplification can be written as: = −gm RD Av = vvds gs 37 The small-signal amplifier can be expressed as two equivalent circuits: • Hybrid Π model: This model involves just a voltage-controlled current source with current : ID = gm vgs , where D gm = kn VOV = V2IOV , and vgs = voltage drop over the current source In this equivalent-circuit model, any ideal voltage sources are replaced by short circuits, and any ideal current sources are replaced by an open circuit - if channel modulation is also taken into account, a resistor is positioned in parallel with the current source, with resistance: r0 = |VIDA | And the current is: 2 ID = 12 kn VOV The voltage gain is defined as: Av = vvds gs The transconductance gm interrelates three design parameters: 38 W L , VOV , ID - any two of these design parameters can be selected, and the third design parameter can be calculated accordingly • T model: In this equivalent-circuit model, a voltage-controlled current source is placed in series from source to drain with a resistor of value g1m , and the gate connects in between - the input resistance connecting to the gate is ideally infinite - output resistance can be modeled as a resistor of value ro in parallel with the combination 5.6: Basic MOSFET Amplifier Configurations General model of a MOSFET amplifier: In an ideal model, the input resistance Ri = ∞, and the output resistance Ro = 0 - in reality: • Input resistance: Rin = vi ii • Open-circuit voltage gain: Avo = vo vi |RL =∞ 39 L • Voltage gain: Av = Avo RLR+R o • Output resistance: Ro = vixx , where ix is a current applied to the output terminal when the input is zero, and vx is a voltage measured over the output terminal • Overall voltage gain: Gv = vo vsig Three basic configurations exist for MOSFET amplifiers: • Common Source (CS): Source is grounded; input voltage is applied at the gate; resistor diode RD is positioned in parallel with MOSFET, and output voltage is read over the resistor The generic model is adapted for the common source amplifier as follows: Rin = ∞ vgs = vi vo = −gm vgs 1 1+ 1 RD Avo = −gm 1 RD ro 1 + r1o ≈ −gm RD Common − sourceamplif iers • Common Gate (CG): Gate is grounded; an input voltage is applied at the source; resistor diode RD is positioned in parallel with MOSFET, and output voltage is read over the resistor • Common Drain (CD) or Source Follower: Drain is grounded; input voltage is applied at the gate; load resistor RL is connected to the source, and output voltage is read over the resistor 5.7: Biasing in MOS Amplifier Circuits Biasing is the process of applying a DC drain current ID and drain-to-source voltage VDS to maintain operation of the MOSFET in saturation mode for small-signal applications Biasing by fixing VGS : This model involves simple biasing by fixing the gate current, such as using a voltage divider - however, this type of biasing is difficult to apply consistently, because some MOSFET properties ( µn , Cox ) can vary by devices even produced using the same process, and Vt , µn are temperaturedependent Biasing by fixing VG : 40 A better model involves fixing VG and connecting a resistor to the source, such that VG = VGS + RS ID - in this case, if VG VGS , then ID is determined by VG and RS , and is therefore predictable and stable; and if VG rises above VGS , the resistor provides negative resistance that reduces the magnitude of changes in the current Biasing using drain-to-gate feedback resistor: Connecting the drain to the gate (through a resistor RG ) causes negative feedback - if the current ID increases, the voltage after the resistor RD is reduced, which reduces VGS and reduces ID - similarly, if the current decreases, the voltage drop over RD is less, and the gate voltage increases to allow more current Biasing using constant current source: 41 When the gates of two MOSFETs are connected and a current is provided through one, the current through the other is determined as a ratio of the W L W2 L2 of each MOSFET: I2 = I1 W - this configuration is known as a current mirror 1 L1 5.8: Discrete-Circuit MOS Amplifiers MOSFETs can be used as a combination of discrete components to build amplifiers, using the concepts provided above - this section covers single-stage amplifiers, based on the following components using a constant-current biasing arrangement The components used solely in the AC portion are connected by capacitors a bypass capacitor connects ground in parallel with the current source, which 42 allows the signal to bypass the input resistance of the current source - coupling capacitors are used to isolate vsig and vo to prevent interference from the DC input component Common-source amplifier: This amplifier exhibits high amplification: G Gv = − RGR +Rsig gm (RD ||RL ||ro ) Common-source amplifier with unbypassed source resistance: 43 The inclusion of a source resistor allows an increase in the gain: (RD ||RL ||ro ) G Gv = − RGR 1 +Rsig +R gm s Common-gate amplifier: Although presenting low gain, this amplifier presents better performance for high-frequency signals Common-drain amplifier (source