EE40458

 
Nonlinearity  &  Noise  
Yet  another  lecture  from  the  road  
 Recap  &  Perspec=ve  
•  An  overview  –  where  have  we  been  recently:  
–  General  linear  system  theory  
•  S-­‐parameters,  but  also  equivalent  small-­‐signal  models  like  the  hybrid-­‐π  
model,  etc.  
•  Give  linear  response  of  circuit/system;  can  use  superposi=on  &  Fourier  
analysis  to  determine  output  for  arbitrary  input  signal  
•  “Gain,  phase”  at  each  frequency;  no  new  frequencies,  no  signal  
components  that  were  not  present  in  the  input  signal  
•  Strictly  applies  only  for  systems  governed  by  linear  differen=al  equa=ons  
(any  order,  but  constant  coefficients)  
•  Approximately  applies  to  most  systems  if  signals  are  small  (equivalent  to  
approxima=ng  func=on  with  first  two  terms  (constant  plus  linear)  in  
Taylor  series)  
 Recap  &  Perspec=ve  (cont.)  
•  Overview  (con=nued):  
–  Nonlinear  effects  
•  Many  important  systems  are  not  well  approximated  with  a  linear  
descrip=on  
•  Examples:  almost  any  component  under  high  power  condi=ons  (e.g.  
hea=ng);  amplifiers  driven  with  large  inputs  (or  designed  for  switching-­‐
mode  opera=on  for  high  efficiency);  devices  for  harmonic  genera=on,  
mixing,  or  detec=on  (nonlinear  response  is  desired)  
•  Our  approach:  Taylor  series  expansion  of  transfer  characteris=c,  resul=ng  
in  polynomial  representa=on  of  response  
–  We  simplified  to  neglect  memory,  history-­‐dependent  effects  (e.g.  hea=ng);  
assumed  output  depends  only  on  instantaneous  input  value.    More  advanced  
approaches  exist  to  handle  this  
•  Conclusions:  saw  harmonic  genera=on  (plus  DC  shi^),  intermodula=on  
products  (e.g.  sum  &  difference  frequencies)  
•  Figures  of  merit:  P1dB  (gain  compression),  PIP3  (third  order  
intermodula=on)  
Nonlinear  Figures  of  Merit  -­‐  Review  
•  Gain  compression  –  P1dB:  
–  Single  input  tone  
–  Map  output  power  (at  input  frequency)  vs.  power  of  input  signal  
–  Linear  theory:  output  power  propor=onal  to  input  power;  nonlinear  
effects  tend  to  cause  satura=on:   # 3 k 2&
vo! (t ) = k1 A %1− 3
A ( cos (ω t )
$ 4 k1 '
–  Note:  input  is  power  in  single  tone;  output  power  “counted”  is  only  
the  power  at  this  frequency  (i.e.,  the  harmonic  power  is  not  included)  
Nonlinear  Figures  of  Merit  -­‐  Review  
•  Gain  compression  –  P1dB:   # 3 k 2&
vo! (t ) = k1 A %1− 3
A ( cos (ω t )
–  Defini=on:  Input  1  dB  compression  point,     $ 4 k1 '
P1dB,  is  input  power  at  which  output  is  1  dB  below  the  linear  case.  

