FRANKFURT UNIVERSITY OF APPLIED SCIENCES Faculty of Computer Science and Engineering Electronics Academic Year 2017/2018 Prof. Dr.-Ing. G. Zimmer Contents 1 2 3 Semiconductor Basics 1.1 Band theory of solids . . . . . . . . 1.2 Intrinsic conductivity . . . . . . . . 1.3 Doping . . . . . . . . . . . . . . . 1.4 Diffusion currents in semiconductors . . . . . . . . . . . . . . . . . . . . . . . . Semiconductor diode and applications 2.1 The pn-junction . . . . . . . . . . . . . . . . . 2.1.1 pn-junction with zero bias . . . . . . . 2.1.2 pn-junction with bias . . . . . . . . . . 2.1.3 Small-signal model of a pn-junction . . 2.1.4 Spice model of a semiconductor diode . 2.1.5 Different types of semiconductor diodes 2.2 Diode applications . . . . . . . . . . . . . . . 2.2.1 Diode as rectifier . . . . . . . . . . . . 2.2.2 Voltage multiplier . . . . . . . . . . . . 2.2.3 Zener diode as voltage regulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The bipolar junction transistor and applications 3.1 The bipolar junction transistor . . . . . . . . . . . . . 3.1.1 Structure and operation principles of a npn BJT 3.1.2 Static input and output characteristics of a BJT 3.1.3 Simple small signal BJT model . . . . . . . . 3.1.4 Advanced small signal BJT model . . . . . . . 3.1.5 SPICE model of a BJT . . . . . . . . . . . . . 3.2 Small signal amplifier . . . . . . . . . . . . . . . . . . 3.2.1 BJT biasing . . . . . . . . . . . . . . . . . . . 3.2.2 Common-emitter amplifier . . . . . . . . . . . 3.2.3 Common-collector amplifier . . . . . . . . . . 3.3 Integrated circuit techniques . . . . . . . . . . . . . . 3.3.1 The differential amplifier . . . . . . . . . . . . I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 2 4 8 10 . . . . . . . . . . 12 12 12 17 21 24 26 29 29 34 36 . . . . . . . . . . . . 41 41 41 44 47 50 50 52 52 55 61 65 65 3.3.2 3.3.3 3.3.4 3.3.5 Current Sources . . . . . . . . Active Load . . . . . . . . . . Level-Shifting Circuits . . . . Complementary Output Stage 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 72 73 74 Chapter 1 Semiconductor Basics Important elements of electric communications systems are devices capable of amplifying the weak received electrical signals making further signal processing possible. Up to the sixties in most purchasable receivers vacuum tubes were used for this purpose. In most pratical applications vacuum tubes were replaced by transistors after the bipolar transistor was invented by Bardeen and Brattain in 1948 and the theoretical prediction of the planar bipolar transistor by Schockley in 1949. Compared to vacuum tubes transistors have an almost infinite lifetime and it is possible to combine a large amount of transistors to form integrated electronic circuit with a very high functionality. To understand the operation principles of semiconductor devices, in the first section their physical basics will be summarized, while in the following sections the devices and their equivalent circuits are discussed. Today the most important semiconducting material is silicon (Si). In contrast to metal the conductivity of a semiductor is quite low but raises with increasing temperature. To understand this strange physical behaviour we will first discuss the atomic structure of a semiconductor and we will have a look on the band theory of solids. 1.1 Band theory of solids In the framework of Maxwells theory the influence of bodies is described by scalare values like the conductivity κ, the permittivity ε and the permeability µ. But they are not subjects of the theory itself. To explain their physical base solid states physics was introduced, which has its own base in atom physics. Since the beginning of the nineteenth century most physicists agree that matter is composed out of atoms, introduced by the greek philosopher Demokrit. Rutherford, an English physicist showed with experiments that an atom consists of a very small positive nucleus (diameter ≈ 10−13 to 10−12 cm) carrying positive charges 2 surrounded by the same amount of negative charges, called electrons, so that the atom itself is neutral. In this classical picture electrons circle around the positive nucleus like the planets circle around the sun. According to Maxwells theory the classical picture of an atom cannot be right, since an electron moving around the nucleus of an atom would losse its energy by emitting an electromagnetic wave, thus losing its energy and dropping into the nucleus. It was Niels Bohr, a Danish physicist, who postulated that an atom does not behave like a classical object, being able to exchange arbitrary amounts of electromagnetic energy with its environment, but only in mulitples of an energy unit ∆W , already introduced by the German physicist Max Planck, to describe the black-body radiation ∆W = h f (1.1) In equation 1.1 h = 6.624 10−34 Ws2 stands for Plancks action quatum or Plancks constant while f describes the frequency of the radiated electromagnetic wave. According to Niels Bohr an electron bound to the nucleus of an atom can only occupy certain levels of total energy, as shown in Fig. 1.1a. If an electron does W 6 x W 6 - W3 W2 a) W1 b) Figure 1.1: Energy levels of an atom and electronic band structure of a crystal lattice not occupy its lowest energy level, it can drop from the energy level W j to the lower energy level Wi by emitting an electromagnetic wave of frequency f = 1 (W j − Wi ) h One says the electron changed its quantum state. If we put a large amount of atoms together we can in principle form a crystal. Due to the Pauli exclusion principle, different electrons may not exist in the same quamtum state. That is the reason why in a crystal the single energy levels of an atom will split up into closely spaced energy levels forming a so-called electronic band structure as shown in Fig. 1.1b. In principle all energy levels within a band may be occupied by electrons, while no electrons may exist at energy levels between the bands. If we cool down a 3 crystal to an absolute temperature of T ≈ 0K, all atoms of the crystal will exist at their ground states and all energy levels within the electronic band structure will be occupied up to a certain level. This level is called Fermi level WF . If the temperature is increased, energy levels above the Fermi level may also be occupied by electrons. The propability p(W ) that a certain energy level W is occupied by an electron is given by the so called Fermi-Dirac distribution []. p(W ) = 1 W −WF exp( )+1 kB T with kB = 1, 38 10−23 J/K (1.2) (Boltzmanns constant) Considering the band with the highest energy one can distinguish between two different cases: 1. Electrons do not occupy all energy levels within the band. As a result there will exist free energy states slightly above the states already occupied. If an electric field is applied, electrons are being accelerated by the field, enhancing their kinetic energy and thus reaching higher energy levels. Electrons will move thru the crystal due to the electric field, resulting in an electric current. This scenario describes the situation within metals as shown in Fig. 1.2a. 2. At the absolute temperature T = 0K all lower energy bands are totally occupied. The occupied band with the highest energy level is called valence band. Normally there will exist a further energy band above the valence band, called conduction band. The energy difference between the highest possible energy state in the valence band and the lowest energy state in the conduction band is called the band gap ∆W of the crystal. If we have ∆W < 5eV one speaks of a semiconductor while for ∆W > 5eV we speak of an isolator. In Fig. 1.2 the band structure of the different materials is shown. Since the band structure shows the energy of the negative electrons inside a crystal the product of the electrostatic potential function Φe and the elementary charge e is up to an arbritray constant equal to the band energy. Thus the following relation holds true: 1 1 (1.3) Φe = − WL +C1 = − WV +C2 e e 1.2 Intrinsic conductivity The semiconductor silicon is a group IV element of the periodic table, thus it possesses four valence electrons and forms a face-centered diamond cubic crystal 4 W 6 0 W W 6 0 6 0 WL WL WL WV WV WV a) metal b) semiconductor c) isolator Figure 1.2: Band structure of a metal, semicondcutor and isolator structure. In the ideal crystal each atom forms covalent bondings with its four neighbours, as schematically illustrated in Fig. 1.3a, which shows a plane model of the crystal. At the absolute temperature T = 0K, all valence electrons are trapped in covalent bondings. Thus considering the band structure, the valence band is totally occupied, while the conduction band is totally empty as shown in Fig. 1.3b. Hence there do not exist free charges inside the crystal and it is an j j j j W 6 j j j j WC j j j j j j j j WV a) j Si-atom b) covalent bonding Figure 1.3: Plane model of Si-crystal and band structure at T = 0K isolator. If the temperature is enhanced the atoms of the crystal will perform a vibration around their mean location. With increasing temperature the thermal movement of the single atoms can become so strong that single covalent bondings will break. Now the valence electron will no longer be trapped to the bonding but can almost freely move within the crystal. This situation is sketched in Fig. 1.4a. In the picture of the band structure the thermal energy of the atom has moved an electron from the valence to the conduction band. If an electric field is applied to 5 j j j j j j u j j j j u j j ? j j ? j W 6 6 Si-atom u j electron u u WV e e E b) a) j WC e hole Figure 1.4: Plane model of Si-crystal and band structure at T > 0K the crystal the free electrons will move against the field direction. But not only the free electrons will move, the electrons trapped in a bonding will move too. Since one bonding electron is missing, other valence electrons may replace the missing electron resulting in a movement of the missing electron in the field direction. The missing electron thus behaves like a positive charge and is called a hole. The thermal induced breaking of a bonding thus results in the creation of an electron-hole pair in the picture of the band structure. Beside the thermal creation of electron-hole pairs there exists a process called recombination. In this process a free electron will be trapped again in a covalent bonding, which is equivalent to the annihilation of an electron-hole pair. In the thermal equilibrium both processes are in balance and for a given temperature we will have a certain density of electrons n and holes p in the crystal. To calculate their values one not only has to take into account the Fermi-Dirac distribution, but also the function D(W ) which describes the density of states in the crystal. If one approximates the Fermi-Dirac distribution by the Boltzmann distribution one finds the following equations describing the electron and hole density inside a crystal []: n = NC exp(− 2πm∗e kB T 3/2 WC −WF ) mit NC = 2( ) kB T h2 (1.4) 2πm∗p kB T 3/2 WF −WV ) mit NV = 2( ) (1.5) kB T h2 Where m∗e denotes the effective electron mass in the conduction band and m∗p the effective hole mass in the valence band. This correction has to be done to reflect the difference between a free particle and an almost free particle in the p = NV exp(− 6 periodic potential inside a cyrstal. The values NC and NV are called effective density of states in the conduction band respectively valence band. Table 1.1 gives some examples for the effective masses of electrons and holes for different semiconductors. As already discussed earlier, the electron and hole density are Semiconductor m∗e /me Si 0,33 Ge 0,22 GaAs 0,067 InP 0,078 m∗p /me 0,56 0,33 0,48 0,64 Table 1.1: Effective masses of electrons and holes for different semiconductors [] equal in an ideal semiconductor. This opens the opportunity to define the so-called intrinsic charge density ni of a semiconductor by: ni = √ np (1.6) With the help of the equations 1.4 and 1.5 and ∆W = WC −WV we find: ni = √ ∆W NL NV exp(− ) 2kB T (1.7) Example: Intrinsic charge density Germanium: ∆W = 0,63 eV, Silicon: ∆W = 1,14 eV, T = 300K ni Ge ≈ 1, 8 1013 1 3 cm 9 ni Si ≈ 2, 6 10 1 3 cm The examples show that we have a much lower intrinsic charge density in silicon at the same temperature, due to its larger band gap. If we expose the semiconductor to an electric field the electrons as well as the holes will move with different mean velocities thru the crystal lattice. This effect is described by the electron mobility µe and the hole mobility µ p . Table 1.2 gives the mobility of electrons and holes for different crystals. With the help of the mobility of electrons and holes and their densities one can formulate the law describing the conductivity of a semiconductor []. κ = e(µn n + µ p p) (1.8) Example: Intrinsic conductivity of germanium and silicon at T = 300 K 7 Crystal Electron Holes Si 1300 500 Ge 4500 3500 GaAs 8800 400 InSb 77000 750 InAs 33000 460 InP 4600 150 Table 1.2: Mobility of electrons and holes for different crystals in cm2 /Vs [] κi Ge ≈ 2.3 10−2 S/cm κi Si ≈ 7.5 10−7 S/cm For example copper at the same temperature has a conductivity of κCu ≈ 5.9 105 S/cm which is by a factor of 107 higher than the conductivity of geramium. 