follower): This amplifier presents no gain (Avo = 1), but enables a reproduction of a high-resistance input signal to a low-output resistance Amplifier response frequency: Amplifiers typically show consistent gain only for signals within a specific frequency range (within the “bandwidth” of the amplifier) - at lower frequencies, the capacitance of the capacitors fails to isolate 44 the low-frequency signal from the zero-frequency DC component; while at higher 1 frequencies, the reactance of the capacitors ( jωC ) becomes significant and reduces the gain 5.9: The Body Effect and Other Topics The basic diagram of the MOSFET indicates that a terminal is connected to the substrate body - this connection is not relevant (or even present) for discrete circuits, but in integrated circuits, the substrate is shared by many densely packed MOS transistors, and the body can exhibit a slight positive voltage that makes it difficult to maintain reverse bias and cutoff - this effect can be reduced by connecting the substrate the most negative power supply in an NMOS circuit, or the most positive supply in a PMOS circuit - however, this voltage can increase the voltage needed to pachieve forward bias: p Vt = Vt0 + γ( 2φf + VSB − 2φf ), where Vt0 = the threshold voltage when VSB = 0 φf , γ are physical parameters determined by the fabrication process ( γ is the “body-effect parameter”) The effect of the body voltage is to create a second voltage difference between the source and body, vbs , which acts as a second gate that also controls the current (“backgate”) - the effect is modeled as: δiD gmb = δv BS Temperature effects: Both Vt and k 0 are temperature-dependent: Vt decreases with rising temperature and increases current; k 0 decreases with temperature and thus decreases current (this effect is more prevalent) Breakdown: Three kinds of breakdown can occur in MOSFETs: • Avalanche breakdown: As the drain voltage increases to a particularly high value (usually in the range of 20 V to 150 V), the drain-to-substrate 45 pn junction exhibits weak-avalanche breakdown • Punch-through: In devices with short channels, a high drain voltage can cause pinch-off to reach all the way to the source, causing a high current increase • Gate oxide breakdown: A high VGS (more than 30 V) can permanently damage the gate oxide - this can occur due to static charge stored on the gate capacitor - this effect can be decreased by adding clamping diodes to the input terminals Velocity saturation: High electric fields can impose an upper limit on the drift velocity of carrier charges; in very-short-channel devices, this can occur below vDS < 1V - this effect causes current to depend linearly on vGS rather than 2 , and can cause gm to become independent of vGS according to vGS Depletion-type MOSFETs: The MOSFETs in this chapter involve a channel that is induced by voltage (“enhancement-type” MOSFETs) - other MOSFETs have a physically implanted channel, where negative voltages applied to the gate and drain deplete the channel of charge carriers, reducing and eventually cutting off the current - some MOSFETs can be operated both in depletion mode and enhancement mode 46 Chapter 6: Binary Junction Transistors (BJTs) BJTs, like MOSFETs, are useful as controlled sources, as switches, and as logic inverters - BJTs were first discovered after MOSFETs, but were more useful for a while until MOSFET manufacturing techniques were improved - however, BJTs are still more useful in some cases, such as extreme environmental conditions and very-high-frequency devices (e.g., emitter-coupled logic); other contexts use MOSFETs in conjunction with BJTs 6.1: Device Structure and Physical Operation Two types of BJTs are possible: pnp BJTs and npn BJTs - the npn BJT is manufactured by creating a highly doped n-type emitter, a lightly doped ptype base, and a lightly doped n-type collector, with metal contacts forming terminals connected to each region (pnp BJTs simply reverse the doping types of the regions) - most of the following discussion will focus on the npn BJT, but the same properties apply to the pnp BJT The BJT has two pn junctions: the emitter-base junction (EBJ) and the collectorbase junction (CBJ) - the relative voltages of the terminals determine whether each junction is forward- or reverse-biased, and thus the operating properties of the device: EBJ Reverse Forward Reverse Forward CBJ Reverse Reverse Forward Forward Mode Cutoff Forward Active Mode Reverse Active Mode Saturation Typically, the forward active mode is used when the transistor is configured as an amplifier, and the saturation and cutoff modes are used when the transistor is configured as a switch 47 Forward-active mode: This mode arises in the npn BJT when vE < vB < vC , exhibiting a conventional current from collector to emitter In the forward active mode, the EBJ is forward-biased and the CBJ is reversebiased - this forward bias creates two types of current: electrons injected from the emitter into the base, and holes injected from the base into the emitter (the doping levels create a much higher