–  Note:  some  data  sheets  will  report  the  output  1  dB  compression  point  
(e.g.  the  y-­‐axis,  rather  than  x-­‐axis).    Depends  on  intended  applica=on  
Nonlinear  Figures  of  Merit  -­‐  Review  
•  Intermodula=on  –  PIP3:  
–  Saw  complicated  rela=onship  with  two  input  tones  
–  Figure  of  merit:  for  comparison  of  components–  PIP3  
–  Two  input  tones,  equal  amplitude  (for  figure  of  merit);  small  enough  
that  gain  compression  can  be  neglected:  v   in (t ) = A !"cos (ω1t ) + cos (ω 2 t )#$
–  Output  power  at  close-­‐in  intermodula=on  product  frequencies  vs.  
power  of  input  signal.    Near  ω1  and  ω2,  have:  
3
vo! (t ) = k1 A "#cos (ω1t ) + cos (ω 2 t )$% + k3 A 3 "#cos ( 2ω 2 − ω1 ) t + cos ( 2ω1 − ω 2 ) t $%
4
Nonlinear  Figures  of  Merit  -­‐  Review  
•  Intermodula=on  –  PIP3  (con=nued):  
–  Purpose  of  PIP3  figure  of  merit:  quan=fy  rela=onship  between  Pin,  Pout  
(at  signal  frequency),  and  intermodula=on  products  
–  Map  power  in  “desired”  frequencies  (first  term)  to  power  in  
intermodula=on  products  (second  term)  
vin (t ) = A !"cos (ω1t ) + cos (ω 2 t )#$
3
vo! (t ) = k1 A "#cos (ω1t ) + cos (ω 2 t )$% + k3 A 3 "#cos ( 2ω 2 − ω1 ) t + cos ( 2ω1 − ω 2 ) t $%
4
1 2 Pin  =  power  at  each  input  tone  
Pin ∝ A
2
1 Pd  =  power  at  each  desired  output  
Pd ∝ k12 A 2 ∝ Pin tone  
2
1 2 9 6 2 9
Pim ∝ k3 A = k3 APin3
2 16 4 Pim  =  power  at  each  intermod  
output  tone  
Nonlinear  Figures  of  Merit  -­‐  Review  
•  Intermodula=on  –  graphically:  
–  Graph  Pd,  Pim  vs.  Pin,  usually  on  log-­‐log  scale  (all  powers  in  dBm)  
•  Reminder:  dBm  =  10*log10(P/1  mW)  
1
Pin ∝ A 2
2 Pd ( dBm ) = Gain ( dB) + Pin ( dBm )
1
Pd ∝ k12 A 2 ∝ Pin
2 Pim ( dBm ) = offset + 3Pin ( dBm )
1 9 9
Pim ∝ k32 A 6 = k32 APin3
2 16 4
 
–  PIP3:  intercept  between    
linear  (Pd)  and  intermod    
(Pim)  terms    
–  Intercept  is  “fic==ous”;  in    
prac=ce,  based  on  low    
power  data  (avoid  gain    
compression);  real  lab  data    
includes  everything…  
 Intermodula=on  Analysis  
•  Measurement:  
–  Measure  Pd,  Pim  at  several  (low)  levels  of  Pin  
•  Can  easily  separate  Pd,  Pim  (on  spectrum    
analyzer)  because  at  different  frequencies  
•  Use  slope  of  1  for  Pd  vs.  Pin  (in  dBm)  
•  Use  slope  of  3  for  Pim  vs.  Pin  (in  dBm)  
•  Find  intercept  point;  input  IP3  (PIP3,  IIP3)  or  output  IP3  (OIP3)  can  
be  projected  
•  Analysis:   2
Pim ! Pin $
–  “Intermodula=on  Ra=o”:     IMR = =# &
Pd " PIP3 %
–  Convenient  rela=on:  
•  Relate  expected  intermodula=on  products  from  (known)  IIP3  and  Pin  
•  Find  IIP3  given  measured  Pim,  Pd  
 Noise  
•  Linear  and  nonlinear  analysis  relates  output  to  input  s=mulus  
–  Linear:  small  signals;  non-­‐linear:  large  signals  
•  Circuits,  systems  also  produce  outputs  independent  of  input:  
noise  
–  Ul=mately,  noise  limits  our  ability  to  resolve/recover/process  very  
small  signals  
•  Noise  is  fundamental  –  cannot  be  eliminated;  but  can  be  
managed  
•  Sources  of  noise:  
–  Thermal  noise:  random  mo=on  of  carriers  (electrons,  holes)  in  
resis=ve  material  
–  Shot  noise:  cause  by  random  =ming  of  events    
•  Current  is  made  of  up  of  flow  of  electrons,  but  they  have  some  “jiler”  in  when  
they  arrive;  this  generates  shot  noise  
–  Flicker  or  1/f  noise:  trapping/detrapping,  o^en  defect  or  surface  
related;  has  ~1/f  noise  power  spectral  density  
 Thermal  Noise  
•  Let’s  look  at  thermal  noise  in  a  resistor:  

en = 0
en2 = 4kTBR

–  Reminder:  k=1.38x10-­‐23  J/K  (Boltzmann’s  const.),  T  (in  Kelvin)  

