Chapter 2

Semiconductor Physics and IC Technology

EXERCISES
E2.2) po = Nd = 1014 cm−3 (a) ANS
no = ni / po = 10 20 / 1014 = 106 cm−3
2
(b) ANS
E2.4) N a >> N d ⇒ holes are majority (p- type) (a) ANS
p = N ′ = N − N = 5 ⋅ 1016 cm−3
o a a d (b) ANS
no = ni / po = 10 / 5 ⋅10 = 2 ⋅ 10 cm −3
2 20 16 3
(c) ANS
E2.6) Since the free electron concentration, no , is on the order of that generated by
thermal equilibrium (i.e., ni ), an expression of no as due to both doping and
thermal generation is needed. By manipulating Eq. (2.14) from the text:
N d + N d 2 + 4ni2
no = = 8ni ⇒ N d 2 + 4 ni 2 = 16ni − N d ⇒
2
N d + 4ni = 256 ni − 32 ni N d + N d ⇒ 32ni N d = 252ni
2 2 2 2 2

∴ N d = 638 ni ANS
E2.8) E < Esat ⇒ v n = − µn E = − (1177)(103 ) = − 118 . M cm / s (a) ANS
Note that µn is extracted from Fig. 2.8 of the text.
J ndr = qnµn E = − qnvd = − (1602
. ⋅10−19 )(1016 )(1177
. ⋅ 106 ) = 1890 A / cm2 (b) ANS
x = v d t ⇒ t = x / vd = (10−4 ) / (1177
. M) = 85.0 ps (c) ANS
k = t / τ c = 8496
. / 01 . = 850 collisions (d) ANS
E2.10) E < Esat ⇒ v d = µp E = (152.9)( 2k) = 306 kcm / s (a) ANS
p = qpv d = (1602
J dr . ⋅10−19 )(1018 )(3058
. k ) = 49.0 kA / cm2 (b) ANS
t = x / v d = 1⋅10 −4 / 3058. k = 327 ps (c) ANS
k = t / τ c = 3270 . / 0.05 = 6540 collisions (d) ANS
E2.12) N d = 10 cm ⇒ µp = 429 cm 2 / (V ⋅ s)
16 -3

Dp = µpVth = (429.3)(2585
. m ) = 111
. cm2 / s
p( x ) = 1018 x cm-4 ⇒ = 1018 cm-4
dp( x )
dx

= − qDp dx = − (1602
. ⋅ 10−19 )(1110
. )(1018 ) = − 178
dp
J diff
p . A / cm2 (a) ANS
Assuming that all of p( x ) is involved in the transport (diffusion) process:
2 ⋅10− 4 ⋅ −4
Qdiff = ∫0 qp( x ) dx = q∫0
210
1018 x dx = 3.20 nC / cm 2
J = dQ dt ⇒ tdiff = Qdiff J diff = 320
. n 1778
. = 180
. ns (b) ANS
Qdiff refers to the total charge per unit area involved in the diffusion.
−4 −4
E2.14) n( x ) = (1015 cm-3 )e− x /2 ⋅10 cm
⇒
dn ( x )
dx = − ( 210
1
⋅ −4
)(1015 )(e − x /210
⋅ cm
) cm-4
N a = 1017 cm-3 ⇒ µn = 727 cm2 / (V ⋅ s) ⇒ Dn = 18.9 cm2 / s
−4
J ndiff ( x ) = qDn ⇒ J ndiff (0) = (1602 ⋅10 −19 )(18.80)( − 210⋅10−4 )e− x /2 ⋅10
dn ( x ) 15

dx .
x =0

PAGE 1
 J ndiff (0) = −151
. A / cm2 (a) ANS
amp
16.
2
cm

Jndiff ( x )

Current Density

amp
0.
2
cm 0 x 4
10. 10 . cm
Distance (b) ANS
Awafer
E2.16) n = ζ , where ζ is the yield factor.
Achip

1st Generation . (π )( 202 ) / 22 = 236 chips

n = 075
2nd Generation . (π )( 202 ) / 15
n = 075 . 2 = 419 chips
3rd Generation . (π )( 202 ) / 12 = 942 chips
n = 075
(a) ANS
p = Cwafer / n , where p is the cost per chip, and Cwafer is the wafer cost.
1st Generation p = 1750 / 236 = $7.42
2nd Generation p = 2000 / 419 = $4.77
3rd Generation p = 2250 / 942 = $2.39
(b) ANS
Note that these costs are marginal costs. There is a large cost involved with
setting up (e.g., lithographic masks) for the first run. In other words, it is
extremely expensive to produce the first batch of chips, but each additional run
is relatively inexpensive. Also, as the technology matures, the yield of good
chips tends to increase due to process improvements.
E2.18) N afinal = Nainitial + N aadditional
Qa = Naadditionalt = ( N afinal − Nainitial )t = (1017 − 1016 )( 01
. ⋅10−4 ) = 9 ⋅ 1011 cm-2 ANS
E2.20) The oxide and polysilicon masks overlap completely, which is not clear from the
layout in Fig. E2.20. The gray pattern of the oxide pattern should be visible
under the dot pattern of the polysilicon mask.
oxide, dark; polysilicon, clear (scale is approximate):

(a) ANS
oxide, clear; polysilicon, dark (scale is approximate):

PAGE 2
 (b) ANS
oxide, dark; polysilicon, dark (scale is approximate):

(c) ANS
oxide, clear; polysilicon, clear (scale is approximate):

(d) ANS
E2.22) Cross section along A-A looks like (scale is approximate):

(a) ANS
Cross section along B-B looks like (scale is approximate):

(b) ANS
Cross section along C-C looks like (scale is approximate):

(c) ANS
(Note that the gate oxide is really thin; its presence is exaggerated in the drawing.)

PAGE 3
 Cross section along D-D looks like (scale is approximate):

(d) ANS
E2.24) Since the thickness remains the same, one needs enough dosage to compensate
the mistake, plus the desired target dosage. ∴ Qd = 2 ⋅1011 cm−2 (a) ANS
N = (1 + 2 ) ⋅10 / 1⋅10 = 3 ⋅10 cm ⇒ µn = 1300 cm / (V ⋅ s)
11 −4 15 -3 2

Rq = (qnµn t )−1 = (q td ' µn t )−1 = (qQd ' µn )−1 = ((1602 ⋅10 −19 )(1011 )(1300))−1
Q
.
Rr = 48.0 kΩ / r (b) ANS
If Qa were never there, then the mobility would change slightly to
µn = 1360 cm2 / (V ⋅ s) , which changes the sheet resistance Rr = 459 . kΩ / r . (b) ANS
Assuming mobility doesn’t change that much, one can add a little bit more
phosphorus to lower the sheet resistance. Qd = 205
. ⋅1011 cm −2 (c) ANS
E2.26) There is a difficulty in solving directly for the donor concentration, since the
mobility will change as the doping concentration changes. Luckily, the mobility
is a relatively weak function of donor concentration; one can guess as to about
what the doping should be, find the corresponding value of mobility, and solve
for the doping as if the mobility were constant. Conclude as follows:
Rq = (N d µn ) (q ⋅ t ) ≈ (N d µn ) (3 ⋅10 22 )≈ 100 Ω
−1 −1 −1

So, N d µn ≈ 3 ⋅10 20 . And µn can vary from 1000’s to 100’s. The mobility
cannot be as high as the thousands, since that would imply N d ≈ 1018 , and that
µn ≈ 300 , which is a contradiction. A good starting point would be to guess the
mobility to be µn ≈ 200. solve for N d and verify the guess. If the answer is not
close enough, one can iterate to a better solution until a tolerable solution is
obtained.
N d = ( Rq µn qt )−1 = ((100 )(200 )(1602
. ⋅ 10−19 )(10 −4 ))−1 = 31
. ⋅ 1018 cm-3
At N d = 31 . ⋅ 1018 cm -3 , µn ≈ 175 . So, one can guess a lower mobility and iterate
the process. N d converges to:
. ⋅1018 cm-3 for which µn = 151 cm2 / (V ⋅ s)
N d = 414 (a) ANS
The two “dog-bone” contacts have a 0.65 square equivalent resistance:
( 2 ⋅ 0.65 + x )(100) = 2.5k ⇒ x = 235
. squares

PAGE 4
 The resistor should have a channel length of 23.5 squares. L = 235
.W.

(b) ANS
E2.28) ∆N d = εNd N d = ( 0.025)(10 ) = ± 2.5 ⋅10 cm
16 14 -3
(a) ANS
For small normalized uncertainties, the total normalized uncertainty squared will
be the magnitude of the sum of the squares of the individual normalized
uncertainties. So:
ε Rq = ε Nd 2 + ε t2 = 0025
. 2 + 0.022 = 0.0360 (b) ANS
Rq = ( qNd µn t )−1 = ((1602
. ⋅10 −19 )(1016 )(1177)(0.5 ⋅10 −4 ))−1 = 10.6 kΩ (b) ANS
∆Rq = εRq Rq = (0.03601)(10.6k ) = ± 382 Ω (b) ANS
E2.30) ε R = εt = 25%
. , ∆R = εR R = ( 0.025)(20 k ) = ± 500 Ω ANS

PAGE 5
Chapter 3
pn Junction and MOS Electrostatics

EXERCISES
E3.2) The total negative charge (magnitude) is equal to the positive charges from the
sheet charges, and thus:
(1000 +1000)(10−7 )(− 2m ) = 2Q ⇒ Q = 200 nC / cm2 (a) ANS
Keep in mind that when an impulse function is integrated, it results in a step
function about the location of the impulse function. So:
250

125

125
E(x) in kV/cm

250
3500 2500 1500 500 0 1000 2000 3000
x in nm
E(x)
(b) ANS
Take a moment for a sanity check: while − 3000 nm < x < − 1000 nm , there is a
positive charge to the left, and a net negative charge to the right; a positive test
charge would go right, and hence the electric field has a positive (constant)
value. At x = 0 nm , a positive test charge remains happy right in the middle
(literally--it’s a point of stable equilibrium for a positive test charge, as the
potential graph will later indicate), so E (0) had better be 0. Along the middle,
the effect would be an intermediate one of the extremes (E-max and E-min), and
the same argument is readily repeated for the other side of the device.
∫
dφ x
For the potential function, note that E ( x ) = − dx ⇒ φ ( x ) = − E ( ~
x ) dx~ + φ (a ) :
a

20

20

phi(x) 40
in volts

60
3500 2500 1500 0 1000 2000 3000 4000
x in nm
phi(x)
(c) ANS
Keep in mind that for − 3000 nm < x < − 1000 nm and 1000 nm < x < 3000 nm ,

PAGE 6
 that φ ( x) is linear since E ( x ) is constant for those intervals. Also, for
− 1000 nm < x < 1000 nm , φ ( x) is quadratic since E ( x ) is linear.

∫
x
E3.4) Using E ( x ) = Q / ε = 1 ρ ( x~ ) dx~ (for E ( x ) = 0 ):
enclosed S εS o
xo

E = 12 (− 1m − 2 m )(0.5 ⋅10 −4 ) / (1036

. ⋅10 −12 ) = − 72.4 kV / cm (a) ANS
E = 12 (− 1m − 3m )(10 −4 ) / (1036
. ⋅10 −12 ) = − 193 kV / cm (b) ANS
1
⋅ − 4 ) + 12 ( 3m + 2 m )( 0 .510
( −1m −3 m )( 0 . 510 ⋅ −4 )
E= 2
⋅ − 12
1. 03610
= − 72.4 kV / cm (c) ANS
One could also note that the electric field in part (c) should equal that in part (a)
by symmetry. Or, note that E( ∞ ) = 0 as well, and make the calculation involve
the enclosed charge within 15. µm < x < 2 µm .
E3.6) Since at the interface, the same amount of charge is enclosed, then:
ε (0− ) = εx = εS E( 0+ ) / E( 0− ) = (1036
. ⋅10=12 )( 015
. / 0.3) = 518 fF / cm (a) ANS
Again, note that ρ ( x ) = ε ( x ) dE ( x )
dx .

(b) ANS
ti + ti + dE ( x ) t i+
Q = ∫t − ρ ( x ) dx = ε (ti ) ∫t − dx dx = ε ( ti )∫t − dE ( x ) = εx ( E ( ti + ) − E (t i − ))
i i i

Q = (518.0 f )(0.3k − 0) = 155 pC / cm 2

(c) ANS
Okay, that was a bit of trickery when it came to the math. But it’s really the
same routine used to derive Eq (3.14). Another way to find Q , is to note that the
charge on the metal is opposite to the charge in the silicon by charge
conservation. That is, Q + ∫0 ρ ( x ) dx = 0 .
500 Å

PAGE 7
 0.351

0.35

0.349
phi(x) in volts

0.348
20 0 20 40 60 80
x in nm
phi(x)
(d) ANS
Note the kink at the interface. This is because the electric field graph is not
continuous, and thus, the potential graph is not smooth at the discontinuity.
E3.8) The graphs are:
16

0 0

16

24
rho(x) in uC/cm^3
32 10
1000 1.106 10 1000 2000 3000 4000 5000 6000
x in nm
rho(x)
(a) ANS
0.05

0 0

0.05

0.1

0.15
phi(x) in volts

0.2

0.25 11
1000 1.08 10 1000 2000 3000 4000 5000 6000
x in nm
phi(x)
(b) ANS
E3.10) This is nothing more than Eq. (3.49):
φo ( x ) = φn − 2ε Sd ( xno − x ) 2 , 0 < x < x n , φo ( x ) = φp + ( x + x po )2 , − x po < x < 0
qN qNa
2 εS

φa = φn = −φ p = (60m )log( N / ni ) = (60 m )log(1017 / 1010 ) = 419 mV

PAGE 8
 ⋅ −19 )(1017 )
= = 7.73 MV / cm 2
qN (1. 60210
2 εS ( 2 )( 1. 036 p )

x no = x po = ( 2ε S ( 2 φ a )
qN )( ) =N
N+N
2ε S φa
qN =
( 2 )( 1. 036 p )( 419 m )
. ⋅10 − 19 )( 1017 )
( 1602
= 736
. nm
φo( x ) = 419m − 7.733M(7.36 ⋅10−6 − x )2 , 0 < x < 736 . nm (a) ANS
φo( x ) = −419m + 7.733M (7.36 ⋅ 10 + x ) , − 736
−6 2
. nm < x < 0 (a) ANS
po ( x ) = ni e−φo ( )/ th ( − x po < x < 0 ), and no ( x ) = nie φo ( x )/Vth ( 0 < x < xno )
x V

17
1 10

16
5 10

concentration in cm^-3
0
70 60 50 40 30 20 10 0
x in nm
po
17
1 10

16
5 10

concentration in cm^-3
0
0 10 20 30 40 50 60 70
x in nm
no
(b) ANS
This plot is the charge-density due to free-carriers (i.e., holes on the left,
electrons on the right). According to the depletion approximation, they are
identically zero throughout the depletion region.

