Professional Documents
Culture Documents
Ayazi
Gain- K N′ S1 I 5 > K P′ S1 I 5
bandwidth
Cc Cc
Upper input I5 > 2I5 I5
common VDD − VTP + VTN − VDD − VTP − −
K P′ S3 K P′ S5 K P′ S1
mode voltage
Slew rate(due I5 = I5
to Cc) Cc Cc
Positive (VDD) no expression needed < no expression needed
PSRR
1
Negative no expression needed > no expression needed
(VSS) PSRR
Phase K N′ S1 I 5 CL K N′ S1 I 5 < K P′ S1 I 5 CL K P′ S1 I 5
90o − tan −1 ( ) − tan −1 ( ) 90o − tan −1 ( ) − tan −1 ( )
Margin ′ C
2 K P S6 I 7 C 2 K P′ S6 I 7 2 K N′ S6 I 7 C
C 2 K N′ S6 I 7
(EXTRA
CREDIT)
2
Problem 2 - (35 points)
The device parameters for the operational amplifier shown below are given in the table.
Ignore the body effect of the MOS transistor and the internal capacitances of all the
transistors.
a. What resistance R in the emitter of Q9 is required to set the first stage bias currents in
the emitters of Q3 and Q4 at 10µA each?
b. Calculate the overall voltage gain of the amplifier, by calculating the effective
transconductances of the differential input stage and the 2nd gain stage, and their
effective output resistances.
c. Calculate the value of the miller capacitance Cc required to obtain a gain-bandwidth
of 2MHz for this op-amp.
d. Calculate the phase margin of this op-amp.
a) RIC9=Vtln(IC8/IC9)
IC8=(20V-1.4V)/50kΩ=0.372mA
IC9=20µA Æ R=3.65kΩ
g m1 g A 1 1
b) GmI = = m1 = 200 × 10−6 note that : re3 = =
1 + g m1re3 2 V g m3 g m1
RI = r07 || r04 (1 + g m 4 × re 2 ) = r07 || 2r04 = 5.65M Ω
VAnpn VApnp
because : r07 = = 13M Ω and r04 = = 5M Ω
IC 7 IC 4
Æ AvI = GmI × RI = 1130
I
R3 I C13 = Vt ln( C10 ) ⇒ I C13 = 100 µ A
I C13
g mm1 A W A
GmII = = 0.663 × 10−3 note that : g mm1 = 2µ Cox ( )mm1 I mm1 = 1.414 ×10−3
1 + g mm1 R2 V L V
3
GmI 200 ×10−6
c) GBW = ⇒ CC = = 15.9 pF
CC 2π × 2 × 10−6
2MHz 2MHz
d) Φ M = 90o − tan−1 ( ) − tan−1 ( )
p2 z
GmII 0.663 ×10−3
p2 = = = 132 × 106 rad = 21.1 MHz
CL 5 × 10−12 sec
GmII 0.663 × 10−3
z= = −12
= 41.69 × 106 rad = 6.64 MHz
CC 15.9 ×10 sec
2 2
⇒ Φ M = 90o − tan −1 ( ) − tan −1 ( ) = 90o − 5.41o − 16.76o = 67.82o
21 6.64
4
Problem 3 - (30 points)
In this problem, you are not asked to provide any numerical calculations. You are only
required to provide expressions. Make sure that your final expression for each section is
clearly identified and legible. For the cascoded two-stage CMOS Op-Amp shown below:
VDD
4
M3 M4 M6
3
1
CC VBP M7
2 VOUT
M1 M2
V− V+ CL
VBN M8
M5
VBIAS M9
VSS
a. Show an expression for the overall voltage gain of this amplifier as a function of
transistor transconductances and output resistances.
b. Show the expressions for the poles associated with the two capacitors CC and CL,
as a function of the two capacitance values, transistor transconductances and
output resistances. Assume that CC and CL include the transistor internal
capacitances at their respective nodes.
1 1
p1 = and p2 =
(rds 4 || rds 2 )Cc RII CL
c. If CL>>CC, provide an expression for the unity gain frequency (GB) of the
amplifier.
g m1 g m 6 (rds 4 || rds 2 )
GBW =
CL
5
d. Show an expression for the pole associated with node 3, in terms of transistors
internal capacitances, transconductances and output resistances. Identify all the
components of the internal capacitances.
