Chapter 2

Electronic III
Bipolar Transistor Models
1
Omid Shoaei, Univ. of Tehran
Simplified Bipolar Operation
When the emitter junction is forward biased, it conducts. It consists of
majority carriers from emitter (electrons here) and majority carriers from
base (holes here).
Since emitter is much more heavily doped than base, injected electrons
from emitter are many more.
Assuming collector voltage is high (collector-base is reversed biased) no
holes from the base will go to the collector.
However electrons that travel from the emitter to the base, where they are
now minority carriers diffuse away from the base-emitter junction due to
the minority carriers concentration gradient in base.
n+ p n- n+
Holes
Electrons
emitter base collector
Fig.2.1 2
Any of these electrons that gets close to collector-base junction will
immediately be whisked across the junction due to the large positive
voltage on the collector, which attracts electrons.
In a properly designed vertical bipolar, the vertical base width W (next
page figure) is small, so almost all of electrons that diffuse from the
emitter to base reach collector-base junction and are swept across
junction.
So the collector current very closely equals the electron current flowing
from the emitter to base.
The much smaller base current very closely equals the current due to the
holes that flow from base to emitter.
3
S C B E
p n+
This image cannot currently be displayed.
n+
WB
n
n+
p-
Fig.2.2.The vertical npn bipolar transistor in IC

4
The Hybrid- small signal model
rB C C = CjCt
B B C
CCS
r vBE gm vBE ro
CjEt CD
E E
Fig.2.3.The hybrid- model of bipolar transistor
5
VBE
I C = I S exp( ) (2 1)
KT / q
2
I = q AEB Dn ni (2 2)
S QB
QB = N B WB is the charge concentration per emitter area in the base (note:

unit is m-2).
N B is base doping concentration (m-3), for NPN: p p 0 = 210 / cm
17 3
WB is the base width (m).

AEB is the EB area (m2).
Dn is the diffusion constant for electrons (m2/sec). It is related to
mobility by Einsteins relation, given by: Dn = n (kT / q ).
ni is intrinsic silicon electron concentration (at a given temperature) (m-3).
(@ room temperature ni = 1.5 1010 / cm 3)
q is the electron charge (unit is Coulombs i.e. C)
Unit: (C)x(m2)x(m2/sec)x(m-3)2/(m-2)=C/sec=A
6
Tranconductance gm
dI C I
gm = = C (2 3)
dVBE kT / q
Directly proportional with IC. Doubling the current doubles the