PAGE 9
 18

12

rho(x)12
in mC/cm^3

18
60 40 20 0 20 40 60
x in nm
rho(x)
(c) ANS
The dips on the edges are the result of this analysis. They bite off about 20% of
the edge in this case.
E3.12) The depletion region in the p-side is so small that virtually all the junction
potential appears in the n-type side. The one-sided diode results in a potential
function that looks like the following:
 φp x<0

φ ( x ) = φn − 2 εS ( x no − x ) 0 < x < x n
qN d 2

 φn x > xn

0.5

phi(x) in volts

0.5
0 20 40 60 80 100
x in nm
phi(x)
(a) ANS
2ε Sφ B 2 qN d φB
E=− , x no = , so E = −
qN d xno
εS qN d εS

φB = φn − φp = 420m + 540m = 960 mV

⋅ − 19 )( 1017 )( 0. 9524 )
E=− = −172 kV / cm
( 2 )( 1. 60210
1. 036 p (b) ANS
2qN d ( φB −VD ) ε E2 (1. 036 p )( −510
⋅ 5 )2
E (VD ) = − εS ⇒ VD = φB − 2 qN
S
d
= 09524
. − ( 2 )(1.60210
⋅ − 19 )( 1017 )
= −713
. V (c) ANS
E3.14) φB = Vth ln( N a N d / ni 2 ) = (2585
. m) ln(1016 ⋅ 5 ⋅1016 / 1020 ) = 756 mV
2 εS (φ B −VD ) ( 2 )(1 .036 p )( 0. 7559+ 2 .5 )
xn = = = 118.47 nm
Na 1016
qNd Nd + N a (1. 602 ⋅10− 19 )( 5⋅ 1016 ) ⋅ 16 +1016
510

2 εS ( φ B −VD ) ( 2 )(1. 036 p )( 0 . 7559+ 2. 5 )

xp = = ⋅ 16
= 592.34 nm (or use
Nd 510
qNa Na + Nd . ⋅10 −19 )( 1016 )
( 1602 ⋅ 16 +1016
510

x n N d = x p Na ...)

PAGE 10
 10

0
rho(x) in mC/cm^3
2
0.6 0.4 0.2 0 0.2 0.4 0.6
x in um
rho(x)
(a) ANS
vd max = 2.5 + 0 .250 = 2.75 V, vd min = 25
. − 0.250 = 2.25 V , so that:
x nmax = 122.93 nm , x pmax = 614.66 nm , x nmin = 11383. nm , x pmin = 569.15 nm

10

rho(x) in
0 mC/cm^3

2
0.8 0.6 0.4 0.2 0 0.2 0.4 0.6
x in um
rhomax
rhomin
(b) ANS
2

2
4

8
rho(x) in mC/cm^3
10
0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8
x in um
rhomin-rho0
(c) ANS

PAGE 11
 10

0
rho(x) in mC/cm^3
2
0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8
x in um
rhomax-rho0
(c) ANS
For the first case, the positive, incremental charge stored on the p-side (which is
equal & opposite to the charge on the n-side by charge conservation) is:
q j = − qN a ( x p min − x p ) = −(1602
. ⋅ 10−19 )(1016 )(569.77 − 592.95)(10−7 )
q j = 4.034 nC cm2 (c) ANS
For the second case, there is a negative incremental charge on the n-side (which
is equal & opposite to the charge on the p-side):
q j = − qN a ( x p max − x p ) = − (1602
. ⋅10 −19 )(1016 )( 61525
. − 59295
. )(10−7 )
q j = 3572
. nC cm 2 (c) ANS
With more than enough significant figures, it should be clear that the stored
charges are not equal and opposite in the two cases. This is because the
depletion region is a nonlinear charge storage element-it does not obey a linear
relationship between stored charge and applied voltage.
C(VD ) = ε S / x d = εS / ( xn + x p ) = (1036
. p) / ((118.5 + 5923
. )(10 −7 )) = 14.6 nF / cm 2
qJ = Cv d = (1457
. n )(± 250m ) = ± 364
. nC / cm2 (d) ANS
The small signal approximation is a reasonable one.
E3.16) For simplicity the diode here will be treated as a one-sided junction:
φB = (60 m )log(N a N d / ni2 ) = (60m )log(1016 ⋅ 1018 / 10 20 ) = 840 mV
qN a εS (1.60210
⋅ − 19 )(1016 )(1036
. p)
Co = 2φ B = (2 )(840 m ) = 314
. nF / cm2 = 0314
. fF / µm 2 (a) ANS
Co A = Cmax ⇒ A = Cmax / Co = 450 / 0314
. = 1430 µm 2 (b) ANS

C(V D ) =
Co
1 − VD / φB
C /A 2
⇒ VD max = φB 1 − Cmax
min / A
. ) 1 − (650
= (0840 450 )
2
( (
= −0.913 V )) ( ) (b) ANS

E3.18) φn = Vth ln( N d / ni ) = (60m )log( 5 ⋅1015 / 1010 ) = 342 mV

VFB = −(φn + − φn ) = − (0.55 − 0342
. ) = − 208 mV (a) ANS
φS = −φn = −342 mV (b) ANS
2 ε S ( 2 φn )
X d max = qN d = ( 2 )(1. 036 p )( 2 )( 0 .392 )
⋅ −19 )( 510
( 1 .60210 ⋅ 15 )
= 421 nm (c) ANS
Cox = εox / tox = ( 3453
. f ) / (150 ⋅ 10−8 ) = 230 nF / cm2
VTp = VFB − 2φn − C1ox 2qεS N d ( 2φn )

PAGE 12
 VTp = −0.208 − ( 2)( 0342
. ) − 2301.2 n ( 2)(1602
. ⋅ 10−19 )(1036
. p )(5 ⋅1015 )(2 )(0.342)
VTp = −1.04 V (d) ANS
E3.20) VTn = VFB − 2φp + C1ox 2qεS N a ( − 2φp ) , VFB = − (φn + − φ p )
VTn = −φn+ − φp + C1ox 2qεS N a (− 2φ p ) , φp = −Vth ln( N a / ni )
Note that φp is a function of N a itself. Also, VTn is overall an increasing
function in N a . But φp won’t change very much... from looking at E3.17, it is
clear that N a must be significantly lower than N a = 1016 cm −3 , maybe half as
much? (Look at the 1
Cox 2qεS N a (− 2φ p ) term.) An iteration should reveal that:
N a = 4.03 ⋅1015 cm −3 , for which φp = −334 mV . (a) ANS
Use information on the depletion width (for the transition region of φ ( x ) in
substrate), stored charge (for the slope of φ ( x ) in the oxide region), and bulk
potential (for the final value of φ (x ) ):
φp = −334 mV
2 ε S ( −2 φ p )
X d max = = = 463 nm
( 2 )(1. 036 p )( 2 )( 0 .3337 )
qN a (1.602⋅10-19 )( 4. 03110 ⋅ 15 )

Q = qN a x d = (1602
. ⋅ 10−19 )( 4.031⋅1015 )( 4627 . µC / cm2
. ⋅10−7 ) = 299
A sketch of E ( x ) may help:
120
100
80
60
40
20
E(x) in kV/cm
0
20
100 0 100 200 300 400 500 600
x in nm
E(x)
So that φ ( x) should look like:
0.6

0.4

0.2

0.2
phi(x) in volts
0.4

0.6
100 0 100 200 300 400 500 600
x in nm
phi(x)
(b) ANS

PAGE 13
E3.22) This device is in accumulation.

(a) ANS
φn = Vth ln( N d / ni ) = (60m )log(7.5 ⋅10 / 10 ) = 413 mV
16 10

VFB = −(φn + − φn ) = − (0.55 − 0.413) = −137 mV

Cox = ε ox / t ox = (3453
. f ) / ( 450 ⋅ 10−8 ) = 76.7 nF / cm2
− QG = Cox (−VGB − ( −VFB )) ⇒ VGB = QG / Cox + VFB = 2 ⋅ 10−7 / 76.74n − 0137
.
VGB = 247. V (b) ANS
Such a large, negative gate charge almost certainly guarantees inversion:

(c) ANS
QP = − QG − qNd X d ,max

= − QG − 2qN d ε S (2φn )
2 εS (2 φn )
QP = − QG − qNd qNd

QP = − (− 5⋅ 10−7 )− ( 2)(1602
. ⋅ 10−19 )(7.5 ⋅1016 )(1036
. ⋅ 10−12 )( 2 ⋅ 0.409 )
QP = 357 nC / cm2 (d) ANS
E3.24) VFB ≈ −09 . V , VT ≈ 0.4 V (a) ANS
Cacc = Cinv ≈ 7.2 fF / µm 2 from the graph
tox = ε ox / Cox = (3453
. f ) / ( 7.2f / (10-4 ) 2 ) = 48 Å (b) ANS
The dip (flatband voltage) begins to appear on the left side, the graph is
discontinuous to the right of the dip. Since VT > VFB , the device has a p-type
substrate. (c) ANS
The minimum capacitance, right before the discontinuity, is when:
Cmin = εox / X d max = εo = 3 fF / µm 2 , φp = −Vth ln( N a / ni )
qN a
2 εS ( −2φ p )

Using iteration: N a = 8 ⋅1016 cm−3 (d) ANS

PAGE 14
Chapter 4
The MOS Field Effect Transistor

EXERCISES
E4.2) VDS = 1 > VGS − VTn = 15
. − 1 = 0.5 ⇒ saturation (a) ANS
ID = 12 µn Cox WL (VGS − VTn ) (1 + λnVDS ) ≈ 12 (50µ )(505 )(15
. − 1) = 62.5 µA
2 2
(b) ANS
E4.4) VTn = VTOn + γ n ( −V BS )
− 2φp − − 2φp = 1 + (0.6) 1 + 0.84 − 0.84 = 126
. V ( )
VDS = 4 − 1 = 3 > VGS − VTn = 35
. − 1 − 1264
. = 124
. ⇒ saturation (a) ANS

(V − VTn ) (1 + λnV DS )
2
ID = 12 µn Cox W
L GS

ID ≅ 12 µn Cox WL (VGS − VTn ) = (50 µ)(804 )(1236

. ) = 764 µA
2 2
1
2 (b) ANS

E4.6) VTn = 1 + ( 0.6) ( 2 + 0.84 − 084

. )
= 146
. V
VGS = 325
. − 2 = 125
. < VTn = 146
. ⇒ cutoff , ID = 0 (a,b) ANS
E4.8) VD = 4 V , VS = 2 V (a) ANS
VTn = 1 + ( 0.6) ( 2 + 0.84 + 084
. )
= 146
. V
VGS = 3 − 2 = 1 < VTn = 146
. ⇒ cutoff , ID = 0 (b,c) ANS
E4.10) VSD = 5 − 4 = 1 < V SG + VTp = 5 − 2 − 1 = 2 ⇒ triode (a) ANS

( ) ( )
− ID = µp Cox WL VSG + VTp − 12 VSD (V SD )1 + λ pVSD
− I D ≅ µp Cox WL (V + V − V )V = (25µ)( )(2 − (1))()1 = 563 µA
SG Tp
1
2 SD SD
45
3
1
2 (b) ANS
E4.12) VTp = VTOp − γ p ( V + 2φ − 2φ )= −1 − ( 0.6)( 1 + 084
BS n . − 0.84 )= − 126
n . V
VSG = 4 − 35
. = 05
. < −VTp = 126
. ⇒ cutoff , − I D = 0 (a,b) ANS
(
E4.14) VTp = −1 − (0.6) 2.5 + 0.84 − 084
. = − 155
. V )
VSG = 2.5 − 1 = 15
. < −VTp = 155
. ⇒ cutoff , − I D = 0 (a,b) ANS
E4.16) VD = 2 V , VS = 4 V (a) ANS
VTp = VTOp − γ p (V BS + 2φn − 2φn = − 1 − ( 0.6) 1 + 084)
. − 0.84 = − 126
. V ( )
VSD = 4 − 2 = 2 > V SG + VTp = 4 − 2 − 126
. = 0.74 ⇒ saturation (b) ANS

( )(1 + λ V )
2
− I D = 12 µ p Cox WL V SG + VTp p SD

(V ) = (25µ )( )(0.74)2 = 110 µA

2
− I D ≅ 12 µp Cox WL SG + VTp 1
2
80
5 (c) ANS
E4.18) Looking at the lower graph, the onset of saturation occurs at VSD = 0.75 V
The onset of saturation is VSD = VSG + VTp = 075
. = 15. + VTp ⇒ VTp = −0.75 V (a) ANS
Since this is a current plot vs. VSD , just check the slope of the current curves in
the saturation region to find λ p :
POINT A: . V , − ID = 300 µA
VSD = 125
POINT B: VSD = 4.5 V , − ID = 350 µA

PAGE 15
 Since − IDo = 300 µA ,
∆(− ID ) ∆ ( − I D) −300 ) µ
slope = go = ( − IDo )λp = ∆VSD ⇒ λp = − I DoVSD = ( 300( 350 . ) = 0.05 V
µ )( 4. 5−125
−1
(b) ANS
. fF / µm 2 .”)
E4.20) (Note: This problem is missing “Let Cox = 142
VTp = VTOp − γ p ( +V BS )
+ 2φn − 2φp = −1 − (0.6) 2 + 084
. − 0.84 = −146
. V ( )
L = Lgate − 2LD = 2 µ − 2 ⋅ 01
. µ = 18
. µm
The device is in saturation, so:
(
gm ≅ µ p Cox WL − VGS + VTp = (25µ )(126.8 )(15) . ) = 141
. − 1461 . µS (a) ANS
γ pg m ( 0 . 6 )(14.1 µ )
gmb = 2 −V SB +2 φn
= 2 2+0 . 84
= 2.51 µS (a) ANS

( ) ( )
−1
ro = λp (− I D ) =  λ p ⋅ 12 µp Cox WL V SG + VTp 
−1 2

ro = (0.05 ⋅ 12 (25µ )( 126.8 )(15

. − 1461
. )2 )−1 = 73.7 MΩ (a) ANS
Cgs = 23 WLCox + WLD Cox = 23 ( 26)(18 . f ) + ( 26)( 01
. )(142 . f ) = 48.0 fF
. )(142 (b) ANS
Cgd = Cov = WLD Cox = ( 26)( 01 . f ) = 284
. )(142 . fF (b) ANS
A = Adrain −diff = Asource−diff = ( 6)( 26) = 156 µm 2
− 19
qεS N a . ⋅10 )(1 .036 p )(10 )
17

Csb = A = (156 )(10−4 )2 = 825

( 1602
2( φ B −V SB ) 2( 2 +0 .9667 ) . fF (b) ANS
− VDB = −VSB + VSD = 2 + 4 = 6 V
− 19
q εS N a . ⋅10 )(1 .036 p )(10 )
17

Cdb = A = (156 )(10−4 )2 = 538

( 1602
2( φB −VDB ) 2( 6 +0 .9667 ) . fF (b) ANS
E4.22) VDS = 35
. > VGS − VTn = 2 − 1 = 1 ⇒ saturation
ID = 2 µn Cox WL (VGS − VTn )2 = 12 (50µ )( 20
1
3 )( 2 − 1) = 167 µA
2
(a) ANS
gm = µn Cox WL (VGS − VTn ) = (50µ )( 203 )(1) = 333 µS
id = g mv gs = (333µ )( 20m ) = 6.67 µA
So id (t ) = 6.67 µ ⋅ cos( 2π 1000t ) (b) ANS
180

175

170

165

iD in160
microamps

155
0 0.5 1 1.5 2 2.5 3
t in msec
iD(t)
(c) ANS

PAGE 16
Chapter 5
Digital Circuits Using MOS Transistors

The channel length modulation parameter ( λ ) is misprinted. It should be

λn = λp = 0.067 V −1@ L = 15
. µm throughout the problems in the chapter.