1
p3 =
(rds 6 || R3 )C3
C3 = Cgs 7 + Csb 7 + Cgd 6 + Cdb 6
rds 7 + g m8 rds8 rds 9 1 g r r
R3 = ≅ + m8 ds 8 ds 9
1 + g m 7 rds 7 gm7 g m 7 rds 7
e. Is there a right half plane zero in this amplifier that would affect the phase
margin? If there is, provide an expression for the position of this zero.
No, because a miller configuration is not usedÆ there is no RHZ that would affect the
phase margin
f. (EXTRA CREDIT) Show expressions for any pole or zero associated with node
4. Identify all the components of the internal capacitances.
− g m3
p4 =
Cx
Cx = Cgs 3 + Cgs 4 + Cdb3 + Cdb1 + Cgd 1
−2 g m3
z4 =
Cx
6
ECE 6412- Spring 2006 Page 1
;;
Homework No. 1 - Solutions
Problem 1 - (10 points)
;;;;
;;; ;;
A top view of a MOS transistor is n+
;;; ;;
shown. (a) Identify the type of
transistor (NMOS or PMOS) and its p+
;;;
;;;;;
value of W and L.
(b.) Draw the cross-section A-A’ A A'
approxi- mately to scale.
Metal
;;;;
(c) Assume that dc voltage of terminal 1
;;;;
;; ;
is 5V, terminal 2 is 3V and terminal 3 is Poly
0V. Find the numerical value of the
capacitance between terminals 1 and 2, 2
and 3, and 1 and 3. Assume that the dc
value of the output voltage is 2.5V and p-well
that the voltage dependence for pn
;;;;;;;;;;;;;;
;;;;
junction capacitances is for both 3 2 1
transistors is -0.5 (this is called MJ in n-substrate
;; ;; ;;
SPICE).
;
IOX IOX
Solution
IOX
;;
(a.) This transistor is an NMOS
;;;
;;
transistor with the drain as terminal 1, the
gate as terminal 2, and the bulk and
F F
;;;
;;
source connected together to terminal 3.
The W = 26µm and L = 4µm . O O
X X
(b.) The approximate cross-section is
shown (vertical scale is magnified by 4 p+ n+ n+
times).
(c.)With VDS = 5V, VGS = 3V and VT =
0.75V, the transistor is in saturation. p-well 21µm
Therefore, the capacitors are:
C12 = CGD = LD(NMOS)xWxCox S00PES1
= 0.45µm·26µm·0.7fF/µm2 = 8.19fF
C23 = CGS = LD(NMOS)xWxCox + 0.67(WxL)X Cox = 8.19fF + 48.776fF
= 56.966fF
C13 requires the area of the drain (AD) and the perimeter of the drain (PD). These values
are AD = 26µmx10µm = 260µm2 and PD = 2(10+26) = 72µm.
[AD·0.33fF/µm2+PD·0.9fF/µm] [260µm2 ·0.33fF/µm2+72µm·0.9fF/µm]
C13 = CBD = =
5 5
1 + 0.6 1 + 0.6
= 49.29fF
ECE 6412- Spring 2006 Page 2
Problem 3
Find the numerical values of all roots and the VDD
midband gain of the transfer function vout/vin of the RL= RL=
differential amplifier shown. Assume that K N ’ = 10kΩ 10kΩ
100µA/V2, V TN = 0.7V, and λN = 0.04V-1. The CL =1pF
- vout +
CL =1pF
values of Cgs = 0.2pF and Cgd = 20fF.
M1 M2
Solution 100/1 100/1
A small-signal model appropriate for this circuit is vin
shown.
1mA
+
vin Cgd vout
Cgs RL CL S03E1P4
2 rds1 2
gm1vgs1 -
Fig. S03E1S4
Summing the currents at the output nodes gives,
gm1vgs1 + sCgd(vout-vin) + (gds1 + GL)vout + sCL vout = 0
(Note: we are ignoring the fact that vout and vin should be divided by two since it makes no
difference in the results and is easier to write.) Replacing vgs1 by vin gives
-(gm1 - sCgd)vin = [(gds1 + GL) + sCL + sCgd] vout
1 - sC g d
vout -(g m1 - sC gd ) -gm1 gm
= =
v in s(C L + C gd ) + (g ds1 + G L ) g ds1 + G L C + C
1 + s gLds1 + Gg dL
gm g ds + G L
∴ MGB = - gm1(rds||RL), Zero = C and Pole = - C + C
gd gd L
1 25
gm = 2·100·100·500 = 3162.3µS and rds = = 500µA = 50 kΩ
λ ID
∴ MGB = -3.162mS·(10kΩ||50kΩ) = -26.35 V/V
3.162x10-3
Zero = = 1.581x1011 radians/sec.