1.3 Doping The property of semiconductors that makes them most useful for constructing electronic devices is that their conductivity may easily be modified by introducing impurities into their crystal lattice. The process of adding controlled impurities to j z u j j W 6 j j j j WC j j z j j j j z u u donor level u h u h u h WF WV a) z donor atom u b) free electron Figure 1.5: Plane lattice and band structure of a n-condcutor a semiconductor is known as doping. The amount of impurity, or dopant, added to an intrinsic semiconductor can variegate its level of conductivity in a wide 8 range. Most useful doping materials are atoms of group 5 of the periodic table of elements like phosphor (P), arsenic (As) and antimony (Sb) and atoms of group 3 like boron (B), aluminium (AL) and indium (In). To clearyfy the influence of doping we will have a look on Fig.1.5. Again Fig. 1.5 shows a plane model of the Si lattice. But in contrast to an ideal Si lattice some of the Si atoms are replaced by atoms having five valence electrons. To build up the crystal lattice only four valence electrons are needed, thus the fifth electron is only weakly bounded to the impurity atom. So only very little thermal energy is needed to free the electron. In the picture of the band structure each impurity atom will contribute its fifth valence electron to the conduction band. If we use ND to denote the volume density of the donator atoms, this will resut in n ≈ ND Hence with the help of the donator atoms, we can influence the density of the free electrons in the semiconductor, which is according to equation 1.4 equivalent to a shift of the Fermi-level NC (1.9) WF ≈ WL − kB T ln( ) ND Since the product np = n2i only depends on the band gap of the semiconductor we have, n2 n2 p = i ≈ i n ND while the conductivity of the n-conductor is essentially given by κ ≈ eµn ND (1.10) In a n-doped semiconductor the elctrons are called majority carrier while the holes are called minority carrier. If we use doping atoms out of group 3 of the periodic table of elements, one valence electron is missing. Due to thermal vibrations this missing bonding can easily move from one atom to the other as shown in Fig. 1.6. But as already introduced, a missing bonding electron is called a hole in semiconductor theory. If we use NA to denote the volume density of the impurities, each so-called acceptor atom will contribute a free hole to the valence band and we have, p ≈ NA and with the help of equation 1.5 we can find the shift of the Fermi-level NV (1.11) WF ≈ WV + kB T ln( ) NA In a p-doped semiconductor the holes are the majority carriers while the electrons are the minority carriers. For the conductivity of a p-type semiconductor we find: κ ≈ eµ p NA 9 (1.12) j z j j W 6 j WC e j j j j j z j z j j j e e atom e h e h e WF WV b) a) z acceptor h e acceptor level free hole Figure 1.6: Plane lattice and band structure of a p-condcutor 1.4 Diffusion currents in semiconductors In contrast to a metal diffusion currents may play an important roll within semiconductors, due to the effect that there might exist electrons and holes within the same volume, forming an electric neutral carrier concentration. To explain the process of diffusion we have a look at Fig. 1.7. It shows a plane section of a crysr r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r r rr r r r r r r r x- Figure 1.7: Diffusion process within a crystal lattice tal lattice in which a uniform concentration drop in the x direction exists, which is represented by a different amount of particles within a certain region. If we assume that due to thermal motion one half of the particles moves to the right and the other half moves to the left, we get a net particle flow in the direction of the concentration drop. So, diffusion does not need external forces to act on a group of particles, but is just driven by their thermal energy. If we define with J pD (x) the one dimensional diffusion current density of the holes and with JnD (x) the diffusion current density of the electrons, the diffusion current is described by the following equations: J pD (x) = −eD p dp dx JnD (x) = eDn 10 dn dx (1.13) The positive sign in the equation of the electron current density reflects the defintion of the positive technical current direction, which is contrary to the movement of the electrons. The constants D p and Dn are called diffusion coefficients and they are related to the mobility of the carriers by Einsteins relation [?, ] Dp = µp kB T e Dn = µn kB T e (1.14) Thus the total current density of the holes J p and of the electrons in a semiconductor is composed of the drift current due to an electric field and the diffusion current. dp J p = eµ p pE − eD p (1.15) dx dn Jn = eµn nE + eDn (1.16) dx 11 Chapter 2 Semiconductor diode and applications 2.1 The pn-junction The simplest semiconductor component fabricated from both n-type and p-type material is the semiconductor diode, a two-terminal device which, ideally, permits conduction with one polarity of applied voltage and completely blocks conduction when the voltage is reversed. For the mathematical despriction of a pn-junction we will assume that changes in the crystal structure only occur in the x-direction while the structure is homogenous in the y- and z-direction. As a result all considered properties will only be functions of the x-coordinate. 2.1.1 pn-junction with zero bias To understand the physical behaviour of a pn-junction we will first consider the junction being separated by an ideal, fictive, infinite thin membrane as shown in Fig. 2.1. In the n-region will exist a huge amount of free electrons, moving arbitrarily thru that region due to their thermal energy. There will also exist the same amount of positive donator atoms being fixed in the crystal lattice. In the adjacend p-region we formally have the same situation but now the holes play the role of the electrons and the donators are replaced by fixed negative acceptor atoms. If we assume the fictive membrane to be removed, due to the difference in concentration, the free holes of the p-region will diffuse into the n-region, while the free electrons of the n-region will diffuse in the p-region and a recombination of electron-hole pairs will occur. As a result a transition region will be established between the pand n-region, were only the fixed acceptor and donator atoms exist but essentially no free carrier. As a further consequence an internal electric field will be built up, 12 e e e e e e a) e e e u e e u e e e e e e e e e b) e u u u u u free hole free electron u u E u u u e u u u u u u u u u u fixed acceptor fixed donator Figure 2.1: pn-junction with and without a fictive membran canceling the diffusion process of the free carriers and also resulting in a potential difference between the end faces of the crystal. This potential difference is called diffusion or build-in voltage UD and is given by the following equation: UD = Φe (∞) − Φe (−∞) (2.1) To calculate the hole distribution p(x) we use equation 1.15 and consider the fact that the diffusion process has stopped (JP = 0) and that one can deduce the electric field by the gradient of the potential function, which is related to the valence band energy WV (x) via equation 1.3: kB T dWV (x) d p(x) = p(x) dx dx The last differential equation can be solved by separation of the variables, while the neccessary constant can be deduced from the boundary condition p(x → −∞) = NA . Thus we get for the distribution of the holes: ) ( WV (−∞) −WV (x) (2.2) p(x) = NA exp − kB T In an analog manner we get for the distribution of the electrons using the boundary condition n(x → ∞) = ND : ( ) WC (x) −WL (∞) n(x) = ND exp − (2.3) kB T 13 To get a unique relation between the potential function Φe (x) and the band energies one uses the condition Φe (x → −∞) = 0. As a result we get the following relation between the potential function and the band energies of the valence and the conduction band: 1 1 Φe (x) = − [WV (x) −WV (−∞)] = − [WC (x) −WC (−∞)] e e (2.4) With the help of the last equations the hole and the electron distribution may be expressed by the potential function and the diffusion voltage: ( ) Φe (x) p(x) = NA exp −e k T B ( ) (2.5) UD − Φe (x) n(x) = ND exp −e k T B The last two equations in combination with equation 1.7, may be used to deduce an expression for the diffusion voltage without knowledge of the potential function Φe (x): NA ND kB T ln( 2 ) (2.6) UD = e ni Example: Diffusion voltage of a pn-junction in silicon NA = ND = 1015 cm−3 , T = 300K ) ( (1015 )2 = 660 mV UD ≈ 25.9 mV ln (2.6 109 )2 According to equation 2.2 and 2.3 the decline of the electron and hole distribuρ(x) 6 −w p eNA eND - wn x Figure 2.2: Charge distribution of an abrupt pn-junction tion follows an exponential function. To calculate the potential function one can approximate the carrier distributions in the transition region by a step function 14 according Fig. 2.2. This approximation is called abrupt pn-junction [2] and considers a constant negative charge distribution NA in the region −w p < x < 0 and a constant positive charge distribution ND in the region 0 < x < wn . Since the pn-junction is electrically neutral, the following equation must hold true: ND wn = NA w p (2.7) To calculate the internal electric field und potential function of an abrupt pnjunction we use the one-dimensional divergence theorem of the electrical field, which results in the following differential equations for the electric field: dE e = − NA dx ε − wp < x < 0 for (2.8) dE e = ND for 0 < x < wn (2.9) dx ε The last equations can be integrated easily and one finds the following functional dependence taking into account that the electric field may only exist in the region −w p < x < wn E(x) = − eNA (x + w p ) ε for − wp < x < 0 (2.10) eND (wn − x) for 0 < x < wn (2.11) ε According to the above equations the value of the electric field will first fall linear reaching its negative maximum at x = 0 and then will rise also linear to reach zero again at x = wn . The negative sign of the electric field reflects the fact that it is directed in the negative x-direction, as already shown in Fig. 2.1b. The potential function can again be evaluated by integration, while the integration constants must be chosen to reflect the following boundary conditions Φe (−w p ) = 0 and Φe (x = 0− ) = Φe (x = 0+ ). E(x) = − Φe (x) = eNA (x + w p )2 2ε for − wp ≤ x ≤ 0 (2.12) eND (wn (w p + 2x) − x2 ) for 0 ≤ x ≤ wn (2.13) 2ε In Fig 2.3 the functional dependence of the electric field and the potential function of an abrupt pn-junction is shown. With the help of the last equation and equation 2.7 the values w p and wn of the depletion zone may be evaluated. Φe (x) = 15 −w p @ @ E(x) 6 wn - x @ @ @ −Emax Φe (x) 6 6 UD ? - x Figure 2.3: Electric field and potential function of an abrupt pn-junction without bias √ wp = √ wn = ND 2ε U e D NA (NA + ND ) NA 2ε U e D ND (NA + ND ) As a result we get for the total width of the depletion zone: √ 2ε NA + ND w = UD (2.14) e NA ND According to Fig. 2.3 the electric field reaches its highest absolute value Emax at the coordinate x = 0. Since the potential function is in the one-dimensional case the integral of the electric field, the easiest way to calculate its value is to evaluate the area under the graph of the function: 1 UD = (w p + wn )Emax 2 Using equation 2.14 we find for the maximum of the electric field: √ 2UD 2eUD NA ND Emax = = w ε NA + ND (2.15) Fig. 2.4 shows the energy band model of a pn-junction at zero bias. Due to the locally fixed acceptor and donator atoms an internal electric field is created within the depletion area, which results in a potential difference between the p- and nconductor called diffusion voltage or build-in voltage UD and in a band bending. 16 W 6 p-conductor PP P n-conductor u PPX z X PX P P P e e e e e e e e e e PP X yP XPeP PP P P ? eUD uuuuuuuuu 6 WC WF WV - −w p wn x Figure 2.4: Energy band model of a pn-junction at zero bias The influence of the electric field on thermally excited electrons can easily be illustrated with the help of the band bending. If a thermally excited electron tries to jump over the potential barrier it behaves like a sphere on a hill, which rolls to the bottom again. In contrast holes act quite different. They behave more like balloons in a water basin, they always bob up to the highest energy value in the valence band, as illustrated in Fig. 2.4. 