Beruflich Dokumente
Kultur Dokumente
Common Data : VDD = 1.8V , Lmin = 0.18µm, µ0N = 350cm2 /V s, µ0P = 100cm2 /V s,
|VT H | = 0.55,tox = 3.8nm, λ = 0.07V −1 , χ = 0.1.
1
Topic Number 1 : VLSI and MOS Fundamentals
Q–1 You are given a MOS Capacitor whose gate material has a work function of 4.1eV ,
oxide (SiO2 ) thickness is 10nm and whose body consists of p-type Silicon with doping
concentration NA = 10 × 1015 /cm3 . Assume that the dielectric constant of SiO2 is 3.9
and that of Si is 11.68. The electron affinity (E0 − Ec ) of Silicon is 4.05eV , and the
band-gap Ec −Ev of Silicon is 1.1eV (thus, the Fermi level and work function of intrinsic
Silicon is 4.6eV ). The magnitude of the electron charge q is 1.6 × 1019 Coulombs. At
room temperature, assume that for Silicon, ni = 1010 /cm3.
Q–2 Plot the transit frequency of NMOS as a function of VGS for the biasing configuration
shown in figure Fig. 1. Calculate the transit frequency of MOSFET for VGS = 1.5V .
Assume W = 1µm, L = 180nm, CGS = 1.2f F and CGD = 1f F . Given VDS = 1.8V .
Note: transit frequency is defined as the frequency at which small signal current gain of
the device drops to unity while the drain and source are held at ac ground.
+ VDS
−
+ VGS
−
Q–3 For an NMOS and PMOS device operating in saturation region sketch W/L versus
V GS − VT H if
a) ID is constant
b) gm is constant
Q–4 Plot the following parameters as function of body to source voltage of an NMOS device.
2
Q–5 An NMOS device operating in subthreshold region has a ξ = 1.2 (ξ is the nonideaelity
factor of subthreshold conduction). What variation in VGS results in a ten-fold change
in ID ? If ID = 8µA, what is gm ? Solution :
ID 8µ
gm = = = 256µA/V
ξVT 1.2 × 26mV
3
Topic Number 2 : Basics of MOS amplifiers
Q–1 For a MOSFET operating in the linear (triode) region (ignore body effect and channel-
length modulation), find an expression for:
Comment upon your findings. Is it desirable for an amplifier to work in the linear
region? Explain.
Q–2 Consider the simple MOS amplifier shown in Figure 2. Assume λ = 0 and VDD = 1.8V.
VDD
RD
Vout
Vin
Assume that the biasing for the gate side of the MOS is taken care of.
(a) For maximum symmetrical swing in Vout and for a drain current of 0.8 mA, find
the required value of RD .
(b) Calculate the maximum symmetrical output swing, and also the small-signal volt-
age gain in this case.
(c) How would the design values and the voltage gain change, if λ = 0.07 V−1 ?
4
Q–3 Consider the circuit in Figure 3.
VDD
RD
Vout
Vin
RS
VDD
Vb
M2
Vout
Vin
M1
5
Q–5 Consider the circuit shown in Figure 5.
VDD
M2
Vout
Vin
M1
(a) What are the maximum and minimum values of Vout required to keep both M1 and
M2 in saturation?
(b) Obtain an expression for the small-signal voltage gain offered by this circuit.
VDD
Vb
M2 ,W/L = 2µm/0.5µm
Vout
Vin
M1 ,W/L = 1µm/0.5µm
(a) What should be the bias voltage Vb if this amplifier is to provide a small-signal
voltage gain of 25?
(b) What are the maximum and minimum values of Vout required to keep both M1 and
M2 in saturation?
(c) How would the voltage gain change if M2 were replaced by an ideal current source
that provides the same current as required in part (a)?
Solution:
6
(a) What should be the bias voltage Vb if this amplifier is to provide a small-signal
voltage gain of 25?