–  Expect  average  voltage  =  0  (no  net  flow,  equal  opposite  flows)  
–  But  variance  ≠  0;  voltage  variance  propor=onal  to  noise  power  
–  B:  bandwidth  of  measurement—how  much  noise  power  you  see  depends  
on  how  wide  of  a  frequency  range  you  look  at  
•  Poten=al  problem:  noise  power  goes  up  as  bandwidth  increases—
no  limit?      
–  Not  really…  this  conclusion  comes  from  simplis=c  assump=on  on  carrier  
sta=s=cs.    But  if  T~300  K,  is  good  approx.  for  frequencies  to  ~1  THz  or  so  
 Modeling  Noise  
•  For  analysis,  need  equivalent  circuit  for  analysis  
•  One  op=on:  model  noisy  resistor  as  noiseless  resistor  with  
associated  noise  source  (Thevenin  or  Norton  op=ons):  
en = 0, in = 0
 
  en2 = 4kTBR, in2 = 4kTB R
•  Can  we  get  power  from  these  sources?    Yes,  but…    Consider:  

en2
–  Power  available:       kTB
=    Independent  of  R  
4R
–  Power  transfer?    If  both  resistors  at  same  temperature,  net  flow  =  0  
(equal/opposite  flows).    If  at  different  temperatures,  power  from  hot  
to  cold  (alempts  to  equilibrate  the  system).    You  knew  that.  
 Modeling  Noise  (cont.)  
•  How  about  complex  impedances?  
–  One  can  show  that:  
Z ( f ) = R ( f ) + jX ( f )
en2 = 4kT ∫ R ( f ) dF
 
B
–  No  noise  from  reactances  (no  loss  or  dissipa=on,  no  noise)  
•  Numerical  example:    
–  Noise  voltage  across  at  1  MΩ  resistor  in  bandwidth  of  100  MHz  (e.g.  
typical  oscilloscope  input)  
4kT  =  1.6x10-­‐20  J  (T=290  K)  
en2 = 4kTBR = 1.6 ×10 −20 ⋅10 6 ⋅108 = 1.6 ×10 −6  V 2

rms voltage :  en2 = 1.26 ×10 −3  V

•  Can  see  why  oscilloscopes  have  minimum  2  mV/div  scales…anything  
smaller  is  just  noise  
 Excess  Noise  
•  Many  components  exhibit  addi=onal  noise,  beyond  the  
thermal  contribu=ons  
–  O^en  modeled  as  if  it  were  thermal  noise,  but  with  modified  
parameters  (i.e.,  fudge  factors)  
•  Two  common  approaches:  
–  Circuit  analysis  approach  (typical):  
2
•  Introduce  Rn  as  fudge  factor,   en = 4kTBRn
•  Rnoiseless  is  actual  R  value  for  circuit  
•  For  pure  thermal  noise  (no  excess  noise),  Rnoiseless=Rn  
–  System  analysis  approach:    
2
•  Use  T  as  fudge  factor:     en = 4kTn BR
•  Tn  no  longer  “thermometer”  temperature;  
 if  excess  noise  present,  Tn>T  
•  Concept  of  “noise  temperature”  is  common,    
and  what  we’ll  (mostly)  use  
 Noise  Temperature  Example  
•  An  example:  antenna  
–  At  resonant  frequency,  antenna  impedance  (Zant)  is  resis=ve  
–  Zant  =  Rohmic  +  Rrad    
•  Rohmic:  from  loss  in  the  conductors,  a  “real”  resistance  
•  Rrad:  accounts  for  conversion  from  input  power  to  radiated  power  (think  
of  antenna  as  broadcas=ng)  
–  Temperatures?  
•  Rohmic:  at  physical  temperature  of  the  antenna—electrons  bouncing  
around  in  conductors  due  to  random  thermal  mo=on  
•  Rrad:  at  an  “effec=ve”  temperature,  TA  
•  TA  is  “fudge  factor”  to  allow  us  to  make  
output  noise  of  antenna  match  the    
power  actually  received  
•  Fun  fact:  “Cosmic  background  radia=on”  
–  1978  Nobel  prize  –  measured  TA  ~  3K,    
when  expected  to  be  0  (dark  sky)  
 Noise  in  Two-­‐Port  Networks  
•  So  far,  everything  has  just  been  about  how  much  noise  a  one-­‐
port  circuit  makes.    But  two-­‐port  networks  are  more  useful—
have  inputs,  outputs  
•  Basic  idea:  two-­‐port  network  does  some  func=on  (amplify,  
mix,  etc),  but  also  adds  some  noise  
–  Schema=cally:  