2.1.2 pn-junction with bias With the help of the energy band diagrams shown in Fig. 2.5 in a first step we now want to discuss qualtively the operation principles of a pn-junction, if a bias is applied. According to Fig. 2.5 a bias voltage is applied to the pn-junction with a direction opposed to the internal electric field. Hence it will lower the potential barrier between the p- and n-conductor. Due to their thermal energy now electrons of the n-conductor as well as holes of the p-conductor are able to surmount the potential barrier and will diffuse into the p- as well as into the n-conductor. Being minority carrier in these regions they will recombine and as a result a current will flow in the direction of the applied voltage. If we change the direction of the applied voltage, the internal electric field will be enhanced, resulting in an enhanced potential barrier. As a result the thermal energy neither of the holes nor of the electrons is high enough to surmount the barrier. So, in principle no carrier exchange between the two regions of the pn-junction will take place. Only due to the intrinsic conductivity there will be a small amount of reverse current flow. Analysis After the qualitative discussion of the operation principles we will now describe the process taking place in more mathematical depth. To deduce the mathematical 17 U HH s s W 6 e(UD −U) uuu X ? XXX Xy "" X XXX XXu u u u u u u u u u X WC 6 6 WF ? e - eeeeeeeX eX XX XX XX XX X zX e e e WV a) - −w p W HH x wn 6 6 H HHu e(UD −U) HH j H HH u u u u u u u u u u u ? H hhh hhhh hh e e e e e e e e e eH HH He H YH HH H HH b) −w p WC WF WV - wn x Figure 2.5: Energy band model a) forward bias b) reverse bias description we will use the following basic assumptions: • The voltage drop along the regions of the p- and n-conductor is neglected and it is assumed that it only takes place along the depletion zone of the pn-junction. • The current due to the minority carrier can solely be described as diffusion current. • In the depletion zone no recombination takes place. As a result the total current thru the diode can be described as the diffusion current of the minority carrier at the boundaries of the depletion zone at each side of the pn-junction. We will start our analysis by considering the density of holes p(x) in the nconductor. The concentration of the electrons n(x) in the p-conductor can be 18 deduced in an equivalent way. Starting from equation 2.5 we get for the hole density at x = wn in dependence of the applied voltage U: p(+wn ) = NA exp( −e(UD −U) eU ) = pno exp( ) kB T kT (2.16) In the last expression the constant pno denotes the hole density in the undisturbed n-region (x → ∞). According to equation 2.16 the hole concentration will rise with U > 0 and decay for U > 0. To calculate the hole distribution in the n-region we use the rate equation ??, extended by the divergence term of the currents [2]: ∂p 1 ∂J p p − pno = − − ∂t e ∂x τ In the stationary case ( ∂ = 0) this expression reduces to: ∂t dJ p p − pno = −e (2.17) dx τ According to our assumption the current J p is solely a diffusion current due to the minority carrier and we get for the region wn < x < ∞ the following differential equation: 1 1 d2 p = (p − p ) = (p − pno ) (2.18) no D pτ dx2 L2p √ In the last term the constant L p = D p τ was introduced, it posseses the dimension of a length and hence denotes the mean length along which a minority carrier can diffuse in its lifetime τ before it will recombinate. The solution of the above differential equation has to reflect that for x = wn the hole density is given by equation 2.16 and hence we find as solution: p(x) = (p(wn ) − pno ) exp(− x − wn ) + pno Lp (2.19) According to equation 2.19 the denisty of the minority carrier in the n-region is governed by an decaying exponential function. Using equation 1.15 we find the following diffusion current density at x = wn : ( ) Dp eU J p (wn ) = e pno exp( )−1 (2.20) Lp kB T In an equivalent way one can also deduce an expression for the diffusion current Jn (w p ) and since we assume that there will be no recombination in the depletion zone we get for the total current thru a pn-junction: ( ( ) ) Dp eU Dn 2 J = Js exp( )−1 with Js = e ni + (2.21) kB T L p ND Ln NA 19 According to equation 2.21 the current density thru the pn-junction will rise exponentially for positive voltages U, while it will decay for negative values. In the limit it will reach a value of Js , hence this value is called reverse biased saturation current density. If we multiply the current density with the area A pn of the pn-junction we get the static I-U characteristic of an ideal pn-junction. ( ) U kB T I = Is exp( ) − 1 with Is = A pn Js and UT = (2.22) UT e In equation 2.22 the constant UT was introduced, which is called thermal voltage. At room temperature (T = 300 K) it shows a vaule of approximately 26 mV. Since the voltage drop along the p- and c-conductor was neglected, equation 2.22 is only valid for small currents. The ohmic behaviour for these regions can in a first step be approximated by a resistor Rs . As a result the ideal pn-junction is only Figure 2.6: Static I-U-characteristic of an ideal pn-junction with Is = 10nA controlled by the reduced voltage U − Rs I. To demonstrate the influence of this resistor in Fig. 2.6 the static I-U-characteristic of an ideal pn-junction with Is = 10nA and of the same diode with Rs = 1Ω is shown. 20 2.1.3 Small-signal model of a pn-junction The equations deduced in the preceding section describe the behaviour of a pnjunction only for almost static time functions. To get an idea of its dynamic behaviour it is useful to study small signal exitation at a given operation point. In principle we will study the circuit given in Fig. 2.7, where a DC current source I is used to setup a certain operation point and a sinusoidal current source i(t) is used to realize the small signal exitation. r I 6 i(t) 6 A D r Figure 2.7: Small signal exitation of a pn-junction Dynamic resistance rD According to Fig. 2.7 we assume the pn-junction to be operated in a given operation point (I, U). Due to the sinusoidal current source with amplitude Iˆ there will also exist a sinusoidal voltage across the pn-junction with amplitude Û. One speaks of small signal exitation as long as the following relations hold true: Iˆ ≪ I and Û ≪ U For a first order approximation, we will describe the current voltage characteristic by its slope at the operation point. Hence for the amplitudes of the sinusoidal time functions the following relation holds true: Û ≈ 1 ˆ dU ˆ I = I = rD Iˆ dI dI dU In the last equation the dynamic resistance rD of a pn-junction was introduced. Assuming the pn-junction is forward biased, we get the following expression for the dynamic resistance using equation 2.22: rD = 21 UT I (2.23) Example: Dynamic resistance of a pn-junction at an operation point of I = 10 mA. According to equation 2.23 we find rD = 25, 9 mV ≈ 2, 6 Ω 10 mA Diffusion capacitance cD As already discussed in section 2.1.2 a forward biased pn-junction will store minority carrier in the n- as well as in the p-region. So each change in voltage ∆u at a given operation point will also result in a change for stored minority carrier. To calculate the stored minority carrier in the n-region we use equation 2.19 and perform an integration over the n-region: ) ∫ ∞( x − wn Q(U) = eA pn ) dx (p(wn ) − pno ) exp(− Lp wn For a differential change of the applied voltage ∆u we can write: ∆q ≈ eAD L p pno U dQ(U) exp( ) ∆u ∆u = dU UT UT (2.24) Equation 2.24 can be used to define the diffusion capacitance cD of a pn-junction. cD = eAD L p pno U τ τ ∆q = exp( ) = I = ∆u UT UT UT rD (2.25) Example: Diffusion capacitance of pn-junction at an operation point of I = 1 mA. In silicon diodes the minority carriers have a lifetime of τ ≈ 2.5 10−3 s cD = 2, 5 ms 1mA ≈ 97 µF 25, 9 mV According to the last example, the diffusion capacitance shows relatively high values. Since the dynamic resistance and the diffusion capacitance are essentially connected in parallel, the storage of the minority carrier in the p- and n-regions inhibits the technical usage of the dynamic resistance at higher frequencies of an ordinary pn-junction diode, since it is short circuited by the capacitance. 22 Junction capacitance cJ To deduce an expression for the junction capacitance we have a look at Fig. 2.3, which shows the electric field distribution inside the depletion zone of the pnjunction. If the applied voltage is changed with time also the electric field will change, resulting in a displacement current density. To find an expression of its value we start with equation 2.15 and assume that the total voltage u pn (t) is given by the sum of a DC voltage Uo and a time varying voltage ∆u(t). Hence we get for the electric field in the pn-junction √ 2e NA ND 1 E(t, x = 0) = − (UD −Uo ) (1 − ∆u(t)) (2.26) ε NA + ND 2(UD −Uo ) and for the displacement current density √ dE(t, x = 0) NA ND 1 d∆u(t) Jv = ε = eε dt NA + ND 2(UD −Uo ) dt (2.27) Since we assume a homogenous distribution across the cross-section of the pnjunction, the total displacement current can be calculated by multiplication with the area A pn of the pn-junction. According to the definition of a capacitance the factor in front of the time differential of the voltage must be the expression for the junction capacitance. √ eε NA ND (2.28) cJ = A pn NA + ND 2(UD −Uo ) To describe the small signal frequency response of a real semiconductor diode in s s LS CP RS s rD cJ s s s Figure 2.8: Small signal equivalent circuit of a real semiconductor diode Fig. 2.8 its equivalent circuit is given. Besides the elements already discussed two further elements are included. This is a series inductance LS accounting for wire bonds and a parallel capacitance CP reflecting the influence of the packaging. 23 2.1.4 Spice model of a semiconductor diode In the preceding section we discussed the behaviour of an ideal pn-junction. As an electric two terminal device it is called semiconductor diode. Since all electronic devices exhibit strong nonlinearities the behaviour of an electronic circuit can only be analysed by using sophisticated simulation tools. Most of todays commercial available tools are based on a simulator called SPICE Simulation programm with Integrated Circuit Emphasis which was developed at the University of Berkley []. Even though we already discussed several effects and parameters of an ideal pnjunction a real diode needs even more parameters to describe its real behaviour. In the following section we will give a short introduction to the equation used to describe a real diode in the Spice simulation tool, while the denotation of the parameters a summarised in Table ?? at the end of this section. Fig. 2.9 shows the equivalent circuit that is used in SPICE. The total time dependent current iD (t) s ?iD RS uD s ?ID A ? s s CJ CD s s Figure 2.9: Spice model of a semiconductor diode thru the diode is calculated using the following equation: iD = ID + CD duD duD + CJ dt dt (2.29) Static diode current ID Forward biased, the static diode current ID is equal to the current of an ideal pnjunction already given in equation 2.22, but with a further parameter N included, called emission coefficient. [ ( ] ) uD IDi = IS(T ) exp −1 (2.30) N ·UT 24 Here the temperature dependence of the saturation current IS(T ) is given by the following expression: ( T IS(T ) = IS · T0 )(XT I/N) ( EG(1 − T0 /T ) · exp N kB T0 ) (2.31) If a diode is reverse biased, experiments show that the real reverse current is higher than that predicted by equation 2.30. To account for this effect an additional current IDc of a so-called correction diode is added: IDc [ ( = ISR exp uD NR ·UT ) ] [( ]M/2 uD )2 −1 · 1− + 0, 005 VJ (2.32) If the reverse voltage of the diode is further enhanced reverse breakdown occurs which is modeled by an exponential function: ) ( −uD − BV (2.33) ID = −IBV exp NBV ·UT Dynamic diode current To account for the dynamic behaviour of a real diode expressions for the junction capacitance and diffusion capacitance have to be considered. According to equation 2.28 the junction capacitance varies proportional to the square root of the applied reverse voltage. For a real diode this expression is slightly modified ( uD )−M CJ = CJO 1 − VJ (2.34) If a diode is forward biased the lifetime of the miniority carrier of the junction has to be considered. In its implementation SPICE uses also equation 2.25 already discussed earlier. diD TT = (2.35) CD = T T duD rD In the following table the essential SPICE parameters used to specify a real diode are summarized IS N ISR NR BV IBV saturation current emission coefficient saturation current of correction diode emission coefficient of ISR reverse breakdown voltage current at break-down voltage 25 NBV RS TT CJ0 VJ M FC EG XT I KF AF coefficient of IBV series resistance minority carrier life time zero-bias junction capacitance junction potential grading coefficient coefficient for forward-bias depletion capacitance formula activation energy temperature exponent of IS flicker noise coefficient flicker noise exponent To include different diodes into SPICE ordinary ASCII-files are used as shown in the following example. Example: SPICE diode data sets *----------------------------------------------------------.MODEL BAT68 D(IS=8N RS=2 N=1.05 XTI=1.8 EG=.68 + CJO=.77P M=.075 VJ=.1 FC=.5 BV=8 IBV=1U TT=25P) *----------------------------------------------------------.MODEL BA592 D (IS=185F RS=.15 N=1.305 BV=70 IBV=.1N + CJO=1.17P VJ=.12 M=.096 TT=125N) *----------------------------------------------------------.MODEL BAS116 D( + AF= 1.00E+00 BV= 7.50E+01 CJO= 1.83E-12 EG= 1.11E+00 + FC= 5.00E-01 IBV= 1.00E-04 IS= 1.48E-13 KF= 0.00E+00 + M= 2.62E-01 N= 1.33E+00 RS= 8.48E-01 TT= 8.66E-09 + VJ= 3.44E-01 XTI= 3.00E+00) *----------------------------------------------------------- 2.1.5 Different types of semiconductor diodes There were developed different types of junction diodes by emphasizing different physical aspects for example by geometric scaling, by changing doping levels or by the use of different semiductor materials. In the following section we will give a short overview of the diodes most often used in electronics. 