Session - 04
Session Speaker
Chandramohan P
V
Current
I source
Current I
sink
V
I
V
©M S Ramaiah School of Advanced Studies - Bangalore 5
Non-ideal current sources / sinks PEMP VSD 528
go is small
go is small
Solution:
Assume VGS= 0.7V
VDD
The value of R, can be found by assuming
ID1= ID2 = 30µA, is determined by solving
IRef
the equation below ID2= Iout
VDD −VDS M1
+
M2
ID1 = VGS
R –
Kn W
ID2 = 30μA = (VGS −VTH)2
2 L
W
30μA =165.5× ×(0.7 −0.379924
)2
L
W
= 1.769
L
Let L1 = L2= 0.36µm, then W2=0.636µm, which gives W1=0.636µm
.include "C:\synopsys\FT07Analog\modelfile018.txt"
R1 Vdd P1 32.5K
M1 P1 P1 gnd gnd CMOSN L=0.36U W=0.7U
M2 P2 P1 gnd gnd CMOSN L=0.36U W=0.7U
Solution
Note that VDS(Sat)4 = 0.8V the source of M4 is at 0.8V, the drain voltage must
be greater than 0.5V to ensure operation in the constant current region.
Setting the W/L ratio of M1and M3 to yield VGS = 0.8V with Iref = 50µA
I REF
VGS = VTn + 50
W μnCox 0.8 = 0.379924 +
W
× 165.5
L 2 L
W W1 W3
= 1.712 ∴ = = 1.712
L L1 L3
We can match the (W/L) of M2 and M4 with M1 and M3, respectively, since
we want IREF = IOUT
W2 W4
∴ = = 1.712
L2 L4
To calculate the small signal output resistance
RS = (gm4 ⋅ ro4)ro2
1 1
ro 4 = ro 2 = ro 4 = ro 2 = = 0.22 MΩ
λn I D 0.09 × 50μ
W
gm4 = 2 μ nCox I REF = 4 ×1.712 ×165.5 × 50 = 238.04
L
I1 Vdd P1 dc 50u
Vdd Vdd Gnd dc 1.8V
V1 P2 gnd dc 0.5V
• Simulation Results
Solution:
We have g m1 = 2 μ nCox (W / L)1 I bias = 1.06mA / V
and
12,000 × 1.6 μm
rds 2 = = 192 kΩ
0.1mA
Vdd 1 0 dc 5
Ibias 2 0 dc 100u
M3 2 2 1 1 pmos W=200u L=2.6u
M2 3 2 1 1 pmos W=200u L=2.6u
M1 3 4 0 0 nmos W=200u L=2.6u
Vin 4 0 dc 0.849 ac 1
.op
.ac dec 10 1k 10000Meg
.plot ac vdb(3)
Use the parameters as in example1 along with the following : Rin=180 KΩ, CL=
0.3pF, Cgs1= 0.2pF, Cgd1= 0.015pF, Cdb1= 20fF, and Cdb2= 36fF.Estimate the –3dB
frequency of the common-source amplifier shown in the figure.
Solution:
[ [ ]
f −3dB = ⎢ ⎥ Rin C gs1 + C gd 1 (1 + g m1 R2 ) + R2 (C gd 1 + C2 )
⎡ 1 ⎤
]−1
⎣ 2π ⎦
= 550kHz.
Vdd 1 0 dc 5
Ibias 2 0 dc 100u
M3 2 2 1 1 pmos W=400u L=2u
M2 3 2 1 1 pmos W=400u L=2u
M1 3 4 0 0 nmos W=150u L=2u
Rin 5 4 180k
Vin 5 0 dc 0.849 ac 1
Cl 3 0 0.3p
.op
.ac dec 10 1k 100Meg
.plot ac vdb(3)
Source Follower Stage with a Current Mirror used to supply bias current
Consider the source follower in the figure where all transistors have W/L = 100μm/1.6μm,
μnCox=90μA/V2, μpCox=30μA/V2, Ibias = 100μA, γn= 0.5V1/2 rds-n(Ω) = 8,000L(μm)/ID(mA).
What is the gain of the stage?