–  TE:  effec=ve  noise  temperature  of  two-­‐port  network  

•  Is  a  func=on  of  Zs,  frequency;  characterizes  the  network,  not  Zs  
 Noise  in  Two-­‐Port  Networks,  cont.  
•  To  include  effects  of  both  two-­‐port  and  termina=on,  use  
“opera=ng  temperature”,  Top  
–  Top  =  TE+Ts  
–  Adding  temperatures:  same  as  adding  powers.    Assumes  no  
correla=on  between  noise  sources  
 
 
 
 

Characteriza=on?    Two-­‐port  network  has  s-­‐parameters,  plus  TE  

 Noise  Factor  
•  Another  common  way  to  characterize  the  noise  added  by  a  
two-­‐port  network  is  the  noise  factor  and  noise  figure  
•  Two  equivalent  defini=ons:  
Input SNR
–  Defini=on  #1:   Noise Factor ≡
  Output SNR Ts =To =290 K
 
 
 
 
 
 
Result:  
Sin Sin
N in N in N out
  F = = =
Sout GA Sin GA N in
  N out N out
 Noise  Factor  &  Noise  Figure  
•  Alterna=ve  view:  
Actual available noise  power (output)
–  Defini=on  #2:   F ≡
Available noise  power if  two − port was noiseless
 
Result:   N out N A + GA N in
 
F= =
GA N in GA N in
 
•  Can  be  framed  in  terms  of  temperatures:  
NA N A / GA kTE B TE
F = 1+ = 1+ = 1+ = 1+
GA N in N in kTO B To
–  So  providing  F  is  equivalent  to  providing  T  
•  Technically,  F=noise  factor  
•  Noise  Figure  is  more  common;  F  converted  into  dB  
–  NF  =  10*log10(F)  

 
 Noise  Figure  
•  Cau=on:  
Input SNR
–  From  defini=on  #1:   Noise Factor ≡
Output SNR Ts =To =290 K
–  Looks  like  noise  figure  should  be  how  much  (in  dB)  the  SNR  degrades  
because  of  the  noise  of  the  two  port  
–  This  is  not  strictly  true:  note  that  this  is  true  only  if  Ts=To    
–  As  we  saw,  Ts  can  be  an  effec=ve  temperature  with  no  obvious  
connec=on  to  “thermometer”  temperatures  (e.g.  if  signal  came  from  
an  antenna,  etc)  
•  So  noise  figure  should  be  thought  of  as  a  “test-­‐based  metric”  
–  In  the  lab,  can  test  the  SNR  with  Ts=To,  and  find  NF  
–  In  real  systems,  Ts  is  almost  never  To,  so  the  actual  SNR  change  can  be  
quite  different  
 
 Noise  Figure  and  LNA  Design  
•  How  is  this  related  to  our  LNA  design  approach?  
•  Recall:     4R Γ S − Γ opt
2
n
F = Fmin + 2
Z o 1+ Γ opt (1− Γ ) S
2