EXERCISES
E5.2)
X Y OUT
0 (OFF) 0 (OFF) 0 (OFF)
1 (ON) 1 (ON) 0 (OFF)
1 (ON) 0 (OFF) 1 (ON)
0 (OFF) 1 (ON) 1 (ON)
In other words, OUT = X Y + XY = X Y + XY , the XOR of X and Y . (a) ANS
Use DeMorgan’s rule to acquire OUT = X Y XY :

(b) ANS
E5.4) Here’s the load line with respect to the MOS variables − ID and VSD :
500

375

250

-ID in microamps
125

0
0 1 2 3 4 5
VSD in volts
-ID(VSG = 2)
-ID(VSG = 3)
-ID (VSG = 4)
-ID (VSG = 5)
-IR (5 k load line)
-IR (25 k load line)

PAGE 17
 That results in the following transfer function of VSD vs. VSG :
5

2
VSD in volts
1

0
0 1 2 3 4 5
VSG in volts
VSD (R = 5 k)
VSD (R = 25 k)
Since VIN = VDD − VSG , and VOUT = VDD − VSD , the graph should be reoriented:
5

VOUT in volts
1

0
0 1 2 3 4 5
VIN in volts
VOUT (R = 5 k)
VOUT (R = 25 k)
ANS
E5.6) ID = µ n C W
ox L (VGS − VTn − VDS / 2)VDS = 50µ ⋅ ( 4 − 1 − 1 / 2)(1) = 250 µA
3
1. 5 (a) ANS
VOUT ≤ 1 V ⇒ ID ≥ (VDD − VOUT ) / R = (5 − 1) / 25k = 160 µA
The device had better sink at least 160 µA with VGS = 4 V .
ID = µn Cox WL (VGS − VTn − VDS / 2)VDS ⇒ WL = I D ( µn Cox (VGS − VTn − VDS / 2)VDS )−1
L ≥ 160µ (50µ (4 − 1 − 1 / 2 )(1)) = 128
W −1
. (b) ANS
. ) / 25k = 188 µA
ID = (5 − 03
L = 188 µ(50 µ( 4 − 1 − 0.3 / 2)( 03. ))−1 = 4.40
W
(c) ANS
ID = (5 − 1) / 5k = 800 µA
L = 800 µ(50 µ(4 − 1 − 1 / 2)(1)) = 6.4
W −1
(d) ANS
E5.8) Note that the symmetry between the PMOS and NMOS inverter is reflected in
the following fashion. In the case of the NMOS inverter, VOL = V MIN = 0.5 V , but
VOH = 5 V (rail) (i.e., the NMOS is in cutoff). So, for the PMOS inverter,
VOH = V MAX = 5 − 0.5 = 45. V , and VOL = 0 V (rail) (i.e., the PMOS in cutoff). In

PAGE 18
 calculating the bias points of the circuit (i.e., DC operating points), the effect of
channel length modulation λ has been ignored to make exercise calculations
tolerable.
For the PMOS device acting as the inverter:
V MAX = 4.5 V ⇒ −VDSp = 0.5 V for VDD = 5 V , for VIN = 0 V ⇒ −VGS = 5 V
( )
− IDp = µ p Cox WL VSG + VTp − VSD / 2 (VSD ) = 25µ ⋅ 156. (5 − 1 − 05
. / 2)(05
. ) = 188 µA
For the NMOS device acting as the current source:
( 2 )(187 .5 µ )
IDn = 12 µn Cox WL (VGS − VTn )2 ⇒ V B = VGS = VTn + =1 + = 2.37 V
2I D
µn Cox W
( 50µ )( 16. 5 )
(a) ANS
L

Assume saturation of both devices:

IDn = − I Dp
( 2 )(187 . 5 µ )
− I Dp = 12 µp Cox WL (VSG + VTp )2 ⇒ VSG = −VTp + =1 + = 2.94 V
2 I Dn
µp Cox W
L ( 25µ )( 16. 5 )

V M = VIN = V DD − VSG = 5 − 2.94 = 2.06 V (b) ANS

rop = ron = (λ p (− I Dp ))−1 = (0.06667 ⋅187.5µ )−1 = 80 kΩ
gmp = µ pCox WL ( I Dn ) = (25 µ )( 16.5 )(1875
. µ) = 137 µS
. µ )( 12 80k ) = − 548
Av = − g mp ( rop // ron ) = −(1369 . V/V (c) ANS
Clearly, VOL = 0 V, and VOH = 4.5 V . Now, VIL and VIH are intersections of the
line through V M of slope Av with VOH and VOL , respectively. The equation for
the line through V M of slope Av is (using point-slope formula
y − yo = m( x − x o ) ):
(VO − VM ) = Av (V I − V M ) ⇒ VI = Av −1 (VO − V M ) + V M
VIL = Av −1(VOH − V M ) + V M = (− 5.477)−1 (4.5 − 2.06) + 2.06 = 161
. V (d) ANS
−1
VIH = Av (VOL − V M ) + V M = (− 5.477) (0 − 2.06 ) + 206
−1
. = 2.44 V (d) ANS
NM L = VIL − VOL = 161 . − 0 = 161
. V , NM H = VOH − VIH = 45 . − 2.44 = 2.06 V (e) ANS
E5.10) The procedures for obtaining the answers here are similar to those of E5.9.

VM =
VTn +
kp
kn (V DD + VTp )= 1 + 10
12 (5 − 1)
= 2.43 V (a) ANS
1+
kp
kn
1+ 10
12

Since inverter is CMOS, VOL = 0 V , and VOH = 5 V . (a) ANS

g m = g mn + gmp = µn Cox (WL )n (VGSn − VTn )+ µp Cox (WL )p VSGp + VTp ( )
g m = (50µ )(16.5 )(2.43 − 1) + (25 µ)(10
. )(5 − 2.43 − 1) = 548 µS
15

IDn = 12 µn Cox WL (VGS − VTn ) = (50µ )(156. )(2.43 − 1)2 = 204 µA

2
1
2

Av = − gm (ron // rop )= − (548µ )(2 ⋅ 0.067 ⋅ 204µ ) = −20.0 V / V

−1
(a) ANS
VIL = V M + Av −1 (VDD − V M ) = 2.43 + (− 20.0) (5 − 2.43) = 2.30 V
−1
(a) ANS
VIH = VM − V M AV = 2.43 − 2.43 (− 200
. ) = 2.55 V (a) ANS

PAGE 19
 (a) ANS
An appropriate SPICE deck is as follows:
*E5.10

MN 3 2 0 0 MNX W=10u L=1.5u

MP 3 2 1 1 MPX W=6u L=1.5u
VDD 1 0 5
VIN 2 0 2.432

.model MNX NMOS KP=50u LAMBDA=0.06667 VTO=1

.model MPX PMOS KP=25u LAMBDA=0.06667 VTO=-1
.DC VIN 0 5 0.001
.probe
.end
A PROBE output of the DC sweep is as follows:

(b) ANS

PAGE 20
 CL (VOH / 2) (200f )(5 / 2 )
t PHL = 2 = = 313 ps (c) ANS
( kn / 2)(V DD − VTn ) ( 1.5 50 µ / 2)(5 − 1)2
6

CL (VOH / 2 ) (200f )(5 / 2 )

t PLH = 2 = 10 = 375 ps (c) ANS
(k p / 2)(VDD + VTp ) ( 1.5 25µ / 2)(5 − 1)2
CDB = Wn Ldiffn CJn + Wp LdiffpCJp + (Wn + 2 Ldiffn )CJSWn + (Wp + 2 Ldiffp )CJSWp
CDB = ( 6)(6 )(01. f ) + (10)( 6)(0.3f ) + (6 + 2 ⋅ 6 )(0.5f ) + (10 + 2 ⋅ 6 )(0.35f )
CDB = 38.3 fF
(200f + 38.3f )(5 / 2)
tPHL = 6 = 372 ps (d) ANS
( 15. 50µ / 2)( 5 − 1)2
(200f + 38.3f )(5 / 2 )
t PLH = 2 = 447 ps (d) ANS
. 25µ / 2 )(5 − 1)
( 10
15

E5.12) Following the rule of Eq. (5.56):

(WL )n = M2 (WL )p ⇒ (WL )p = M2 (WL )n = 22 (153. )= 13.5 ANS
And for M = 3, (WL )p = 23 (153. ) = 152. ANS
E5.14) To minimize illustration clutter, parasitic capacitances have been omitted
(though they are required for dynamic logic operation).
The static inverters are implemented as follows:

For Y = ABC :

(a) ANS
For Y = BC + AC :

PAGE 21
 (b) ANS
For Y = (A + B )+ (C + D):

(c) ANS

PAGE 22
E5.16) Transmission gates can be used to implement the desired function:

ANS

PAGE 23
Chapter 6
The pn Junction Diode

EXERCISES
E6.2) The law of junction under low-level injection states:
− φ /V
pn ( x n ) = Na e j th , for φ j = φB − VD
φB = Vth ln( N a N d / ni 2 ) = ( 26m )ln(1017 ⋅1017 / 1020 ) = 838 mV
pn ( x n ) = 1017 e −( 0.838 −0.7 )/26 m = 4.9 ⋅1014 cm-3 << N d (a) ANS
The same applies to np ( − x p ) . (a) ANS
 Dp Dn  VD /Vth
ID = qni2 A +  (e − 1) = k ( Dp + Dn ) , for some particular k ,
 N dWn N aWp 
when biased at VD = 700 mV , since N d = Na , and Wn = Wp . It is clear that the
fraction of total current which hole diffusion is responsible for is:
Dp
f = , for N d = Na = 1017 cm -3 , Dn = 188
. cm2 / s , Dp = 8.21 cm 2 / s
Dp + Dh
8.205
f = = 30.3% (b) ANS
8.205 + 1880 .
qni2 A
ID = ( Dp + Dn )(eVD /Vth − 1)
NW
(1602
. ⋅10 −19 )(10 20 )( 25 ⋅10−4 )2
ID = (8.805 + 18.80)( e0 .7/ 26m − 1) = 68 µA (c) ANS
(1017 )( 2 ⋅10 −4 )
E6.4) φB = Vth ln( N a N d / ni 2 ) = ( 26m )ln(1017 ⋅1019 / 1020 ) = 959 mV (a) ANS
2εS (φB − VD ) N d εφ Nd φ
xp = = S B for VD = B
qN a Nd + Na qN a N d + Na 2
(1036
. ⋅10 −12 )( 0959
. ) 1019
xp = = 78.4 nm (b) ANS
(1602
. ⋅ 10 −19 )(1017 ) 1019 + 1017

2ε S φB Nd (2)(1036
. ⋅10 −12 )(0959
. ) 1019
= = = 111 nm
(1602 ⋅ 10−19 )(1017 ) 1017 + 1019
x po (b) ANS
qN a N d + N a .
The injected minority concentration looks like:

(c) ANS
PAGE 24
E6.6) It was found in E6.2 that at a forward bias of VD = 700 mV , ID = 68 µA .
rd = gd −1 = Vth / I D = 26m / 68µ = 382 Ω (a) ANS
C jo ε A qεS N a N d
Cj = , where C jo = S = A
1 − VD / φB xdo 2( N a + N d )φB
But for a significant forward bias (as in this case), C j = 2 Cjo .
From E6.2 φB = 838 mV .
(1602
. ⋅10−19 )(1036
. ⋅10 −12 )(1017 )(1017 )
C jo = ( 25⋅ 10 )
−4 2
= 440 fF
2(1017 + 1017 )(0.838)
C j = 2C jo = 2 (440f ) = 622 fF (b) ANS
qA
Cd = ((Wp − x p ) npo + (Wn − xn ) pno )eV D /Vth
2Vth
Since N d = Na , it follows that x p = x n and pno = n po and we can write that:
qA
Cd = (Wn − xn ) pnoeVD /Vth
Vth
Under forward bias, the depletion layer shrinks so that Wn − xn ≈ Wn :
Cd = (qAWn / Vth ) pnoeVD /Vth
Since pno = ni2 / N d = 1020 / 1017 = 103 cm-3 ,
Cd = (1602
. ⋅ 10−19 ⋅ ( 25 ⋅10−4 )2 ⋅ 2 ⋅ 10 −4 / 2585
. m )103 e 0.7 /25.85m = 4.60 pF (c) ANS
In this forward-biased case, the diffusion capacitance is nearly an order of
magnitude larger than the depletion capacitance.
E6.8) The diode is one-sided since N d = 100 Na . Making the approximation:
N a = 1017 cm-3 and Dn = 188. cm2 / s
 Dp Dn 
2
qn ADn (1602 . ⋅10−19 ⋅1020 )(20 ⋅10−4 )2 (1880
. )
Io = qni2 A +  → i =
NW dNW 
n a NW p a p 10 ⋅10
17 −4