Since the λ of both transistors are equal, we have ro1 = ro2 = ro . Thus,
ro
Av = gm1
2
2ID 1
= .
VGS − VT H 2λID
VA
=
VGS − VT H
Substituting the values, we find, VGS − VT H = 0.57V. Thus, the current through
M1 is approximately 0.2mA. The same current should flow through M2 as well.
So we obtain the VGS for M2 to be approx. -1 V. Thus, Vb = VDD − VSG = 0.8 V.
(b) What are the maximum and minimum values of Vout required to keep both M1 and
M2 in saturation?
For M2 to be in saturation, we need VDS < VGS − VT H , i.e Vout − VDD < −0.45 ⇒
Vout < 1.35 V. Thus, the maximum allowed Vout is 1.35 V. For M1 , we need
Vout > VGS − VT H ⇒ Vout > 0.57V. Thus, the minimum allowed Vout is 0.57V.
7
(c) How would the voltage gain change if M2 were replaced by an ideal current source
that provides the same current as required in part (a)?
An ideal current source would have an infinite resistance. Thus, there would be no
ro2 as offered by M2 . The gain expression would thus have reduced to Av = gm ro ,
hence for the same current as in part (a), the gain would now be 25 × 2 = 50.
8
Topic Number 3 : Cascode Amplifiers
Q–1 Design Cascode amplifier (Fig. 7 ) for a gain AV > −50V /V , V outmax of 1.3V, V outmin
of 0.4V. The slew rate with 5pF load should be at least 10V /µsec. Supply voltage is
1.8V and maximum power dissipation is 200µW .
VDD
Vbp
Vbn2
Cbig
Vin
Rbig
Vbn1
9
Q–2 Calculate the transconductance and output resistance of the cascode circuit shown in
Fig. 8 . Assume both transistors are in saturation and gm = 2mA/V and r0 = 20kΩ.
Ignore body effect. Repeat the calculation including body effect and observe the per-
centage of error.
VDD
Vbn2
Cbig
Vin
Rbig
Vbn1
Q–3 Calculate the output impedance of the circuit shown in Fig. 9 and state the advantages
and disadvantages of the circuit.
10
VDD
Vbn2 +
A
−
Cbig
Vin
Rbig
Vbn1
Q–4 Find the output resistance for the active cascode circuit shown in Fig. 10 excluding
resistor R. What is the maximum allowable output swing?
VDD
R
Ibias
Vin
Q–5 Determine input resistance, transconductance, output resistance and maximum open
circuit voltage gain of cascode if ID = 300µA, W/L = 200.
11
VDD
Vbn
Vin
Q–6 Consider the cascode amplifier shown in Fig. 12. Given gm1 = 1mA/V , gm2 =
0.5mA/V , calculate the gain of the amplifier and the swing on node X. Ignore channel
length modulation and body effect.
12
VDD
RD
Vbn gm2
Vin
gm1
13
Topic Number 4 : Differential Amplifier
Q–1 Calculate the unloaded voltage gain Vo /Vi and output resistance of the circuit in Fig.
13 Assume (W/L)N = 50, (W/L)P = 1, Itail = 200µA. Ignore body effect.
VDD
Itail
Q–2 Derive the expression for VIC (max) and VIC (min) for the p-channel input differential
amplifier.
Q–3 Design the current and W/L values of the current mirror load differential amplifier to
satisfy the following specification VDD = −VSS = 0.9V , slew rate > 10V /µs, CL = 3pF ,
f3dB ≥ 400KHz, and a small signal gain of 60V /V . −0.5V ≥ ICMR ≥ 0.5V , and
Pdiss ≤ 0.2mW .
