Operational Amplifiers I (Ch. 24)p p

김 영 석김 영 석

충북대학교 전자정보대학

2011.3.1. .

Email: kimys@cbu.ac.kr

전자정보대학 김영석 24-1

Contents24.1 The Two-Stage Op-Amp

24.2 An Op-Amp with Output Buffer

24.3 OTA

24 4 G i E h t24.4 Gain Enhancement

전자정보대학 김영석 24-2

Block Diagram of OP-AMPOp-amp (Operational Amplifier) is a fundamental building block in Analog IC Design

Diff-amp + Gain Stage + Buffer

Design Spec:Design Spec:

Open-loop Gain

CMR

CMRR

PSRR

Output Voltage Range

Current Sourcing/Sinking

Power DissipationPower Dissipation

전자정보대학 김영석 24-3

24.1 Two-Stage Op-AmpLow-Freq Open-Loop Gain VVrgrrgA opmpoponmnOLDC /832)||( ≈⋅⋅⋅=Input Common-Mode Range

opmpoponmnOLDC

VVVVDDVmVVVV ssGSCMMIN

93028035014501.035.022,1

++

=+=Δ⋅+=

Power Dissipation: 30uW

Output Swing and Current Source/Sinking

mVVVVDDV THNSGCMMAX 93028.035.013 =+−=+−=

Output Swing and Current Source/Sinking

mVVVDDVmVVV

OUTMAX

OUTMIN

9001002

7

8

≈Δ−=≈Δ⋅=

Offset=Random+Systematic

AIAI SINKOUTSOURCEOUT μμ 10,10 ,, =>

Stability?

전자정보대학 김영석 24-4

Stability1) Rz=0, Cc=100fF

fFCCCC

MggVAgg

krRkrrR

mmpmmn

opopon

613

)7(/150

333,111||

21

21

=++=

====

Ω=≈Ω==

μ

kHzCRR

f

fFCfFCCC

fFCCCC

LgdL

gsgddg

287)(2

1

100@56.101

6.13

1

82

7241

==

==+=

=++=

MHzCCCCCC

Cgf

CRgR

cc

cm

cm

210)(2

)(2

2211

22

2211

=++

=π

π

MHzC

gfc

mz 240

22 ==

π

전자정보대학 김영석 24-5

Stability2) Rz=0, Cc=2.4pF

cc

mun

OL

pFCMHzC

gf

MHzfAPMgetTo

=⇒==

°−==∠°=

4.2102

100)10(,80

1

π c

ffffRHP zerotodueUnstablePMtionBut Simula

≈↑↓

≈=>

:!! ,0

21 unz ffffRHP zerotodue ≈↑↓ ,,: 21

전자정보대학 김영석 24-6

Stability3) Rz=6.5k, Cc=2.4pF

kRg

sistorNullingZeroofAddition

zmn

5.61Re

Ω==

−

i titdi ibllRB t

StablePM

1

! ,90°≈

iationprocesstodueimpossiblenearlyRg

But zmn

var , =

전자정보대학 김영석 24-7

Stability4) Rz=MOSFET, Cc=2.4pF

To Overcome Process Variations

Use MOSFET in Triode

271

1)16.9(1,

z

SGzSGPSGSGP

gWR

VVVV

==

==

Problem: Power Dissipation

1mzp

gVL

WKP Δ

전자정보대학 김영석 24-8

Stability5) SF, Cc

To remove RHP Zero, Use Source Follower + Cc

RHP Zero due to Signal Forwarding Through Cc

> R Si l F di Th h C=> Remove Signal Forwarding Through Cc

:Direction Reversenot But :back gFeedbackin

21

12

vSFCcvvCcSFvvout

=>=>=>=>=>=>=

SF ofRout low todue Impossible 21

전자정보대학 김영석 24-9

Stability6) CG, Cc

To remove RHP Zero, Use Common Gate Amp + Cc

RHP Zero due to Signal Forwarding Through Cc

> R Si l F di Th h C=> Remove Signal Forwarding Through Cc

:DirectionReversenotBut)(:back gFeedbackin 12

vCcMCGvvCGMCGCcvvout

=>=>=>=>=>=>=

MCG of Source Drain to frompath signal little todue Impossible :Direction Reversenot But 21 vCcMCGv =>=>=>

전자정보대학 김영석 24-10

Stability6) CG, Cc - Continued

ssgCsRgRg

vv mcg

cmm

s

out

(...)(...)(...)

)1(

2

2211

++

+−=

FreqHigherCgCgf

ZeroRHPnotZeroLHPC

gf

cmcm

c

mcgz

:))6621((

) ,!!! (2

22 >=

=π

FreqHigherZeroLHPtodueOKNowfgf

FreqHigherCCCCCCCC

f

m

cc

:)(

Set

:))66.21()(2

()(2

1

2211212

≈

++>=

ππ

fFCMHzf

FreqHigherZeroLHPtodueOKNowfC

gf

cun

zc

mun

240@100

:) (2

1

==

≈=π

전자정보대학 김영석 24-11

Stability7) CG, Cc

Same as 6)

4Di iRB)(4:back gFeedbackin 12

CBMvCGBMCcvvout =>=>=>=

No Additional Power Dissipation

M4B of Source Drain to frompath signal little todue Impossible 4:Direction Reversenot But 21 vCcBMv =>=>=>