Io = 0121
. fA , ID = Io e = 01209
. f ⋅ e 0 .720 /26 m = 128 µA
VD /Vth

rd = Vth / ID = 26m / 128µ = 203 Ω (a) ANS

qε S N a N d (1602
. ⋅10−19 )(1036
. ⋅10 −12 )(1017 ⋅109 )
C jo = A = (20 ⋅ 10−4 )2
2( N a + N d )φB 2(1017 + 1019 )(0.953)
C jo = 372 fF , and with forward bias, C j = 2 Cjo = 2 (3715
. f ) = 525 fF (b) ANS
The diffusion capacitance is calculated for a one-sided diode and the depletion
width x p (on the order of nm ) is also neglected in comparison with Wp (on the
order of µm ). Using Eq. (6.62) in the case of N d >> N a :
ID  Wp  128 µ  (10−4 )2 
2

Cd =  =   = 2.62 pF (c) ANS

Vth  Dn  26m  18.80 
dp' p' dp' p'
E6.10) = gl − , steady -state ⇒ = 0 ⇒ gl = or p' = glτ p
dt τp dt τp
p' = (1020 )(250n ) = 2.5 ⋅1013 >> po ⇒ p = 25
. ⋅1013 cm −3 (a) ANS

PAGE 25
 As shown in Example 6.4 of the text, the solution to this type of differential
equation involves a decaying exponential which has a time constant of τ p . Or:
1 − e− t /τ p = 0.9 ⇒ t = −τ p ln(01
. ) = − (250 n )ln( 01
. ) = 576 ns (b) ANS
By symmetry, the excess decay also obeys an exponential transient:
e− t /τ p = 01
. ⇒ t = 576 ns for τ p = 250 ns , as before. (c) ANS
E6.12) The background equilibrium concentration,
n po = ni 2 / N a = 1020 / 1017 = 103 cm-3 , is negligible compared to n p ( x ) near
x = 0. So, the graph is of the form n p ( x ) = n p ( 0)e − x /Ln . Note that
n p ( Ln ) = n p (0)e − x /Ln x= Ln
= np ( 0)e−1 = 0367
. np (0 ) . And
0367
. . µm = Ln .
np (0 ) = 4.4 ⋅ 1015 cm-3 occurs near x = 15 (a) ANS
Ln = Dnτ n ⇒ τ n = Ln / Dn = (15
. ⋅10−4 )2 / 18.80 = 120
2
. ns (b) ANS
E6.14) The side that will first fail low-level injection is the n-type side, because the
minority carriers are being injected from a heavier-doped side. Note also that
the n-type side does not need as many p-type carriers to violate low-level
injection as does the p-type side.
pn = N a e− φj / th = 101 N d , φ j = φB − VD *
V

φB = Vth ln( N a N d / ni 2 ) = ( 26m )ln(1016 ⋅1015 / 1020 ) = 660 mV

VD * = φB − Vth ln(10 Na / N d ) = 0.660 − (26 m )ln(10 ⋅ 1016 / 1015 ) = 540 mV (a) ANS
Note that lolw-level injection fails at low bias for this diode due to the light
doping.
Ln = Dnτ n = ( 30.4)(100 n) = 174 . µm << Wp = 50 µm
Lp = Dp τ p = (19.4)(100 n ) = 109
. µm << Wn = 50 µm
The long base diode equation applies:
 D Dp 
Io = qAni2  n + 
N L N L 
a n d p

 ( 304
. ) (19.4) 
Io = (1602
. ⋅10−19 )(100 ⋅10 −4 )2 (1020 ) 16 −4 + −4 
 (10 )(17.4 ⋅10 ) (10 )(19.4 ⋅10 )
15

Io = 18.8 fA (b) ANS

There is absolutely no effect on the saturation current in doubling W since
Wp/ n >> Ln / p already! (b) ANS
E6.16) VO = Vth ln( ISUP / Io1 ) − Vth ln( ISUP / Io 2 ) = Vth ln( Io 2 / Io 1 ) = kT
q ln( Io 2 / Io 1 )
Now, Io 2 / Io1 = 10 / 10 = 10 , and
−16 −17

k / q = (1381
. . ⋅ 10−19 ) = 86 µV / K
⋅10−23 ) / (1602
So Vo = (86 µ )(10)T = 860T µV / K , where T is in Kelvins, gives a linear output
of voltage with respect to temperature.

PAGE 26
 0.4

0.2

VO in volts

0
250 300 350
T in Kelvins
VO
(a) ANS
To make two diodes with saturation currents of 10:1 ratio, the most obvious way
would be to scale identically-processed diodes by area. In other words, they are
both manufactured identically, except that one has ten times the cross sectional
(layout) area as the other. The tolerance on the saturation currents themselves
would be limited by lithographic accuracy, assuming that the vertical dimensions
and doping were identical. (b) ANS

PAGE 27
Chapter 7
The Bipolar Junction Transistor

EXERCISES
E7.2) IC < βF I B ⇒ saturation (a) ANS
VCE ≈ VCE( sat ) = 01
. V (b) ANS

(c) ANS
E7.4) The emitter current I E is positive, and thus, current is entering the emitter.
Also, VBC ≈ 07
. V , and VEC > V EC( sat ) , so the device is reverse-active. ANS
E7.6) − IC = 50 µA < βF ( − IB ) = 50(2.5µ ) = 125 µA ⇒ saturation (a) ANS
VEC ≈ VEC( sat ) = 01
. V (b) ANS

(c) ANS
E7.8) IB = 0 ⇒ cutoff (a) ANS

(b) ANS
PAGE 28
E7.10) Normally, for a pnp transistor, the current should be entering the emitter
junction. In this case, IE < 0 , i.e., the current is leaving the emitter junction.
Furthermore, VCB ≈ 0.7 V is the approximate diode-drop on voltage, and
VEC > V EC( sat ) , so the transistor is reverse active. ANS
E7.12) The minority carrier sketches look like:

ANS
E7.14) Please see E7.12 for characteristics of minority-carrier plot. (a) ANS
N aB = 1017 cm ⇒ DnB = 188
. cm 2 / s , N dE = 7.5 ⋅1018 cm-3 ⇒ DpE = 197
-3
. cm2 / s
n pBo = ni 2 / NaB = 10 20 / 1017 = 103 cm-3
qDnBn pBo AE (1602
. ⋅10 −19 )(18.80)(103 )(10 ⋅10−4 )2
IS = = = 0120
. fA
WB 0.25 ⋅10 −4
IC = I S eV BE / Vth ⇒ V BE = Vth ln( IC / IS ) = ( 25m )ln( 50µ / 01205
. f ) = 669 mV (b) ANS

( ) = (1 + ) = 0.999126
−1 −1
DpE N aBWB ⋅ −4 )
(1. 974 )(1017 )( 0. 2510
αF = 1 + DnB N dE WE ⋅ −4 )
⋅ 18 )( 0 . 410
(18. 80)( 7. 510

βF = α F (1 − α F ) −1 = 0.999126 / (1 − 0.999126) = 1140

forward active ⇒ IB = βF −1 IC = (1143)−1 (50 µ ) = 437 . nA (c) ANS
Or, do an algebraic manipulation as in EX7.1, and calculate:
⋅ 18 )( 0 . 4 )
βF = = = 1140
DnB N dE WE (18. 80)( 7. 510
DpE N aBWB . )( 1017 )( 0 . 25)
( 1974
(c) ANS
E7.16) gm = IC / Vth = 50µ / 25m = 2.00 mS (a) ANS
rπ = β F / g m = 1140 / 200
. m = 570 kΩ (b) ANS
ro = V An / IC = 20 / 50µ = 400 kΩ (c) ANS
E7.18) They’re identical transistors, except one has more emitter area. So:
IC2 I C1 = AE 2 AE1 = (9 × 7) (2 × 7) = 9 2 . (a) ANS
Since the currents are scaled, so is the base charging capacitance:
Cb 2 / Cb1 = 9 / 2 . (b) ANS

PAGE 29
 The base-collector junctions are ratioed like:
Cµ2 Cµ1 = (17 × 7) (10 × 7) = 17 10 = 17
. . (c) ANS
E7.20)
Ca rrier Co nc./Mo del Pa ra meter Increa se No Cha ng e Decrea se
minority elec. conc. at B-E junc. n p ( 0) ü
minority elec. conc. at B-C junc. n p (WB ) ü
majority hole conc. at B-E junc. p p ( 0) ü*
majority hole conc. at B-C junc. p p (WB ) ü
minority hole conc. at B-E junc. pn (− x BE ) ü
mnty. hole c onc. at ohmic cont. ü
pn (− x BE − WE )
transconductance gm ü
input resistance rπ ü
output resistance ro ü
base-charging capacitance Cb ü
base transit time τF ü
* p p (0) = p po + p p ′ (0) = N aB + n p ′ (0) ANS
E7.22) βF = α F (1 − α F )−1 = 0 .96 / (1 − 0.96) = 24 , βR = 0545 . / (1 − 0.545) = 120
. (a) ANS
− IE = − I ES ( e VEB /Vth
− 1) + α R ICS ( eVCB /Vth
− 1)
− I C = α F I ES ( eVEB /Vth
− 1) − I CS ( eVCB /Vth
− 1)
IB + IE + IC = 0 , and VEC = VEB − VCB
Since an expression in − I B is desired, mix it into Ebers-Moll early on:
− IB = IE + IC = (1 − α F ) I ES ( eVEB /Vth − 1) + (1 − α R ) ICS ( eVCB /Vth − 1)
Note that this is an expression in VEB and VCB . Modify this expression to terms
of VEB and VEC using KVL relations above, and then find another expression in
VEB and VEC later on to find the desired equation. Note
e −V EC = e −VEB +VCB = e −VEB eVCB :
− IB = (1 − α F )I ES eVEB /Vth + (1 − α R ) I CS eVCB /Vth − ((1 − α F ) IES + (1 − α R ) ICS )
− IB = (1 − α F )I ES eVEB /Vth + (1 − α R ) I CS e −VEC /Vth eV EB /Vth − ((1 − α F ) I ES + (1 − α R )I CS )
− IB = ((1 − α F ) I ES + (1 − α R ) ICS e −VEC /Vth )e VEB /Vth − ((1 − α F ) I ES + (1 − α R ) I CS )
Now is the time to find another expression involving − IC , VEC , and VEB :
− IC = α F IES (eV EB /Vth − 1) − ICS ( eVCB /Vth − 1) = α F I ES eVEB /Vth − ICS e VCB /Vth + I CS − α F I ES
− IC − ICS + α F IES = α F I ES eVEB /Vth − ICS e VCB /Vth = α F I ES eVEB /Vth − ICS e −VEC /Vth eVEB /Vth
− IC − ICS + α F IES = (α F IES − ICS e −VEC /Vth )eVEB /Vth
Great! Now there’s two expressions relating − I B , VEC , VEB , and − IC , VEC ,
VEB . Combine these two expressions by eliminating VEB to get a final

PAGE 30
expression in − I B and − IC . The first expression gives:
IB + ((1 − α F )I ES + (1 − α R ) ICS )
eVEB /Vth =
(1 − α F )I ES + (1 − α R ) ICS e − EC / th
V V

So substitute this into the second expression:

− IC =
( )
(αF I ES − ICS e −V EC /V th ) I B + ((1 − αF ) IES + (1 − αR ) I CS )
+ ICS − α F I ES (a) ANS
(1 − α F ) I ES + (1 − α R ) ICS e −V EC /V th

PAGE 31
 160

120

80

-IC in uA

40

0
0 1 2 3 4 5
VEC in volts
-IC (-IB=0.5 uA)
-IC (-IB=1 uA)
-IC (-IB=2 uA)
-IC (-IB=5 uA)
(b) ANS
0

IC in uA

12

16
5 4 3 2 1 0
VEC in volts
-IC (-IB=0.5 uA)
-IC (-IB=1 uA)
-IC (-IB=2 uA)
-IC (-IB=5 uA)
(c) ANS
E7.24) gmp = − I C / Vth = 25µ / 25m = 1.00 mS (a) ANS
N dB = 1016 cm-3 ⇒ DpB = 10 cm2 / s , N aE = 4 ⋅1017 cm-3 ⇒ DnE = 10 cm 2 / s
⋅ 7 )(1 .5 )
βF = = = 60 , rπ = β F / g m = 60 / 100
. m = 60 kΩ
DpB N aEWE (10 )( 410
DnE N dBWB (10 )(1016 )( 1)
(b) ANS
ro = V Ap / ( − IC ) = 25 / 25µ = 1 MΩ (c) ANS

PAGE 32
Chapter 8
Single Stage Bipolar/MOS Transistor Amplifiers

EXERCISES
 RL   Rin 
E8.2) v out =   ⋅10v i , v i =  v
 RL + Rout   RS + Rin  in
 RL   Rin 
⇒ v out =    ⋅10v in
 RL + Rout   RS + Rin 
vout  RL   Rin   10k   10 k 
= 10 ⋅    = 10 ⋅    = 980
. V/V (a) ANS
vin  RL + Rout   Rin + RS   10k + 100   10k + 100 
10v i  10   Rin 
iout = − = −  v
Rout + RL  Rout + RL   Rin + RS  in
Remember that iout is defined as the small signal current entering the port, and
thus the minus sign is needed above.
iout  10   Rin   10   10 k 
= −   = −   = − 980 µS (b) ANS
v in  Rout + RL   Rin + RS   100 + 10 k   10 k + 100 
Alternatively, one can use earlier results in solving this part instead:
iout = − vout / RL , so voutin = − voutin ⋅ R1L = − (9.80)(10k)-1 = −980 µS
i v
(b) ANS
10v i 10(iin Rin )
iout = − =− , since v i = iin Rin
RL + Rout RL + Rout
iout 10 Rin (10)(10k )
=− =− = − 990
. A/ A (c) ANS
iin RL + Rout 10k + 100
Or, v in = iin ( RS + Rin ) , = ( Rs + Rin ) = ( −980 µ )(100 + 10k ) = − 9.90 A / A
iout iout
iin v in (c) ANS
RL
v out = ⋅10 vi , v i = iin Rin ⇒
Rout + RL
v out 10 Rin RL (10)(10k )(10 k )
= = = 99.0 kΩ (d) ANS
iin Rout + RL 100 + 10k
= ( RS + Ri ) = ( 980
. )(100 + 10k ) = 99.0 kΩ
vout vout
Also valid, iin vin (d) ANS
E8.4) VOUT = 0 ⇒ IOUT . / 500µ = 5 kΩ
= 0 ⇒ I RC = I C , RC = VRC / IRC = VCC / IC = 25 (a) ANS
IC = I S eVBE /Vth ⇒ VBE = Vth ln( I CS ),
I