Q–4 For a folded cascode current mirror differential amplifier shown in Fig. 14 assume
W/L = 200µm/1µm for all the transistors.
a) Find the maximum input common mode voltage VICmax , and the minimum com-
mon mode voltage VICmin .
b) What is the ICMR?
c) Find the small signal gain Vo /Vi if Vi = V1 − V2 .
d) If a 20pF capacitor is connected from the output to ground, what is the −3dB
frequency for Vo (jω)/Vi (jω) in hertz
14
Vout
Vi1 Vi2
100µA
Load
Vip Vin
Vip Vin
Load
Q–5 Consider the NMOS and PMOS input stages as shown in Fig. 15. Given that VDD =
1.8V , VGSp = VGSn = 0.6V , VDSsatn = VDSsatp = 0.2V ans the saturation voltages of the
current sources is also VDSsat determine
Solution: The common mode input range for the Pchannel input stage is given by
0 < Vcommon < VDD − VDSsat − VGSp Thus the ICMR is 0 < Vcommon < 1V .
The common mode input range for the Nchannel input stage is given by
VDSsat + VGSn < Vcommon < VDD Thus the ICMR is 0.8 < Vcommon < 1.8
15
Vbn
Vip Vin
Vout
Vip Vin
For the complementary input stage the ICMR is rail to rail, i.e. 0 to 1.8V , due to the
overlap of the two ICMR’s of the complementary stage.
16
Topic Number 5 : Current sources and References
Q–1 Design a simple MOS current mirror of the type shown in Fig. 17 to meet the following
constraints.
VDD
Iin
Iout
Vin M1 M2 Vo
a) Transistor M2 must operate in the active region for values of Vout to within 0.3
from ground.
b) The output current must be 30µA
c) The output current must change less than 5% for change in output voltage of 0.4V
Make M1 and M2 identical. You are to minimize the total device area within given
constraints. The device area is the total gate area (W × L). Assume Xd = 0.
Q–2 Design a voltage reference using the Beta-multiplier. Using spice simulation show the
temperature performance of your design, by plotting reference voltage vs temperature
Q–3 Consider the circuit of Fig. 18 assuming (W/L)1−3 = 15µm/0.18µm, IREF = 0.1mA,
and γ = 0.
17
VDD
P
Iref
Iout
Vbias
M3
X
Y
VIN M1 M2
c) If the circuit fed by cascode current source changes VP by 0.3V by how much does
VY change?
Q–4 Consider the circuit of Fig. 19 with (W/L)1−5 = 50µm/0.5µm and ISS = 0.5mA
VDD
M3 M4
F Vout
1.5V 1.5V
M1 M2
Vbias
18
a) Calculate the deviation of Vout from VF if |VT H3 | is 1mV less than |VT H4 |
b) Determine the CMRR of the amplifier
Q–5 Calculate the quiscent curent and temperature coefficient of the circuit shown in Fig.
20 where (W/L)1 = 25, VDD = 1.8V and R = 100kΩ. The resistor has a temperature
coefficient of 1400ppm/o C. α( dV
dT
t
= 2 × 10−3 V /oC.
VDD
I2
I1
Solution :
We have
s
VT 1 1 1 2VT 1 1
I2 = + 2
+ +
R βR R β1 R βR2
19
β1 = µN Cox (W/L)
= 7.9 × 10−3
β1 R = 795
β1 R2 = 79.5 × 106
Hence
I2 = 5.8µA
20
Topic Number 6 : Frequency Response
Q–1 Use the Miller approximation to calculate the −3dB frequency of the small signal voltage
gain of a MOS common source stage as shown in Fig. 21, using Rs = 20kΩ, RL =
10kΩ, ID = 0.5mA, and the following NMOS transistor data. Ignore channel-length
modulation.
W = 50µm, L = 1um, Cgd = 5f F .
RL
VO
Rs
Vi
Q–2 Calculate the −3dB frequency of the circuit in Fig. 22 assuming the following parameter
values.
VDD
RL1 RL2
Vo
M1 M2
Vi
Rs = 6kΩ, RL1 = 8kΩ, RL2 = 6kΩ, Cgs1 = 3pF , Cgs2 = 8pF , Cgd1 = Cgd2 = 1.5pF ,
Cdb1 = Cdb2 = 2.5pF , gm1 = 4mA/V , gm2 = 8mA/V .
