You are on page 1of 76

Microelectronic Circuits

Common source amplifier

Single stage MOS amplifier


with drain-to gate feedback bias

Figure 4.38 Example 4.10: (a) amplifier circuit; (b) equivalent-circuit model.
Bits, pilani

CSA with current mirror bias

Bits, pilani

CSA with potential divider bias

Ac model with dc bias


No change in ac model with different biasing schemes So, for ac analysis, biasing circuit is not considered

T model

Bits, pilani

Common Source

Bits, pilani

Why do we need a capacitor Cs?


Without Cs. Vgs vin, As a result a.c current becomes small [gm (vgs-vx)] Hence drop across Rd reduces Gain reduces
Bits, pilani

Bits, pilani

Pmos CSA

Common Source
Bias point Given--Kn= 80uA/v2, Vt=1v, Vdd= 5v Suppose, we want Id= 100uA, Vds= 2.5v, This gives ---Rd= 25k
Bits, pilani

Design of common source for gain


To set gain

Av= 30 = - gm [ RD || ro ]
ro=1/ Id = 1M ohm; = .01V-1 gm= 1.2 mA / V = 2 Id / Vov Hence, Vov= 0.166V [designers choice depends on VDD, Small value for small supply ] gm= 1.2 mA / V = Kn (w/L) Vov W/L= 90.3, Vgs= 1.166V
Bits, pilani

Practical design
Sketch VTC Choose bias current, choose a w/L Find corresponding Vgs, Vds, re calculate Rd, w/L Draw the schematic in eda tool Set voltages by applying sources Do .dc analysis Check all dc current and voltages Apply ac , check ac output. Do .ac analysis

gm/ Id
transconductance generation efficiency
It is strongly related to the performances of analog circuits 2) It gives an indication of the device operating region. 3) It provides a tool for calculating the transistors dimensions The gm/ID ratio is a measure of the efficiency to translate current (hence power) into transconductance; i.e., the greater the gm / ID value, the greater the transconductance we obtain at a constant current value.

Intrinsic gain

How to maximize gain in active region?

Av increases with W/L, VRD, or by dec ID


Bits, pilani

Trade offs
W/L increased keeping ID and VRD constant ---greater device capacitance at output , time constant increases, speed of response is affected ---less overdrive voltage reqd. ( as Id constt.) so vomin reduces output swing range increases, bias point will shift ----gain increases Bits, pilani

Bias point for increased (w/L)--- ID and V

RD

constant

(w/L) 2 large (w/L) 1 small

vgs 2 small

vgs 1 large

Bits, pilani

Id =
Vdd/Rd

(w/L) 2 > (w/L) 1 ID and VRD constant

Q1, Q2 (w/L)2 large Vov2 less


Less lower swing

vgs1 vgs2

(w/L)1 small Vov1 more


More lower swing

vov2 less

vov1 more

Vdd VdS

VRD increased keeping Id and

W/L constant
Gain increases Vdd - VRd inc., or Vout max reduces, output swing range decreases, MOS shifts towards triode region, Voltage swing is limited
Bits, pilani

Bias point for increased VRD ------ID and w/L

constant

vgs 1,2 Bits, pilani

Id=
Vdd/Rd

(VRD)2 > (VRD)1 Id and W/L constant

(VRD)2 large

Q2 vgs Q1
Less lower swing

(VRD)1small
More lower swing

vds2 less

vds1 more

Vdd VdS

ID dec. keeping VRD and W/L const.


ID reduces---to keep VRd constant Rd must increase, So, Rout shd. be increased time constant increases, So, speed of response is affected Vgs must dec. Gain increases
Bits, pilani

Bias point for decreased (ID)--- (w/L) and V

RD

constant

vgs 2 vgs 1 small large

Bits, pilani

Id=
Vdd/Rd

VRD and W/L constt.

(Id)2 < (Id)1

Q1 vgs1 (Id) 2 small Vov2 less Q2 Less swing ((Id)1 large Vov1 more More swing vgs2

vov2 less

vov1 more Bits, pilani

Vdd VdS

Maximum gain possible (intrinsic


gain)

Av= - gm ro Make Rd infinite But this will make VRd=Vdd, MOS goes to cutoff How?---Rd replaced by current source
Bits, pilani

GAIN and SWING COMPATIBILITY


What if gain is high but not sufficient voltage swing ? Output will be distorted as MOSFET slips in LINEAR/ CUTOFF region Hence, output voltage swing calculation is important

Output voltage swing under


dynamic state
Constraint---MOS must remain in saturation Vo max Vdd Vomin Vgs-Vt Output DC level [Vdd-IdRd]
Vdd= 3V Reduced upper swing Dc level Reduced lower swing For symm sigdc level shd. be in the middle for max swing

Vgs-Vt = 0.2

How to configure an Active current source ?

Bits, pilani

Current source load


Pmos current mirror

Rout= ro1 || ro2

-Swing [ Vdd- Vov2] to Vov1 -MOS Cap. at output increases due to M2 RoutCout product inc.
Bits, pilani

Output Voltage swing


Vomax---- Vdd - Vov2 Vomin---- Vov1 For Vdd=3v, Vgsp = Vgsn =.9v, Vt=0.7v Vomax---- 0.2v Vomin---- 2.8v

2.6v

Bits, pilani

Penalty paid
Fixed Vgs required to support current bias No roll back flexibility for Q point Remedy--Place a current mirror at source also. Prob.---Then we have to use Cs to provide bypass path Cant we remove Cs??