VBIAS = −VCC + V BE = Vth ln ( )− V

IC
IS CC ( ( ))− 25. = −183
= (25m ) ln 500 µ
10− 15
. V (a) ANS

Note that the voltage drop across the base resistor has been wisely neglected in
the above calculation.
V 25m
Rin CE = rπ = βo th = (100 ) = 5.00 kΩ (b) ANS
IC 500µ
V
Rout CE = ro // roc = A // RC , VA → ∞ ⇒ Rout CE = RC = 5 kΩ (b) ANS
IC

PAGE 33
 First, express the output voltage in terms of the controlled source and other
relevant circuit parameters. Then make required substitutions as follows:
 Rin 
v out = −( Gm vin )( Rout // RL ) = − ( gm )  ( v )( R // RL )
 Rin + RS  s out
IC 500µ v  Rin 
gm = = = 200
. mS , out = −( gm )  ( R // RL )
Vth 25m vs  Rin + RS  out
 5.00k 
 (10k // 5k ) = −65.4 V / V
v out
= (− 20.0m ) (c) ANS
vs  500
. k + 100
Note that the second term is nearly unity. When making rapid calculations, it
should be approximated as such because in practice, many other uncertainty
terms are likely to come into play.
 Rout   Rin 
iout = g mv in   , v in =  v
 Rout + RL   Rin + RS  s
To do the last part of the problem, the output current has been expressed in terms
of the controlled current source and the current divider consisting of Rout and
RL . Substituting and rearranging:
iout  Rout   Rin   20.0m ⋅ 5k   500 . k 
= gm  =   = 654
. mS (d) ANS
vs  Rout + RL   Rin + Rs   5k + 10 k   5.00k + 100 
(Or, just note that − iout RL = vout ⇒ =−
iout vout 1
vs vs RL .)
E8.6) Although the device in this problem is pnp, the configuration is the same, so the
small signal analysis does not change. Large signal analysis changes slightly,
since proper signs must be introduced in calculating the bias voltage.
RC = 5 kΩ , VBIAS = VDD − V BE = 5 − 0815
. = 418
. V (a) ANS
Rin = 2.59 kΩ , Rout = 5 kΩ (b) ANS
v out / vs = −621. V/V (c) ANS
iout / v s = 6.21 mS (d) ANS
E8.8) VOUT = 0 ⇒ IOUT = 0 ⇒ I RD = ID , RD = V RD / I RD = VDD / ID = 2.5 / 500 µ = 5 kΩ (a) ANS
Assuming saturation:
( 2)(500µ )
ID = 12 µn Cox WL (VGS − VTn ) ⇒ VGS = VTn +
2 2 ID
= 1+ = 2.41 V
µn Cox L
W
( 50µ )( 303 )
VBIAS = VGS − VDD = 2.414 − 2.5 = −85.8 mV (a) ANS
Rin CS → ∞ , Rout CS
= ro // roc = ( λn I D ) −1 // RD , but λn → 0 ⇒ Rout CS = RD = 5 kΩ (b) ANS
vout = − Gm ( Rout // RL )v in = − gm ( Rout // RL )v s
Note that v in = v s because Rin → ∞ .
gm = 2 µn Cox WL ID = (2 )(50µ )( 303 )(500 µ ) = 707 µS
v out
. µ )( 5k // 10k ) = −2.36 V / V
= − gm ( Rout // RL ) = − (7071 (c) ANS
vs

PAGE 34
  Rout   Rout 
iout = g mv in   = gm v s  
 Rout + RL   Rout + RL 
iout  Rout   5k 
= gm . µ )
 = ( 7071  = 236 µS (d) ANS
vs  Rout + RL   5k + 10k 
E8.10) The analysis here is almost identical to the previous two problems. The only
difference would be a few signs in the large signal analysis:
RD = 5 kΩ , − VGS = 294
. V / V , so that means VBIAS = 206
. V (a) ANS
Ri → ∞ , Rout = 5 kΩ (b) ANS
v out / vs = −149
. V/V (c) ANS
iout / vs = 149 µS (d) ANS
2ID (2)(100µ )
E8.12) VGS = VTn + = 1 + = 163
. V
L µn Cox ( 505 )(50µ )
W

VBIAS = −VDD + VGS = −2.5 + 1632

. = −0.868 V (a) ANS
Rin CS → ∞ , Rout CS
= ro // roc = ( λn I D ) // roc
−1
(b) ANS
roc → ∞ ⇒ Rout = ( λn I D ) −1 = (0.03 ⋅100µ ) −1 = 333 kΩ (b) ANS
v out / vs = − gm ( Rout // RL ) , gm = 2 µn C W
ox L ID = (2 )(50µ )( )(100µ ) = 316 µS
50
5

v out / vs = ( − 316.2µ )(3333

. k // 5k ) = − 156
. V/V (c) ANS
Again, it is advisable to make quick approximations as appropriate. Note that
( 333k // 5k ) ≈ 5k , and that this particular approximation would have resulted in
a negligible 2% error.
 Rout   3333 . k 
iout / vs = gm   = (316.2µ )  = 312 µS (d) ANS
 Rout + RL   3333
. k + 5k 
The second term is near unity and has a negligible effect on the intrinsic
transconductance.
E8.14) The large signal analysis in the PMOS case is similar to the last two exercises,
except for a few signs. The idea is just the same , and so:
− VGS = 180
. V , so VBIAS = 0705
. V (a) ANS
Watch out! Solving for the output resistance here is a little bit different:
Rout CS = ro // roc , roc → ro
⇒ Rout = 12 roc = ( 2λp ISUP )−1 = (2 ⋅ 0.02 ⋅100µ ) −1 = 250 kΩ (b) ANS
And of course, Rin → ∞ (b) ANS
The analysis of the rest of the problem does not change:
v out / vs = − gm ( Rout // RL ) = ( −182.6µ )( 250k // 5k ) = −0.895 V / V (c) ANS
iout / vs = ( gm Rout ) / ( Rout + RL ) = (182.6µ ⋅ 250k ) / ( 250k + 5k ) = 179 µS (d) ANS
E8.16) Finding the DC bias here is similar to E8.4a, except that there is no collector
resistor: VBE = V th ln( I C / IS ) = (25m )ln( 200µ / 10 −16 ) = 0.708 V
VBIAS = −VCC + V BE = −179 . V (a) ANS

PAGE 35
 Rin CE = rπ = βoVth / I C = (100)( 25m ) / ( 200µ ) = 125
. kΩ (b) ANS
Rout CE
= ro // roc = ro // ro = 12 ro = V A / 2 IC = ( 20) / (2 ⋅ 200µ ) = 50 kΩ (b) ANS
The analysis here is similar to that of E8.4c:
gm = IC Vth = 200µ 25m = 8.00 mS
vout  Rin 
= − gm ( Rout / / RL ) 
vs  Rin + RS 
v out  12.5k 
= ( − 8.00m )(50 k / /5 k )  = − 26.0 V / V (c) ANS
vs  12.5k + 5k 
iout  Rin   Rout   12.5k   50k 
= gm    = (8.00m )   = 519
. mS (d) ANS
vs  Rin + RS   Rout + RL   12.5k + 5k   50 k + 5k 
−1
E8.18) IC = βo IB ⇒ IBIAS = βo I SUP = 100−1 ⋅ 200 µ = 2 µA (a) ANS
Rin CE = βoVth / IC = 100 ⋅ 25m / 200 µ = 125
. kΩ (b) ANS
Rout CE
= ro // roc = VA / (2 I C ) = 25 / (2 ⋅ 200 µ ) = 625
. kΩ (b) ANS
 Rout   Rout   RS   Rout 
iout = Gmv in   = Gmiin Rin   = g m Rin   i
 Rout + RL   Rout + RL   RS + Rin   Rout + RL  s
iout I C  RS   Rout  200 µ  5k   62.5k 
= Rin   = 12.5k  
is Vth  RS + Rin   Rout + RL  25m  5k + 12.5k   625
. k + 5k 
iout is = 265
. AA (c) ANS
−1 iout
iout = − vout / RL ⇒ v out / is = − R L is = − (5k ) −1 ⋅ (26.5) = − 529
. mΩ (d) ANS
A note about this circuit (and others like it): There are subtleties involved with
simulating this circuit--the IBIAS required to set VOUT to zero in a SPICE deck is
significantly different from what was just calculated. The value (as calculated
by SPICE) of I B , the current that enters the transistor, is correct. This is because
there is a DC voltage across the “fake” source resistance RS (“fake” in the sense
that it is really associated with just the small signal source, not the large signal
bias), and that SPICE correctly computes a “meaningless” current through it,
part of which must be supplied by IBIAS for the circuit to simulate properly in the
context of a specified problem. The solution is to do as the text says--“do a DC
sweep of ( IBIAS or VBIAS ) to tune VOUT (or whatever) to zero (or whatever...)”.
By doing so, one can simulate the circuit properly. This situation is likely to
cause confusion in simulating circuits, and the explanation should be pointed out
where appropriate.
E8.20) VBE = Vth ln( I C / IS ) = ( 25m )ln( 200µ / 10−15 ) = 0.651 V
VBIAS = V BE + V RE − VCC = VBE + IC R E − VCC = 0.651 + (200µ )(1k ) − 25
. = − 165
. V (a) ANS
. mS, D = (1 + gm RE ) = (1 + (8.00m )(1k )) = 9.00
gm = I C / Vth = 80
Rin CE = Drπ = D( βo / g m ) = (9)(100) / 8.00m = 113 kΩ (b) ANS
deg

PAGE 36
 Rout CEdeg
= Dro // roc = Dro // ro = ro (D−1 + 1) = VA
−1
(I (D
C
−1
)
+ 1)

Rout CEdeg
(
= 25 200 µ(9.00−1 + 1) = 113 kΩ ) (b) ANS

 Rin  g
v out = − Gm vin ( Rout // RL ) = − Gm   ( Rout // RL ) vs , Gm = m
 Rin + RS  D
v out g  Rin  8.00m  112.5k 
=− m  ( Rout // RL ) = −   (1125
. k // 10 k )
vs D  Rin + RS  9.00  112.5k + 1k 
v out
= −8.09 V / V (c) ANS
vs
iout v 1 − ( − 8.09)
= − out = = 809 µS (d) ANS
vs vs RL 10k
E8.22) VOUT = 0 ⇒ IOUT = 0 ⇒ I BIAS = − I SUP = − 100 µA (a) ANS
−1
Rin CB = g m . m / 100 µ = 259 Ω
= Vth / I C = 2585 (b) ANS
Rout CB
= ro (1 + gm (rπ // RS )) // roc = ro (1 + gm ( rπ // RS )) // ro
K = 1 + gm (rπ // RS ) = 1 + ( IC / Vth )((βoVth / IC ) // RS )
K = 1 + (100µ / 25.85m )((100 ⋅ 2585
. m / 100 µ ) // 10 k) = 28.9
Rout = ro ( K + 1) = (VA / IC )( K + 1)−1 = (25 / 100 µ )(28.89−1 + 1)−1 = 242 kΩ
−1 −1 −1
(b) ANS
 Rout   RS   Rout 
iout = Aii in   = − is   
 Rout + RL   RS + Rin   Rout + RL 
iout  10 k   2416 . k 
= −   = −0.936 A / A (c) ANS
is  10k + 258.5  2416
. k + 10k 
−1
iout = − vout / RL ⇒ vout / is = − RL (iout / is ) = −10 k−1 ( −0.9360) = 93.6 µΩ (d) ANS
E8.24) IBIAS = − ISUP = −100 µA (a) ANS

( 2µ C ) = ( (2 )(75µ )( )
−1 −1
Rin CG = gm −1 = n
W
ox L ID 10
5 )(100µ ) = 577
. kΩ (b) ANS
Rout CG
= ro (1 + gm RS ) // roc = ro ((1 + gm RS )−1 + 1)−1 since roc = ro
Rout = (λn I D ) −1 ((1 + gm RS ) −1 + 1)−1 = ( 0.02 ⋅100 µ )−1 ((1 + 1732
. µ ⋅10k )−1 + 1)−1
Rout = 366 kΩ (b) ANS
iout  RS   Rout   10k   366.0k 
= −   = −   = − 0.617 A / A (c) ANS
is  RS + Rin   Rout + RL   10 k + 5.774k   3660
. k + 10 k 
v out 1 iout
=− = − (10 k )−1 ( 06171
. ) = 617. µΩ (d) ANS
is RL is
E8.26) VOUT = 0 ⇒ IOUT = 0 ⇒ ISUP = IC
. m ) ln(100 µ / 10 −15 ) = 0.655 V
VBE = Vth ln( IC / IS ) = (2585
VBIAS = VBE + VE = VBE + VOUT = VBE = 0.655 V (a) ANS
Of course, the base voltage drop has been neglected in calculating the voltage
drop. VE refers to the DC voltage at the emitter (or output), which is zero.
Rin CC = rπ + βo ( ro // roc // RL ) , roc = ro , so Rin = rπ + βo ( 12 ro // RL )

PAGE 37
 Rin = βo + βo ( 2 IAC // RL ) = 100( 25100 µ + ( 2 ⋅100 µ // 100)) = 35.8 kΩ
Vth V . 85m 20
IC (b) ANS
−1
Rout CC
=g m + RS / βo = IC / Vth + RS / βo = 2585
. m / 100 µ + 1k / 100 = 269 Ω (b) ANS
Inspecting the two-port model for the common-collector amplifier:
 RL   Rin  v  RL   Rin 
v out = vin   , v in =   vs , so out =   
 RL + Rout   Rin + RS  v s  RL + Rout   Rin + RS 
v out  100   35.84k 
=   = 0.264 V / V (c) ANS
v s  100 + 268.5  35.84k + 1k 
iout v 1
= − out = − 2.64 mS (d) ANS
vs vs RL
2 I SUP ( 2)(200 µ )
E8.28) VBIAS = VTn + = 1 + = 189
. V (a) ANS
µn Cox WL ( 50µ )( 20
2 )