20x10-15
-1
Pole = = -1.1176x108 radians/sec.
1.02x10-12(10kΩ||50kΩ)
ECE 6412- Spring 2006 Page 4
Problem 4
Find the voltage transfer function of the common-gate amplifier VDD
shown. Identify the numerical values of the small-signal voltage
gain, vout/vin, and the poles and zeros. Assume that
RD =
ID =250µA, K N ’ = 100µA/V2, V TN = 0.5,
10kΩ
λ ≈ 0V-1, Cgs = 0.5pF and Cgd = 0.1pF.
20 +
Solution 1 vout
The small signal transconductance is, VBias RS = I -
D
1kΩ
gm = 2·KN·(W/L)ID = 2·100·20·250 = 1mS
vin
rds = ∞
gmvgs
The small signal Rs S04E1P3
model is, - +
vin Cgs vgs Cgd vout
+ RL
-
S04E1S3
-(gm+Gs) -10-3+10-3
p2 = C = -12 = -4x109 radians/sec.
gs 0.5x10
ECE 6412- Spring 2006 Page 5
Problem 5
Draw the electrical schematic using the proper symbols for the transistors. Identify on your
schematic the terminals which are +5V, ground, input, and output. Label the transistors on
;
the layout as M1, M2, etc. and determine their W/L values. Assume each square in the
;;;;
layout is 1 micron by 1 micron. Find the area in square microns and periphery in microns
for the source and drain of each transistor.
;; ;
;;;;
Metal n+ p+ p-well Poly Contact
;;; ;;;
M1 Ground
;;;;
;;;
;;; ;;;
;;;
M2
+5V
;;;;
;;;;;;
Input
Output
N-Substrate
+5Volts
D1
G1 M1 B1 AS1 = AD1 = 40x8 = 320µm2
W1
= 10
L1 PS1 = PD1 = 8+8+40+40 = 48µm
Input S1 Output
S2
M2 AS2 = 2AS1 = 640 µm2
G2 B2 W2 AD2 = AD1 = 320µm2
= 20
L2 PS2 = 2PS1 =192µm
D2 PD2 = PD1 = 96µm
Ground S01PES1
ECE 6412 - Spring 2006 Page 1
+ +
Vin C1= R2= V1 R3= C3=
V1 Vout
10pF 10kΩ 100 20kΩ 10pF -
-
S02E1P3
Solution
The midband gain is given as,
Vout 20kΩ 10kΩ
V in = - 100 11kΩ = -181.82V/V
To find the –3dB frequency requires finding the 3 open-circuit time constants.
RC10:
RI
+
Vin
RL Vout
-
S02E1P4
Solution
The best approach to this problem is a direct analysis.
Small-signal model:
gmVgs gmVs
RI RI
- + + +
Vin Vgs Cgs Cgd Vin Vs Cgs Cgd
RL Vout RL Vout
+ - - -
S02E1S4
1 V in-V s
Vout = gmZLVs where ZL = sR C +1 and RI = gmVs +
L gd
sCgsVs
Solving for Vs from the second equation gives,
V in
Vs = 1+g R +sC R
m I gs I
Substituting Vs in the first equation gives,
V in Vout 1 1
V out = gm ZL 1+g R +sC R → V = gm sR C +1 1+g R +sC R
m I gs I in L gd m I gs I
gmRL 1 1
= 1+g R sR C +1 sC R = MBG 1 1
m I L gd gd I 1 - s s
+ 1 p 1 - p
1+gmRI 1 2
gmRL 1x9
∴ MBG = 1+g R = 1+1x3 = 2.25V/V
m I
1 1 1+gmRI 1+3 8
p1 = -R C = - 9x1ns = 1.1e8 rad/s and p2 =- RICgs = - 3x4.5ns = -2.9x10 rad/s
L gd
ECE 6412 - Spring 2005 Page 1
Power dissipation:
Pdiss = 5(36 + 36 + 60)µW = 660µW (6)
ICMR:
Vin ,max = 2.5 − VT 1 − V ON1 − VON 5 = 0.51V (7)
Slew Rate:
Slew rate under no load condition can be given as,
I
SR = 5 = 6V / µs
CC
W 9 1 W 10
K’9 L = 1 and K’10 L = very small (12)
9 10
The dominant pole could be calculated as,
(g g )
p1 = − ds4 + ds2 =640HZ .