26 Zener diodes The ordinary junction diode will be destroyed, if a reverse voltage is applied, extending their maximum reverse voltage and breakdown occurs. Zener diodes in this sense are special diodes that will not be destroyed when the breakdown occurs. Furthermore it is possible to controll the breakdown voltage or Zener voltage of the diode very precisley. Fig. 2.10 shows the current voltage characteristic of an ideal Zener diode, which will be conducting as soon as the applied reverse voltID −UZ0 6 - UD Figure 2.10: I-U characteristic of a ideal Zener diode age exceeds the Zener voltage UZ0 . In practical applications these diodes are used to stabilize a voltage to a certain level. Schottky diode From a historical point of view not the pn-junction but the crystal detector was the first electronic device already used at the end of the 18th century. In principle it consists of thin sharpened metal wire pressed against a crystal, thus forming a metal to semiconductor contact. Today this kind of diode can also be constructed using semiconductor technology and is called Schottky diode. But in contrast to a pn-junction no minority carrier is essential for the nonlinear behaviour and they tend to show a much lower junction capacitance. Thus they can be used up to very high frequencies as mixers and detectors []. Varactor diodes As already discussed in section 2.1.3 if reverse biased, each junction diode shows a certain capacitive value, that depends on the applied reverse voltage. Furthermore the value of capacitance and its voltage dependence can be controlled using certain doping profiles. Thus varactor diodes can be used to replace a capacitor, with the advantage of being adjustable by an applied voltage. One of the main practical application are their use in voltage controlled oscillators. 27 Photo detector If a pn-junction is reverse biased only a small reverse current exists, due to thermal creation of electron hole pairs within the depletion region. But if the pn-junction is exposed to light and the photon energy is high enough to surmount the band gap energy of the semiconductor ∆Wg they can create electron hole pairs. This process is called absorption. If no external voltage is applied, the photodiode operates in the mode of a solar cell, converting optical into electrical energy. If the diode is reverse baised, it operates in the mode of a photo detector and can be used to sense light. In this case the reverse current, called photo current I ph , is proportional to the incident optical power Popt and the proportional constant is called responsivity Rsp of the photo detector. I ph = Rsp Popt (2.36) Light emitting diodes (LEDs) The fundamental physical principle LEDs are based on is called spontaneous emission []. If an electron of the conduction band recombines with a hole of the valence band, the energy may be emitted as photon of a certain wavelength or frequency, depending on the bandgap ∆Wg of the semiconductor. λ = C0 h ∆Wg f = ∆Wg h (2.37) But this process may only take place in certain semiconductors, called direct bandgap semiconductor. Unfortunatly silicon is no direct-band gap semiconductor. So more sophisticated materials like GaAs have to be used. All LEDs produce incoherent, narrow-band light. Laser diodes In a crude approximation a laser diode is a LED-like structure with an additional optical resonator, formed by the endfaces of the semiconductor crystal itself. Due to this resonator the bandwith of the light due to spontaneous emission is reduced and stimulated emission takes place resulting in light with a high coherence []. Laser diodes are commonly used in optical storage devices and for high speed optical communication. 28 2.2 Diode applications 2.2.1 Diode as rectifier In the previous sections we discussed intensively how to describe and model the electrical behaviour of a semiconductor diode. For the basic understanding of diode applications such as a rectifier circuit these models are even far to complicated. So here we will introduce the simplest possible model of a diode. From Fig. 2.6 we know that a semiconductor diode has a very strong nonlinear behaviour. Essentially there will be no current flow, if it is reverse biased, but arbitrarilly high currents if it is biased in the froward direction. In Fig. 2.11 the static I-UID 6 ideal diode→ ← diode with threshold voltage Uth - Uth UD Figure 2.11: Diode modeled as a voltage sensitive switch characteristic of an ideal diode is given. Essentially an ideal diode will behave like a voltage sensitive switch. If the voltage UD across the diode is negative the diode will show an infinite resistance, thus it behaves like an open switch. If on the other hand the voltage across the diode is positive, it shows a very low resistance or the switch is closed. Specially for discussion of the following basic diode applications this model is appropriate for their principle understanding. Especially, when dealing with small voltages, the model with a certain threshold voltage Uth can be used, also shown in Fig. 2.11. For normal silicon diodes the value of the threshold voltage lies in the range from 0.6 V to 0.7V. The slight differences in behaviour of real diodes can be examind using simulation tools. Almost in all electronic equipment DC voltages of different values are needed for their operation. Since the electric power distribution system uses AC voltages of 230 V nominal they usually have to be transformed to a lower level and converted to DC. This process is called rectification and in most practical application this is done with the help of semiconductor diodes. In the following sections we will discuss different circuits that are used for rectification. 29 Half-wave rectifier uD r uS H H - r RL ? uR ? r r Figure 2.12: Circuit schematic of a half-wave rectifier Fig. 2.12 shows the circuit schematic of a half-wave rectifier. It consists of an alternating source delivering a sinusoidal voltage uS (t), a diode and a load resistance RL . It should be noted that the nominal output voltage UN of a transformer is the effective value of the sinusoidal time √ function, so one always has to remember, that the amplitude Û is by a factor of 2 higher than the nominal value UN . To understand the behaviour of the circuit we introduce the voltage uD (t) across the diode and the voltage uR (t) across the load resistance. According to KVL the following equation holds true: −uS (t) + uD (t) + uR (t) = 0 Since the diode is essential for the operation of the circuit we first have a look on the voltage across the diode uD (t) = uS (t) − uR (t) = uS (t) − RL iD (t) (2.38) Starting with a positive half cycle all voltages are zero and so is the diode current iD (t). If now the source voltage becomes positive, the diode voltage becomes positive too and according to our model the diode will switch into its on state. As a result the source voltage will drop across the load and the voltage across the diode will essentially be zero. If now the negative half cycle will start, at first again the diode current will be zero and as a result the voltage across the diode will become negative. According to our model the diode will now switch into its off state. No current iD (t) will exist and thus there will be no voltage drop across the resistor, but the whole voltage of the source will drop across the diode. As an example Fig. 2.13 shows the time function across the resistor as result of a simulation with SPICE. The amplitude of the sinusoidal voltage source was chosen to be 5V, with a load resistance of 500Ω and the diode BA592. Essentially it shows the half-wave of the exiting voltage source, but there is a remarkable 30 Figure 2.13: Output voltage of a half-wave rectifier difference. While the model of an ideal diode would propose an amplitude of 5V, the simulation shows that there will be a voltage drop of about 0.8V across the diode at the peak voltage of the half cycle. For practical applications it is neccessary to choose an appropriate diode for the application. Thus one has to consider certain maximum ratings of a diode, which are usally given in their data sheet. Two crucial parameters are the maximum reverse voltage URmax and the maximum forward current IFmax . In case of a half-wave rectifier we must fulfill the following conditions: URmax > Û = √ 2UN and IFmax > Û RL (2.39) Of course the voltage shown in Fig. 2.13 is not yet a DC voltage but still a periodic time function, with a DC part given by the following equation. √ 2 Û = UN (2.40) UDC = π π To further smooth the ouput voltage a capacitor may be used as shown in Fig. 2.14. Fig. ?? shows the simulation results of the same half-wave rectifier where according to Fig. 2.14 a capacitor of 100µF was included for smoothing, also shown is the time function without a capacitor. So even if there seemed to be only 31 Figure 2.14: Half-wave rectifier with smoothing capacitor a little change in the circuit due to the capacitor there is an significant change in the maxium ratings the diode now has to withstand. At first we will have a look on the simple equation for the diode voltage 2.38. In the limit of high load resistances RL the maximum voltage will become nearly equal to the amplitude Û of the AC voltage. So, according to equation 2.38 the maximum reverse voltage may reach a value of 2Û, thus the following condition must be fulfilled, in case of a half-wave rectifier with smoothing capacitor. URmax > 2Û (2.41) But not only the diode must withstand a two times higher reverse voltage, but also the maximum possible forward current is significantly changed due to the capacitor. This is because at swichting time a capacitor behaves like a short circuit. Thus, if the rectifier is not switched on at a zero crossing, but at a certain positive voltage value of the alternating source, the maximum forward current is only limited by internal resistances and can reach fairly hight values. To circumvent this problem, it is sometimes neccessary to include a resistor in series to the diode to limit the maximum possible forward current. Even though the circuit of a halfwave rectifier is very simple, it is also very inefficient for power transfer, since only one half-cycle is used. 32 Full-wave rectifier The circuit that allows us to use every half-wave of a cycle is called full-wave rectifier. Fig. 2.15 shows its circuit schematic. To realise the two equal voltage sources, in pratice a transformer is used whose secondary winding is split into two with a center tap connected to the ground. In principle the upper and lower part r uS (t) H H r uD1 (t) uR (t) RL ? r ? uS (t) ? uD2 (t) r H H Figure 2.15: Circuit schematic of a full-wave rectifier of the circuit each work like a half-wave rectifier, but if the anode of diode one is positive, due to the grounding of the sources, the anode of diode two is negative and vice versa. Since now each half-wave will be rectified, we get for the DC part of the voltage: √ 2 Û UN (2.42) UDC = 2 = 2 π π while the maximum reverse voltage will reach a value of 2Û and thus the following condition must be fullfilled. URmax > 2Û (2.43) One disadvantage of this kind of full-wave rectifier is the costly transformer, due to its center tap. To over-come this a so-called bridge rectifier as shown in Fig. 2.16 may be used. With the help of this circuit the costly transformer is omitted by the expense of two further diodes. During the positive half cycle D1 and D4 will be conducting, while diodes D2 and D3 are reverse biased. Thus the current will flow in the direction of diode D1 thru the resistor RL . If the polarity of the cycle changes, now diodes D3 and D2 are conducting, while diodes D1 and D4 are reverse biased. Now the current will flow in the direction of D3 thru the resistor, but this direction is identical to that during the positive half cycle. Thus independent of the polarity of the half cycle, the current will always flow in the same direction thru the load resistance. Fig. 2.17 shows the simulation results, again using the 33 D2 H H uS (t) ? r D1 H H r r D4 H H r RL - D3 H H uR (t) Figure 2.16: Circuit schematic of a bridge rectifier diode BA592 and a 500Ω load resistance. Comparing the maximum amplitude, with the simulation given in Fig. 2.13 shows, that in the case of the bridge recticfier the peak voltage is further reduced, since in the rectification process two diodes are involved always. Of course as in the case of the half-wave rectifier, also in the case of the bridge rectifier a smoothing capacitor may be connected in parallel to the load resistance. The result using a cpacitor of 100µF parallel to the load resistance is also shown in Fig. 2.17. Series and parallel connection of diodes Under certain circumstances there may exist a neccessity to use diodes that for example cannot withstand the occuring reverse voltage or forward current. In the first case diodes can be connected in series to reach the necessary reverse voltage capability as shown in Fig. 2.18a, but with the help of two parallel resistors it must be assured that the voltage will drop equally across the diode to compensate for differences in their saturation currents. To enhance the forward current capability, two diodes may be connected in parallel, as shown in Fig.2.18b. But here series resistors have to be used to compensate for differences in current distribution. 