No Additional Power Dissipation

전자정보대학 김영석 24-12

Slew-Rate Limitations1) Miller Comp

nsmVpFA

CISRvofIncreaseLimitsiBut

vvONMvV.V.h: Low-to-Hig

SSLHoutC

outGp

c/3.8

4.220:

2,9050 7

≈==

↑↓=>=>↑=>=>

μ

AIvCurrentSinkMvCv

ONMMMvV.V.w: High-to-Lo

pFC

outoutcG

p

c

10 8 )2( , )1(

4,3,1,5090

4.2

7 ↓=>↑=>↑=>

=>↓=>=>

μ nsmVpFA

CCISR

Lc

DHL /3

4.310

)( 8 ≈=

+=

μ

2) Indirect Comp: Faster

nsmVpFA

CCISR

Lc

DHL /8

24.110

)(8 ≈=

+=

μpCC Lc .)(

전자정보대학 김영석 24-13

CMRRCMRR is determined by the diff stage

dBAA

AAAA

AAfACMRR

c

d

c

d

c

OL 50||log20||log20|)(|log202

2

2

≈===

전자정보대학 김영석 24-14

PSRR

77433 0,:

/)(

outSGDDG

out

OL

vvvvvvvfreqlowat

vvfAPSRR

≈↑=>↑===↑=>

=

+

++

77

/)(

)1(:

out

OL

outDGc

vvfAPSRR

gainvvvvshortCvfreqhighat

=

==≈==>↑=>

−−

++

78

78

1:

:

mBmoutc

oBmout

ggvshortCvfreqhighat

rgvvfreqlowat

+≈=>↑=>

+≈↑=>

−

−

전자정보대학 김영석 24-15

Increasing Input CMRVIN,CM low => PMOS diff ON

VIN,CM high => NMOS diff ON

전자정보대학 김영석 24-16

24.2 An Op-Amp with Output BufferWhen Driving Resistive Load, Output Buffer is required

)||)((),||(

84800,156.31500

2244

22211

21

oopoonmopmonomomm

OLDC

rrggArgrggA

dBAAA

+−=−=

==⋅==

전자정보대학 김영석 24-17

CompensationUse Indirect Comp

fFCPMMHzf cun 240@70,100 =°=≈

전자정보대학 김영석 24-18

Effect of Device ShrinkL=100nm => 50 nm

Speed Enhancement

But, gain=84dB => 44dB due to CLM

전자정보대학 김영석 24-19

24.3 The Operational Transconductance Amp (OTA)OTA=Amp where all nodes are low impedance except input/output nodes

KVVrrgKvv

vA oomnp

outv 1@/65.16)||( 54 ==⋅⋅=

−=

MHzpF

VAC

gf

gKg

L

mnun

mmOTA

1

2412

/1502

=⋅

==

⋅=

πμ

π

MHzCrr

fLoo

dB 4.1)||(2

1

543 ==

π

전자정보대학 김영석 24-20

Increasing the OTA Output Resistance

54/500)||( dBVVRRgA ==⋅=

85

47)||(2

1

54/500)||(

3

PM

kHzCRR

f

dBVVRRgA

LocaspocasndB

ocaspocasnmnv

°=

==

===

π

)4.1,4.24: (85

3 MHzfdBAOTASimplePM

dBv ==

Important Note:

UnStableCAmpOp

StableMorefCOTA unL

↑=>

↓=>↑=>

:

:

UnStableCAmpOp L ↑=>− :

전자정보대학 김영석 24-21

OTA with an Output Buffer (An Op-Amp)1) Cascode OTA with CS Output Buffer

전자정보대학 김영석 24-22

OTA with an Output Buffer (An Op-Amp)2) Cascode OTA with Class-AB Output Buffer:

=> 100uA Lower Current than OTA with CS Output Buffer

전자정보대학 김영석 24-23

Comparison with CascodeCascode

M2-MOP: Fast

But, Limited CMR

OTA with Class AB

M1-M31-M3-M51-M5-MON: Slow

Better Input CMR/Speed

=> Folded-Cascode

전자정보대학 김영석 24-24

The Folded-Cascode OTA/Op-AmpBetter Input CMR and Speed

LdB

ocaspocasnmnv

CRRf

RRgA

π )||(21

)||(

3 =

⋅=

L

mnun

Locaspocasn

Cgf

CRR

π

π

2

)||(2

=

전자정보대학 김영석 24-25

Folded-Cascode Op-Amp with Class AB Output Buffer

전자정보대학 김영석 24-26

½ ShrinkL=100nm => 50nm

fun=100MHz => 400MHz

AOL=70dB => 50dB (CLM)

td 10 > 5 @CL 1 Ftd=10ns => 5ns @CL=1pF

전자정보대학 김영석 24-27

Wide-Swing Op-Amp

전자정보대학 김영석 24-28

24.4 Gain EnhancementShrink => Faster Speed, But Gain Reduction

=> Use Gain-Enhancement(GE) Technique

GEOLDCGEOLDC

rrggRRgAAAA

⋅+⋅⋅=

⋅=

)]||()[()]||([,

cGEocaspocasndB

oponmpmnocaspocasnmnOLDC

CARRf

rrggRRgA

⋅⋅=

⋅+⋅⋅=

)||(21

)]||()[()]||([

3 π

전자정보대학 김영석 24-29

Gain EnhancementGE Amps

C ti GE A

OKworksGEffIf ampopdBGEun , ,3, −>

Compensating GE Amp

RingingOvershootNofFCcGE

/2400 =>=

RingingOvershootNo /

전자정보대학 김영석 24-30