•  This  shows  explicitly  how  F  depends  on  Zs  (Zs  <-­‐>  Γs)  
–  This  is  important  if  you’re  doing  detailed  circuit  design  (e.g.  of  an  
amplifier  to  meet  a  specific  noise  figure  target)    
•  But  for  system  design  or  analysis,  the  Zs  is  usually  already  
defined  and  fixed  
–  Block  diagram-­‐level  interconnec=ons;  not  re-­‐designing  the  individual  
components  
–  Example:  making  a  system  by  interconnec=ng  available  “50  Ω”  
components  
•  In  this  case,  TE  is  sufficient  (if  TE  is  for  the  Zs  in  ques=on)  
 Noise  Figure  and  Loss  
•  A  special-­‐case  two-­‐port  is  the  matched  alenuator  
•  Passive  device—just  a  resistor  network.    Typically  designed  to  
have  input  &  output  impedance  matched  to  Zo,  with  a  
specified  alenua=on  (e.g.,  3  dB,  6  dB,  etc).  

•  What  is  it’s  noise  figure?  

–  Since  network  is  just  passive,  at  output  appears  as  Zo  termina=on  at  
physical  temperature;    N out = kTB
–  But  regular  two-­‐port  equa=ons  also  apply:    N out = kTB ⋅ GA + GA ⋅ N added
–  Loss=1/Gain,  so  :   TE = Tatt ( L −1)
–  If  Tal=To,     T T
F = 1+ E = 1+ att ( L −1)
TO TO
then  F=L  
 System  Noise  Analysis  
•  Common  situa=on:  want  to  evaluate  noise  performance  of  a  
system  consis=ng  of  several  building  blocks  cascaded  
together  

•  Overall  noise?  
TE 2
–  Can  show:   TE = TE1 +
GA1
TE F −1
F = 1+ = F1 + 2
TO GA1

–  Careful:  G’s  are  available  power  gains,  TE’s  must  be  for  actual  
impedances  presented  
 System  Noise  Example  
•  Consider  two    
amplifiers  in    
cascade:  

•  And  if  we  reverse  the  order  of  the  amplifiers?  

–  Gain  is  the  same,  but  what  about  noise?  

–  Note:  usually  want  the  lowest  TE  amplifier  in  front.    But  not  always—
the  gains  also  play  a  role  
 Receiver  Sensi=vity  
•  For  radio  receivers,  sensi=vity  is  limited  by  noise  floor  
•  Define:  minimum  detectable  signal  (MDS)  for  given  SNR  
–  O^en  choose  0  dB  as  the  threshold  (though  other  choices  are  possible  
depending  on  the  system)    
•  Example:  

 
 

•  Results:  
–  NF  =  8  dB  à  F=6.31  à  TE  =  1540  K  
–  Top  =  290  K  +  1540  K  =  1830  K  
S S
–  Nout  =  k  Top  B  GA;  Sout  =  Sin  GA;   SNRout = out = in
N out kTop B
–  Se•ng  SNRout=1  à  Sin  =  5.3x10-­‐17  W  =  -­‐132.8  dBm  
–  Effec=ve  noise  floor  of  the  receiver;  for  reference,  thermal  noise  at  290  K  
=  -­‐174  dBm/Hz  
 Noise  Figure  Measurement  
•  Basic  idea:  measure  output  noise  power  for  two  different  
source  temperatures  
–  From  this,  can  separate  contribu=on  from  source  and  from  two-­‐port  