2.2.2 Voltage multiplier Before we will discuss the circuit of a voltage multiplier according to Greinacher, we will again have a look on the simple circuit of a half-wave rectifier shown in Fig. 2.19 where the positions of the capacitor and diode are changed with respect to the ground and compared to the circuit of Fig. 2.14. Using KVL we get for the ouput voltage uo (t) of the circuit uo (t) = uS (t) + uC (t) 34 Figure 2.17: Output voltage of a brigde rectifier without and with smoothing capacitor Rs r r H H RP r r H H r r r Rp r Rs a) b) H H r r H H Figure 2.18: Combined diodes to enhance reverse voltage or forward current capability Fig. 2.20 shows the simulation result for the time function uo (t) according to the circuit of Fig. 2.19. As source voltage uS (t) a sinusoidal time function with a 5 V amplitude was used. Roughly spoken the output voltage uo (t) shows a maximum amplitude of approximately 10 V, which is two times the source amplitude, because the capacitor is charged to 5 V. Of course the output voltage may be used as an input of a further half-wave rectifier as shown in Fig. 2.21a. The principle of the voltage doubler shown in Fig. 2.21a was extended by Greinacher to reach even higher voltage levels by adding further stages, as shown in Fig 2.21b. In principle the voltages of the capacitors in the lower line will add up to the final voltage level U0 . The time function of a two stage voltage multiplier ist given in Fig. 2.22. According to our crude approximation, with two stages we should reach a voltage level of 20 V. As the simulation shows we only reach a value of 35 C uC (t) uS (t) A ? uo (t) ? r Figure 2.19: Half-wave rectifier with interchanged capacitor and diode Figure 2.20: Output voltage uo (t) approximately 17 V. If we would assume a voltage drop of approximately 0.7V across each diode, this would sum up to a value of 2.8V, which may essentially explain the difference. 2.2.3 Zener diode as voltage regulator In the circuit shown in Fig. 2.23 a zener diode is used to stabilize the output voltage U0 to the Zener voltage of the diode. To describe the performance of a Zener diode usually the current and voltage directions given in Fig. 2.23a are used. These results in the I-U characteristic of a Zener diode given in Fig. 2.23b. In contrast to the very sophisticated models that can be used with SPICE, we will 36 r uS (t) ? H H r r r uS (t) Uo A r ? r ? r r r r r A A A A r r r r r U0 Figure 2.21: Voltage doubler and multiplier circuits Figure 2.22: Output voltage of a two stage voltage multiplier restrict our considerations to idealized Zener diodes. As shown in Fig 2.23b we will describe the diode by its Zener voltage UZ0 and a resistance rZ , which will become zero in the limit of an ideal Zener diode. Of course the circuit of Fig. 2.23 is only able to stabilize the output voltage to the Zener voltage as long as the relation Ui > UZ0 holds true. One crucial parameter of a Zener diode is its maximum possible dissipation power Pmax , which will limit the maximum current IZmax thru the diode and we have: Izmax ≈ Pmax UZ0 (2.44) But for proper operation at least a certain minimum current IZmin must flow thru the diode. To deduce an expression for the series resistor we use the loop equation 37 r Ui Rs r ?IZ A ? r IZ 6 Io r r - UZ ? Uo ? r UZ0 - UZ Figure 2.23: Simple circuit to regulate the ouput voltage of the circuit and solve it for the resistor: Rs = Ui − Uo Io + IZ (2.45) In the practical operation of the circuit there are two extreme cases possible: • The input voltage reaches its minimum value Uimin while the maximum output current Iomax is drawn. Under these circumstances it has to be sure that IZ must not fall below IZmin , thus resulting in an upper limit for the series resistor. Uimin − Uo Rs < (2.46) Iomax + IZmin • The input voltage reaches its maximum value Uimax while only a minimal value of output current is drawn Iomin . Under these circumstances it has to be sure that IZ must not exceed its maximum value IZmax , thus defining a lower limit for the series resistor. Rs > Uimax − Uo Iomin + IZmax (2.47) Only if the two inequalities are both fulfilled, the circuit according to Fig. 2.23 is realisable with the chosen Zener diode. To compare different circuits to stabilize the output voltage we define the following stability factor S S = dUi /Ui ∆Ui /Ui ≈ ∆Uo /Uo dUo /Uo (2.48) In principle the last form of equation 2.48 allows us to deduce expressions for the stability factor using small signal approximations. Of course, if we would assume an ideal Zener diode with rZ = 0, the stability factor S would become infinte since a variation of the input voltage would not result in a variation of the ouput voltage 38 at all. If we now, in a first order approximation consider the Zener diode to have a non zero rZ , a change in the input voltage Ui will also result in a change of the output voltage Uo . According to the circuit schematic of Fig. 2.23a the following equations are valid: Ui = Rs I + rZ IZ + UZ0 and Uo = rZ IZ + UZ0 If there is a change in the input voltage dUi there will also be a change in the current I and the current IZ , so we have, dUi = Rs dI + rZ dIZ and there will also be a change in the output voltage dUo = rZ dIZ So we find for the ratio dUi /dUo : Rs dI + rZ dIZ Rs dI dUi = ≈ dUo rz dIZ rZ dIZ for Rs ≫ rZ In a first order approximation we can neglect a current change due to the change of the output voltage and we have dI = dIz and we get: dUi Rs = dUo rz So we get as final result for the stability factor of the circuit according to Fig. 2.23a: Rs Uo (2.49) S ≈ rz Ui Example: The current thru a load may vary between 0 mA and 100 mA, while the voltage should be kept stable at 15 V and the input voltage may vary between 27 V and 33 V. A diode with IZmax = 200 mA and IZmin = 20 mA is used. Find the value of Rs and the stability factor. According to the equations 2.46 and 2.47 we get for the series resistance the following relations, Rs < 100Ω and Rs > 90Ω so the ratings of the Zener diode are sufficient and the series resistor may be chosen to be RS = 95Ω. From the data sheet of the Zener diode one finds the maximum dynamic resistance rZ to be 7Ω, so we get for the stability factor S = 95 15 ≈ 6.8 7 30 39 Stabilization of low voltages Usually Zener diodes are built for breakdown voltages above 3 V. So, if one has to stabilize an output voltage below this value one has to use an other circuit. One possible simple circuit is shwon in Fig. 2.24. Here the series connection of diodes Rs r r r A Ui Uo ? r A r ? r Figure 2.24: Circuit to stabilize low voltages is used to stabilize the output voltage. Roughly spoken each diode needs a voltage of approximately 0.6 V to become conducting. So the output voltage is a multiple of this value. 40 Chapter 3 The bipolar junction transistor and applications 3.1 The bipolar junction transistor We will now discuss the bipolar junction transistor (BJT), which started the age of electronics. Since its invention in 1948 a lot of different electronic devices have been realized capable to amplify weak electric signals. Even though today the most commonly used transistor is the field effect transistor, we will start our discussion with the BJT since its operation principles are based on the the behaviour of a pn-junction, we already discussed. 3.1.1 Structure and operation principles of a npn BJT Fig. 3.1a shows the simplified physical layout and 3.1b the circuit schematic of a npn BJT. It consists of a highly n-doped conductor called emitter (E), followed by a thin p-doped zone, called base. The adjanced zone is called collector, which is again formed by a n-doped conductor. In Fig. 3.1c an example of a cross sectional view of a npn-BJT is given, which is realized with the help of SBCtechnique (Standard Buried Collector ) [?], [?] inside an integrated circuit. The realisation process starts with weak p-conducting silicon crystal. With the help of gas phase epitaxy a weakly doped n-conductting layer is formed, realizing the collector (NDC ≈ 1015 cm−3 ). With the help of the p-zones on both sides the single transistor is isolated to the adjanced ones. With the help of an oxidation process a silicon oxid layer is formed, in which a window defining the base is etched. In the following diffusion process the base is formed using Bor atoms with a concentration of approximately NAB ≈ 1017 cm−3 . In a further oxidation and etching process the window for the emitter is formed and finally with the help 41 a) E s N P N s C s E s p+ B b) E s sC @ @ I @ B s Bs + n % ++ n p sC n + p & p-Silizium c) Figure 3.1: a) simplified physical layout, b) circuit schematic and c) cross sectional view of a npn-BJT of a diffusion process a donator concentration of approximately NDE ≈ 1022 cm−3 is realized in the emitter zone, leaving a thin p-conduction layer, which forms the base of the transistor. To discuss the principle of operation of a BJT we have a look on Fig. 3.2. In the upper part a simple cross sectional view of the different layers of the npn BJT ist given. Since the volatge UBE > 0 the E-B junction is forward biased and since the voltage UBC < 0 the B-C junction is reverse biased. Also sketched are widths of the depletion zone of the two junctions. Since the emitter is highly doped the depletion zone of the E-B junction extends wider into the base and since the base is normally higher doped than the collector, here the depletion zone extends wider into the collector. In the lower part of Fig. 3.2 the energy band diagram under typical basing conditions is shown. Under these conditions the emitter-base-diode is forward biased and thermally excited electrons are able to surmount the potential barrier to the base, in which they will diffuse. Since they are minority carriers some of them will recombine and result in a base current. But if the diffusion length is much longer than the thickness of the base, the majority of electrons entering the base from the emitter will diffuse thru the base and enter the depletion zone between base and collector. Since this diode is based in reverse direction there will exist an electric field, accelerating the electrons into the collector. Hence creation of an emitter base current will result also in an emitter collector current. Thus with the current thru the emitter base diode the current from the emitter to the collector may be controlled. This is essentially the principle of operation of a npn BJT. To reach this state of operation the following conditions must be met: • The current thru the emitter base diode must be essentially an electron current. According to equation 2.21 this is only valid for highly doped emitters. • The majority of electrons entering the base are only capable to reach the 42 B UBE > 0 UBC < 0 s ? s ? s E C W * u- u u u u u u PP PP P 6 uuuu Q c Qs cQ c H ? ee c c HH H c c c cc u u u u u - WC HH WF cc WV - x Figure 3.2: npn BJT forward baised E-B junction reversed baised B-C junction collector if the diffusion length Ln inside the base is longer than the base thickness dB . • The reverse current of the base collector diode has to be negligible small. In Fig. 3.3 the current distribution inside a BJT is shown qualitatively. The directions of the currents IE , IB and IC where chosen to give the technical current directions, which is opposed to the movement of the electrons. The thinner arrows denote the unwanted hole currents between the emitter and the base as well as the reverse current of the base collector diode. As a result of Fig. 3.3 it is clear that the collector current is proportional to the emitter current. IC = α0 IE (3.1) The parameter α0 of the last equation is called static current gain in a common base circuit, despite the fact that due to the recombination of electrons in the base its value is always lower than one (αo ≈ 0,9 · · · 0,999). Since the BJT is a node 43 IE @ @ s E B C - - IC s s 6IB Figure 3.3: Current distribution in a npn BJT under typical operation conditions we can apply Kirschhoffs current law: IE = IB + IC If we use the last equation to give the collector current as function of the base current we will get: α0 IB = β0 IB (3.2) IC = 1 − α0 The parameter of equation 3.2 is denoted as static current gain in a common emitter circuit. Depending on the transistor β0 can reach values between approximately 30 and 500. 3.1.2 Static input and output characteristics of a BJT According to the arragement of the layers a transistor can be represented by two diodes which are connected at their p-layer. Such a circuit would of course not act as a transistor because the anode of the emitter base diode is also the anode of the base collector diode in a physical sense, but not only in an electrcical sense as modeled by the equivalent circuit. To account for the transistor effect, according to [?], a current controlled current source has to be included parallel to the base collector diode, which represent the electron current from the emitter to the collector of a real transistor. As a result we get the equivalent circuit of a transistor given in Fig. 3.4 under typicall operation condictions, describing its static behaviuor. The currents ISE and ICE representing the saturation currents of the emitter base and base collector diode, while the resistances of the semiconductor layers are neglected. ) ( UEB )−1 (3.3) IE = ISE exp(− UT 44 IE H H s s H H s αo IE IC s s UEB UCB ? s ? s s Figure 3.4: Simplified equivalent circuit according to Ebers-Moll under typical operation conditions ( ) UCB IC = αo IE − ISC exp(− )−1 (3.4) UT As shown in Fig. 3.5 the equivalent circuit of Fig. 3.4 can also be given in a common emitter configuration. With the help of equations 3.3 and 3.4 we will H H s s # # I s B- @ @ R UBE ? s s UCE ? s UBE ? s IC s s α0 IE s A I ?E s UCE ? s Figure 3.5: BJT in common emitter configuration and its Ebers-Moll-model now deduce an expression for the input characteristic IB = f (UBE ,UCE ) and for the output characteristic IC = f (UCE ,UBE ). Staring point is again KCL for the BJT as a whole: ) ( UCB )−1 IB = IE − IC = (1 − α0 )IE + ISC exp(− UT Accounting the different definitions of the voltages UEB = −UBE and UCB = UCE −UBE we get for the base current: ( ) ( ) UBE UCE −UBE IB = (1 − α0 )ISE exp( ) − 1 + ISC exp(− )−1 UT UT For typical conditions of operation we have UCE ≫ UBE and the last term may be neglected, resulting in the following expression for the base current: ) ) ( ( UBE UBE ) − 1 = ISB exp( )−1 (3.5) IB = (1 − α0 )ISE exp( UT UT 45 The form of the last equation is equivalent to that of a normal pn-junction, which has a saturation current of ISB = (1 − α0 )ISE . Thus the input characteristic of a transistor is equivalent to that of a pn-junction already shown in Fig. 2.6. With the help of equation 3.4 we get as expression for the collector current IC , ( ) ) ( UBE UCE −UBE IC = α0 ISE exp( ) − 1 − ISC exp(− )−1 (3.6) UT UT or with reference to the saturation current ISB ( ) ( ) UBE UCE −UBE α0 ISB exp( ) − 1 − ISC exp(− )−1 IC = 1 − α0 UT UT (3.7) Example: Theoretical output characteristic of a BJT As example Fig. 3.6 shows the output characteristic of a BJT according to equa- Figure 3.6: Theoretical output characteristic of a BJT with UBE as parameter tion 3.6, where the following values were used for its calculation: ISE = ISC = 1 nA, UT = 43 mV, α0 = 0, 999. Comparing the theoretical predicted output characteristic shown if Fig.3.6 with that of a real BJT shows that the current IC of a real BJT rises with rising voltage UCE . Is effect is called Early-effect [?]