Lecture 3

Common Source Amplifier

Small-Signal Model

Jayant Charthad
Stanford University
jayantc@stanford

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-1

 Let's Build Our First Amplifier

• One way to amplify

– Convert input voltage to current using voltage controlled
current source (VCCS=“trans-conductance”)
– Convert back to voltage using a resistor (R)
• "Voltage gain" = DVout/DVin
– Product of the V-I and I-V conversion factors

ΔVin ΔI ΔVout
VCCS R

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-2

 Common Source Amplifier

• MOS device acts as VCCS

VDD

Vo

Vi
Id T

1 W 1 W
I D = µCox (Vi − Vt )2 Vo = VDD − µCox (Vi − Vt )2 ⋅ R
2 L 2 L

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-3

 Biasing

• Need some sort of "battery" that brings input voltage into useful
operating region
• Define VOV=VI-Vt, "quiescent point gate overdrive"
– VOV=VGS-Vt with no input signal applied

VDD
Vo

VO +ΔVo VO DVo
DVi
I D+ΔId

"Signal" ΔVi VOV

"Bias" VI

VI

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-4

 Relationship Between Incremental Voltages

• What is DVo as a function of DVi? Note: Vgs=Vi=(VI+DVi)

1 W
VO + ΔVo = VDD − µCox (VOV + ΔVi )2 ⋅ R
2 L
1 W
[
ΔVo = − µCox R ⋅ (VOV + ΔVi )2 − VOV 2
2 L
]
1 W
[
= − µCox R ⋅ 2VOV ΔVi + ΔVi 2
2 L
]
2I & ΔVi #
= − D ⋅ R ⋅ ΔVi $1 + !
VOV % 2VOV "
• As expected, this is a nonlinear relationship
• Nobody likes nonlinear equations; we need a simpler model
– Fortunately, a (1st order) linear approximation to the above
expression is sufficient for 90% of all analog circuit analysis
A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-5
 Small Signal Approximation (1)

2I D & ΔVi #
ΔVo = − ⋅ R ⋅ ΔVi $1 + !
VOV % 2VOV "
• Assuming DVi << 2VOV, we have
2I D
ΔVo ≅ − ⋅ R ⋅ ΔVi
VOV

• If we further pretend that the input voltage increment is infinitely

small, we can find this result directly by taking the derivative of
the large signal transfer function at the "operating point" VI

dVo 2I D W
=− ⋅ R = −µCox VOV ⋅ R = Av
dVi Vi =VI
VOV L
Small-signal voltage gain

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-6

 Small Signal Approximation (2)

• Graphical illustration:

Notation:

dVo/dVi = vo/vi = Av
VO

VOV

VI

• The slope of the above tangent is the so called "small-signal

voltage gain" of our amplifier (Av)

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-7

 Notation

Quiescent Incremental
Total quantity point value change

Vo = VO + vo
Alternatively:
(IEEE standard)
Quiescent Incremental
Total quantity point value change

vO = VO + vo
A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-8
 Small Signal MOS Model
• Fortunately we don't have to repeat this analysis for every single
circuit we build
• Instead, we derive a linearized circuit model for the MOS
transistor and plug it into arbitrary circuits
ID+id id

+
+
VDS +vds gm⋅vgs
+ + vds
-
VGS+vgs vgs -
- Conditions: -
VDS>VGS-Vt
Id vgs<<VGS-Vt id

gm
gm
ID vgs

VGS Vgs

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-9

 Transconductance
• The parameter that relates small signal gate voltage to drain
current is called transconductance (gm), or y21 in two-port
nomenclature
• The transconductance is found by differentiating the large signal
I-V characteristic of the transistor at its operating point

1 W
I D = µCox (VGS − Vt )2
2 L
id ∂I W W
gm = = D = µCox (VGS − Vt ) = µCox VOV
v gs ∂VGS L L

2I D
gm =
VOV

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-10

Small-Signal Equivalent of CS Amplifier
VDD

R
VO +ΔVo

I D+ΔId +
+ gm⋅vgs
ΔVi vo
vi vgs -
VI -

• Use large signal I-V law to compute operating point (ID, VO, gm)
– Make sure device operates in proper region; consider
desired “signal swing”
• Now perform rest of calculations in “small-signal land”
– Gain, bandwidth (more later), …
A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-11
 Example (1)