Effect of technology scaling on intrinsic gain (C E SCALING)


MOS scaled down by constant electric field scaling strategy All dimensions scale down by a factor : ( > 1: usually =1.33) All voltages scale down by a factor
Bits, pilani

Effect of technology scaling on intrinsic gain (C E


SCALING)

Keeping W/L constt----- Vdd decreases by L scaled down by , W also decreases to keep ratio constant, Cox, increases by -----Id decreases by ---gm remains constt. increases by , Id decreases by ---ro constt. Hence intrinsic gain does not change
Bits, pilani

Guideline--To increase intrinsic gain by design


Keep L large Keep Id small Example--Keeping (w/L) constant L= 4L, W=4W, I = I/4 Intrinsic gain= 1/ [ x ]= 8 times increase Intrinsic gain [8] original intrinsic gain by properly choosing values
Bits, pilani

Effect of increasing only L

Keeping I constt----L scaled up by , ro increases L scaled up, W also increases to keep ratio constant, so gm remains same Hence intrinsic gain increases by
Bits, pilani

Effect of increasing L on CSA gain

Keeping I constt.----L1, L2 scaled up by , (ro1, ro2) increase W1, W2 also increase to keep ratio constant, so gm remains same Hence CSA gain increases by

Bits, pilani

----Active load CS amplifierother types of load


Diode connected load--bad load small impedance, leads to small gain
I

I V

desirable

Vgs

Bits, pilani

Equivalent Resistance calc.

Bits, pilani

Active load amplifier


pmos

Small value Gain depends on device dim for same I. Also on p if load is pmos

Gain improvement technique for diode connected loadinc. I1


I1/4

But Vgs of M1 will increase to carry 4I current. Swing reduces

Double gain

Gain depends only on device dim. ---so, linear amplification, less distortion For lin. Amp. ----Gain shd. be independent of bias voltages and current Reason--|Av|= gm Rd changes as gm changes with the input signal swing (though close to Q point)

Non linearity due to bias dependence


V vi Voeach input point is amplified by a slightly different gain value

Bad drawing

Triode load

Bits, pilani

CSA----

Bits, pilani

ADV. / DISADV.
No bypass capacitor---------good High gain, high swing--------good Large capacitance at the output--------bad A fixed Vgs is reqd .----worst --- bias point shift with input signal variation, temp. , process variation We need to keep Vgs constant. (or make gain a weaker function of gm) So, allow source node to swing. ---Again place a current source at source node or put a resistor Rs
Bits, pilani

Circuit trans-conductance Gm, Av CSA

=gm of MOS

Rout = ro1 || ro 2
Vout Vout Iout Av = = Vin Vin Iout Av = R out G m = g m R out
Bits, pilani

Common source amplifier


with Source Degeneration

Motivation- remove Cs
Aim----How to remove Cs ? To allow source node to swing
Put a current mirror sink at source but it has high ro So, replace current mirror by low Rs in place of current mirror sink and remove Cs. Ibias is decided by current mirror load --Can we get minimum (ideally NO) reduction in gain??

CSA with Source Degeneration

vout

Bits, pilani

Performance variation
Vgs < vin, gain reduces Output impedance will change Non linearity decreases as Id varies smoothly w.r.t vin variations
Without Rs With Rs

Bits, pilani

Nonlinearity Error

Circuit trans-conductance Gm

ID

gm of MOS
Bits, pilani

ID

Circuit trans-conductance

Voltage Gain

Bits, pilani

Trans-conductance variationcomparison
Gm=

Without Rs

With Rs

Bits, pilani

Gm, Rout from approximate small signal model

= gm (Vin-Vx) Vx= Iout Rs

Gm, Rout from small signal model with ro and gmb

Bits, pilani

gm

For Rs=0, Gm = gm For Rs 0, Gm < gm

Bits, pilani

Rout
Rout= Rd || Rout

Rout
Bits, pilani

Rout

Rout

Bits, pilani

Why Rout increases due to Rs?


A high impedance at a node means if voltage at that node is changed slightly, current through that node does no change much. Here we achieve this at Vout node through negative feedback in vgs.--If Vout ---Id (effect comes through ro)
Bits, pilani

If Vout ---Id (effect comes through ro) At the same time, due to Vout , a voltage vs appears at source Vs causes reduction in Vgs, due to which Id falls more due to square dependence on Vgs Hence Id changes little in comparison to change in vout So Rout increases
Bits, pilani

vs

Voltage gain (from ac model)

If ro >> Rs, Rd

Using GmRout

Gm
Bits, pilani

Rout

Gain rewritten (simplified form)

Bits, pilani

High / low impedance node


Look at R= V/ I

Bits, pilani

Resistance looking into the source


Considering ro to be large Vx / ix

Bits, pilani

Rout by inspection

Bits, pilani

Bits, pilani

Gain by inspection

Replace by nmos

?
Bits, pilani

Ideal Current source load--problem???


Av Effect of Rs is negated as I0 (ideal) can not sustain current variations
Can not change

We want non ideal active load


Bits, pilani

Push-pull amplifier

Bits, pilani

Power efficiency
Class A amplifierconducts for full cycle

Bits, pilani

Power Efficiency ---power delivered to load


For equal swing in both directions, Vo shd. be biased at Vdd/2 and then current through RL= Vdd/2RL
PL = 100 % PS Now _ when _ up transition ; Vo max V o = Vdd ; for max Vo Ps = V dd I Vdd I PL = 2 2
Bits, pilani

Vdd Vdd , IL 2 RL 2

( )

rms peak vout

-----for max. swing Vdc= vdd/2, IL=I/2 In up (or down) Transition Vomax=vdd (or 0), extreme value Peak vo amplitude= vdd/2 Peak current supplied by vdd = I/2 = 25%
Bits, pilani

You might also like