electron current than hole current) the injected electrons diffuse through the base region, with the concentration steadily diminishing across the length of the base (reduced a bit further by recombination) - the electron concentration just beyond the EBJ is defined as: vBE np (0) = np0 e VT , where: np0 is the thermal equilibrium value of majority carriers in the base region This injection also gives rise to an electron diffusion current: dnp (x) n (0) In = AE qDn dx = AE qDn (− pW ), where AE = the cross-sectional area of the emitter-base junction Because electrons flow left-to-right (emitter-to-base), this diffusion gives rise to a conventional current from right-to-left Forward-active mode collector current: Because the CBJ is reverse-biased, injected electrons that diffuse across the base to reach the CBJ are swept across the junction due to the barrier voltage - this gives rise to a collector current, which is defined as: VBE iC = IS e VT , where IS = saturation current, which is defined as: A qDN n2i n IS = AE qDN Wp0 = ENA w Notably, the collector current iC is independent of VCB ; as long as this value is positive (to maintain reverse-bias), the saturation and collector currents are not dependent on the actual voltage - the linear dependence on the area enables otherwise-identical devices to exhibit currents as a ratio (e.g., if the same voltage is applied to two BJTs with a 1:2 area ratio, the current through each BJT will also have a 1:2 ratio) 48 Forward-active mode base current: The base current iB has two components: iB1 , the injection of holes from the base region into the emitter region, and iB2 , the supply of holes from the terminal in response to the loss of holes from recombination - the total base current is defined as: iB = iB1 + iB2 VBE Since iB is proportional to e VT , just like iC , the base current is always proportional to the emitter current: iB =iC , where β β = the common-emitter current gain, which is a transistor parameter that is usually in the range of 50 to 200 - it is desirable to keep β high - this factor is inversely proportional to the length of the base region (hence W is usually kept small), and directly proportional to the doping ratio of the base and emitter NA ) regions ( N D Forward-active mode current relationships: The current through the npn BJT in forward-active mode creates a conventional current from collector to emitter - the total emitter current is defined as: VBE β+1 VT iE = iC + iB = β+1 i = I e S β C β To simplify this equation, we can define a parameter α as the common-base current gain: β α α = β+1 (or conversely, β = 1− , such that iE = iS αe vBE VT Forward-active mode equivalent-circuit models: When a positive VBE voltage is applied to an npn BJT, the collector emits a constant current that is independent of VCB as long as the CBJ remains reverse-biased ( VC B ≥ 0 ) because the base current is ideally very small, iC ≈ iE - the npn BJT therefore operates as a constant-current source, where the current is defined by VBE based on this behavior, four equivalent circuit models are possible: 49 This model is a voltage-controlled current source - note that the current in this model does not scale linearly, but exponentially, with the measured voltage This model is a current-controlled current source - again, the current in this model does not scale linearly, but exponentially, with the measured current the gain between iE (as input) and iC (as output) is determined by α, which is why α is called the common-base current gain 50 This model is a two-port voltage-controlled current source This model is a two-port current-controlled current source - the gain between iB (as input) and iC (as output) is determined by β, which is why β is called the common-emitter current gain A typical npn BJT is formed as follows: 51 Unlike a MOSFET which is structurally symmetric and reversible, the structural asymmetry makes this device non-reversible, with α ≈ 1 and a large β value Reverse-active mode: This mode arises in the npn BJT when vE > vB > vC - the BJT in reverse-active mode operates similarly to forward-active mode, but with a much lower value of β, making this mode undesirable Cutoff mode: This mode arises in the npn BJT when vE > vB < vC , such that both junctions are reversed-biased - in this mode, the reverse biases of the EBJ and CBJ effectively prevent the flow of current Saturation mode: This mode arises in the npn BJT when vE < vB > vC , such that both junctions are forward-biased - note that the forward-active mode of the npn BJT can be maintained as long as vCB ≥ −0.4 V (since forwardbias of the pn junction only occurs at +4 V) - below this value, the npn BJT is in saturation mode - the collector current is reduced, and the base current increases, dependent upon vCB : vBC vBE iC = IS e vVT − ISC e vVT BE BC iB = IβS e VT + ISC e VT This relationship causes the value of β to change based on vBC ; i.e., vBC “forces” β to a value lower than the constant value of β in forward-active mode - the resulting new value of β is called βforced , and is defined