Tranconductance.
gm q 1
= = @ room temperature larger than the MOST counterpart.
I C kT 26 mV
2 I DSQ gm 2 1
gm = for Veff = 0.2V : = =
VGSQ VTH I DSQ 0.2V 100 mV
7
Input Resistance r
The ratio of the AC VBE and the AC IB
is the AC input resistance. It is called
r as follows: r
dVBE dVBE dI C AC
=r = =
dI B dI C dI B gm gm C f
F fT
Its relationship with and gm is
depicted in figure2.4.
rB fB
g m r = AC ( 2 4)
Fig.2.4
8
Output Resistance ro
dVCE
ro is the ratio of the AC VCE to the AC IC: ro =
dI C
VBE
I C = I S exp( )
KT / q
2
I = q AEB Dn ni
S QB
VCE = VCB + VBE = - VBC Since VBE = cte
dVCE dVBC dV dQ related to the BC
=ro = - = - BC ( B ) (2 5)
dI C dI C dQB dI C junction cap per area
Increasing VBC by VBC increases the width of the depletion layer as shown
in the next figure. The depletion layer charge increases by QB.
So less QB is left (the base charge is decreased by the same amount QB).
9
(VBC + VBC=
) VCB + VCB
vBC
C jC
E B C
QB QB
Fig.2.5. Relation between QB and VBC through CjC
Also the variation of the depletion layer charge QB is linked to the depletion
capacitance CjC (in F/cm2) as:
QB dVBC q
q = C jC = (2 6)
VBC dQB C jC
dVBC dQB
=
The second term in ro - ( ) (2 5) is also derived as follows:
dQB dI C
q AEB Dn ni2 dI C q AEB Dn ni2 IC
IC = exp( VEB / VT ) =- exp( VEB / VT ) = - ( 2 7)
QB dQB QB2 QB
10
dVCE dV dQ q Q q QB
r=
o = (- BC )( B=) (- )(- B=)
dI C dQB dI C C jC IC C jC I C
Defining Early voltage as:
VA q QB q ABC QB
ro = VA = = (2 8)
IC C jC C jC t
C jC t = ABC C jC
Represented by C in the hybrid- model
ABC is the total base-collector area.
C jC t is the total base-collector junction capacitance.
11
Small-Signal Voltage Gain Av
A v = g m ro
IC V VA V
gm = , ro = A Av = = A ( 2 9)
VT IC VT kT / q
1 kT
Also one can rewrite : ro = where =
g m qVA
( )
For VA = 50 V = 5.2 10 4 A v 2000 V/V or 66 dB
Comparing gain of BJT with that of MOST stage :
VE L VA 1/
Av = = =
(1 / 2)Veff (1 / 2)Veff (1 / 2)Veff
12
Base Transit Time F
i =
QF
c
F
We can write the current in BJT as follows: 2
F = WB
2 Dn
QF is the total integrated charge in the base which depends on the forward bias VBE.
The more injected minority charge (QF) into base, the more current to collector.
The faster injected charge reach collector (smaller F ), the higher collector
current is.
WB
As it will be seen: F = where sat is the saturation velocity in base
vsat
region.
This leads to the physical interpretation that F is the average time in which
the electrons diffuse through the base from the emitter side to the
collector side.
1
It is a measure of the max. frequency: fT max = (2 10)
2 F
Note: Similarly in existing short channel MOS, with reduction of L the current is governed by
above velocity saturation (sat) equation and so max. frequency is: f max =
1 vsat
2 Leff 13
Base Transit Time F
vBE QF 2 Dn
I
C = I exp( ) = = QF
F WB2
S
KT / q
2
I = q AEB Dn ni
S QB
QF, the dynamic charge is not same as QB.

QF is the total integrated charge in the base which depends on the forward
bias VBE.
QB=NBWB is the charge concentration density per emitter area that is
physically present in the base.
They dont have same dimension either. QB, number of carrier per cm2,
whereas QF, total charge in Coulombs.
14
Capacitances
C = Cbe = C j + C D
Where
CD is a Diffusion Capacitance (discussed later), and
Cj is the depletion capacitance of the base-emitter junction. For forward
biased junction:
AE C jE 0
C jEt = , m jE = 1 / 3 (2 11)
V m
(1 - BE ) jE
jE 0
C jEt 2 AE C je 0 (2 12)
GM 15
Diffusion Capacitance CD
A variation in base-emitter voltage vBE causes a variation in injected
charge QF.
This is the charge of minority carriers in base (electrons in npn). This

variation causes a variation in majority carriers charge in collector (as
explained in BJT operation).
QF/vBE has the dimension of capacitance and is called the diffusion
capacitance.
dQF dQF dI C IC F V
CD = = = F gm = F = I S exp( BE ) (2 13)
dVBE dI C dVBE KT / q KT / q KT / q
WB2 WB2 IC WB2 I C

CD = F g m = gm = = (2 14)
2 Dn 2 n kT / q kT / q 2 n (kT / q ) 2
So physically CD is directly proportional with WB (width of the base region

2
squared) and the collector current!