• Given: VI=1.5V, W=20µm, L=1µm, R=5kW, VDD=5V

• Technology parameters: µCox = 50µA/V2, Vt=0.5V
• Calculate: ID, VO, gm, Av

VDD
1 µA 20 2
I D = ⋅ 50 2 ⋅ ⋅ (1.5V − 0.5V ) = 500µA
R
2 V 1
VO +vo
VO = 5V − 5kΩ ⋅ 500µA = 2.5V
ID+i d

vi VDS = VO = 2.5V $
# ⇒ Saturation
VI VGS − Vt = VI − Vt = 1V "

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-12

 Example (2)

R 2I D 2 ⋅ 500µA
gm = = = 1mS
VOV 1.5V − 0.5V
+
+ gm⋅vgs
vo
vi vgs
-
- Av = − g m R = −1mS ⋅ 5kΩ = −5

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-13

 Getting Started with HSpice

• The above circuit was easy to analyze

– And it is unlikely that we made a mistake
• In general, we want to be able to compute circuit characteristics
both manually and by using a circuit simulator
– Both hand calculation and simulation is important; one
does not “replace” the other
– “Double book keeping is important in design and analysis to
detect flaws in assumptions and understanding
• Let’s see how we can duplicate this result using HSpice…

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-14

 HSpice Input File (1)
* Common source amplifier
* B. Murmann, Fall 2008

*** device model

.model my_nmos nmos kp=50u vto=0.5

*** useful options

.option post brief nomod

vdd vdd 0 5

*** input voltage

vi vi 0 dc 1.5 *** value for .op analysis
+ ac 0.1 *** amplitude for .ac analysis
+ sin 1.5 0.1 1k *** sinewave for .tran: V_I=1.5V, v_i=0.1V, f=1kHz

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-15

 HSpice Input File (2)
*** d g s b
mn1 vo vi 0 0 my_nmos w=20u l=1u
R vdd vo 5k

*** calculate operating point

.op

*** large signal analysis (sweep Vi)

.dc vi 0 5 0.01

*** small signal analysis (sweep frequency)

.ac dec 10 100 1k

*** transient analysis (sweep time)

.tran 1u 5m
.end

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-16

 .op Output
**** mosfets
element 0:mn1
model 0:nmos114_
region Saturati
id 500.0000u
vgs 1.5000
vds 2.5000
vbs 0.
vth 500.0000m
vdsat 1.0000
vod 1.0000
beta 1.0000m
gam eff 527.6252m
gm 1.0000m
gds 0.
…

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-17

 .dc Output

5.5

5
4.5
4

3.5
]
V 3
Vo [V]

[
o 2.5
V
2

1.5

0.5

0
0 1 2 3 4 5
V i [V]

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-18

 .ac Output

1
]
V
|vo| [V]

[
0.5
|v o
|

0 Av = vo/vi
2 3
10 10
f [Hz] = -0.5V/0.1V
] = -5
Phase (vo) [Degrees]

V
v
[ 300
(
e
) o 200
s
a
100
h
p
0 2 3
10 10
f [Hz]

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-19

 .tran Output
3.5

2.5

2
]
[V]

V
[ 1.5

0.5 Vi
Vo
0
0 0.5 1 1.5 2 2.5 3
t [sec] x 10
-3

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-20

 Another Run

• Now with the following stimulus

*** input voltage

vi vi 0 dc 1.5 *** value for .op analysis
+ ac 1000 *** amplitude for .ac analysis
+ sin 1.5 1000 1k *** sinewave for .tran: V_I=1.5V, v_i=0.1V, f=1kHz

• 1000V input amplitude applied to the circuit!

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-21

 .tran Output
6

3
]
V
[V]

[ 2

0 Vi
Vo
-1
0 0.5 1 1.5 2 2.5 3
t [sec] x 10
-3

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-22

 .ac Output

10000

9000

8000

7000
]
V 6000
[
|vo| [V]

5000
|v o
| 4000

3000
5000V output!
2000

1000

0 2 3
10 10
f [Hz]

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-23

 Important to Remember

• Once a small-signal model of the circuit is constructed, all large

signal information is lost
– The small-signal (.ac) circuit transfer function is linear and
extends from –¥ to + ¥
– Features such as finite voltage range, signal clipping, etc. are
lost and completely meaningless in a small-signal analysis (or
.ac simulation)
• The input amplitude in the .ac statement is irrelevant and can be
set to any number (other than 0)
– Best to use 1V, in which case the output amplitude
corresponds to the circuit gain

A. Arbabian, R. Dutton, B. Murmann EE114/214A L03-24