as: C βforced = iiB |saturation ≤ β This provides two ways to determine whether the BJT is in saturation mode: C either vCB < −0.4 V, or iiB <β The depth of the saturation mode can be expressed as: vCEsat = vBE − vCE , where vCEsat = 0.3 V indicates that the transistor is at the edge of saturation, and vCEsat ≤ 0.2 V indicates that the transistor is deep in the saturation region The saturation-mode BJT can be modeled as follows: The pnp BJT: 52 The pnp BJT works in a similar manner as the npn BJT, except that the current is mainly conducted by holes, and conventional current is from the emitter to the collector - also, a base current runs out of the BJT rather than into the base region 6.2: Current-Voltage Characteristics npn and pnp BJTs exhibit current flow in opposite directions, and are represented by the direction of the arrow (always pointing from positive to negative) - currents are always measured in the positive direction - the following diagrams show the currents of npn and pnp BJTs respectively: 53 Collector-base reverse current: The reverse current across the collector-base junction, with the emitter open-circuited, is denoted ICBO ; the value is typically in the nA range, and is dependent upon VCB , doubling for each 10◦ C increase BJT iC − vBE Characteristic: Since iC varies logarithmically with vBE and is temperature-dependent, the BJT exhibits the following relationship: Early effect: The voltage vCE has a small but positive relationship with iC , which also scales with higher values of vBE - iC − vBE is a straight line, and, extrapolating backward, we find a common point −VA that defines the relationship: This relationship occurs because of the base-width modulation effect: raising vCE increases the reverse-bias voltage, increases the width of the depletion region, and decreases the effective base width (i.e., the width of the transistor 54 1 acting as the base region); since IS W , a shorter effective base width increases the current - the Early effect can be modeled as follows: VBE ) iC = IS e VT (1 + vVCE A This can also be modeled as output resistance: CE ro = VA +V IC 0 We can also define IC as the collector current with the Early effect neglected, and that ro = VI 0A C δiC ; i.e., vary iB by a little, and notice the We can also measure β as β = δi B change in iC - this property actually exhibits differences in alternating-current and direct-current circumstances, but we can disregard this difference for now In saturation, ICsat = βforced IB , where βforced < β - as a result, the Early effect is attenuated in saturation, but leads to significant differences in the active mode due to larger values of β BJT Circuits at DC Examples of BJT circuits operating in direct-current scenarios: Presume that VBE = 0.7V; VCEsat = 0.2V when saturated; and disregard Early effect - the first question is: Which mode is the transistor in? • First presume that the BJT is in forward-active mode, and verify that behavior matches VCB > −0.4V • If not, presume that the BJT is in saturation mode, and verify that βforced < β 55 IC IB = 6.4: Applying the BJT in Amplifier Design Natively, the BJT operates as a transconductance amplifier (i.e., a voltagecontrolled current source): vBE controls IC - these types of amplifiers can easily be converted to a voltage amplifier (i.e., a voltage-controlled current source) by inserting a resistor RC at the collector and measuring its voltage drop the relationship between input voltage vBE and output voltage iC RC does not change linearly (the response depends on whether vBE places the BJT in the cutoff mode, active mode, or saturation mode), but can be converted to a linear amplifier over a small range Simple amplifier and its voltage transfer characteristic (vBE → vCE : This transistor has output vCE = VCC − iC RC - the active mode shows strong gain (steep slope) with output: vBE vCE = VC C − RC IS e VT While this range, described broadly, is nonlinear, it becomes more linear on a small scale - we can make use of this region as a linear amplifier by confining our input to this range Biasing the BJT: A point Q along this curve can be selected in the middle of the range, and a constant voltage source VBE can be applied to bias the base up to this point: 56 At this point, the BJT exhibits output: VBE IC = IS e VT Then we apply to the base a second voltage source, vbe , which carries a smallsignal variation - the total instantaneous value applied to the base is: vBE (t) = VBE + vbe (t) This variance produces linear output vCE as long as vbe is kept small; linearity decreases with higher magnitude of vbe The voltage gain AV of the amplifier is: dvCE AV = dv |vBE ≈VBE BE The voltage gain can also be expressed as: AV = − VICT RC This indicates that the voltage gain is proportional to the voltage drop of RC , so we can increase the gain by increasing RC - however, if VCC is constant, increasing RC correspondingly decreases VCE and pushes the quiescent point toward the cutoff region, thus reducing the maximum swing (specifically, head room) of the small signal Determining the