2.π . AV2 .C C
It can be noted that in this problem, the product of the open-loop gain and the
dominant pole is approximately equal to the load pole. Thus, the gain bandwidth is
approximately equal to 2.8 MHz and the phase margin would be close to 45 degrees.
ECE 6412 - Spring 2005 Page 4
SPICE file:
. modelnmos nmos vto=0.7 lambda=0.04 kp=110u
.model pmos pmos vto=-0.8 lambda=0.05 kp=50u
vdd 1 0 dc 2.5 ac 0
vss 10 0 dc -2.5 ac 0
vinp 5 0 dc 0 ac 1
*vinn 4 0 dc 0 ac 0
.op
.ac dec 10 1 100meg
.option post
.end
Operating points:
**** mosfets
subckt
element 0:m8 0:m5 0:m1 0:m2 0:m3 0:m4
model 0:pmos 0:pmos 0:pmos 0:pmos 0:nmos 0:nmos
region Cutoff Cutoff Cutoff Cutoff Saturati Saturati
id -35.3708u -34.8506u -17.4107u -17.4399u 17.4107u 17.4399u
ibs 0. 0. 0. 0. 0. 0.
ibd 14.6292f 11.4726f 28.7676f 28.3314f -9.7598f -10.1959f
ECE 6412 - Spring 2005 Page 5
subckt
element 0:m7 0:m6 0:m9 0:m10
model 0:pmos 0:nmos 0:nmos 0:pmos
region Cutoff Saturati Linear Cutoff
id -61.7971u 61.7971u 0. 0.
ibs 0. 0. 0. 0.
ibd 24.9901f -25.0099f 0. 0.
vgs -1.4629 1.0196 2.4990 -2.5010
vds -2.4990 2.5010 0. 0.
vbs 0. 0. 0. 0.
vth -800.0000m 700.0000m 700.0000m -800.0000m
vdsat -662.9217m 319.5939m 0. 0.
beta 281.2376u 1.2100m 110.0000u 500.0000n
gam eff 527.6252m 527.6252m 527.6252m 527.6252m
gm 186.4385u 386.7225u 0. 0.
gds 2.7467u 2.2471u 197.8911u 850.4951n
gmb 64.5917u 133.9802u 0. 0.
cdtot 5.753e-18 1.152e-17 1.727e-16 17.2658f
cgtot 1.1698f 2.3660f 3.463e-16 34.6349f
cstot 1.1511f 2.3021f 1.727e-16 17.2658f
cbtot 1.301e-17 5.233e-17 9.769e-19 1.033e-16
cgs 1.1511f 2.3021f 1.727e-16 17.2658f
cgd 5.753e-18 1.152e-17 1.727e-16 17.2658f
M7 M5
M6
1/1 1/1 5/1
vout 0.5gm1gm4
∴ vin = (gds1+gds3)(gds4+gds5)
-5V
2KNW1ID1
gm1 = L1 = 24·2·4·10 x10-6 = 43.82µS
2KPW4ID4
gm4 = L4 = 2·8·10·100 x10-6 = 126.5µS
Output resistance:
1 106
R o u t = g +g = 1+2 = 333kΩ
ds4 ds5
3.6
14.25 uA
3.6
14.25e3
9.12
14.25e3
0.54
4.75
1.8 4.75
3.6
4.75 =2.59
2590
719
3.6
401,000
is changed to 50ohm
+ x50ohm=11.8+251x0.05=24.35k
16.37
Req
377 16.37
4.487
97ohm
1+REgm17
(1+REgm17)
79.14
4.487 79.1
97
2703 78.4K
79.14 78.4K
0.097K
404,610
ECE 6412 - Spring 2005 Page 5
βF 1 20 VA 25V
rπ 6 = g = 52kΩ rds7 = = 50µA = 0.4MΩ and ro6 = I = 50µA = 0.5MΩ
m6 λPID C
Solution
Problem 6 - Continued
Thus, the pole at the drain of M4 is
g
p D 4 = ! m 4 = !280 MHz.