16
The diffusion capacitance is much larger than the base-emitter junction
capacitance. It increases exponentially with VBE, whereas CjEt increases
only with the inverse of the square root of (1-VBE/0).
5 pF
CD
C jEt
C jEt 0
0
j
VBE
0 0.5 1V
Fig.2.6
17
Collector Junction Capacitance C
C models the depletion capacitance of the collector-base junction
(normally in reverse bias). Since this is a graded junction:
AC C jC 0
C = C jCt = C jCb = (2 15)
V
(1 + CB )1/ 3
C 0
Where AC is the effective area of the collector-base interface.
Junction Potentials :
kT N N
EB = ln( E 2 B ) 0.95 V
q ni
kT N N
CB = ln( C 2 B ) 0.73 V
q ni
N E = 1019 / cm 3 , N B = 2 1017 / cm 3 , N C = 2 1015 / cm 3
18
Small Signal Model for Forward-Biased Diode
rd Cj CD
VT
rd = , Ct = C D + C j (2 16)
ID
ID
CD = F , C j 2 C j0 (2 17)
VT
Where :
q si N D N A q si N A
C j0 = =
2 0 N A + N D 2 0
N D >> N A
2
W
and F = n
2 Dn
19
Base Resistance rB
The active region of the bipolar transistor is located directly underneath the
emitter. The base region is contacted by means of ohmic regions that add
series resistance as well as additional capacitance.
For example, a series base resistance rB is present between the base
contact metal and the active base. It is the most important parasitic series
resistance and is included in the hybrid- model.
C B rB E
n+ p n+
n+
Fig.2.7 20
Collector-Base Resistance
GM
Represents
recombination of holes
and electrons
in the base and is
proportional to the
minority-carrier charge
Q, in the base.
21
A Simple High-Freq. Model for CE with Current Drive
vCC
In first example we want to
realize a current-gain amplifier. RL
For this purpose consider a BJT iOut
amplifier, which has a large input
Rs
source resistance RS.
The AC signal is amplified, then is C
short circuited by a large cap C vin
to ground, through which we can

measure iOut .
Fig.2.8-a
C iOut= iC
B rB B
iin RS r vBE gm vBE

C
Fig.2.8-b
22
A Simple High-Freq. Model for CE with Voltage Drive
vCC
In 2nd example we want to
realize a voltage-to-current RL
amplifier. For this purpose iOut
consider a BJT amplifier, which
Rs
has a small input source
resistance RS. C
The AC signal is amplified, then is vin
short circuited by a large cap. C

to ground, through which we can
Fig.2.8-a
measure iOut .
Zin C iOut= iC
B rB B
vin r vBE gm vBE

C
Fig.2.9
23
Since output is short-circuited ro can be left out. A current iC= iOut flows in
the output short circuit.
RS >> r + rB therefore the transistor is current driven. Its input current is
approximately iin = vin /RS .
RS = , so from Fig.2.8-b one could say since the input impedance of device
is very small compared to RS so the input current is almost equal to vin/RS.
C iOut= iC
B rB B
iin r gm vBE
C vBE
Fig.2.8-c
24
Short circuit current gain AI = iout /iin
A common indicator for the speed of a BJT is the frequency at which the
transistor current gain drops to unity, when the collector is connected to a
small-signal ground: fT = Unity-Gain Frequency
So for RS >> r + rB :
r
1 s (C + C ) r
=vB ' E iB=
[ r || ] iB= iB
s (C + C ) r +
1 1 + s r (C + C )
s (C + C )
i g m r
= =
iC g m vB ' E (neglecting C current) C
This is true for iB 1 + s r (C + C )
frequencies
iC AC 1
=
near and = where f = and AC g m r
smaller than fT iB 1 + j f / f 2 r (C + C )
if:
fT << gm / C iC g m r
Note: ( ) = for f >> f
iB (C + C ) r
iC g gm
for ( ) =
1 T = m or fT = (2 18)
iB C + C 2 (C + C ) 25
iOut AC AI
( j ) = AI = = (2 19)
iin 1 + j f log AC
f
- 3 dB
fT slope 1 or
f = (2 20)
AC 20 dB/decade
gm
fT = Unity - Gain Frequency
2 (C + C )
f = f -3dB for current gain AI log 1 f
f fT
AI
0 f
- 45
- 90
Fig.2.10
26
gm gm
fT = =
2 (C + C ) 2 (C D + C jEt + C )
and C D = g m F
gm 1 C C C C C C C
fT = T = T = + = D + jEt + = F + jEt +
2 ( g m F + C jEt + C ) T gm gm gm gm gm gm gm
1 1 1 IC
= fT =
2 F C jEt + C 2 F I C (C jEt + C )
1+ IC +
gm F gm F
1 IC
=
2 F I + (C + C ) VT