VTC by graphical analysis: 57 We can determine the VTC by graphing vCE vs.iC , for various values of vBE , and superimposing a line representing VRC - this line (the “load line”) begins at the horizontal axis at vCE = VCC , and has a negative slow − R1C - the intersection of the load line with each value represents the voltage of VCE for each value of vBE 6.5: Small-Signal Operation and Models When evaluating circuit models for small-signal amplification, first set the small signal vbe to zero, and evaluate the circuit to determine its direct-current bias VBE - applying a small-signal voltage over the direct-current bias results in the following models: vBE = VBE + vbe Collector current: The collector current in the small-signal model is defined as: VBE vbe iC = IS e VT e VT If vbe << VT , then the following approximations can be used: iC ≈ IC (1 + vVbe ) T iC = IC + VICT vbe This latter term is the signal component of the combined signal: ic = VICT vbe This can be rewritten using transconductance: gm = VICT ic = gm vbe Transconductance defines a linear, positive relationship between vbe and iC : Base current: As usual, the base current is a factor of the collector current, but 58 we can now separate the base current into the direct-current component and the small-signal component: C vbe = IβC (1 + vVbe = IβC + vbeβgm iB = iβC = IβC + IβV T T The small-signal component of the base current is: ib = gβm vbe The small-signal input resistance is defined as the resistance between the base and the emitter, looking into the base: rπ = vibe = gβm = VIBT b Emitter current: The total emitter current can also be broken into components: iE = iαC = IαC + iαc The small-signal emitter resistance is defined as the resistance between the base and the emitter, looking into the emitter: re = vibe = VIET = gαm ≈ g1m e The base resistance rπ and emitter resistance re are related as follows: vbe = ib rπ = ie re rπ = iieb re Voltage Gain: In the small-circuit model, the amplifier is operating as a transconductance amplifier, where vbe causes a proportional current ic = gm vbe to flow - the amplifier can be converted to a voltage amplifier by positioning a resistor at the conductor: vCE = VCC − iC RC = VCE − ic RC The voltage gain of the amplifier is measured as: C AV = vvce = −gm RC = − ICVR T be Hybrid π models: The small-signal BJT amplifier can be equivalently represented by a hybrid π model of the circuit, with separate branches for ic and ib , and the collector current represented as either ic = gm vbe or ic = βib - this model explicitly includes the base resistance, rπ 59 T model: The small-signal BJT amplifier can also be equivalently represented by a T model, with the collector and base feeding into the emitter - this model explicitly includes the emitter resistance, re 60 Application of small-signal equivalent circuits: The process for evaluating a circuit using a small-signal model is as follows: • Determine the DC operating point and the DC collector current IC • Calculate the parameters of the small-signal model: gm = β VT α gm ; re = IE = gm IC VT ; rπ = • Substitute a small-signal equivalent circuit • Analyze the circuit to determine the required quantities Early effect in small-signal models: The models above disregard the Early effect, but this can be modeled by supplementing each model with a resistor ro , with the following value: r0 = VICA The Early effect reduces both vo and AV : vo = −gm vbe (RC ||ro ) 6.6: Basic BJT Amplifier Configurations Three basic BJT amplifier configurations are possible - the small-signal version of each amplifier can be substituted with this model: 61 Each amplifier has several properties: • Input resistance: Rin = vi ii • Output resistance, when vsig = 0: : Ro = • Voltage gain of the amplifier: Av = vx ix vo vi • Voltage gain of the amplifier under open-circuit conditions: Avo = • Overall voltage gain: Gv = vo vsig Common-emitter (CE) amplifier: The equivalent circuit for the common-amplifier BJT is: • Input resistance: Rin = rπ • Output voltage: vo = −gm vπ )(RC ||ro ) 62 vo vi |RL =∞ • Open-circuit voltage gain: Avo = −gm (RC ||ro ) • Output resistance: Ro = RC ||ro rπ gm (RC ||RL ||ro ) • Voltage gain: Av = −gm (RC ||RL ||ro ) Overall voltage gain: Gv = − rπ +R sig For the common-emitter amplifier, input resistance tends to be low (increasing the input resistance decreases the gain), and output resistance and open-circuit voltage gain are high The properties of the amplifier change if a resistor is positioned at the emitter - this changes the circuit in the following ways: • The input resistance is then reflected as: Rin = (β + 1)(re + Re ) The input resistance thus increases by a factor of Rree - this property is termed the “resistance-reflection rule”: i.e., the input resistance reflects the resistance of the emitter • The voltage gain is reduced by a factor of gm Re - the gain can be maintained by increasing the input signal proportionally by a factor of gm Re The effect of the emitter resistor is to reduce the dependence of Gv on β, and to improve the high-frequency