1Structured Electronic DesignStructured Electronic Design

ET 8016

5 ECTS credits

2Structured Electronic DesignStructured Electronic Design

Design methodology Analysis and Synthesis

Applied network theory

Fundamental research Free-swinging intellect

Some keywords:

3An amplifierAn amplifier

10k

1R

1. What type of amplifier is this?2. When the gain is 10, what is the value of R1?

4Another amplifierAnother amplifier

10k

2R

3. When the gain is 100, what is the value of R2?

5And another amplifierAnd another amplifier

10k

3R

4. When the gain is 2, what is the value of R3?

6And anotherAnd another

10k

4R

5. When the gain is 20, what is the value of R4?

76. How do you get the crab out?

8

9Fundamental researchFundamental research

Why do it this way? (philosophy)

How to design an amplifier? (techniques)

10

11

br

r c

c

or

7. What is this?

12

8. Will the lamp light up?

13

9. Will the lamp light up?

14

YES NO

15

Tetrahedron Hexahedron Octahedron

Dodecahedron Icosahedron

Platonic solidsPlatonic solids

Edges and angles are all congruent

Edges and angles are all congruent

16

17

10.Is this a good model?

It correctly predicts the orbit of all planets you can see with the naked eye.

18

ModelingModeling

A correct model gives a correct prediction

19

Tetrahedron Hexahedron Octahedron

Dodecahedron Icosahedron

Platonic solidsPlatonic solids

The five Pythagorean regular polyhedra dictate the structure of the universeand reflect God's plan through geometry; June19, 1595, Johannes KeplerThe five Pythagorean regular polyhedra dictate the structure of the universeand reflect God's plan through geometry; June19, 1595, Johannes Kepler

20

ModelingModeling

A correct model gives a correct prediction

Never confuse models with the truth

21

Back to the amplifierBack to the amplifier

6. What is the current through the load?

10k

10k

1V 1k

Kirchoffs first law is not obeyed!Kirchoffs first law is not obeyed!

22

Ludwig Wittgenstein(1889-1951)

Language is a labyrinth of pathsLanguage is a labyrinth of pathsDesign problems arise from bad formulationsDesign problems arise from bad formulations

23

Amplifier designAmplifier design

24

An amplifier with a nullorAn amplifier with a nullor

2R

1R

10k

4R

25

vin=0 vout

iout

NullorNullor

Input current and input voltage of the nullor are made zero by the output signals of the nullorInput current and input voltage of the nullor are made zero by the output signals of the nullor

iin=0

26

Inside the NullorInside the Nullor

NorratorNullator

NullatorNorrator

0?0?

iv==

0!0!

iv==

0v =

0i =

Input current and input voltage of the nullor are made zero by the output signals of the nullorInput current and input voltage of the nullor are made zero by the output signals of the nullor

27

What is the transfer T of this amplifier?What is the transfer T of this amplifier?

2R

1R

28

2R

1R

The transfer T of this amplifierThe transfer T of this amplifier

2

1

RTR

= 2R

1R

outv

29

Another transfer T of this amplifierAnother transfer T of this amplifier

2anotherT R= 2R

1R

outvini

30

Joseph HenryJoseph Henry

The seeds of great discovery are constantly floating around us, but they only take root in minds well prepared to receive them

31

2R

1R

2

1

RTR

= 2anotherT R=

in

in

vi

? out

out

vi

The chain matrix of this amplifierThe chain matrix of this amplifier

1

2

2

0

1 0

in out

in out

v vi

RR

Ri

= ??again anotherT =

32

The two-port and its chain matrixThe two-port and its chain matrix

+ +

_ _vin vout

iin iout

=

out

out

in

in

iv

DCBA

iv

33

The two-port and its chain matrixThe two-port and its chain matrix

0out

in

out i

vAv =

=0out

in

out v

vBi =

=0out

in

out i

iCv =

=0out

in

out v

iDi =

=

=

out

out

in

in

iv

DCBA

iv+ +

_ _vin vout

iin iout

34

+

+_

_

vin=0 vout

iin=0 iout

NullorNullor

Input current and input voltage of the nullor are made zero via the output signals of the nullor