Assume Cgb ||Cgs ∼ Cgs .
21
IIN
IOU T
Vi
Q–3 Use zero value time constant method to estimate the small signal dominant pole for the
current gain of the MOS cascode current mirror shown in Fig. 23. Assume an input ac
current source in parallel with IIN and a zero load impedance with Vout = Vgs3 + Vgs4.
The bias current IIN = 100µA. Compare your answer with the fT value of the devices.
Device parameters are W = 8µm, L = 1µm. Ignore channel length modulation and
substrate effect. Given Cgs = 35f F , Cgd = 2f F , Csb = Cdb = 7f F .
Q–4 The ac schematics of a common source stage and a common source common gate(cascode)
stage are shown below, with Rs = 6kΩ and RL = 12kΩ. Using the transistor and oper-
ating point data as ID = 0.5mA. Given Cgs1 = 3pF , Cgs2 = 8pF , Cgd1 = Cgd2 = 1.5pF ,
Cdb1 = Cdb2 = 2.5pF , gm1 = 4mA/V , gm2 = 8mA/V .
Assume Cgb ||Cgs ∼ Cgs .
M2
VO
VO
RS RS RL
Vi M1 RL Vi M1
a) Calculate the low frequency small signal voltage fain for each circuit
b) Use the zero value time constant method to calculate and compare the −3dB
frequencies of teh gains of the two circuits
22
c) Estimate the 10% to 90% rise times for each circuit for a small step input and
sketch the output voltage waeform over 0 to 300ms for a 1mV step input
Q–5 Find an expression for the transfer function of an NMOS source follower amplifier (Fig.
25). Ignore body effect.
VDD
Vin Rs X
Y Vout
CL
Solution : Consider the ac equivalent of the source followe as shown in Fig. 26.
Applying KCL at the outout node we can write
Vin Rs
CGD CGS
V1 gm V1
Vout
CL
V1 CGS s + gm V1 = Vout CL s
Thus
CL s
V1 = Vout
gm + CGS s
Applying KVL for the input section we can write
23
From above two equations we can write the transfer function as
Vout gm + CGS s
(s) =
V in Rs (CGS CL + CGS CGD + CGD CL )s2 + (gm Rs CGD + CL + CGS )s + gm
Note : Note that the transfer function has a zero. Thus the estimation of bandwidth
cannot be done using the dominant pole approximation. i.e. the zero value time constant
technique cannot be used for circuits with zero’s in the transfer function. Such circuits
can be identified by inspecting for a direct capacitive path from the input to output. The
pole value estimated by the zero value technique will be correct but will not correspond
to the bandwidth of the circuit due to the zero.
24
Topic Number 7 : Opamps
Q–1 Consider the amplifier in Fig. 27, where W/L of the PMOS transistors are 20µm/1µm
and that of NMOS are 10µm/1µm. Given Iref = 200µA and I = 20µA.
VDD
Iref I
M1 M2 CA
Vin
B
A
Vbias CB
M3 M4
Solution :
25
ro = λI1D
1
ro2 = ro3 = 0.07×100µ = 143kΩ
Threfore PA = 27.97Mrad/sec
The pole at node B is given by
PB = gCm4
√B √
2ID µP Cox (W/L) 2×20µ×90µ×20
PB = CB
P B = 0.1p
PB = 2.68Grad/sec
The net gain of the amplifier is given by
Av = Av1 Av2
Av = pgm2 (ro2 ||ro3 ) × 1
Av = √ 2ID µP Cox (W/L) × 71.5k
Av = 2 × 100µ × 90µ × 20 × 71.5k
Av = 42.9 = 32.65dB
Next we find the gain at the second pole frequency as follows.