. The first term of equation 3.7 can be identified as current gain β0 of the common emitter configuration: α0 (3.8) β0 = 1 − α0 46 while the second term describes nothing else but the dependence of the base current IB on the voltage UBE . ( ) UBE IB (UBE ) = ISB exp( )−1 (3.9) UT With the help of the introduced parameters equation 3.7 can be rewritten as ( ) UCE −UBE IC = β0 IB (UBE ) − ISC exp(− )−1 (3.10) UT Since under normal operation conditions the base collector diode is based in reverse direction, the last term of equation 3.10 can be neglected, which reduces this equation to IC ≈ β0 IB (UBE ) (3.11) describing essentially the behaviour of a BJT in a common emitter configuration. Equation 3.9 and equation 3.11 can be used to setup a large signal model of a BJT operating in common emitter configuration, as shown in Fig. 3.7 It consists of the β 0 IB r C B r -IB A r rE Figure 3.7: Large signal model of a BJT under normal operation conditions base emitter diode carrying the current IB and an ideal current controlled current source being responsible for the collector current. 3.1.3 Simple small signal BJT model One of the applications of transistors is the amplification of small signals. To use a transistor for amplification one has to operate the transistor under certain DC conditions UCE , IC , called biasing. In principle one uses voltage or current sources to establish the DC conditions under which the transistor shows the described transistor effect. Fig. 3.8 shows a BJT circuit in common emitter configuration. The DC voltage sources UBE and UCE are chosen to establish the typical operation conditions of the transistor, emitter base diode forward biased and base collector diode biased in reverse direction. In the input circuit an additional sinusoidal 47 voltage source uBE (t) is used, with an amplitude ÛBE fulfilling the small signal condition ÛBE ≪ UBE . In the ouput circuit a load resistance RL is included to allow an alternating voltage uCE (t) to exist, also fulfilling the small signal condition ÛCE ≪ UCE . A mathematical exact analysis of the circuit shown in Fig. 3.8 is us (t) UBE RL @ R @ ? ? r U0 ? Figure 3.8: Common emitter circuit of a BJT with small signal excitation very complicated due to the nonlinear behaviour of the equations 3.9 and 3.10 and only possible using advanced simulation tools. Since we are at the moment only interested in the small signal behaviour we linearize equations 3.9 and 3.11 in the vicinity of the point of operation and thus establish a small signal equivalent circuit of the BJT at the point of operation. In mathematical sense we will perform a Taylor approximation at the opertaion point. ∆IB = UBE | IBE | ISB exp(− )∆UBE ≈ ∆UBE UT UT UT (3.12) In analogy to the pn-junction one can introduce the dynamic resistance rBE of the base emitter diode. UT rBE = (3.13) | IB | Also linearizing equation 3.11 yields the equivalent circuit shown in Fig. 3.9, where the resistor rCE was additionaly included to account for the Early effect, already mentioned above. Since the equivalent circuit was deduced from equations 3.9 and 3.11 describing the static behaviour of the transistor, it is only valid for low frequencies. h-parameter To describe the small signal behaviour of BJT at low frequency also the h-parameters are used []. In this case these parameters are real and they describe the influence 48 βiB iB uBE (t) ? r - rBE r r r r r CE r r RL uCE (t) ? r Figure 3.9: Simple common emitter small signal equivalent circuit of a BJT of the output voltage uCE (t) and input current iB (t) on the input voltage uBE (t) and the output current in a more formalized way given by the following equation: uBE = hie iB + hre uCE (3.14) iC = h f e iB + hoe uCE From equation 3.14 it becomes clear that the single parameters are defined according to the following equations: hie = uiBE |uCE =0 B BE | hre = uuCE iB =0 i h f e = iC |uCE =0 B C | hoe = uiCE iB =0 short circuit input resistance open circuit reverse voltage ratio (3.15) short circuit forward current gain open circuit output conductance Since the first equation of 3.14 defines the small signal voltage uBE it may be interpreted to be the result of Kirchhoff’s voltage law at the input of the transistor while iC of the second equation is the result of Kirchhoff’s current law at its output. Using these interpretations one finds the equivalent circuit shown in Fig. 3.10. Comparing the equivalent circuit of Fig. 3.10 with that already given in Fig. 3.9 iB hie h f e iB s uBE ? s iC ? s s hoe hre uCE ? s s uCE s ? s Figure 3.10: Small signal equivalent circuit according to the h-parameters reveals the following equivalences: hie = rBE hfe = β 49 hoe = 1/rCE (3.16) On the other hand, the parameter hre finds no equivalent element in Fig. 3.9, because it was deduced from the static behaviour and the parameter hre decribes the feeback of an alternating voltage at the output on the input voltage, which of course cannot be deduced using static equations. Nevertheless we will use the parameters given in Fig. 3.9 for a first order analysis of small signal amplifier, because of their physical significance. If a more detailed analysis is needed this can today be done using advanced simulation tools. Nevertheless h-parameters are often specified in data sheets of single transistors for a certain operation point. 3.1.4 Advanced small signal BJT model If we want to have a more accurate equivalent circuit describing the behaviour also up to higher frequencies, we have at least to account for the capacitve behaviour of all encountered pn-junctions. According to Fig. 3.7 we must consider the capacitance cBE of the base emitter diode, which essentially will reflect minority storing, due to the diffusion process. Even if we neglected the reverse biased base collector diode, to reach the simplified equivalent circuit of Fig. 3.7, its junction capacitance cBc has to be considered. A further effect we also did not consider up RBB′ cB′C r r r ? I B r β0 I B ? U BE ? r cB′ E r r r rCE rB′ E r r r U CE ? r Figure 3.11: Advanced common emitter small signal equivalent circuit of a BJT to now is the bulk semiconductor material between the base and the active base emitter junction, giving rise to the resistor RBB′ . Of course all considered voltages or currents are now given in phasor notation. The extended equivalent circuit of a BJT shown in Fig. 3.11 is referred to as full hybrid π model. 3.1.5 SPICE model of a BJT Fig. 3.12 shows the so-called Gummel-Poon-model of a npn BJT as it is implemented in the spice circuit simulator. As in the model already given in Fig. 3.4 also in the Gummel-Poon model a transistor consists essentially of the base emitter diode DBE and the base collector diode DBC and current controlled current sources iF and iR . In contrast to the Ebers-Moll model they are controlled by the base currents iBE and iBC . The current voltage characteristic of this diode is described by the exponential law already used to describe a real pn-junction (2.30). 50 r Kollektor RC r uBC Basis 6 RBB r r CsBC r CdBC r CsBE uBE ? r CdBE r r r A DLC A DBC r r A DLE A DBE r r r r ?iF 6iR r r RE r Emitter Figure 3.12: Gummel-Poon-model of a npn BJT The diodes DLE and DLC are incorporated to describe leakage currents which have no influence on the controlled current sources, and the capacitances C jBE , CdBE , C jBC , CdBC account for the junction und diffusion capacitance of the doides DBE and DBC . In the following table the SPICE data set of the BJT BFP420 is given: SPICE data set: ********************************************************** .MODEL BFP420 NPN( + IS = 17.7E-18 RB = 9.47 CJC = 380E-15 BF = 117 + IRB = 0.5E-3 VJC = 1.0 NF = 0.98 RBM = 5.47 + MJC = 0.5 VAF = 45 RE = 0.948 XCJC = 0.18 + IKF = 0.15 RC = 4.4 TR = 5.0E-9 ISE = 4.5E-12 + CJE = 130E-15 CJS = 0 NE = 2.31 VJE = 1.0 + VJS = 0.8 BR = 1.0 MJE = 0.5 MJS = 0.33 + NR = 1.0 TF = 9.6E-12 XTB = 0 VAR = 1000 + XTF = 0.457 EG = 1.16 IKR = 1000 VTF = 0.413 + XTI = 3.0 ISC = 0 ITF = 41E-3 FC = 0.78 + NC = 2.0 PTF = 0 ) 51 ************************************************************ 3.2 Small signal amplifier In order to use a BJT as small signal amplifier at first certain DC values UCE and IC have to be established ensuring that the base emitter diode is forward biased, the base collector diode is reversed biased and the transistor effect can take place. For single transistors a certain point of operation is often recommended on its data sheet. Under these circumtances small signals may be applied and the transistor can be used as an amplifier. 3.2.1 BJT biasing At first we now want to discuss different possiblities to reach this point of operation. For calculation we use the equivalent circuit of the BJT shown in Fig. 3.7, were we assume that the base emitter diode may be modeled by an ideal diode with a fixed threshold voltage UBE according to Fig. 2.11 and the BJT has a constant current gain β0 . Constant base current biasing One of the simplest possible circuit to reach a certain point of operation is shown in Fig. 3.13. Since the whole circuit is driven by the DC voltage source U0 , so we r r RC RB UBE ? r r @ @ R r r U0 UCE ? ? r Figure 3.13: Constant base current circuit get for the collector resistance RC : RC = U0 −UCE IC 52 (3.17) Due to the given current gain β0 of the transistor the base current IB is also known and we can setup an equation for the resistor RB . RB = U0 −UBE β0 IC (3.18) Usually the value of the DC voltage U0 is much higher than the voltage UBE and further-more the base current IB is very low, so the value of RB becomes very high and essentially it behaves like a current source. So the base current is essentially constant and the temperature dependence to the operation points is only due to the temperature dependence of β0 (T ). A very crucial drawback of the circuit is that the value of RB depends directly on the value of β0 , which usually shows a high variation for single transistors. A circuit with which one can over-come the considered problems is shown in Fig. 3.14. For the collector resistor we get: r RB IB ? RC r ?IC U0 UCE UBE ? r @ R ? @ r ? r Figure 3.14: Constant base current circuit with feed back RC = U0 −UCE IC (1 + 1/β0 ) (3.19) And the equation for the resistor RB reads: RB = UCE −UBE β0 IC (3.20) Although the equation for RB does not exhibit an obvious decrease in β0 dependence, the feedback does tend to stabilize the operation point. For exmaple, if the transitor has a much higher β0 than the nominal value used to calculate RB , the collector current will be higher and the collector voltage lower than the design values. With the lower collector voltage there is less voltage across RB and therefore less base current, thus at least partially compensating for the higher β0 . 