•  “Y  factor”  measurement:  output  powers  

–  NH=k(TE+TH)BGA  
–  NC=k(TE+TC)BGA  
N T +T T −YTC
Y = H = E H ; TE = H
N C TE + TC Y −1
 Noise  Figure  Measurement  
•  For  “hot”  and  “cold”  input  termina=ons,  could  use  resistors  at  
different  temperatures  
–  For  best  measurements,  want  largest  possible  difference  between  hot  
and  cold  temperatures  (cryogenic  resistor,    hot  resistor)  
•  Inconvenient  in  prac=ce,  o^en  use  “noise  diode”  instead  
–  Diode  off  (no  bias):  room  temperature  resistor  (thermal  noise)  
–  Diode  “on”  (biased  in  reverse  breakdown):  avalanche  breakdown  
process  is  very  noisy,  acts  like  “hot”  resistor    
•  Characterized  by  the  “excess  noise  ra=o”  
" T − 290 %
ENR = 10 log10 $ H '
# 290 &
•  Typical  ENR  ~15  dB;  TH  ~  9461  K!  (Solar  surface  ~6000  K).    Much  bigger  
temperature  difference  than  possible  using  “thermometer”  temperatures  
–  A  caveat:  we  know  that  amplifier  noise  figure  depends  on  Γs:  so  if  
diode  has  different  impedance  in  “on”  and  “off”  states,  
measurement  can  be  off—o^en  use  “low  ENR”  diode  to  avoid  this.    
Just  a  regular  “high  ENR”  diode,  followed  by  an  alenuator  
 Dynamic  Range  
•  For  systems,  very  important  considera=on  
•  Dynamic  range:  range  of  input  signal  amplitudes  
for  which  the  system  has  “acceptable”  
performance  
–  System-­‐level  considera=ons  dictate  what  counts  as  
“acceptable”  –  some  systems  can  tolerate  more  
distor=on  than  others,  etc.  
–  Limited  for  large  input  signals  by  nonlineari=es  
–  Limited  for  small  input  signals  by  noise  
–  So  we’ll  be  combining  the  last  few  topics  together  
 Dynamic  Range  Defini=on  
•  Basic  no=onal  system  picture:  

•  With  this  framework  in  mind,  can  define  dynamic  range:  

max. usable input power Pmax

DR = =
min. usable input power Pmin
•  Usually  expressed  in  dB—since  these  are  powers,  10  log10(DR)  
 Spur  Free  Dynamic  Range    
•  One  common  (but  not  universal!)  choice  for  “acceptable”  
performance:  spur  free  dynamic  range  (SFDR)  
•  SFDR  =  DRf  (same  thing,  different  terminology)  
–  For  SFDR:  Pmin=minimum  detectable  signal  (MDS)  
•  But  what  is  the  MDS?    Smallest  Pin  that  will  provide  a  specified  SNR  
at  the  output.    O^en  this  “reference”  SNR=1  
–  For  SFDR:  Pmax  =  input  signal  level  at  which  3rd-­‐order  in-­‐band  products  
are  equal  to  Pmin  (=MDS)  
•  Easier  to  understand  (I  think)  as  a  picture  
 Spur  Free  Dynamic  Range    
•  SFDR  in  pictures:  
(think  of  2-­‐tone    
intermod  
measurement)  
 

–  Pmin=minimum  detectable  signal  (MDS)  

–  Pmax  =  input  signal  level  at  which  3rd-­‐order  in-­‐band  products  are  equal  
to  MDS  
–  Basic  idea:  intermodula=on  is  never  larger  than  the  noise—no  “spurs”  
•  In  microwave-­‐speak,  “spur”  is  short  for  “spurious  signal”.    Not  poking  horses.  
 Spur  Free  Dynamic  Range    
2
P ! Pin $
•  Analy=cally:  using   IMR = im
=# &
  Pd " PIP3 %
from  our  previous  analysis,  can  work  out  SFDR,  etc.      
•  Basic  approach:  consider  IMR  when  Pin  =  Pmax  

2
Pim ! Pmax $
IMR = =# &
Pd " PIP3 %
Pim = Pmin ⋅ G;  Pd = Pmax ⋅ G
2
PminG Pmin ! Pmax $
IMR = = =# &
PmaxG Pmax " PIP3 %
 Spur  Free  Dynamic  Range    
2
! Pmax $
•  Since   IMR = PminG = Pmin =# &
PmaxG Pmax " PIP3 %
 
it  follows  (just  re-­‐arranging)  that:  
3
Pmax 1/3 2/3
Pmin = 2 ;  Pmax = Pmin ⋅ PIP3
  PIP3
 