C4
The total capacitance at the drain of M8 is
C8 = C gd 8 + Cbd 8 + C gs10 + C gs12
or, C8 = 9 + 51 + 34 + 34 = 128 fF
1
R2 + = 3.4 K!
g m10
Thus, the pole at the drain of M8 is
1
pD 8 = " = "366 MHz.
#% 1 &(
R + C
$ 2 gm10 ' 8
For a phase margin of 60 o , we have
/ ) # GB & # # # ,2
PM = 180 o " 190o " *tan"1 % ( + tan"1% GB &( + tan"1% GB &( + tan"1 % GB &( -
0 + $ pS7 ' $ pS12 ' $ pD4 ' $ pD 8 ' .43
Solving the above equation
GB ! 65 MHz.
And, Av = 6925 V/V
Thus, p1 = 9.39 KHz, and CL " 1.54 pF
ECE 6412 - Spring 2006 Page 1
g m1 1 1
acm = − (ro1 (1 + 2 g m1ro 5 ) ro 3 )= = 0.01, where :
1 + 2 g m1ro 5 g m 3 1 + 2m × 50 K
g m 3 = 5000 × 0.2 = 1mS
ro5 = 50 K
Because of adding the 20Kresistors, the adm becomes smaller but acm becomes much
smaller. The effective impedance of M3 and M4 is now 1/gm, which is much smaller than
roÆ adm/acm=1538 which is much higher than 100 in 12.4.1
Applying the half-circuit principle, it can be seen that each ½ circuit consists of a cascade of two
common-source (CS) stages – the first with a diode connected PMOS load and the other with an
NMOS load.
2 ⎛2⎞ K P g m2 5
i n 5 = 4kT ⎜ ⎟ g m 5 +
⎝3⎠ COX (WL )5 f
2 ⎛2⎞ K N g m2 7
i n 7 = 4kT ⎜ ⎟ g m 7 +
⎝3⎠ COX (WL )7 f
Similarly, the noise at the gate of M1 due to M1 and diode connected M3 can be expressed as:
2 in21 + in23
v in (1,3) =
g m2 1
2 ⎛2⎞ K N g m2 1 2 ⎛2⎞ K P g m2 3
Where, i n1 = 4kT ⎜ ⎟ g m1 + and i n 3 = 4kT ⎜ ⎟ g m 3 +
⎝3⎠ COX (WL )1 f ⎝3⎠ COX (WL )3 f
Therefore the total noise referred to the input (gate of M1) is:
2 2
v in = v in (1,3) +
vo21
=
( ) (
in21 + in23
+
)
in25 + in27 × g m2 3
. Therefore for one half-circuit,
Av21 g m2 1 g m2 5 × g m2 1
2
v in =
( ) ( )
g m2 5 × in21 + in23 + in25 + in27 × g m2 3
g m2 5 × g m2 1
Considering only the thermal noise, the total input referred noise is:
⎡ ⎛2⎞ ⎤ ⎡ ⎛2⎞ ⎤
g m2 5 ⎢4kT ⎜ ⎟( g m1 + g m3 )⎥ + g m2 3 ⎢4kT ⎜ ⎟( g m5 + g m 7 )⎥
2 ⎣ ⎝3⎠ ⎦ ⎣ ⎝3⎠ ⎦
v in (THERMAL ) = 2×
g m 5 × g m1
2 2
Considering only the flicker noise, the total input referred noise is:
⎡ K N g m2 1 K P g m2 3 ⎤ 2 ⎡ K P g m2 5 K N g m2 7 ⎤
g m2 5 ⎢ + ⎥ + g m3 ⎢ + ⎥
2
v in ( FLICKER ) = 2× ⎣ COX (WL )1 f COX (WL )3 f⎦ ⎣ COX (WL )5 f COX (WL )7 f⎦
g m 5 × g m1
2 2
Equating the thermal noise and flicker noise to find the flicker noise corner frequency (fC), we
have:
⎧⎪ 2 ⎡ K N g m2 1 K P g m2 3 ⎤ 2 ⎡ K P g m5
2
K N g m2 7 ⎤ ⎫⎪ 1