=

out

out

in

in

iv

DCBA

iv 0 0

0 0in out

in out

v vi i

=

35

Information from source to load

Signal power is enlarged

Information stays unaltered

A,B,C,D constantA,B,C,D accurately knownA,B,C,D constantA,B,C,D accurately known

A BC D

Accurate amplificationAccurate amplification

36

A voltage-to-voltage amplifierA voltage-to-voltage amplifier

1

2

2

0

1 0

RR

R

2

1v

RAR

=

2R

1R

outvinv

37

Amplification factor becomes inaccurate.

2R

1R

outvinv

sR

2

1v

s

RAR R

+=

38

2R

1R

outvinv

sR

nv

( )2

1

1 24 //n

vs

v s

RAR R

S kT R R R

= += +

OptimizationOptimization

/ Information is disturbed/ Amplification factor still inaccurate.

39

2R

1R

outvinv sR

Change topologyChange topology

1 2

1

1 24 ( // )n

v

v s

R RAR

S kT R R R

+== +

OrthogonalizationOrthogonalization

40

2R

1R

outvinv sR

1 2

1

1 2

1 2

4 ( // )n

v

v s

outfb

R RAR

S kT R R R

viR R

+== += +

41

Orthogonalization

Optimization

Orthogonalization

Optimization

2R

1R

outvinv sR

2R

1R

outvinv

sR

42

outv

inv sR

n1

4

0n

v

v s

fb

A nS kTR

i

===

1 2

1

1 2

1 2

4 ( // )n

v

v s

outfb

R RAR

S kT R R R

viR R

+== += +

43

Right choice

Orthogonalization

Optimization

Right choice

Orthogonalization

Optimization

44

The right choice: Start? (as usual?) Analyze? Optimize? Orthogonality? Specifications? Criteria?

The right choice: Start? (as usual?) Analyze? Optimize? Orthogonality? Specifications? Criteria?

45

NNSBC += log2

The criteriaThe criteria

1. Noise

2. Signal power (distortion)

3. Bandwidth

1. Noise

2. Signal power (distortion)

3. BandwidthAnd what about supply voltage, current, power consumption, technology etc.?

46

Optimization

Orthogonalization

Correlated propertiesCorrelated properties

1. Noise

2. Signal power (distortion)

3. Bandwidth

1. Noise

2. Signal power (distortion)

3. Bandwidth

47

Properties are generally not orthogonal

But design as if

Make it right

Orthogonal properties (2)Orthogonal properties (2)

48

When optimizing non-linear behavior (distortion): No dynamic effects No noise

When optimizing bandwidth: Small-signal models (no non-linearity) No noise

When optimizing noise behavior: Small-signal models (no non-linearity) No bandwidth details

As ifAs if

49

bandwidth noise distortionbandwidth distortion - noisenoise - bandwidth distortionnoise distortion - bandwidthdistortion - noise bandwidthdistortion bandwidth - noise

The most orthogonal sequence?The most orthogonal sequence?

50

Is it a cow? Is it green? Is it an animal?

Know which problem to solve firstKnow which problem to solve first

51

See the structure: Create Hierarchy!See the structure: Create Hierarchy!

Detail

Limiter

Limiter

Vlaag

Vhoog

Memory Vc

Global

52

The right design step at the right time

A design step occurs only once

Orthogonality, hierarchy, classification

Efficient designEfficient design

53

Reduce the search space as fast as possible.

Look for fast criteria: Necessary but not sufficient (e.g. LP -product)

Extensive calculations only when it makes sense Simplest model that suffices

Efficient design (2)Efficient design (2)

54

AttitudeAttitude

Know exactly what you want

Use second best, if you cant have it

Know the penalty

55

Voltage amplifier, whats in the nullor?Voltage amplifier, whats in the nullor?