C jEt
F
1 IC
f = (2 21)
2 F I C + I C f T
T
I = (C + C )
kT / q
(2 22)
C f T jEt
F
27
fT is reached to its max. value at medium and high currents. The transition
current at which this occurs is denoted by ICfT :
fT 1
1 2 fT
2 F kT
slope ( C jEt + C )
q
0 IC 0 1/ IC
IC f T
Fig.2.11
Finally, fT is specified for a bipolar transistor with a short-circuited (for AC
signal) collector. If an Ohmic series resistance rC is present, the output can
be shorted but there is still some collector resistance that remains.
1 kT / q
T = = F + ( C jEt + C ) + rC C (2 23)
2 fT IC
28
fT declines as IC increases
The decline in fT at high collector currents is not predicted by
this simple theory and is due to an increase in F caused by
high-level injection and Kirk effect at high currents. These are
the same mechanisms that cause a decrease in F at high
currents as described in Section 1.3.5.
GM
29
rc resistance
rc, the Ohmic resistor between the
collector and base contacts, is minimal if
the collector contact is made right
C
underneath the emitter in the collector rc3
rc1
area (the imaginary node represented by rc2
C in the figure)!
rc = rc1 + rc 2 + rc 3
30
Inclusion of rc resistance
GM
31
Example:
Calculate f and fT for IC=0.01 mA, 0.1 mA, and 1 mA ?
If rC=30 , what is the fTmax ?
What is the value of transition current (ICfT) ?
= 100 , F = 0.25 ns , C jEt = 5 pF , C = 1 pF
Solution:
1 kT / q
= F + ( C jEt + C )
2 fT IC
1 25 mV
I C = 0.01mA = 0.25 10 9 + (5 + 1) 10 12 = 15.25 10 9
2 fT 0.01 mA
fT = 10.6 MHz
for I C = 0.1 mA fT = 88 MHz

I C = 1 mA fT = 393 MHz
1 1
fT max = = = 636 MHz
2 F 2 0.25 10 9
32
Addition of rC
1
= F + rC C
2 fT max
1 1
fT max = = = 568 MHz
2 ( F + rC C ) 2 (0.25 10 9 + 30 110 12 )
kT / q 25 mV
I C f T = ( C jEt + C ) = (5 + 1) 10 12 0.6 mA
F 0.25 10-9
fT = 318 MHz = 1/ 2 fT max
IC = 0.6 mA
33
iB C rC iOut= iC
rB B C
r vBE gm vBE
C
Modification:
g m vBE ic g m sC
1) =
iC g m vBE ic v=
BE rC ic =
ic vBE
sC 1 + srC C
g m sC
2) ib = ( g + sC ) vBE + ( g m vBE ic ) = g + sC + g m vBE
1 + srC C
(
Note: For rC = 0 ib = g + s ( C + C ) )
g sC 1 + srC C
ib= g + sC + g m m ic
1 + srC C
g m sC
ib g + sC + g m + srC C ( g + g m ) ( g m sC )
2
Ignoring s components
ic g m sC

ic g m sC g m r

ib g + s ( C + C ) + rC ( g + g m ) C gm
1 + sr ( C + C ) + rC ( g + g m ) C

C 34
=
1
, T
( = g m r )
r ( C + C ) + rC (1 + g m r ) C r ( C + C ) + rC (1 + g m r ) C
1 C + C
+ rC C
T gm
Another approach (Lemma):
The voltage across C :