⎨ g m5 ⎢ + ⎥ + g m3 ⎢ + ⎥⎬ =
⎪⎩ ⎣ COX (WL )1 COX (WL )3 ⎦ ⎣ COX (WL )5 COX (WL )7 ⎦ ⎪⎭ f C
⎡ ⎛2⎞ ⎤ ⎡ ⎛2⎞ ⎤
g m2 5 ⎢4kT ⎜ ⎟( g m1 + g m3 )⎥ + g m2 3 ⎢4kT ⎜ ⎟( g m5 + g m 7 )⎥
⎣ ⎝3⎠ ⎦ ⎣ ⎝3⎠ ⎦
Numerical Calculations:
⎛W ⎞
Using g m = 2 µCOX ⎜ ⎟ I D , we obtain:
⎝L ⎠
gm1 = gm2 = 1.768 mS
gm3 = gm4 = 0.5 mS
gm5 = gm6 = 1 mS
gm7 = gm8 = 2.646 mS
Using the above, we obtain, the following values for the thermal and flicker noise powers
2 2
v in (THERMAL ) = 2.247 ×10 −17 V
Hz
Assuming tox= 100A°, we obtain COX = 34.53 x 10-4. Therefore the total flicker noise is given by:
2 3.2417 ×10−8 V 2
v in ( FLICKER ) =
f Hz
Equating the noise powers to find the flicker noise corner frequency, we obtain: fC = 1.44 GHz.
Problem 3 – (10 points)
Assumptions:
• VOD = VGS – VTH
• Only thermal noise of drain current considered
2I D 2I 1
gm = = D and ro =
VGS − VTH VOD λI D
Since from the above, we see that both Gm and Rout are dependent on the over-drive voltage, we
need to consider the effect of variation of VOD on both. Substituting for gm and Rout in terms of
VOD, we obtain the following expressions for Gm and Rout
2I D
Gm = g m1 =
VOD
2 1
Rout = ×
5 VOD λ2 I D
The major noise contributors in the folded cascode amplifier are: M1, M3 and M5. Therefore we
first obtain the noise contributions of each of these noise sources at the output.
2 ⎡ ⎛ 2 ⎞⎤ 2 2
v n−out (1) = ⎢4kT ⎜⎜ ⎟⎟⎥ × g m1 Rout
⎣ ⎝ 3 g m1 ⎠⎦
2 ⎡ ⎛ 2 ⎞⎤ 2 2
v n−out ( 3) = ⎢4kT ⎜⎜ ⎟⎟⎥ × g m3 Rout
⎣ ⎝ 3 g m 3 ⎠⎦
2 ⎡ ⎛ 2 ⎞⎤
v n − out ( 5 ) = ⎢ 4 kT ⎜⎜ ⎟⎟ ⎥ × g m2 5 Rout
2
⎣ ⎝ m 5 ⎠⎦
3 g
Therefore taking the superposition of these noise sources, we have the output referred noise given
by:
⎛ 2⎞
v n−out = 4kT ⎜ ⎟ × [g m1 + g m3 + g m5 ]× Rout
2 2
⎝ 3⎠
Therefore substituting expressions for gm and Rout into the above, we obtain the total output
referred noise power as:
2 ⎛ 2 ⎞ ⎛ 32 ⎞ 1
v n−out = 4kT × ⎜ ⎟ × ⎜ ⎟ × 3 4
⎝ 3 ⎠ ⎝ 25 ⎠ VOD λ I D
Therefore the output referred noise voltage can be expressed as a function of VOD as:
1
vn−out = B ×
(VOD ) 2
3
DR1 = DD
B
After the over-drive voltage changes to 75% of its original value, the new dynamic range (DR2) is
given by:
3
⎛3 ⎞
(VDD − 3VOD )⎜ VOD ⎟
2
DR2 = ⎝4 ⎠
B
Therefore, finding the difference between the initial and final dynamic ranges, we can find the
variation in the dynamic range caused by 25% reduction in VOD
(V ) 2 ⎡VDD − 7VOD ⎤
3