Rin CD → ∞ (b) ANS

( 2µ C ) = ( ( 2)(50 µ )( )
−1 −1
Rout CD
= gm −1 = n
W
ox L ID 20
2 )(200 µ ) = 2.24 kΩ (b) ANS
v out / vs = RL / ( RL + Rout ) = 100 / ( 2236 . k + 100 ) = 428. mV / V (c) ANS
. m ) / 100 = − 428 µS
iout / vs = − v out / ( v s RL ) = ( − 428 (d) ANS
E8.30) Again, the small signal analysis here is almost identical to that to E8.28. There
is only a slight difference in signs with the DC bias analysis.
VBIAS = −197 . V (a) ANS
Rin → ∞ , Rout = 316. kΩ (b) ANS
v out / vs = 30.7 mV / V (c) ANS
iout / vs = 307 mS (d) ANS

PAGE 38
Chapter 9
Multistage Amplifiers

EXERCISES
E9.2) The overall voltage gain of the CS-CS stage is:
( )
v s = gm1ro1 gm 2ro 2 RL + Rout 2 , Rout 2 = ro 2
vout RL

And for the CS-CD stage:

( )
v s = gm1ro1 RL +Rout 2 , Rout 2 = g m2
vout RL −1

So, whether or not if one topology is better than the other depends on:
gmro ( ) vs. (
RL
RL +ro
RL
RL + g m −1 )
Again, this is a very similar situation to E9.1. In fact, the rest of the analysis is
identical. So, one could compare these unlike terms and draw conclusions
accordingly:
gmro ( )= 1m ⋅100 k(
RL
RL +ro
100
100+100 k ) = 99.9m > ( RL
RL + g m− 1 )= ( 100
100 +1 m − 1 )= 90.9m
So, in the first case, CG-CS is better. (a) ANS
gmro ( )= 1m ⋅100 k(
RL
RL +ro
10 k
10k +100 k ) = 9.09 > (R +Rg
L
L

m
−1 )= ( 10 k
10k +1 m −1 )= 909 m
The second case also presents CG-CS as the better transresistance amplifier. (b) ANS
( )= 100 µ ⋅10M ( . m > (R + g
)= 999 )= ( )= 9.90m
RL 100 R L 100
gmro RL +ro 100 +10 M L m
−1
100 +100 µ −1

In the third case, the CG-CS configuration is more appropriate, again. (ca) ANS
gmro ( )= 100 µ ⋅10M (
RL
RL +ro
10 k
10 k +10 M )= 999m > ( RL
RL + g m −1 )= ( 10 k
10k +100 µ −1 )= 500 m
Again, here, the CG-CS configuration is better as a transresistance amplifier. (cb) ANS ANS
E9.4) To match two stages to form an amplifier with the highest intrinsic current gain
would first require a CG-amp for the input/first stage. Then a CS amp follows to
provide high gain. (a) ANS
The overall current gain would be:
iout RS
( ) Rout 2
( −1
)
is = RS + Rin1 Rout 1Gm 2 Rout 2 +RL ; Rin 1 = gm , Rout1 = (1 + gm RS )ro , Gm 2 = g m ,

R out 2 = ro
iout
is =( 1k
1k +1 m −1 )⋅100k ⋅ 1m ⋅ ( 100 k
100 k +100 )= 50.0 A / A (b) ANS
=( )⋅100k ⋅ 1m ⋅ ( 100 k +10k ) = 455
iout 1k 100 k
is 1k +1 m −1
. A/ A (c) ANS
E9.6) For the voltage amplifier, have the CE as the front end--the degradation caused
by a finite rπ (which is equal to βo / (1+ βo ) ) is outweighed by the huge voltage
gain caused by large gm and ro (larger in comparison with the MOS). Then,
protect this gain by putting a CS stage next. Then, because the load resistance is
so low, the last stage should be a CC. (a) ANS

PAGE 39
 Because the source resistance is pretty low, a CB input stage makes sense.
Then, it is followed by a CS stage for gain, and then a CC stage to handle the
low load resistance. (b) ANS
E9.8) VOUTmin = 0.5 V is the lowest that the output can go with ISUP2 on the verge of its
constant-current region. This leads to a (plausible) VIN min = 2.2 V . However, in
the upper end, the NMOS is first to go triode, with VDS = 05 . V . The output
would be VOUTmax = 5 − 0.5 − 0.7 = 38 . V . The input would be
VIN max = 5 − 0.5 + 15. = 6 V , this is not possible if the input could not make 6 V .
So, VIN max = 5 V , which leads to a VOUTmax = 5 − 15 . − 0.7 = 28. V. ANS
E9.10) VOUT min = V IBIAS + VCEsat + (VGS − VTn ) = 0.5 + 0.2 + 12
. − 1 = 0.9 V , assuming the
proper gate/base biases. ANS
E9.12) ID = I REF = 12 µn Cox WL (VOUT − VTn )2 = (VDD − VOUT ) / RREF
L = 2(VDD − VOUT ) / ( R µn Cox (VOUT − VTn ) )
W 2

W
. ) / (10k ⋅ 50µ ⋅ (15
L = 2(5 − 15 . − 1)2 ) = 56 (a) ANS
rS = gm−1 = ( µn Cox WL (VOUT − VTn )) −1 = ( 50µ ⋅ 56 ⋅ (15
. − 1))−1 = 714 Ω (b) ANS
(Actually, the incremental source resistance will be less than this. The resistor
used in generating IREF lowers rT to 667 Ω .) (b) ANS
To make rS > 500 Ω , note that VOUT won’t change drastically, though it will
decrease. So, WL = ( µn Cox rT (VOUT − VTn ))−1 = (50µ ⋅ 500 ⋅ (15
. − 1))−1 = 80 . (c) ANS
2 Rµn Cox L (VOUT − VTn ) + (VOUT − VTn ) − VDD + VTn = 0
1 W 2

A quadratic equation solves VOUT = 142 . V. (d) ANS

Same deal with the quadratic--VOUT = 139 . V for VDD = 45 . V. (e) ANS
E9.14) (Part (b) misprinted VOUT2 . It should be VOUT2 = 2.0 V .)
. m )ln( 200µ / 10 −15 ) = 673 mV
VBE = VOUT1 = Vth ln( I C / IS ) = ( 2585 (a) ANS
( 2 )( 200 µ )
VGS = VTn + = 0.7 + = 159
2 I REF
µn C ox WL ( 50 µ )( 20
. V,
2 )

VOUT 2 = VGS + VOUT1 = 1594. + 0.673 = 227. V (a) ANS

VOUT 3 = VOUT 2 + VBE = 2.267 + 0.673 = 294 . V (a) ANS
L = 2 I REF / ( µn Cox (VOUT 2 − VOUT 1 − VTn ) )
W 2

L = ( 2)( 200µ ) / (50µ ⋅ (2.0 − 0.673 − 1) ) = 75

W 2
(b) ANS
VOUT 3 = VOUT 2 + VBE = 2 + 0673 . = 2.67 V (c) ANS

( 2µ C ) = (2µn Cox I REF rT )

−1
2 −1
E9.16) rT = n
W
ox L IREF ⇒ W
L = (2 ⋅ 50µ ⋅ 100µ ⋅1k 2 )−1 = 100 (a) ANS
(WL )2 = 3(WL )1 = 300 for I2 = 3 I REF (b) ANS
ro = (λn ID )−1 = ( 0.03 ⋅ 300µ )−1 = 111 kΩ (c) ANS
E9.18) By proportionality, (WL )1 = (WL )REF = 10040 (102 ) = 42 .
IOUT 1
I REF (a) ANS
(WL )2 = IOUT 2
I REF (WL )REF = 100
100 ( 2 ) = 2
10 10
(b) ANS
(WL )3 = II
OUT 3
REF
(WL )REF = 250
100 ( 2 ) = 2
10 25
(c) ANS

PAGE 40
 (WL )1 = II
OUT 1
REF
(WL )REF = 4020 (102 ) = 202 (da) ANS
(WL )2 = II
OUT 2
REF
(WL )REF = 10020 (102 ) = 502 (db) ANS
(WL )3 = II
OUT 3
REF
(WL )REF = 25020 (102 )= 1252 (dc) ANS
(WL )1 = II
OUT1
REF
(WL )REF = 100 ( 2 )= 122
40 30
(ea) ANS
(WL )2 = II
OUT 2
REF
(WL )REF = 100
100 ( 2 ) = 2
30 30
(eb) ANS
(WL )3 = II
OUT 3
REF
(WL )REF = 250
100 ( 2 ) = 2
30 75
(ec) ANS
E9.20) Inverting the circuit:

ANS

PAGE 41
Chapter 10
Frequency Response

EXERCISES
YR2
E10.2) Io = YC +YR1 +YR2 IS (current divider), YR = R −1 , YC = jωC
R2 −1
= = R2 = = 110 µ + jω1 (10−12 ) = 1+ j ( ω )
Io Vo Io 1 100 k 11
Is R1− 1 +R2 − 1+ jωC
, Is IS R1 − 1 + R2 − 1 + jωC 110 M

Factoring the numerator and denominator into the forms of ∑ (1+ j ωωn ) makes it
n

easy to see the corner frequencies of the zeroes and poles. The first circuit has
11 = 79 dB and has a 3dB point at ω = 110 Μrad / s , or at
an initial DC gain of 100 k

f = 17.5 MHz . The following are appropriate magnitude (in dB) and phase (in
degrees) plots:
80

60

Magnitude
40 in dB-ohm
6 7 8 9 10
1 10 1 10 1 10 1 10 1 10
Angular Frequency in rad/s, log scale
VO/IS
(a) ANS
10
0
10
20
30
40
50
60
70
Phase 80
in degrees
90
100 6 7 8 9 10
1 10 1 10 1 10 1 10 1 10
Angular Frequency in rad/s, log scale
VO/IS
(a) ANS
Evaluating the transfer function for the second set of parameters:
Is = R −1 + R −1 + jωC = 10. 01m+ jω 10−12 = 1+ j( ω )
Vo 1 1 100 k 1001
1 2 ( ) 10 .01G

PAGE 42
 40

20

Magnitude
0 in dB-ohm
8 9 10 11 12
1 10 1 10 1 10 1 10 1 10
Angular Frequency in rad/s, log scale
VO/IS
(b) ANS
10
0
10
20
30
40
50
60
70
Phase 80
in degrees
90
100 8 10 11 12
1 10 1 10 1 10 1 10 1 10
Angular Frequency in rad/s, log scale
VO/IS
(b) ANS
And for the third set of parameters:
Is = R −1 + R −1 + jωC = 110 µ+ jω 10−12 = 1+ j( ω )
Vo 1 1 100 k 11
1 2 ( ) 110 G

20

40

Magnitude
60 in dB-ohm
9 10 11 12 13
1 10 1 10 1 10 1 10 1 10
Angular Frequency in rad/s, log scale
VO/IS
(c) ANS
10
0
10
20
30
40
50
60
70
Phase 80
in degrees
90
100 9 10 11 12 13
1 10 1 10 1 10 1 10 1 10
Angular Frequency in rad/s, log scale
VO/IS
(c) ANS
PAGE 43
 =
Vo Io
Note that Is IS R2 . All the phase plots are identical to those they correspond to
of E10.1. However, all the magnitude plots will be shifted up by a constant
value, namely 20 log R2 = ( 20)log(100k ) = 100 dB . (Remember that
log xy = log x + log y .) Another change that isn’t reflected in the bode plots is
the inherent units of the transfer function, which would now be in ohms.
A (1+ jωτ )( 1+ jωτ )
E10.4) H( j ω ) = (1+ jωτ3 )(1+ j1ωτ4 )(1+2jωτ5 ) , where τ1 and τ 2 are the time constants for the
zeroes, and τ 3 , τ 4 , and τ 5 are the time constants for the poles (i.e.
τ = ωo −1 = (2π fo ) −1 ), and the initial low-frequency gain is A = 500 (a) ANS
τ1 = 15.92 µs , τ 2 = 159.2 ns , τ 3 = 1592
. µs , τ 4 = 15.92 ns , τ 5 = 1592
. ns (a) ANS
100

80

60

Magnitude in dB

40
1 103 1 104 1 105 1 106 1 107 1 108 1 109 1 1010 1 10111 1012
Angular Frequency in rad/s, log scale
H
(b) ANS
60

30

30

60

Phase 90
in degrees

120
1 103 1 104 1 105 1 106 1 107 1 108 1 109 1 1010 1 10111 1012
Angular Frequency in rad/s, log scale
H
(b) ANS
A (1+ jωτ3 )( 1+ jωτ4 )(1+ jωτ5 )
In this case, H( j ω ) = (1+ jωτ1 )( 1+ jωτ2 ) , where the τ ’s and A are as above (ca) ANS

PAGE 44
 40

60

80

Magnitude in dB
100 3 4 5 6 7 8 9 10 11 12
1 10 1 10 1 10 1 10 1 10 1 10 1 10 1 10 1 10 1 10
Angular Frequency in rad/s, log scale
H
(cb) ANS
120

90

60

30

Phase 30
in degrees
60 3 4 5 6 7 8 9 10 11 12
1 10 1 10 1 10 1 10 1 10 1 10 1 10 1 10 1 10 1 10
Angular Frequency in rad/s, log scale
H
(cb) ANS
2 µn Cox WL I D ( 2 )( 50 µ )( 50/ 2 )( 1µ )
E10.6) f T = (2π )−1 Cgsm = (2π ) −1 23 WLCox +WCov = (2π )−1 23 ( 50)( 2 )(2 .3 f )+( 50 )(0.5 f ) = 44.6 MHz
g
(a) ANS
2µ C W
I ( 2 )( 50µ )( 50 /2 )( 100 µ )
f T = (2π )−1 Cgsm = (2π ) −1 23 WLCn oxox+WC = (2π )−1 23 ( 50)( 2 )(2 .3 f )+( 50 )(0.5 f ) = 446 MHz
g L D
(b) ANS
ov

2µ C W
I ( 2 )( 50µ )( 50 /2 )( 100 µ )
f T = (2π )−1 Cgsm = (2π ) −1 23 WLCn oxox+WC = (2π )−1 23 ( 50)( 2 )(2 .3 f )+( 50 )(0.5 f ) = 4.46 GHz
g L D
(c) ANS
ov

2µ C W
I ( 2 )( 50µ )( 25/2 )( 100 µ )
f T = (2π )−1 Cgsm = (2π ) −1 23 WLCn oxox+WC = (2π )−1 23 ( 25)( 2 )(2 .3 f )+( 25 )(0.5 f ) = 631 MHz
g L D
(d) ANS
ov