53 Conversely, if β0 is lower than the nominal value, the collector current is less than the design value giving a greater voltage across RB and hence results in a higher base current, again partially compensating for the low β0 . Biasing using a base voltage divider Fig. 3.15a shows a further possiblity to bias the transistor to a certain operation point UCE and IC . In contrast to the preceeding circuit now the voltage UB is r r RC R1 r IT ? R2 I -B UB ? IC ? @ R @ UCE U0 IE ? ? RE UE r ? ? r Figure 3.15: Base voltage divider with current feedback essentially kept constant. Without the resistor RE this circuit would be very problematic, because of the self heating of the transistor. But with the resistor RE a feedback due to the current IE is introduced. For example, assume the current IE will increase due to a rise of temperature. Since the voltage UB is fixed, the voltage UBE driving the base current IB will be reduced and hence the current IE is reduced again. Typically the voltage UE across RE is chosen to be in the range of 1 V to 3 V. For the voltage devider represented by the resistors R1 and R2 to act as a voltage source the transverse current IT has to be large compared to the base current IB . Here typically a factor of 10 is chosen. As for the preceeding circuit, we assume to know the transistor parameter UBE and β0 . From the given voltage UE we first can calculate the value of the resistor RE . RE ≈ UE IC (3.21) With the help of KVL in the output circuit, we find for the resistor RC . RC = U0 −UCE −UE IC 54 (3.22) The voltage UB can also be calculated as a result of KVL in the lower base circuit, UB = UBE + UE and if we assume the transverse current IT to be 10 times the base current, we find for the resistor R2 , UB UBE + UE R2 = = (3.23) IT 10 IC /β0 and finally we get for the resistor R1 R1 = U0 − UB U0 − UBE − UE = IT + IB 11 IC /β0 (3.24) Example: Biasing of the transistor BC548C Data: U0 = 12V, UCE = 5V, IC = 2mA, β0 = 400, UBE = 0.65V and UE = 2V RE = 2 Ω = 1kΩ 2 10−3 12 − 5 − 2 Ω = 2.5kΩ 2 10−3 As a result the voltage UCE will become: RC = chosen RC = 2.7kΩ UCE = U0 − RC IC − UE = 4.6V 0.65 + 2 Ω = 53kΩ chosen R2 = 47kΩ 10 · 5 10−6 As a result the transverse current will become IT = 56µA wich is approximately 11 times the base current, so we get for the resistor R1 R2 = R1 = 3.2.2 12 − 2.65 ≈ 156kΩ chosen 12 · 5 10−6 R1 = 150kΩ Common-emitter amplifier One of the most commonly used configurations to realize a small signal amplifier using a BJT is the common-emitter amplifier shown in Fig. 3.17. It is called common-emitter configuration, because the emitter of the transistor belongs to the input as well as to the output of the circuit. In this circuit the operation point of the transistor is realized using a base voltage divider. The signal of the source 55 r RS r C1 r r @ @ R R2 C2 RL r R1 RC CE RE r r r r r C∞ r 6 U0 Figure 3.16: Common-emitter amplifier is AC coupled with the help of capacitor C1 to the base to the transistor. This must be done to not disturb the operation point by the internal resistance of the signal source. For the same reason the capacitor C2 is used to AC couple the output signal to the load resistance RL . As already mentioned in section 3.2.1 the resistor RE will result in a current feedback, which will reduce the overall amplification of the circuit, thus the capacitor CE is used to allow the AC current to by pass the resistor. Even so the DC supply voltage source U0 has a zero internal resistance theoretical on a practical circuit board the additional capacitor C∞ has to be used to realised the AC short circuit on the circuit board. AC equivalent circuit Of course all currents and voltages at the transistors are a superposition of DC and AC values. Since we already discussed how to realise a certain operation point the DC values are no longer of interest and the transistor may be replaced by its small signal equivalent circuit shown in Fig. 3.9. Further-more we want to assume IB RS US ? r - r R2 r r - β IB ? U BE R1 r rBE ? r r rin r RC U CE r r r RL ? r rout Figure 3.17: AC equivalent circuit of a common-emitter amplifier 56 that the capacitors C1 , C2 and CE are sufficiently high-valued to have negligble reactance at the frequencies of interest. With the help of these considerations, we can setup the AC equivalent circuit of the common-emitter amplifier shown in Fig. 3.17, where the small signal output resistance rCE of the transistor is also neglected. Small signal voltage gain Normally a small signal amplifier is used to amplify weak input signals. Of course one is interested in the value of the output signal, if a certain input signal is applied. The ratio of these two voltages is called voltage gain v which usually is a function of the frequency. Using the circuit shown in Fig. 3.17 we define the voltage gain of this stage to be: vu = U CE U BE (3.25) Introducing a resistor R′L RC RL RC + RL being equal to the parallel connection of RC and RL , we can write for the output voltage U CE U CE = −β R′L I B R′L = According to Fig. 3.17 we have the following relation between the base current I B und the base voltage U BE , U I B = BE rBE and hence we get for the output voltage, U CE = −β R′L U BE rBE and finally for the voltage gain of the common-emitter stage: vu = U CE R′ = − Lβ U BE rBE (3.26) Input and ouput resistance According to the equivalent circuit shown in Fig. 3.17 the input resistance rin of the stage is given by the parallel connection of R1 , R2 and rBE , rin = 1 1 1 1 + + R1 R2 rBE 57 (3.27) while the ouput resistance rout is equal to the collector resistance RC . rout = RC (3.28) Choosing the values of the capacitors All capacitors included in the common-emitter amplifier circuit shown in Fig. 3.17 will contribute to its highpass character. So the amplifier will only properly work above a certain lower frequeny limit we call flow . Instead of a rigorous treatment of the equivalent circuit incorporating all capacitors we will discuss the influence of each capacitor separately in a more heuristic manner. Capacitor CE The purpose of this capacitor is to reduce the influence of the resistor RE on the AC gain of the amplifier. Since it is connected in parallel to the resistor, with increasing frequency the total reactance will become smaller and smaller and for high frequencies there will exist no current feedback at all. If already for the lower frequency limit flow there shall be no remarkable current feedback, the following relation must hold true, 1 2π flowCE ≪ RE which gives a lower limit how to chose the value of CE . CE ≫ 1 2π flow RE (3.29) Capacitor C1 If the capacitor CE is chosen high enough, the resistors RS , rin and the capacitor C1 will form a first order high-pass filter. Its corner frequency fc being given by: 1 fc = 2πC1 (RS + rin ) Choosing the corner frequency equal to the lower frequency limit flow gives a lower limit for the capacitor C1 C1 ≥ 1 2π flow (RS + rin ) (3.30) Capacitor C2 The capacitor C2 in the ouput circuit is equivalent to the capacitor C1 in the input circuit. Thus it will also form a first order high-pass filter with 58 the resistors RC and RL , if the capacitor CE is chosen high enough. So we get an equivalent lower limit for its value analog to the limit of capacitor C1 . C2 ≥ 1 2π flow (RC + RL ) (3.31) Example: Common-emitter amplifier with the transistor BC548C As an example we will calculate the gain to be expected from a common-emitter amplifier using the transistor BC548C with a biasing according to the last section. We will choose the load resistance RL to be equal to RC and we assume a source with internal resistance RS of 50Ω. According to equation 3.13 we get for the dynamic resistance rBE , rBE = 0.026 Ω = 5.2kΩ 0.002/400 and hence for the expected voltage gain v = −1.35kΩ 400 ≈ −104 5.2kΩ and for input resistance: rin = 1 kΩ = 4.54kΩ 1 1 1 + + 47 150 5.2 If we want the amplifier to work above a lower frequency limit of flow = 50 Hz we have to choose the capacitor CE according to equation 3.29: CE ≫ 1 F = 3.18 µF 2π 50 · 103 chosen CE = 300µF Using the equations 3.30 and 3.31 we find for the cpacitors C1 C1 ≥ 1 F = 0.69µF chosen C1 = 1µF 2π 50 · 4.59 103 and C2 1 F = 0.59µF chosen C2 = 1µF 2π 50 · 5.4 103 To verify the prediction above of the gain and the values of the chosen capacitors Fig. 3.18 shows the simulated frequency response of the common-emitter amplifier with the transistor BC548C. The source amplitude was set to be 1mV. With C2 ≥ 59 Figure 3.18: Simulated frequency response of the common-emitter small signal amplifier the help of the marker function we can read out an output amplitude of almost 40dBmV, which corresponds to an absolute value of a small signal gain of 100, which is very close to the predicted value. The 3 dB cut-off frequency of the highpass can be read out to be approximately 110Hz, which is considerably above the envisaged lower frequency limit of 50 Hz. But the lower frequency limit can now easily be adjusted by increasing the values of the coupling capacitors C1 and C2 . It is interesting to notice the limitation of the gain at higher frequencies predicted by the simulation. Of course the SPICE model of the transistor is far more elaborated than the simple small signal model of Fig. 3.17 and the internal capacitances of the transistor will limit the frequency range of operation and result in a low-pass characteristic for high frequencies with a 3 dB cut-off frequency of approximately 14 MHz. If one liked to use the ampflier as the input stage of an audio amplifier, one should reduce the cut-off frequency to lower values using additional capacitors in the circuit. Summarizing the properties we can say that a common-emitter state shows • a high voltage gain • a high current gain 60 • a high power gain • a moderate input resistance approximately equal to rBE • a moderate output resistance approximately equal to RC 3.2.3 Common-collector amplifier A further possibility to realize a small signal amplifier with the help of a BJT is the common-colllector circuit shown in Fig. 3.19. It is called common-collector RS r C1 r r @ R @ R2 C2 r r RE R1 r C∞ RL r r r r r 6U 0 Figure 3.19: Common-collector amplifier amplifier, since the collector of the BJT belongs to the input as well as to the output circuit of the amplifier. Also the term emitter-follower is used. To establish the operation point again a biasing base voltage divider is used, and established with the help of the resistors R1 , R2 and RE . In contrast to the common-emitter amplifier the resistor RC is missing. The AC input signal is coupled to the base using the capacitor C1 while the capacitor C2 is used to couple the output signal to the load resistance RL . AC equivalent circuit As in the case of the common-emitter amplifier we will assume that the capacitors C1 and C2 are sufficiently high-valued to have negligible reactance at the frequencies of interest. With the help of these considerations, we can setup the AC equivalent circuit of the common-collector amplifier shown in Fig. 3.20, where the small signal output resistance rCE of the transistor is also neglected. 61 RS US ? r - I B rBE r R2 r IE r r R1 U BC rin r r RE U EC βI B 6 r ? r r - ? r RL r rini Figure 3.20: AC equivalent circuit of a common-collector amplifier Input resistance We will start our analysis calculating the input resistance rini of the intrinsic common collector stage first excluding the contribution of the resistors R1 and R2 . As a first step we introduce the resistor R′L which reflects the parallel connection of the resistors RE and RL . 1 R′L = 1 1 + RE RL Using KVL for the input circuit we find, U BC = [rBE + (β + 1)R′L ]I B and hence we get for the intrinsic input resistance of a common-collector stage: rini = U BC = rBE + (β + 1)R′L IB (3.32) Comparing rini of a common-collector stage with the intrinsic input resistance rini of the common-emitter stage, which is equal to rBE , shows that it is considerable larger, since the total load resistance R′L is multiplied by the factor β and added to rBE , according to equation 3.32. So the resistors of the base voltage divider may not be neglected calculating the input resistance of the total stage. rin = 1 1 1 1 + + rini R1 R2 Small signal voltage and current gain The voltage gain of the common-collector amplifier is defined by: vu = U EC U BC 62 (3.33) According to the equivalent circuit shown in Fig. 3.20 we get for the output voltage U EC , U EC = R′L I E while we get for the current I E and hence for the current gain I E = (β + 1)I B I E = (β + 1) → vi = IE = (β + 1) IB (3.34) U BC β+1 = U rini (β + 1)R′L + rBE BC So we find for the voltage gain of the common-collector amplifier (β + 1)R′L < 1 (β + 1)R′L + rBE vu = (3.35) which is always lower than one. Output resistance To deduce an expression for the output resistance of the common-collector stage we again consider Fig. 3.20 and introduce the resistor R′ S which accounts for the parallel connection of the resistors RS , R1 and R2 , assuming the voltage source to be turned off. 1 R′ S = 1 1 1 + + RS R1 R2 According to Fig. 3.20 the output resistance routi is given by: routi = − U EC IE A voltage UEC applied at the output will result in the following base and emitter current: U U → I E = −(β + 1) ′ EC I B = − ′ EC R S + rBE R S + rBE as a result we get or the output resistance: routi = R′ S + rBE β+1 63 (3.36) Choosing the values of the capacitors All capacitors included in the common-collector amplifier circuit shown in Fig. 3.19 will contribute to its highpass character. So the amplifier will only work properly above a certain lower frequency limit we call flow . Instead of a rigorous treatment of the equivalent circuit incorporating all capacitors we again will discuss the influence of each capacitor separately in a more heuristic manner. Capacitor C1 Again the resistors RS , rin and the capacitor C1 will form a first order high-pass. Its corner frequency fc being given by: fc = 1 2πC1 (RS + rin ) Choosing the corner frequency equal to the lower frequency limit flow gives a lower limit for the capacitor C1 C1 ≥ 1 2π flow (RS + rin ) (3.37) Capacitor C2 The capacitor C2 in the ouput circuit is equivalent to the capacitor C1 in the input circuit. Thus it will also form a first order high-pass filter with the resistors rout and RL . So we get an equivalent lower limit for its value in analogy to the limit of capacitor C1 . C2 ≥ 1 2π flow (rout + RL ) (3.38) Summarizing the properties we can say that a common-collector state shows • a voltage gain which is lower than one • a high current gain • a moderate power gain • a high input resistance equal to rBE + (β + 1)R′L ′ • a low output resistance of R S + rBE β+1 64 3.3 Integrated circuit techniques In designing integrated circuits the economic rules of discrete component circuit design are reversed. Active devices are inexpensive since they can be realized on a much smaller area than resistors or capacitors. Hence, every effort must be made to minimize the total resistance in a circuit and certainly replace passive components with transistors or diodes wherever possible. In the following section we will start with the conventional design of the operational amplifier and we will discuss different possibilities to make it a circuit with almost no passive elements. 