Solution:
We have g m1 = 2 μ nCox (W / L)1 I bias = 1.06mA / V
We have 1.06
AV = = 0.86V / V
1.06 + 0.16 + 1 / 128 + 1 / 128
Vdd 1 0 dc 5
Ibias 1 2 dc 100u
M3 2 2 0 0 nmos W=80u L=2u
M2 3 2 0 0 nmos W=80u L=2u
M1 1 4 3 0 nmos W=80u L=2u
Vin 4 0 dc 2 ac 1
.op
.ac dec 10 1k 1000Meg
.plot ac vdb(3)
Using the parameters in example2 and assume Rin=180 KΩ, CL= 10pF, Cgs1=
0.2pF, Cgd1= 15pF, Csb1= 40fF, and Cin= 30fF.Find ω0, Q, and the frequency of
the zero for the source follower.
Solution:
C s = C L + C sb1 = 10.04 pF
−π / 4 Q 2 −1
%overshoot = 100e = 8%
− g m1
Zero frequency ωz = = 844 MHz , and thus it can be ignored.
C gs1
Vdd 1 0 dc 5
Vss 2 0 dc -5
Ibias 3 2 dc 100u
Rin 4 0 180k
Cin 4 0 30f
Cl 3 0 5p
M1 1 4 3 2 nmos W=150u L=2u
Iin 4 0 pulse(0 -5u 10n 0 0)
.op
.tran 0.5n 300n
.plot v(3)
Using the parameters as in example above find the compensation network and the resulting first
order and second order poles of the source follower
The capacitor is a reasonable value to be realized on chip. The resistor can be realized by a MOS
transistor biased in the triode(ie., linear) region. Assuming the compensation network is used, the
poles of the transfer function then become
Gin
p1 ≅ = 2π × 3.61MHz
C gs1 + Cin
'
And
g m 1 + G s1
p2 = = 2π × 19 .3 MHz
C gs + C L
The speed penalty paid for using the compensation network is quite high, because the pole frequency
without compensation was around 8 MHz whereas here the dominant pole is at 3.6 MHz.
Vdd 1 0 dc 5
Vss 2 0 dc -5
Ibias 3 2 dc 100u
Rin 4 0 180k
Cin 4 0 30f
Cl 3 0 10p
M1 1 4 3 2 nmos W=200u L=2u
Iin 4 0 dc 0 ac 1
C1 4 5 0.17p
R1 5 0 49.3k
.op
.ac dec 10 1k 1000Meg
.print vdb(3)
.lib C:\Documents and Settings\Desktop\SPICE\CMOS 1.2UL3.lib
.end
©M S Ramaiah School of Advanced Studies - Bangalore 56
PEMP VSD 528
The Bode Plot of a Source Follower with compensation network
-20dB/decade
-40dB/decade
Design Considerations:
Constraints Specifications
Power supply Small-signal gain
Technology Frequency response (CL)
Temperature ICMR
Slew rate (CL)
Power dissipation
• Procedure:
– Pick ISS to satisfy the slew rate
knowing CL or the power
dissipation
– Check to see if Rout will satisfy
the frequency response, if not
change ISS or modify circuit
– Design W3/L3 (W4/L4) to satisfy
the upper ICMR
– Design W1/L1 (W2/L2) to satisfy
the gain
– Design W5/L5 to satisfy the
lower ICMR
– Iterate where necessary
Solution Step1:- To meet the slew rate, and maximum Power Dissipation
2
Rout is given by Rout =
(λN +λP )⋅ I5
2
I5 = = 34 .888 μA
(0.09 + 0.09 ) ⋅ 318471 .33
⇒ I 5 ≥ 35μA LL (3)
From Eqns. (1), (2) and (3) We can pick the I5 as approx. 100µA
Step3:- The maximum input common mode voltage gives
⎛W ⎞
2 ×165.5 × ⎜⎜ 1 ⎟⎟ ⎛ W1 ⎞
⎝ L1 ⎠ ⎜⎜ ⎟⎟ = 48.9425 ≅ 49
100 =
1.8 × 50 ⎝ L1 ⎠
©M S Ramaiah School of Advanced Studies - Bangalore 63
PEMP VSD 528
CMOS Differential Amplifier
W2 W1
∴ = = 49
L2 L1
W5 I 100
= ' 25 = = 5.18
L5 K nVDS 5( sat ) 165.5 × (0.3415) 2
W5
=5
L5
W6
=5
L6
• SPICE code
CMOS Differential Amplifier with PMOS Current Mirror
• Simulation Results