(Actually, it’s okay to use 2Cov instead of just Cov in the above equations to
include the effect of Cgd . Then transition frequency will be slightly lower.)
E10.8) This is different from E10.7 in that µn Cox changes to µp Cox .
− Av = gmro = 2µ pCox WL ISUP ( λn I SUP ) −1 = (2 )(25µ )( 502 )( 50µ )(0.05 ⋅ 50µ )−1
− Av = 100 V / V (a) ANS
CM = (1 − Av )Cgd = (1 − Av )CovW = (1 + 100 )(0.5f )(50 ) = 253
. pF
Cgs = 23 WLCox + WCov = 23 (50)( 2)(2.3f ) + (50)( 0.5f ) = 178 fF
C = CM + Cgs = 253. p + 1783 . f = 2.70 pF , R = RS = 10 kΩ
ω3dB = ( RC )−1 = (10 k ⋅ 270
. p) −1 = 37 Mrad / s (b) ANS
ω3dB = b1−1 = ( RS Cgs + RS (1 − Av )C gd + ro Cgd ) −1
ω3 dB = ( RS Cgs + RS (1 − Av )Cgd + (λn I SUP )−1 Cgd )−1
ω3 dB = (10k ⋅178.3f + 10 k(1 + 100)( 05
. f )(50) + (0.05 ⋅ 50 µ )−1 (0.5f )( 50))−1
ω3dB = 27 Mrad / s (c) ANS

PAGE 45
E10.10) The gains work out as follows:
v s = Rin +RS
vout Rin
( RL
RL + Rout )( )
Rin = rπ + βo RL = βo (Vth / I C + RL ) ( ro >> RL )
Rout = gm −1 + RS / βo = Vth / IC + RS / βo
vout
vs =( βo (Vth / IC +RL )
βo (Vth / IC + RL ) + RS )( RL
RL +Vth/ I C +RS / βo )
=( )( )= 0.642 V / V
vout (100 )( 25m /100 µ+500 ) 500
vs ( 100 )( 25 m/100 µ +500 )+1k 500 +25m /100 µ +1 k /100 (a) ANS
=( )( )= 0.681 V / V
vout (100 )( 25m /100 µ+1k ) 1k
vs (100 )( 25 m /100 µ +1 k ) +10 k 1 k +25m /100 µ+10k /100 (b) ANS
=( )( )= 0.408 V / V
vout ( 100 )( 25m /100 µ+ 250 ) 250
vs ( 100 )( 25m /100 µ +250 )+5 k 250+ 25m /100 µ+5 k /100 (c) ANS
From the Miller approximation:
(
ω3 dB = (RS // ( rπ + β o RL )) 1+ Cgmπ RL + Cµ ( ))
−1

((R // ( β ( R + V ( ))
−1

th / I C )))
C je + I C τ F / Vth
ω 3dB = S o L 1 + I C R L / Vth + Cµ

( ( ))
−1
15f +100 µ⋅50p /25 m
ω3 dB = (1k // (100( 500 + 25m / 100µ ))) 1+100 µ⋅500 / 25m + 10f = 125
. Grad / s (a) ANS

= ((10k // (100(1k + 25m / 100µ )))( + 10f ))

−1
15f +100 µ⋅ 50p/ 25m
ω3 dB 1+100 µ⋅1 k /25 m = 2.04 Grad / s (b) ANS

= ((5k // (100(250 + 25m / 100µ )))( + 10f ))

−1
15 f +100 µ ⋅50p /25 m
ω3 dB 1+100 µ⋅250 /25 m = 189
. Grad / s (c) ANS
For those who used the OCTC formula, the ω3dB ’s calculate out to:
ω3dB = 8.46 Grad / s , ω3dB = 152 . Grad / s , ω3dB = 167 . Grad / s (a,b,c) ANS
E10.12) v out / vin = ( RL / ( RL + Rout )) ⇒ Rout = (1 − vout / v s ) RL / ( v out / v s )
−1
Rout = gm − = 2µn Cox WL I D ⇒ W = L( 2µn Cox I D Rout )−1
1 2

L( vout / v s )2 ( 2µ )(0.3)2
W= = = 147 µm
2µn Cox I D (1 − v out / v s )2 RL2 (2 )(50µ )(100µ )(1 − 0.3)2 (500 )2
So, with a device size of W
L = 147
2 ( Rout = 117
. kΩ ), Eq. 10.86 evaluates to
∑τ C = 623 ps , which certainly meets the frequency response spec. So, W
L = 147
2 . (a) ANS

L( vout / v s )2 (2µ )( 0.3)2

W= = = 36.7 µm
2µn Cox I D (1 − v out / v s )2 RL2 (2 )(50µ )(100µ )(1 − 0.3)2 (1k )2
With a device size of W
L = 362.7 (and other parameters as given/calculated), Eq.
10.86 evaluates to ∑τ C = 1192 ps ⇒ ω3dB = 839 Mrad / s , which meets the
frequency response spec. So, W
L = 362.7 . (b) ANS
L( vout / v s )2 (2 µ )(0.3)2
W= = = 588 µm
2µn Cox I D (1 − v out / v s )2 RL2 (2 )(50µ )(100µ )(1 − 0.3)(250 )2
This time, ∑τ C . ns ⇒ ω3 dB = 109 Mrad / s barely clears spec.
= 917 (c) ANS
E10.14)
iout
is =− ( RS
RS + Rin )( Rout
Rout + RL )= −( RS
RS +g m −1 )( ro (1+ g m RS )
ro ( 1+g m RS )+ RL )
−
gm = 2 µn Cox WL ISUP = ( 2)(50 µ )( 100
2 )(200 µ ) = 1 mS , gm = 1 kΩ
1

ro = (λn ISUP )−1 = (0.05 ⋅ 200µ )−1 = 100 kΩ

PAGE 46
 ω3dB = (Cgs ( gm −1 // RS ) + RLCgd ) −1
Cgs = 23 WLCox + WCov = (100)( 23 ⋅ 2 ⋅ 2.3f + 05
. f ) = 357 fF
Cgd = Cov W = ( 05
. f )(100) = 50 fF
iout
is =− ( )( RS
RS + g m−1
ro ( 1+g m RS )
ro ( 1+g m RS ) +RL )= −( 100
100 +1 k )(100100k ( 1k+(11m+1⋅100 ) +10 k ) = −833
m ⋅100 )
. mA / A (a) ANS
ω3dB = (356.7f (1k // 100 ) + 10 k ⋅ 50f )−1 = 188
. Grad / s (a) ANS
iout
is =− ( )( RS
RS + g m−1
ro ( 1+g m RS )
ro ( 1+g m RS ) +RL )= −( )( 1k
1 k +1 k
100 k (1+1 m⋅1 k )
100 k ( 1+1 m⋅1 k )+100 k )= −0333
. A/A (b) ANS
ω3dB = (356.7f (1k // 1k ) + 100 k ⋅ 50f )−1 = 193 Mrad / s (b) ANS
iout
is =− ( )( RS
RS + g m−1
ro ( 1+g m RS )
ro ( 1+g m RS ) +RL )= −( 500
500 +1k )(100100k(k1+(11+m1m⋅500⋅500) +)5k )= −0.333 A / A (c) ANS
ω3dB = (356.7f (1k // 500 ) + 5k ⋅ 50f )−1 = 2.71 Grad / s (c) ANS
E10.16) Neglecting interstage loss,
(
vs = g m1 rπ 1 + RS
iout rπ 1
)(
ro 2 (1+ gm 2 ( ro 1 //rπ 2 ))
)
ro 2 (1 +g m 2 ( ro1 / /rπ 2 ))+ RL , gm1, 2 = I SUP / Vth = 100µ / 26m = 385
. mS
rπ1 ,2 = βo / gm1,2 = 100 / 385
. m = 26 kΩ , ro1 ,2 = VA / ISUP = 25 / 100 µ = 250 kΩ
iout
vs . m )(2626k +k1 k )
= (385 ( 250 k (1+3 .85 m ( 26k// 250k ))
250 k (1+3. 85 m ( 26k // 250 k )) +50k )= 372
. mS ANS
For the cascode, Ccs and CL appear in parallel to Cµ 2 , so that:
ω3 dB = ( RS ( Cπ1 + 2Cµ1 ) + ( Cπ 2 / gm2 ) + RL ( Cµ 2 + Ccs + CL )) −1
Cπ 1,2 = C je + gm1,2τ F = 15f + (3.85m )(50 p) = 208 fF
ω3dB = (1k (208.4f + 2 ⋅10f ) + (208.4f / 385
. m ) + 50k ⋅ (10f + 50f + 01
. p ))−1
ωdB = 121 Mrad / s ANS
E10.18)
iout
vs =g ( ro 2 ( 1+ gm 2 ro 1 )
m1 ro 2 (1 +g m 2 ro1 )+ RL )
gm1 ,2 = 2µn Cox WL I SUP = (2 )(50µ )( 100
2 )(100µ ) = 707 µS

ro1 ,2 = ( λn I SUP )−1 = ( 0.05 ⋅100 µ )−1 = 200 kΩ

iout
vs . µ)
= ( 7071 ( 200 k (1+( 707 .1 µ )( 200 k ))
. µ )( 200 k ))+50 k
200 k (1 +( 7071 )= 706 µS ANS
ω3dB = ( RS ( Cgs1 + 2Cgd1 ) + (Cgs 2 / gm2 ) + RL ( Cgd + Cdb + CL ))−1
Cdb = (WLdiff )Cjn + (W + 2 Ldiff )Cjswn = (100 ⋅ 6)( 01 . f ) + (100 + 2 ⋅ 6 )(0.5f ) = 116 fF
Cgs1,2 = 23 WLCox + WCov = (100 )( 23 ⋅ 2 ⋅ 23
. f + 0.5f ) = 357 fF
Cgd1,2 = WCov = (0.5)(100 ) = 50 fF
ω3 dB = (1k(356.7f + ( 2)(50f )) + (356.7f / 7071
. µ ) + 50 k(356.7f + 116f + 01
. p ))−1
ω3dB = 338. Mrad / s ANS

PAGE 47
Chapter 11
Differential Amplifiers

EXERCISES
E11.2) IBIAS = 100 µA ⇒ I1 = I2 = 50 µA , RC = VRC / I1/ 2 = 2.5 / 50 µ = 50 kΩ (a) ANS
IBIAS = 100 µA ⇒ I1 = I2 = 50 µA , RC = VRC / I1/ 2 = 2.5 / 50 µ = 50 kΩ (b) ANS
IBIAS = 500 µA ⇒ I1 = I2 = 250 µA , RC = VRC / I1/ 2 = 2.5 / 250µ = 10 kΩ (c) ANS
E11.4) gm = I1,2 / Vth = 50µ / 2585 . mS , RC = 2V + / IBIAS = 2(2.5) / 100µ = 50 kΩ
. m = 193
adm = − gm RC = − (1934
. m )(50 k ) = −96.7 V / V ,
−g R −( .
acm = (1 +2 gmm rCob ) = 1+2(1.934m )( 25/100 µ ) = − 999
1 934
m )( k ) 50
. mV / V (a) ANS
CMRR = = −96. 70
= 968 = 59.7 dB
a dm
a cm −99. 90m (a) ANS
gm = 9.67 mS , RC = 10 kΩ , adm = −967
. V / V , acm = −99.9 mV / V (b) ANS
CMRR = 968 = 59.7 dB (b) ANS
(Note that adm , acm , and CMRR end up being independent of bias current.)
. / 50k = 50 µA
E11.6) I1 = I 2 = VRC / RC = 25
Rid = 2rπ = 2βoVth / I1 ,2 = ( 2)(100 )(25m ) / 50µ = 100 kΩ (a) ANS
Rod = 2 RC = ( 2)(50k ) = 100 kΩ , for ro >> RC (b) ANS
Gmd = gm / 2 = I1,2 / (2Vth ) = 50µ / (2 ⋅ 25m ) = 1 mS (c) ANS
v od  Rid   2 RL 
=  (− Gmd Rod ) 
vid  Rid + 2 RS   2 RL + Rod 
v od  100   2 ⋅ 50 
=  ( −1m ⋅100k )  = − 49 V / V (d) ANS
vid  100 + 2  2 ⋅ 50 + 100
E11.8) Since µp Cox is deficient from µn Cox by factor of two, W
L can be multiplied
accordingly to retain a similar analysis.
( WL ) p = 2( WL )n = 400
2 ⇒ Cgs = 2Cgso = 0.54 pF ⇒ Cgd = 2Cgdn = 0.2 pF

⇒ CM = 10 pF ⇒ f3 dB = 302 . MHz ANS

. / 100µ = 25 kΩ
E11.10) RC = VRC / I1, 2 = 25 (a) ANS
adm = − gm RC / 2 = − I1,2 RC / (2Vth ) = −(100µ )( 25k ) / ( 2 ⋅ 25m ) = −25 V / V (b) ANS
− gm RC − I1, 2 RC /Vth −100 µ⋅ 25k / 25m
acm = (1+ 2 g mrob ) = (1+2 I1, 2V A/ (Vth IBIAS )) = ( 1+25/25 m ) = − 100 mV / V (c) ANS
v i1 + vi 2
v id = v i1 − v i2 = 10m − 12m = −2 mV , v id = = 10 m 2+12 m = 11 mV 2

v o = voc + v od = adm v ic + acm v id = (− 25)( −2 m ) + ( − 100m )(11m ) = 489 . mV

v o / vid = 489 . m / −2m = − 24.5 V / V (d) ANS
E11.12) io / vid = Gm = − gm1 = − 2µn Cox WL I1,2 = − (2 )(50µ )( 100
2 )(100µ ) = − 707 µS (a) ANS
Rod = ro 2 // ro4 = (λn 2 I 2 )−1 // ( λ p 4 I4 )−1 = ( 002
. ⋅100µ ) −1 // ( 001
. ⋅100µ ) −1 = 333 kΩ
. µ )(3333
avd = − Gm Rod = − ( − 7071 . k ) = 236 V / V (c) ANS

PAGE 48
 gm3 = 2µp Cox WL I3 = ( 2)( 25µ )( 504 )(100µ ) = 250 µS
avc = − (2 gm3 rob )−1 = −( 2 ⋅ 250µ ⋅100k ) = − 20 mV / V (d) ANS
E11.14) io / vid = Gm = − gm1 = − I1 / Vth = −100 µ / 2585 . m = − 389 . mS (a) ANS
Rod = ro 2 // ro4 = (V A / I2 ) // ( λp 4 I 4 ) = (25 / 100µ ) // ( 0.01⋅100 µ )−1 = 200 kΩ
−1
(b) ANS
avd = − Gm Rod = − ( − 389
. m )( 200k ) = 774 V / V (c) ANS
gm3 = 2µp Cox WL I3 = ( 2)( 25µ )( 504 )(100µ ) = 250 µS
avc = − (2 gm3 rob )−1 = ( 2 ⋅ 250 µ ⋅100k )−1 = − 20 mV / V (d) ANS
E11.16) (The solution is given for the case where RC = 10 kΩ , W
L = 100
2 .)