VB'C' Vbe ' ( g m rCVbe
= ') Vbe ' (1 + g m rC )
So one can simply replace C with (1 + g m rC )C in previous equation for f T (2-18)
gm gm
=fT = fT
2 (C + C ) 2 (C + C + g m rC C )
1
fT = (2-24)
C + C
2 ( + rC C )
gm 35
CE Configuration with Voltage Drive
Short cicuit tranconductance AG = iout /vin
The configuration is the same as figure below. Here RS is much smaller
than r . The small-signal model is shown below (for zero RS).
vCC
RL C
rB iOut= iC
B
Rs
C vin r vBE gm vBE

vin C
Fig.2.12-a Fig.2.12-b
Low-Frequency:
Neglect all capacitances.
iOut
for rB << r = gm
vin
36
High-Frequency:
gB
=
vBE vin
g B + [ g + s ( C + C ) ]

iOut g m vBE (neglecting C current)
This is true for
gm g B
frequencies
near and i i v gB ( g B + g )
Out = Out BE = g m =
smaller than fT
vin vBE vin g B + [ g + s ( C + C ) ] C + C
if: 1+ s
fT << gm / C g B + g
for g B >> g or rB << r g B + g g B
iOut gm i gm
AG = Out = (2 25)
vin C + C vin 1 + j f / f B
1+ s
gB
1
where f B (2 26)
2 rB (C + C )
37
f B = f 3dB for transconductance current - gain
iOut
AG =
vin
log g m
AG rolls off with slope of unity
- 3 dB gain i.e. fB.gm
slope
20 dB/decade
log 1 f
fB
Fig.2.13
Unity-gain frequency for AG (Transconductance) is meaningless, i.e. we are

not interested to know where gm=1 A/V!!
38
g m F
rB (C + C )= rB (CD + C jEt + C )= rB (C jEt + C ) 1 +
C jEt + C

Defining: B = rB (C jEt + C ) (2 27)
g m F
rB (C + C
= ) B 1 + ()
C jEt + C
g m F I C F
Besides: 1 + = 1+
C jEt + C KT
(C jEt + C )
q
I Cft ( C jEt + C )
KT / q
and recall the definition: =
F
g m F I
1+ = 1+ C (I )
C jEt + C I Cft
from (I) and (II) we can rewrite f B as follows:
1 I Cft
=fB = (2 28)
2 rB (C + C ) 2 B ( I C + I Cft )
39
Note: This equation can be modified as follows too (by adding collector
resistor related time constant):
B rB ( C jEt + C )+ rC C
=
1 1
=
f B max f= =
2 B 2 [ rB ( C jEt + C ) + rC C ]
B I =0
C
IC f T
from f B = plot of f B vs. I C is as follows:
2 B ( IC + IC f T )
fB
1
2 B
Decreasing IC increases the corner
frequency fB (pole of transconductance)
but reduces the value of gm=IC/VT (see
both in (2-25)).
IC Fig.2.14
I CfT
40
1
f B 2 rB ( C + C ) r
= = so f B > f if r > rB
f 1 rB
2 r ( C + C )
So voltage drive BW is larger than current drive BW for r > rB .
(This is almost the case; i.e. always r > rB ).
1
f B 2 rB ( C + C ) 1
= =
fT gm rB g m
2 ( C + C )
1
So f B > fT if rB g m or rB < .
gm
1
rB could be smaller than for small g m (or small I C ) values.
gm
kT / q
Definition: for =
I C I=
CTB we have=
fT f B because=
rB 1 / g m .
rB
41
The max. value of f B is reached for currents smaller than I CfT .
1 1
for I C 0 f B max = =
2 B 2 [ rB (C jEt + C ) + rC C ]
This maximum depends on the base resistance and both collector and
emitter junction capacitances, and as can be noted it is independent of the
base transit time F!
In general, it can be verified that only rB limits the high frequency
performance of bipolar junction transistor. For rB=0 infinite BW can be
achieved!!!
42
Example:
Calculate fT , f B , I CTB for I C = 0.01mA, 0.1mA, 1mA?