A(PB ) = 32.65 − [20log(PB /PA )]
A(PB ) = 32.65 − [20log(2.68G/27.97M)]
A(PB ) = −7dB
Hence we can conclude that the gain touches 0dB at a slope of 20dB/decade. Hence
we calculate the unity gain frequency as
ωunity
20log( 27.97M = 32.65
Hence
ωunity = 42.9 × 27.97M
ωunity = 1.2Grad/sec
Now we can calculate the phase margin as
ω ω
φ = −tan−1 ( unityPA
− tan−1 ( unity
PB
−1 1.2G −1 1.2G
φ = −tan ( 27.97M − tan ( 2.68G
φ = −112.78o
Therefore the phase margin is
P M = 180 − φ = 67.22o
b) To redesign the amplifier for a phase margin of 55o , we first find the desired position
of PB and the corresponding value of gm4 . Since PM=55o , we have φ = 125o . We
can write
1.2G
−125 = − − tan−1 ( 27.97M − tan−1 ( 1.2G
PB
−36.33 = −tan−1 ( 1.2G
PB
PB = 1.63Grad/sec
PB = gCm4
B
gm4 = p
1.63G × 0.1p = 163µS
gm4 = 2ID µP Cox (W/L)
26
Therefore
ID = 7.4µA
will give the required phase margin.
Q–2 An amplifier has forward gain of A = 500 with two poles at P1 and P2 frequencies.
a) For P1 = 1MHz, find the phase margine with unity gain feedback if
i) P2 = P1
ii) P2 = 2P1
iii) P2 = 4P1
b) If the unity gain closed loop amplifier exhibits a frequency peaking of 60% near
the gain crossover frequency, then what is the phase margine?
VDD
M3 M4 M9
M11
Vbias3
M7 M8 M10
Rz CC
Vo
Vbias2
M5 M6
M1 M2
Vin
Vbias1
M13 M12
27
b) Determine the ICMR of the opamp
c) Derive the maximum output swing
d) Derive the value of Rz needed for a phase margine of 60o at unity gain feedback
when Cc = 1pF
e) Determine the slew rate of the opamp
WP WP WP WP
WPP
VBp VBn
W P /2 W N/2
WN WN
Vin
W NN
RL CL
WN WN
Q–5 Design a PMOS input stage two stage opamp. The specifications to be met are
VDD = 1.8V , GBW = 10MHz, CL = 1pF , Ao > 2000V /V , Slew rate > 10V /µs, ICMR
0.3V to 1.5V, Pdiss ≤ 2mW
Assume gm1 = gm2 and stability needed is P M = 60o .
28
Q–6 Consider the amplifier shown in Fig. 30 employing a cascode stage ans a CS amplifier.
Assume node A is dominant. Sketch the frequency response and explain how the circuit
can be compensated.
VDD
A B Vo
CA
VB2
Vin
29
a) If the circuit is modified as shown in Fig. 31, then sketch the frequency response
for
VDD
Cc
A B Vo
VB2
Vin
30
b) If the circuit is modified as shown in Fig. 32, then sketch the frequency response
for CC = CA and
VDD
Cc Rz
A B Vo
VB2
Vin
31
Topic Number 8 : Noise
Q–1 Consider circuits shown in Fig. 33. Which of these source degenerated circuits have
higher input referred noise voltage, and by what factor?
VDD VDD
RL RL RL RL
Vout
Vout
M1 M2 M1 M2
Vin Vin
RS RS 2RS
M3
Vbias Vbias
M3 M4
(a) (b)
VDD
RL
Ibias
Vout
M1 Vbias
Vin
M2 M3
Co
32
Q–3 The circuit in Fig. 35 is designed with (W/L)1 =40/0.4 and ID1 =0.4mA. If an output
swing of 2 V is required, estimate the dimension of M2 and M4 , such that the input
referred thermal noise current is minimum.
VDD
M5 M4
Ibias
Vout
I2
M1 Vbias
Vin
M2 M3
Co
Q–4 Assume (W/L)1 =(40/0.4), ID1 =0.2mA for Fig. 36. If the contribution of M2 to the
input referred noise voltage must be one-fifth of that of M1 , what is the maximum
output voltage swing of the amplifier?