•  Remember:  
–  Pmin  comes  from  noise  analysis,  so  is  known  
–  SFDR  =  Pmax/Pmin  
•  Final  result:  
2/3
! PIP3 $
SFDR = # &
" Pmin %
 Spur  Free  Dynamic  Range    
•  Careful:  previous  page  was  all  in  MKS  (or  similar)  units  
•  Usually  specify  these  things  in  dB:  
2/3
! PIP3 $
SFDR = # &
" Pmin %
Becomes:  
  2"
SFDR(dB) = # PIP3 ( dB) − Pmin ( dB)$%
  3
•  Not  complicated—just  be  careful  
•  One  final  note:  the  book  gets  the  same  results,  but  from  
another  path;  Pozar  does  the  analysis  from  the  point  of  view  
of  the  output  power  (vs.  input  like  done  here).      
 Congratula=ons  
•  You  made  it  to  the  end.  

•  We’ll  have  a  review  session  at  our  regular  

class  period  on  Thursday  –  come  with  your  
ques=ons  about  the  final  exam  
 Topics  Covered  This  Year    
•  RF  models  –  lumped  element  RLC   •  Fun  with  Smith  charts  
models  of  common  components  (l<   •  Lumped-­‐element  matching  network  
λ/10)   design  
–  Behavior,  resonanaces  (series,  parallel),  
some  uses  of  these  parasi=c   •  Distributed  circuits  
phenomena   –  λ/4  transformers  
•  Electromagne=c  analysis  of   –  Series-­‐line  matching  networks,  single-­‐
stub  matching  networks,  double-­‐stub  
transmission  line  structures  
–  Limita=ons,  design  procedures,  detailed  
–  Maxwell’s  eq.,  boundary  condi=ons,   understanding  
assump=ons  made  for  solu=on  
–  Bandwidth  effects  in  matching  
•  Transmission-­‐line  models   networks  
–  Deriva=on  from  lumped-­‐element   •  Network  analysis  
sec=ons,  parameters  (α, β, γ),  
significance  of  each   –  S,  Y,  Z  parameters  –  matrix  
representa=ons,  defini=ons,  finding  
–  Meaning  and  role  of  Zo,  vp,  λ,  etc.   matrix  elements  from  circuits  
–  Source,  load  mismatch  effects;   –  Circuit  analysis  using  matrix  
reflec=on  coefficients,  impedance   representa=ons  
transforms,  VSWR,  transla=on  along  
lines,  boundary  condi=ons   –  Flow  graphs,  Mason’s  rule  
–  Power  transmission   –  Generalized  s-­‐parameters    
–  Stub  impedance/admilance;  origin,  
uses  
 Topics  Covered  This  Year  (2)    
•  Amplifier  design  cases  as   •  Nonlinear  effects  
considera=ons   –  Deriva=ons,  founda=ons  
–  Simultaneous  conjugate  matching   –  Gain  compression  
condi=ons  –  maximum  gain   –  Intermodula=on  
–  Design  for  specified  gain;  gain  circles,   –  Defini=ons:  PIP3,  P1dB,  IMR,  blocking,  
trade-­‐offs  (e.g.  for  bandwidth)   desensi=za=on,  cross  modula=on  
–  Design  for  noise  figure;  noise  figure   –  Input/output  spectra,  frequencies  
circles,  interac=on  with  gain  circles   present,  etc.  
–  Detailed  understanding  and  ability  to  
design  circuits   •  Noise  
–  Thermal  noise  in  resistors/passives  
•  Power  gain  defini=ons  &  use   –  Available  noise  
–  Opera=ng  power  gain  (GP),  available  
power  gain  (GA),  transducer  power  gain   –  Effec=ve  noise  temperature,  noise  
(GT)   resistance  
–  Defini=ons,  significance  of  each   –  Noise  in  2-­‐ports:  TE,  F,  NF  
–  What  is  each  good  for?   –  Noise  measurement  (Y  factor)  
–  Cascaded  noise  figure  
•  Stability   –  Receiver  sensi=vity  (MDS)  
–  Meaning/significance  of  stability  
–  Dynamic  range  (SFDR)  
–  Source,  load  stability  circles  
–  Interpreta=on  of  the  circles  
–  k-­‐Δ,  µ  tests  and  what  they  mean  
–  Uncondi=onal  stability:  defini=on,  
concepts