fT max if rC= 30?
=100, F =0.25ns, C jEt + C jCt =6pF, rB =100
from previous example: fT = 10MHz, 88MHz, 393MHz respectively for

I C = 0.01mA, 0.1mA, 1mA.
25 103
( C jEt + C )
I Cft =
KT / q
F
6 1012
=
0.25 109
600 A
=
=
B rB (C jEt + C ) + rC C=
100(6 1012 ) + 30 11012= 630 1012 sec
1 I CfT
fB =
2 B I C + I CfT
fB =
248MHz, 216MHz, 95MHz respectively for I C 0.01mA, 0.1mA, 1mA .
1
=
f B max = 252.6MHz
2 B
43
1 1
f B= = = 252.6MHz
I =0
C
(
2 B 2 630 1012 )
The I C current at which f B reduces to one half of its maximum :
1 0.6
I C = I CfT = 0.6mA f B = 12
= 126.3MHz
2 630 10 1.2
The I C current at which f B = fT :
kT / q 25mV
=
I CTB = = 0.25mA
rB 100
0.6mA 1 0.6
f B ( IC =
0.25mA) = = 252.6MHz 178.3MHz =
fT
0.25mA + 0.6mA 2 630 1012 0.85
Also recall:
1 kT / q
F + ( C jEt + C )
=
2 fT IC
1 26 mV
I C= 0.25 mA = 0.25 109 + (5 + 1) 1012 = 0.874 109
2 fT 0.25 mA
44
Conclusion
f, fT and fB are the most important frequency parameters of a bipolar
junction transistor (BJT). They all depend on the parameters shown in
Fig.2.4: r
i.e. , g m , F , rB and junction capacitances C jEt , C .

gm C f
F fT
rB fB
Fig.2.4
and F are determined by technology.
g m can be varied by varying the current I C .
rB can be varied only by taking different (larger) layouts.
45
CC and CB Configurations
ZIN ZIN
RS rB
RS vBE
ZOUT r C
vOut ZOUT
vin C vOut
vin IE gm vBE
Fig.2.15-a Fig.2.15-b
CC: two bias sources: 1- bias voltage at base

2- emitter current source
low-frequency
IE with 1/ margin error is equal to IC. Having a fixed IC requires a constant
VBE , so only a DC shift from base to emitter and vOut = vin and ZIN = .
46
Impedance is converted from high to low :
RS + rB + r
ROut LF = (2 29)
1 + AC
for AC >> 1 we have :
1 RS + rB
ROUT LF +
gm AC
1 1
Note : For MOST ROUT LF =
g m + g mb + g o g m
i.e. source follower
47
High-Frequency
At high frequencies caps have to be taken into accounts:
rB 2 vBE
r C
1
RS C
+
gm vBE vx ix
- Fig.2.16
( 1 ) ix + g m vBE + ( g + s C ) vBE =
0
1
( 2 ) ( g + s C ) vBE + ( g + s C ) ( vBE =
+ vx ) 0 where
= g
rB + RS
(2) vBE g + g + s ( C + C ) = vx ( s C + g )
( s C + g )
= vBE vx into (1)
g + g + s ( C + C )
48
( g m + g + s C ) ( s C + g )
ix = vBE ( g m + g + s C ) = vx
g + g + s ( C + C )
vx g + g + s ( C + C )
Z OUT = +
=
ix ( g + g m + s C ) ( g + s C )
C + C
1+ s
g + g g + g
= +