response of the amplifier Common-base (CB) amplifier: For common-base amplifiers, the Early effect is typically small and is disregarded to simplify the analysis • Input resistance: Rin = re 63 • Output voltage: v0 = −αie Rc • Open-circuit voltage gain: Avo = gm RC • Output resistance: Ro = RC • Voltage gain: Av = gm (RC ||RL ) - note that the low input resistance reduces the magnitude of the input signal, and the achievable gain C ||RL • Overall voltage gain: Gv = α RRsigh +re The gain of this amplifier is very small, but it maintains a good frequency response at higher frequencies, so it is often used in combination with other amplifier types Common-collector (CE) amplifier: The common-collector amplifier functions as an emitter follower, and is useful when a voltage signal is presented with a high input resistance, such that attaching a load directly to the input causes a very small signal across the load - the CE amplifier uses the input signal at the base to drive a current through the emitter with a much lower output resistance • Input resistance: Rin = β + 1)(re + RL ) (another example of resistance reflection) • Open-circuit voltage gain: Avo = 1 • Output resistance: Ro = re 64 • Voltage gain: Av = RL RL +re • Overall voltage gain: Gv = (β+1)RL (β+1)(RL +re )+Rsig 6.7: Biasing in BJT Amplifier Circuits A BJT amplifier is biased to ensure that the gain of the amplifier remains constant across all frequencies of interest - two biasing solutions that do not work well: applying a fixed VBE (very small deviations in VBE lead to large changes in the output), and providing a constant base current (variations in β in different circuits lead to significant differences in gain) Classical discrete-circuit bias: If only one voltage source is available, a fourresistor network is used as a voltage divider to provide a fixed voltage the the base This circuit can be evaluated by substituting the left half with its Thévenin equivalent, and then applying KVL from the base voltage to the emitter ground - this circuit can be made more resistant to variations in temperature and β by RB making VBB >> VBE and RE >> β+1 - this arrangement provides negative feedback: if RE increases, the rise in VE causes a decrease in VBE , which reduces the flow of current through the transistor; conversely, if RE falls, the lower VE and higher VBE cause more current to flow across the resistor Two-power-supply version: If two power supplies VCC , −VEE are available, the base can be grounded (resistor RB is only needed if the signal is capacitively coupled to the base) 65 Collector-to-base feedback resistor: A resistor connecting the collector and base can provide negative feedback - this circuit can be analyzed by applying KVL over VCC , RC , RB , VBE , andtheemitterground : − VCC + IE RC + IB RB + VBE = 0 Constant-current source: A constant current source can be applied from the emitter Current mirror: Two identical transistors can be connected at the base and emitter, and a positive voltage is applied to RC of one transistor - this creates a reference current IREF through R that must be matched by the current I in the other transistor Since the transistors are identical, the base currents and emitter currents are identical; the reference current IREF is determined by applying KVL from VCC 66 through the collector-base connection, over VBE , and to −VEE : +VEE I = IREF = VCC −VBE R 6.8: Discrete-Circuit BJT Amplifiers The constant-current biasing model can be combined with the amplifier types discussed in 6.6: When combined with any model, capacitors are used to connect the AC components for the amplification (the capacitors connected to the input and output terminals are called “coupling” capacitors; the capacitor coupled to the ground is called a “bypass” capacitor) Biased common-emitter circuit: 67 Biased common-emitter circuit with unbypassed resistance: Biased common-base circuit: 68 Biased common-collector (voltage follower) circuit: Frequency response: Common-emitter amplifiers typically demonstrate constant gain over a significant frequency range (the bandwidth of the amplifier) - at lower frequencies, the gain and frequency response are diminished because the impedances of the coupling and bypass capacitors begin attenuating the lowfrequency signal; at higher frequencies, the capacitance of the pn junction becomes significant 6.9: Transistor Breakdown and Temperature Effects The maximum voltage applicable to a BJT is limited by the breakdown voltage of the reverse-biased junction - the breakdown voltage is typically around 50 V breakdown of the collector-base junction is typically not destructive, but breakdown of the emitter-base junction causes a runaway current that permanently reduces β; however, the EBJ can still be used as a zener diode, which does not depend on β 69 One complexity that has been disregarded is that β is dependent on the bias current, but the bias current is typically chosen to make β as large as possible - β is also temperature-dependent, which affects the transistor when operating at high power levels 70