The PMOS version looks like:

gm = 2µp Cox WL ID1 = (2)(25µ)(100

2 )(
250µ ) = 791 µS
v o vin = gm RC 2 = (7906
. µ)(10k ) 2 = 395
. V/V ANS
( 2 )( 250 µ )
E11.18) VGS = VTn + =1 + = 145
2 I1, 2
µn Cox WL ( 50µ )( 100
. V
2 )

V X = V I 1,2 − VGS = 0 − 145

. = − 145
. V > −VDD + VIBIAS = − 2.5 + 0.5 = − 2 V (a) ANS
VO1,2 = VDD − I1, 2 RD = 2.5 − (250 µ )(5k ) = 125
. V (a) ANS
All transistors are in their constant-current (saturation) region. (a) ANS
VGS = 145
. V , V X = VG − VGS = 2 − 145. = 553 mV , VO1,2 = 125. V as before (b) ANS
(Also make sure: VO1,2 − V X = V DS = 0.697 V > VDSsat = 0447
. V .)
All transistors are in their constant-current (saturation) region. (b) ANS
IBIAS 500 µ
= 1 = 0.632 < VID = 1 − 0 = 1 ⇒ one of the devices cutoff
Kn ( 2 )(50 µ )(100
2 )

( 2 )( 500 µ )
VO1 = 2.5 − (500 µ )(5k ) = 0 V , VO2 = 2.5 V , VGS1 = 1 + ( 50 µ )( 100
= 163
. V (c) ANS
2 )

V X = V I 2 − VGS 2 = 1 − 163
. = −.632 V > −VDD + VIBIAS = − 25
. + 05
. = −2 V (c) ANS
Also check: VDS 1 = VO1 − V X = 0.632 V = VDSsat ... so, M1 is on the edge of
triode/saturation, M 2 is cutoff, and current source is okay. (c) ANS
VGS1 = 163
. V for same reason as in E11.18c.
V X = VG − VGS 1 = 0 − 163
. = − 163
. V , so current source is okay (d) ANS
So, M1 is in saturation, M 2 is cutoff, and the current source is fine. (d) ANS
( 2 )( 100 µ )
VGS = 1 + ( 50µ )( 100
= 128
. V
2 )

PAGE 49
V X = V1,2 − VGS = 0 − 1288
. = −128
. V (e) ANS
VO1,2 = VDD − I1, 2 RD = 2.5 − (100 µ )(5k ) = 2.00 V (e) ANS
All three devices are in saturation. (e) ANS

PAGE 50
Chapter 12
Feedback and Operational Amplifiers

EXERCISES
E12.2) T = Rmd f = (− 1M )( −2 ⋅10−4 ) = 200
Rmf = Rmd (1 + T )−1 = ( −1M)(1 + 200 )−1 = − 498
. kΩ ≈ f −1
= (2 ⋅ 10−4 )−1 = − 5 kΩ (a) ANS
Rin , f = Rin (1 + T )−1 = ( 50k )(1 + 200 )−1 = 249 Ω (b) ANS
Rout , f = Rout (1 + T )−1 = ( 2M )(1 + 200) −1 = 995
. kΩ (c) ANS
id = 750 nA ⇒ v o = Rmd id = ( −1M )(750n ) = −750 m
ii,max = id + fvo = 750 n + ( −2 ⋅ 10−4 )( −750 m ) = 151 µA (d) ANS
E12.4) f = T / aid = 20 / 350 = 571 . mA / A (a) ANS
Aif = aid (1 + T ) = 350 / (1 + 20) = 16.7 A / A
−1
(b) ANS
Rin , f = Rin (1 + T )−1 = ( 25k )(1 + 20 )−1 = 119
. kΩ (c) ANS
Rout , f = Rout (1 + T ) = (2 M)(1 + 20) = 42 MΩ (d) ANS
E12.6) v i = v d + fvo = v d + f (avd v d ) = (1 + avd f )v d
v d / vi = (1 + avd f )−1 (a) ANS
ii Rin , f = vi ⇒ ii = v i / Rin , f (b) ANS
From Eq. (12.60):
a' = avd ( Rin ( R1 + R2 )
Rin ( R1 + R2 + Ro )+ R1 ( R2 + Rout ) )= ( 3 ⋅10 )(
5 ( 2 .8 M )( 5 k +45k )
( 2. 8 M )( 5 k +45 k +50) +( 5 k )( 45 k +50) )= 299 kV / V
(The “loaded” amp is virtually undisturbed by loading from the feedback net.)
f = 5 / ( 45 + 5) = 01
. , v d / vi = (1 + 299.2k ⋅ 01 . µV / V
. )−1 = 334 (d) ANS
(d) ANS
“Ideal op-amp” assumptions are as follows: v IN + ≈ v IN − (note that
v d vi = 334. µV / V ), and iin ≈ 0 . (Look at the huge input resistance...)
E12.8) To = 1M / 25 . k −1 = 399 ; f = To / α ido = 300 / 200 = 15
. (a) ANS
Aifo = 200 / (1 + 399 ) = 05
. (attenuation) (b) ANS
To = 399 from above (c) ANS
E12.10) The magnitude bode plot is:
0
20
40
60
80
100
120
Magnitude
140 in dB-mho 3 4 5 6 7 8 9 10
10 100 1 10 1 10 1 10 1 10 1 10 1 10 1 10 1 10
Angular Frequency in rad/s, log scale
G
(a) ANS

PAGE 51
 The phase plot is:
30
0
30
60
90
120
150
180
210
Phase240
in degrees
270
300 3 4 5 6 7 8 9 10
10 100 1 10 1 10 1 10 1 10 1 10 1 10 1 10 1 10
Angular Frequency in rad/s, log scale
G
(a) ANS
f = 500 ⇒ G = 0.02 = − 54 dB (S) for unity gain
G = − 54 dB (S) ⇒ ω = 114
. Mrad / s (b) ANS
PM = 60°⇒ ∠Gmd ( jω ) = −120°⇒ ω = 1 Mrad / s ⇒
G = −50 dB (S) = f = 340 Ω (c) ANS
E12.12) The offset voltage results in a shift in the overall transfer function:
VO = (VI − VOS )avd
Differential Amplifier Transfer Function
2

0.002 0.001 0 0.001 0.002

Output Voltage VO (volts)

2

Input Voltage VI (volts)

VO (VOS=+2mV)
VO (VOS=-2mV)
(a) ANS
VO = (VI − VOS ) vd
a
1 + a vd f

PAGE 52
 Diff. Amp. X-fer Function, with Feedback
1

0.2 0.1 0 0.1 0.2

Output Voltage VO (volts)

1

Input Voltage VI (volts)

VO (VOS=+2mV)
VO (VOS=-2mV)
(b) ANS
The output voltage is much closer to zero for vi = 0 V. However, the gain has
also been reduced as well. Once again, this is an example of a trade-off—some
of the gain is sacrificed for an apparently smaller effect of VOS . Still, attention
has to be paid to “disturbance” terms that appear early in a feedback path (i.e., at
the inputs, as VOS , or uncertainty in f ). One still can’t differentiate 2 mV from
the input v i and the offset VOS , so if v i is a couple of millivolts in amplitude,
then the circuit would be problematic.
E12.14) All current paths consume half their initial currents at 20% below the initial
supply voltage. So, the ratio is ( 12 )(0.8):1 = 0.41 :. (a) ANS
VIC (min) = −VDD + V BE ( on ) + VTp = − 2 + 0.7 − 1 = −2.3 ⇒ V IC(min) = − 2 V (lower rail) (b) ANS
( 2 )( 50µ )
VIC (max) = VDD + VDS ( sat ) + VTp = 2 − ( 25 µ )(150 / 3) − 1 = 0.7 V (b) ANS
VOUT(min) = −2 + 02. = −18 . V (c) ANS
( 2 )( 50 µ )
VOUT(max) = 2 − ( 25µ )( 150 /3 ) = 17
. V (c) ANS
E12.16) gm1 = (2 )(25µ )( 150
3 )( 25µ ) = 250 µS , ro 4 = (λn ISUP )
−1
= ( 013. ⋅ 25µ ) −1 = 1200 kΩ
3 )(50 µ ) = 500 µS , ro 5 = ( λn I SUP )
gm5 = (2)( 50µ )( 150 −1
= ( 03.1 ⋅ 50µ )−1 = 600 kΩ
avdo = g m1 ( 12 ro 4 ) gm5 ( 12 ro5 ) = ( 250µ )(500 µ )( 12 1200k )( 12 600k ) = 2250 V / V (a) ANS
The device sizes didn’t change, so C1 = 112 . pF , CL = 535
'
. pF .

Cc ' = CL ' ( )(1 + )1 +

Gm 1
Gm 2
C1
CL ' 1+ ( )(
Gm 2
G m1
C1 / CL '
1+C1 / C L ' ) , Gm 1
Gm 2 =
gm 1
gm5 = 1
2

. p )(12 )(1 + 112

Cc ' = (535 .
(
5 .35 ) 1 + 1 + ( 2)(1 +112
. / 5. 35 ) = 7.0 pF
112
. /5 .35
) (b) ANS
Since Cgd 5 = 75 fF , then Cc = 6.93 pF . Because of the lower supply current, the
unity-gain frequency is lower at f unity = 4 .3 MHz, vs. 107
. MHz from Eq.
(12.147) from hand-calculations. Under SPICE, the compensation capacitor
increases about 10 pF to have a 45° phase margin under unity gain. Take care
to properly enter device model parameters in setting up the decks. (c) ANS

PAGE 53
E12.18) The critical frequency ω * is the frequency such that a vd ( jω *) f = 1. Or, in the
. , the critical frequency ω * is when avd ( j ω *) = 10 = 20 dB .
case of f = 01 (a) ANS

Under unity gain consideration, Cc '= 980 fF from the text. But if Avf = 10 (i.e.,
f = 01
. ), that means that the first pole can be moved ten times higher than that
for Avf = 1. The compensation capacitance is approximately ten times smaller,
that is Cc '= 98 fF . (b) ANS
This is left as an exercise to the reader. The results should be on the order of
hand-calculations. (c) ANS

PAGE 54
Chapter 13
MOS Memories

EXERCISES
E13.2) S = 16 M2 bits , S = 2 N + M +P , where N = M = P
S = 23 N ⇒ N = 13 log 2 ( S ) = 13 log2 (16M 2 ) = 8 ; N = M = P = 8 ANS
E13.4) The PMOS array looks like:

ANS
E13.6) Since the original drawing in Fig. 13.9 obeys design rules clearances, and since
design rule clearances scale along as well, then both parts of this problem are
addressed by using scaled versions of the original figure. Consequently, for
L = 1 , the cell dimensions are 4 µm × 6 µm , and for L = 0 . 75 , the cell
W 3 W 2 . 25

PAGE 55
 dimensions are 3 µm × 4.5 µm .

(a,b) ANS
E13.8) RWORD = 5 kΩ is unchanged
CWORD / n = (4.5)(15 . f ) + (15
. )( 23 . )(15 . µm
. f ) = 15.7 fF (unchanged from 15
. )( 006
NMOS example), or CWORD = ( 256)(15.66f ) = 4 pF , and τWORD = 20 ns (a) ANS
However, CBIT changes as CJn ≠ CJp and CJSWn ≠ CJSWp :
CBIT / n = (4.5)(4.5)( 03
. f ) + ( 45
. + 2(4.5))(0.35f ) + ( 45 . f ) = 131
. )( 05 . fF
Or CBIT = (256 )(1305. f ) = 334. pF (b) ANS
E13.10) From EX 13.1, IDn = 330 µA . That makes:
C BIT ∆ VBIT
t READ = 12 τWORD + = 12 (5n ) + ( = 10 ns
5 p )( 0. 5 )
I Dn 330 µ (a) ANS
Note that RON ∝ ( WL )−1 , so that as W
L = 115.5 is 3
4 of W
L = 120.5 in the example, then
RON = 1500 Ω . Thus, τ BIT = RON CBIT = (1500 )(5p) = 7.5 ns, and
tCYCLE = t READ + 3τ BIT = 10n + (3)(7.5n ) = 33 ns (b) ANS
E13.12) τWORD does not change since the number of cells along the word line is
unchanged
CBIT doubles since the number of cells along the bit line doubles
τWORD = 4 ns , CBIT = 34 . pF (a) ANS
This time, τWORD increases by a factor of four since τWORD ∝ n ( n =# of cells )
2

CBIT is unchanged; τWORD = 16 ns , CBIT = 17 . pF (b) ANS

Since the number of word cells is increased by a factor of eight, then τWORD
increases by a factor of sixty-four. Also, CBIT doubles as well, since the bit line
geometry is doubled. So, τWORD = 32 ns and CBIT = 34 . pF (c) ANS
E13.14) tWRITE = 12 τWORD + 1n = 2n + 1n = 3 ns , where 1 ns is worst-case delay due to CPQ
Since τWORD is unchanged, then tWRITE = 3 ns . (a) ANS

PAGE 56
 Since τWORD is quadrupled, then tWRITE = 8n + 1n = 9 ns (b) ANS
Since τWORD is increased by a factor of sixty-four, τWORD = 128n + 1n = 130 ns (c) ANS
E13.16) τ = RON ( CS // CBIT ) = RON CS for CBIT >> CS ; crudely calculated:
RON = (µn Cox WL (VWORD − VTn )) = ((50µ )( 13.5 )(5 − 1)) = 2.5 kΩ
−1 −1

τ = (50f )(2.5k ) = 125 ps (a) ANS

RON = ((50 µ )(13.5 )(3 − 1)) = 5 kΩ
−1

τ = (50f )(5k ) = 250 ps (b) ANS

RON = ((50 µ )(11 ..55 )(5 − 1)) = 5 kΩ
−1

τ = (50f )(5k ) = 250 ps (c) ANS

(In reality, the times may be significantly longer, since it may be that
VGS < VWORD .)
E13.18) Since the devices are being operated as switches, one approach is to swap all the
array NMOS’s with PMOS’s and drive them with complementary signals.

ANS
E13.20) CL = 46 CIN = 46 WLCox = 46 (3)(15 . f ) = 40 pF
. )( 23 ANS

PAGE 57
PAGE 58