Nullor

Source

Load

56

Inside the NullorInside the Nullor

NorratorNullator

57

Nullator? Norrator?Nullator? Norrator?

NorratorNullator

0v =

0i =Current sensor

Voltage sensor

Current source

Voltage source

58

practical nullor implementationspractical nullor implementations

i

vv

i

A B

C D

59

ChooseChoose

Nullor

Source

Load

i

vv

i

A B

C Di

vv

i

60

The experts with practical experienceThe experts with practical experience

v

61

Loopgain with B-type blockLoopgain with B-type block

Source

v

Load1R 2

RLR LC

1

1 2

1

1

1B

L

L L

L

L L

Rj R CL R R R R

j R C

+ = + + +

62

Loopgain with A-type blockLoopgain with A-type block

1R 2R

v

Source

Load

LR LC

1

1 2

1A

RLR R

= + 1

1 2

1

1

1B

L

L L

L

L L

Rj R CL R R R R

j R C

+ = + + +

63

Which transistor stage?Which transistor stage?

1R 2R

v

Source

Load

LR LC

+

+

+ ++ +

-

-

- - - -

+

+

+ ++ +

-

-

- - - -

CE-stage CB-stage CC-stage

Differential CE-stage Differential CB-stage Differential CC-stage

64

Voltage amplifier with A-type blockVoltage amplifier with A-type block

1R 2RSource

Load

LR LC

1R 2R

v

Source

Load

LR LC 1

1 2

1A

RLR R

= + L

65

Back to the B-type solution (told you so.)Back to the B-type solution (told you so.)

1R 2RSource

Load

LR LC

1

1 2

1 12

1

1

L

L L

L

E L

L

C

Rj RL CR R R

j RB

C

R

+ + + + = >

66

The expert solutionThe expert solution

1R 2RSource

Load

LR LC

2 or

Miller effect Distortion Miller effect Distortion

1

1 2

1 1lim2

1B

1Do

L

L Lr LCE

L

C

L

C

Rj R CL R R R R

j R C

+ = + + + BA CECE

or=

( )

( )

1 2

1 2

1 2

1

1

1 2

2

1

1

12

1A A

D

1

L

L L

L

L L

L

L Lo CC

L

L

CE

L

CC

R R Rj R

RL

CR R Rj R C

R R Rj R Cr R R R

R

j R C

R

+ + = +

+ + + + + + + + +

ro should be low???

ro should be low???

67

Expert versus logicExpert versus logic

1R 2RSource

Load

LR LC

2 or

1R 2RSource

Load

LR LC

1

1 2

1

1

1 12B D

L

CE CC

L L

L

L L

Rj R C

R R Rj

R

R

L

C

+ + + + =

no miller effect less distortion/ Load dependent loopgain

no miller effect less distortion/ Load dependent loopgain/ miller effect first stage

/ distortion and clipping first stage Loopgain not(less) load dependent

/ miller effect first stage/ distortion and clipping first stage Loopgain not(less) load dependent

( )

( )

1 2

1 2

1 2

1

1

1 2

2

1

1

12

1A A

D

1

L

L L

L

L L

L

L Lo CC

L

L

CE

L

CC

R R Rj R

RL

CR R Rj R C

R R Rj R Cr R R R

R

j R C

R

+ + = +

+ + + + + + + + +

68

Which solution when?Which solution when?

1R 2RSource

Load

LR LC

2 or

1R 2RSource

Load

LR LC

no miller effect less distortion. Load dependent loopgain

no miller effect less distortion. Load dependent loopgain/miller effect first stage

/ distortion and clipping first stage Loopgain not(less) load dependent

/miller effect first stage / distortion and clipping first stage Loopgain not(less) load dependent

69

AlternativesAlternatives

70

ConclusionsConclusions

Know ideal case

Know the penalty for choosing second best

Beware of experts !

Always remember the ideal case!

The seeds of great discovery are constantly floating around us, but they only take root in minds well prepared to receive them