3.3.1 The differential amplifier One serve disadvantage of the amplifier circuits discussed in the previous section is that single stages have to be coupled using capacitors which is inpratical for integrated circuit technology. Another drawback is that also due to the coupling it is not possible to amplify voltages of very low frequencies down to DC. To overcome these problems the circuit shown in Fig. 3.21 can be used, which is called differential amplifier. Essentially it consists of two identical transistors coupled r r RC r r Up ? r U0 RC UTo r r U1o @ @ R - r U2o r? ? r r IE1 @ @ r ? r Un IE2 I0 ? r −U0 Figure 3.21: Differential amplifier by an ideal current source of constant current I0 . In contrast to conventional amplifiers a positive as well as a negative supply voltage is used and two voltages U p and Un are used as input signal. To get an ouput signal to equal load resistors RC are used in the collector branch of each transistor. Due to the symmetry of the circuit an output voltage UTo will only exist, if there is a difference between the 65 input voltages. Thus it is convenient first to define the difference voltage Ud by: Ud = U p − Un (3.39) It is interesting to point out that according to KVL the difference voltage Ud is also equal to the difference of the base emitter voltages of the two transistors. Ud = UBE1 − UBE2 (3.40) To further analyse the behaviour of the circuit especially in dependence of an applied difference voltage, we start with the node equation at the current source, which forces the sum of the two currents IE1 and IE2 to be constant. IE1 + IE2 = I0 (3.41) Of course the emitter current of each transistor may also be expressed according to equation 3.3 as a function of its base emitter voltage, ) ( ) ( UBE2 UBE1 IE2 = IS exp IE1 = IS exp UT UT assuming identical transistors at the same temperature. Using equation 3.41 we get [ ( ) ( )] UBE1 UBE2 I0 = IS exp + exp UT UT and the last expression can be rearranged to: ( )[ ( )] UBE1 UBE2 −UBE1 I0 = IS exp 1 + exp UT UT According to 3.3 the first term of the last equation is equal to the emitter current of the first transistor and in the last term, the difference of the base emitter voltages may be expressed by the difference voltage Ud . As a result we get for the emitter current IE1 : I0 ( ) IE1 = Ud 1 + exp − UT By analogy one can deduce the equivalent equation for the emitter current IE2 . IE2 = I0 ( ) Ud 1 + exp UT 66 Since we have IC1 = α IE1 and IC2 = α IE2 we get for both collector currents: IC1 = αI0 ( ) Ud 1 + exp − UT and IC2 = αI0 ( ) Ud 1 + exp UT (3.42) Fig. 3.22 shows the normalised collector currents as a function of the applied difference voltage Ud . According to Fig. 3.22 for a zero difference voltage the 1 0.9 I I C2 C1 0.8 0.6 IC/(α I0) → 0.7 0.5 0.4 0.3 0.2 0.1 0 −5 −4 −3 −2 −1 0 Ud/UT → 1 2 3 4 5 Figure 3.22: Normalised collector currents as function of the applied difference voltage differential amplifier is in balance and the total current I0 is equally distributed on both transistors. For a positive difference voltage transistor 1 becomes more conductive carrying more current, while the conduction of transistor 2 is reduced. Similarly for a negative difference voltage the current I0 is steered more towards 67 transistor 2 while now the conduction of transistor 1 is reduced. To deduce an expression for the voltage UTo we use KVL and apply it to the top mesh of the circuit shown Fig. 3.21: UTo = RC (IC1 − IC2 ) With the help of equation 3.42 we can rewrite the equation above as: UTo = RC α I0 1 1 ( )− ( ) Ud Ud 1 + exp − 1 + exp UT UT Using the definitions of the functions cosh(x) and sinh(x), the last expression can be manipulated to: ( ) Ud sinh UT ( ) (3.43) UTo = RC α I0 Ud 1 + cosh UT Fig. 3.23 shows the normalised output voltage as function of the difference voltage Ud . For a difference input voltage Ud < 2UT ≈ 50mV we have an almost linear dependence of the output voltage on Ud . But as the difference voltage increases the ouput voltage will reach a saturation value of ≈ RC I0 . Of course this value may not become larger than the supply voltage U0 , which can be used to get a relation between the current I0 and the value of RC . RC I0 < U0 → RC < U0 I0 Small Signal Gain As already mentioned above, for low values of the difference input voltage Ud the differential amplifier behaves linear and can be described by the differential voltage gain vd . Approximation of equation 3.43 yields: UTo ≈ RC α I0 Ud 2UT → vd = RC α I0 RC I0 ≈ 2UT 2UT (3.44) Using the deduced relation above for RC we find vd < U0 2UT 68 (3.45) 1 0.8 0.6 → 0.4 Uout/(RCα I0) 0.2 0 −0.2 −0.4 −0.6 −0.8 −1 −8 −6 −4 −2 0 U /U d T → 2 4 6 8 Figure 3.23: Normalised output voltage as function of the difference voltage With the last equation we found an upper limit for the attainable differential voltage gain of a single stage BJT differential amplifier with a resistive load. It is interesting to notice that this limit is essentially independent of the used BJT as long as its current gain β is sufficiently high and that it depends strongly on the used level of supply voltage. Realizing a Simple Current Source The simplest way to realize the current source of the differential amplifier is shown in Fig. 3.24. Here the ideal current source is just replaced by the resistor R0 . To understand why this resistor acts like a current source, one should again consider Fig. 3.23 which showed that already a very small differential voltage Ud is sufficient to drive the output voltage into the region of saturation. On the other hand this implies that the voltage level of the base of the transistors is very close to 69 r r RC r r Up ? RC UTo r r U1o @ @ R r U0 U2o r? - r ? r r @ @ IE1 r ? r Un IE2 R0 I0 ? r −U0 Figure 3.24: Differential amplifier with a resitor used as current source the ground level and as a result the voltage drop across the current source will be equal to the supply voltage U0 . Applying these considerations to the circuit shown in Fig. 3.24 gives the following approximation between the current I0 and the resistor R0 . U0 (3.46) I0 ≈ R0 Common Mode Rejection A differential amplifier should only respond to the difference between the input signals Ud . But of course the input signal U p and Un can also show a common mode voltage Uc , defined as: Uc = 1 (U p + Un ) 2 (3.47) As a consequence there may also exist a common mode gain vcm , defined by vcm = ∆Uout ∆Uc (3.48) Considering the circuit according to Fig. 3.24 an increase in common mode voltage ∆Uc will result in a change of the current thru the resistor R0 : ∆I0 = 70 ∆Uc R0 Assuming an operation near the balance point of the amplifier, both collector currents will increase ∆I0 ∆Uc = 2 2 R0 Now one must distinguish between two cases: ∆Ic = 1. In a double-ended output defined by the voltage UTo , given that the circuit is truly symmetrical, both collector currents will alter by the same amount and the differential ouput voltage will still be zero and no common-mode gain will occur. 2. In a single-ended mode, defined by the output voltages U1o or U2o there is an ouput response and for the common-mode gain we will get: |vcm | = ∆U1o Rc = ∆Uc 2 R0 (3.49) A useful figure of merit for the differencing performance of the differential amplifier may be defined as the ratio of differential voltage gain vd to common-mode gain vcm . This is called the common-mode rejection ratio (CMRR). With the help of equations 3.44 and 3.49. CMRR = 3.3.2 2UT vcm = vd R0 I0 (3.50) Current Sources Current sources are very important electronic circuits providing for example biasing functions such as the current source of a differential amplifier. According to the simple large signal equivalent circuit shown in Fig. 3.7 a BJT essentially acts like a current source, with the advantage of the current being adjustable by the base current. Thus in principle the constant base voltage circuit with current feedback shown in Fig. 3.15 can also be used to realise the current source of a differential amplifier as shown in Fig. 3.25a, which can be used to replace the resistor. Of course the voltage UCE of the BJT may not drop below approximately 0.7 V, which is the onset of saturation. The advantage of the BJT as current source, is its higher ouput resistance, when compared to the single resistor. But with the expense of using three additional resistors, which will be a severe drawback for the realisation as integrated circuit. A circuit to overcome these problems is shown in Fig. 3.25b. It is called current mirror. In contrast to the previous circuit here only one resistor is needed to define the current I0 . A further advantage of the integrated version of the circuit is that both BJTs may be realised in close vicinity, 71 R1 r r r r ? I0 ?I1 R UCE @ R @ ? r R2 @ @ RE r r r r r 2IB r I0 ? ? r @ R @ r -U0 r -U0 Figure 3.25: Single BJT as current source and current mirror so that process variations will not cause severe differences between the transistors and on the other hand they will be closely related in temperature so that thermal tracking will take place. If we assume strictly equal transistors with a static current gain β0 , we will have the following relation between the currents I0 and I1 , which shows that for high static current gain both currents are almost identical. I0 = 3.3.3 β0 I1 β0 + 2 (3.51) Active Load A further step in reducing the number of passive components in a differential r R @ T 3@ r r r T1 @ @ R @ @ r U0 T4 r r @ T 2@ r r r Figure 3.26: Differential amplifier with current mirror as active load amplifier is to replace the resistors RC by a pnp current mirror as shown in Fig. 72 3.26, which is called active load. This load circuit is very economical in area since T1 provides the current I1 for the current mirror and no resistors are used. By using an active load, a high-impedance ouput load can be realized without excessively large resitors and hence large power-supply voltage. As a result, for a given power-supply voltage, a larger voltage gain can be achieved using an active load than would be possible, if a resistor would be used as load. For example, if a 50 kΩ load would be used with a bias current of 1 mA, a resistive-load approach would require a power-supply voltage of at least 50 V. An active load makes use of the possibilities of a transitor to create simultaneously a large bias current and a large small-signal output resistance rCE . 3.3.4 Level-Shifting Circuits Despite using circuit techniques which avoid the use of high-valued decoupling capacitors, there still remains the problem of coupling one circuit to another. Of course direct coupling between amplifier stages removes the requirement for coupling capacitors but it is difficult because of the different DC-levels. So DC levelshifting circuits must be introduced. In principle the DC level may be shifted by using an ideal voltage source and due to its zero internal resistance no AC-signal attentuation would occur. In section 2.2 we already studied how to use normal diodes to stabilize an ouput voltage. Figure 3.27a shows the principle layout of r r r A r r n r A uin r ? r r R1 r uin R2 r uout ? r ? r @ R @ r r r r uout ? r Figure 3.27: Level-shifting circuit with diodes and an amplified diode a circuit to shift the DC level by an integer number times the threshold voltage Uth ≈ 0.7 V of a diode. Using KVL we find for the ouput voltage of the first circuit: uout (t) ≈ uin (t) − nUth To deduce an expression for the second circuit we must first consider the circuit consisting of the two resistors and the transistor. Assuming a transistor with a high 73 current gain β0 the current IT thru the voltage divider R1 and R2 will not change and we will have: → UR2 = UBE = R2 IT UR1 = R1 IT = R1 UBE R2 So, we get for the total voltage across the voltage divider and hence across the transistor UCE ) ( R1 UBE (3.52) UCE = 1 + R2 Which can be adjusted to be a non-integer fraction of the diode voltage UBE and is thus called amplified diode. With the help of the last equation we find for the output voltage of the circuit with amplified diode: ( ) R1 uout (t) = uin (t) − 1 + UBE (3.53) R2 3.3.5 Complementary Output Stage In Fig. 3.28 the circuit schematic of a complementary emitter-follower output stage is shown. In contrast to the input stages of an operational amplifier the output U0 r Uin r r @ R @ r Uout - RL @ R @ r −U0 Figure 3.28: Complementary emitter-follower output stage stage must be able to drive other circuits and so the main emphasis does not lie on the voltage gain of the stage but on the current the stage is able to deliver to the circuits to be driven. As already discussed in section 3.2.2 an emitter follower has a voltage gain of nearly one but a high current gain and provides a small output resistance. In contrast to the conventional emitter-follower, the transistors shown 74 in Fig. 3.28 will only amplify on half of the sine wave. So the npn transistor will only conduct for a positive signal, while the pnp transistor will only contact for a negative signal. 75 Bibliography [1] Jackson J.D.: Classical Electrodynamics, John Wiley & Sons, New York, 1999. [2] Sze, S.M.: Physics of Semiconductor Devices, John Wiley & Sons, 1981. [3] John, D; Martin, K.:Analog Integrated Circuit Design, JoŽhn Wiley & Sons, 1996. [4] Vladimirescu, A.: The Spice Book, John Wiley& Sons Inc., 1993. [5] Kittel, C.: Introduction to Solid State Physics, John Wiley & Sons, 2004. 76