( g m + g ) g C C
1 + s 1 +
g m + g g
1 1
+
r RS + rB 1 + j f / fZ
=
1 1 ( 1 + j f / f P1 ) ( 1 + j f / f P 2 )
gm +
r RS + rB
1
fZ = for RS = 0 & rB << r f Z f B
2 ( C + C ) [ r || ( rB + RS ) ]
49
g m + g gm
f P1 = fT
2 C 2 C
g 1
fP2 =
=
Very high frequency, that can be ignored.
2 C 2 ( RS + rB ) C
1 1
+
r RS + rB R +r +r 1 RS + rB
= S B + = ROUT LF
1 1 1+ gm
m
g +
r RS + rB
So Provided g m ( RS + rB )C <
C i.e. f p1 <
f p 2 .:
1 1 IC f T
= B
f =
2 rB ( C + C ) 2 B I C + I C f T
=
B rB ( C jE t + C ) + rC C
1+ j f / fB
Z OUT = ROUT LF , 1 IC
= 1 + i f / fT fT (2 30)
2 I
F C + I C f T
kT / q 1 kT / q
IC f T = ( C jE t + C ) , F ( C jE t + C )
=+ + rC C
F 2 fT IC
50
IC
mA
10 fB fT KT / q
=
fT f B where=
I C I=
CTB
rB
1 I CfT = 0.62 mA
1
I CTB = 0.25 mA fT =
0.1 I c
2 F
0.01
0.1 1 10 100 1000 MHz f
ZOUT 1
Lower I C I C = 2.5 A fB = These are the bode plot the actual
1
10 k I c 0
2 B value of o/p impedance at fT is
rB < capacitive almost 3dB smaller (shown later)
gm
25 A f
fT < f B 1 k 1+ j
fB
=
Z OUT ROUT LF
I CTB = 0.25 mA g m rB = 1 resistive f
100 1+ j
fT
Fig.2.17.
Higher I C 2.5 mA
10 Position of pole and zero & bode diagram of ZOUT of
inductive
rB >
1 emitter follower for =100, rB = 100 , F = 0.25 ns,
gm
CjEt + C= 6 pF.
fT > f B 0.1 1 10 100 1000 MHz f 51
In previous figure, the asymptotic values of fB and fT are plotted versus IC.
This plot gives the positions of the zero and the pole with IC as such is
called pole-zero position plot.
@ ICTB, gm rB = 1 : fT = fB so a pure resistance results, i.e. rB.
At lower IC, fT (pole) < fB (zero) so the output impedance rolls off vs.
frequency (capacitive).
At higher IC, fB < fT so there is a region in which output impedance
increases with frequency. This region is called an inductive region.
This inductance could cause instability if combined with parasitic cap. at
the output terminal, so it is safer to reduce the biasing current.
52
fT
1+ j
1+ j f / fB fB
ZOUT ( jf ) = ROUT LF Z OUT = ROUT LF
1 + j f / fT f = fT f
1+ j T
fT
RS + rB 1 fT
neglecting ZOUT ( jf ) =
f 2
m fB
g gm
2 C + C
gm 1+ ( )
1 2 ( C + C ) 1
= 2 r C + C
B( )

m 1 1
g Z OUT ( jfT ) =
2 rB ( C + C ) gm 2
1 1 + ( g m rB )
2
= ( g m rB ) = rB= 1 rB
ZOUT ( jf ) f for g m rB >> 1: Z OUT ( jfT )
m
g gm 2 2
So for every IC the output impedance of the emitter follower at very high
frequency is equal to rB.
1 VT
Also note that at ICBT we have rB = = .
gm I C = I C TB I C TB
53
Note previous results obtained for
RS << rB << r f Z f B
However, for (perhaps more practical case)

1
rB << r << Rs f z f =
2 r ( C + C )
moves to a lower frequency
g 1
f=
p1 fT unchanged; and f=
p2
C Rs C
moves to a higher frequency
54
CB Configurations
The input impedance of Collector-
Base is exactly the same as that of
the output impedance of the Emitter-
Follower. So the same pole-zero
position plot can used for CB input
impedance.
Practically the previous assumption

for the source impedance in CB
Configuration could be more
practical: RS << rB << r f Z f B
55
Comparison between MOSTs and Bipolar transistors
Maximum Frequency of Operation:

fT is assumed as the parameter (unity-gain of amplifier).
from bipolar: 1 WB
fT max = where F =
2 F vsat
for MOST, it can be shown that (shown later):

Leff
F =
vsat
F (transit time) is likely to be smaller in a bipolar transistor than that in a