VDD VDD
M2 Vbias M1 Vin
Vout Vout
M1 Vin M2 Vbias
(a) (b)
Q–5 Consider the circuit of Fig. 37 - Fig. 38. To decrease overall noise of circuit,
(a) Whether transconductance of M2 should be kept high or low in two cases of Fig. 37?
33
VDD VDD
M2 Vbias M2
Vout Vout
M1 Vin M1 Vin
(a) (b)
VDD
RL
Vout
M1 Vin
Rs
34
VDD
Vbias
Vin+ M1 M2 Vin-
Vout
M3 M4
Vn1 Vn2
Vin+ M1 M2 Vin-
Vout
Vn3 Vn4
M3 M4
2
Vno (f ) = 2(gm1 Ro )2 Vn1
2
(f ) + 2(gm3 Ro )2 Vn3
2
(f )
35
Step 3: Find input referred thermal noise voltage by dividing output noise value with
gain gm1 Ro which results in
2
2 2 2 gm3
Vneq (f ) = 2Vn1 (f ) + 2Vn3 (f ) (1)
gm1
1
where, Vni2 (f ) = 4kT ( 23 )( gmi )
2 16 1 16 gm3
⇒ Vneq (f ) = kT 2
+ kT 2
3 gm1 3 gm1
Ki
Step 2: Substitute noise sources with spectral density Vni2 (f ) = Wi Li Cox f
.
2 K1 µn K3 L1
⇒ Vni2 (f ) = +
Cox f W1 L1 µp W1 L23
(i) Taking L3 longer greatly helps due to inverse squared relationship in the second term.
(ii) The input noise is independent of W3 , and therefore it could be made large to maxi-
mize signal swing at the output.
(iv) Taking L1 longer increases the noise because the second term is domnant (due to
µn > µp . Thus, large L1 decreases the input-referred noise of p-channel drive transistors
which are not the dominant noise source, but, increases the input-referred noise of n-channel
load transistors, which are dominant noise sources.
36
Topic Number 9 : Oscillators
VDD
RD RD RD
VC
a) Derive an expression for small signal open loop gain of the circuit in terms of Vc .
b) From the above expression derive the frequency of oscillation
Q–3 A three stage ring oscillator is made with inverters that have asymmetric delay for the
rising and falling edge, due to difference in channel mobilities of NMOS and PMOS
devices.
Given td(0−1) = 1ns and td(1−0) = 0.8ns. Will this circuit oscillate. If yes calculate the
frequency of oscillation and the duty cycle of the oscillator output.
Q–4 Consider the small signal equivalent of an oscillator as shown in Fig. 42. The inductor
has a series resistance of rs . Estimate the maximum permissible value of rs in terms of
L, C1 , C2 and gm , for sustained oscillations.
C0
Q–5 Consider the oscillator shown in Fig. 43. Capacitance of the varactor is given by 1+α/Vc
.
α << Vc
37
L
C2
C1
VC
Q–6 Consider the oscillator shown in Fig. 44. Treating the transistors to be ideal switches
(the output voltage has only fundamental harmonic), sketch the spectrum of the current
through the inductor, capacitor and resistor.
Solution : Given sinusoidal output voltage, fundamental component of the current
through inductor and capacitor are equal and out of phase. The current through the
transistors is square. This square current will have all odd harmonics. The LC tank
together presents open circuit for the fundamental frequency. Hence this current flows
through the resistor. The higher harmonics flow through the capacitor as it presents
a low impedance as compared to the inductor and resistor. The bias current for the
transistor flows only through the inductor as it has zero DC resistance.
Fig. 45 shows the sketch spectrum of the currents through the inductor, capacitor and
resistor.
38
VDD
R L L R
C C
freq
freq
freq
f 2f 3f 4f 5f
39