MOST because the vertical WB is easier to make smaller than the lateral
Leff. [for 0.1 m, this time constant is about 1 ps <=> fT 160 GHz]
Recall: vsat 105 m/sec in silicon
56
fT
1 vsat
Bipolar 2 WB
MOST 1 vsat
2 Leff
I
Fig.2.18
I MB
When saturation velocity and so the drain saturation current happens well
before pinch-off!
I D = vsat Qd = vsat W Cox ( VGS VTH ) (2 31)

g m ,sat = W Cox vsat (2 32)
1 g m ,sat 1 W Cox vsat 1 vsat Leff

fT max = = (2 33)
As defined before: F =
2 CGS 2 W Leff Cox 2 Leff vsat
57
In long-channel since the quadratic relations already exist:

1 W W
ID = n Cox ( VGS VTH ) 2 g m = n Cox ( VGS VTH )
2 Leff Leff
1 gm 1 n Cox W / Leff ( VGS VTH ) 1 n

fT max = = = ( VGS VTH ) (2 34)
2 CGS 2 W Leff Cox 2 L2eff
In short-channel considering short-channel effects (i.e. mobility
degradation
1 W ( VGS VTH ) 2 1
I D = n Cox where = (2 35)
2 L 1 + ( VGS VTH ) Leff EC
As shown in the Fig.2.18 increasing ID beyond some point makes fT

saturated. This can be somehow explained by equations 2-32 & 2-35.
Increasing ID is done by increasing (VGS-VTH). For very large ID and so (VGS-
VTH) if (VGS-VTH)>>1 then 2-35 becomes:
1 W
0 Cox
ID = 2 L ( V V ) = 1 E W C ( V V ) , = vsat (2 36)

GS TH n C ox GS TH n
2 EC
58
Fig.2.19
0 EC E
59
Tranconductance-Current Ratio:
gm
1
I kT / q bipolar
Drive capability
1
n kT / q
MOST
Fig.2.20
weak strong velocity
invertion invertion saturation I
Bipolar offers a better current drive capability. Less input voltage is
required to drive a larger output current!
60

Chapter 2

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Chapter 2

Hochgeladen von

Copyright:

Verfügbare Formate

Electronic III

Bipolar Transistor Models

emitter base collector

Fig.2.2.The vertical npn bipolar transistor in IC

Fig.2.3.The hybrid- model of bipolar transistor

QB = N B WB is the charge concentration per emitter area in the base (note:

WB is the base width (m).

Directly proportional with IC. Doubling the current doubles the

QF, the dynamic charge is not same as QB.

This is the charge of minority carriers in base (electrons in npn). This

WB2 WB2 IC WB2 I C

So physically CD is directly proportional with WB (width of the base region

squared) and the collector current!

Where AC is the effective area of the collector-base interface.

to ground, through which we can

iin RS r vBE gm vBE

short circuited by a large cap. C

vin r vBE gm vBE

for I C = 0.1 mA fT = 88 MHz

Another approach (Lemma):

The voltage across C :

C vin r vBE gm vBE

Unity-gain frequency for AG (Transconductance) is meaningless, i.e. we are

Calculate fT , f B , I CTB for I C = 0.01mA, 0.1mA, 1mA?

from previous example: fT = 10MHz, 88MHz, 393MHz respectively for

i.e. , g m , F , rB and junction capacitances C jEt , C .

CC: two bias sources: 1- bias voltage at base

However, for (perhaps more practical case)

Practically the previous assumption

Maximum Frequency of Operation:

for MOST, it can be shown that (shown later):

F (transit time) is likely to be smaller in a bipolar transistor than that in a

I D = vsat Qd = vsat W Cox ( VGS VTH ) (2 31)

1 g m ,sat 1 W Cox vsat 1 vsat Leff

In long-channel since the quadratic relations already exist:

1 gm 1 n Cox W / Leff ( VGS VTH ) 1 n

As shown in the Fig.2.18 increasing ID beyond some point makes fT

Das könnte Ihnen auch gefallen