Lecture 57 - Power Amplifier Design - III

IIT Roorkee July 201
61 followers
Mar 4, 2024
Concepts Covered: Cascode Output Stage, Large Signal Impedance Matching, Load Pull Measurement, Basic Linearization Techniques.
[Music]
[Music]
uh welcome to RAB transer design course
and we are discussing about the power
amplifier design we have defined a
different class A B C D E F in the last
lecture now we will see
that uh how uh we will see in this class
the about the linearity and how uh we
can do the impedance
matching and one uh widely used uh
topology which is a cascode output
stages so cascode is also uh used in in
the output stage and what are the
advantages and disadvantage using the
cast code
stages so here when we use the C code
stage uh one can choose the vdd vdd
which is uh equal to the half of the
maximum uh tolerable voltage of the
transistor means which is a half
voltage and uh what will happen that but
there is these are two penalties the
lower headro uh limits the linear
voltage range of the
circuit and the proportionally higher
output current for a given output power
leads to Greater loss in the output
matching Network reducing the efficiency
so we we need to have a two penalty that
we need to give it will reduce the
efficiency and it will have a lower
voltage headro limits the linear voltage
range of the circuits so if we see the
VX which is full swing and v y which is
a VB minus V VB minus V
threshold so the ccod device uh Shields
the input transistor as VX Rises uh and
uh keeping the drain Source voltage of
M1 less than a VB minus VT
H2 this that is a V3
two now if we try try to see the
linearity consideration for this uh the
uh for circuit a the we can get the
swing which is vdd when minus V of the
Cascade minus vds2 that is a drain to
Source voltage of two plus vth equal to
V +
VM and for circuit B it is a vdd minus
so for uh circuit a it is vdd minus VP
uh that is vdd minus VP of this casket
stage minus
vds2 plus vth equal to V plus VM and for
circuit B which is this uh V is a a
input voltage swing and uh VM is a uh uh
mid voltage at the
input for circuit B vdd minus v v p
cascet plus VT H1 equal to v+ VM so what
it follows that v p common source
amplifier equal
to uh VP of uh Cascade plus vds2 so this
vds2 is added and this is a v course so
for circuit a we can write it
down
vdd
minus
VP
CC minus
vds2 plus v t
h equal
to this is for this circuit and for
this
for this we can write it down
vdd
minus VP common
source plus VT
H1 equal
to so if we threshold one and uh this is
same we can get a VP common source which
which is equal to VP cascet plus vds2 if
the common source stage remain linear
Across The Wider output voltage range
than the cascode circuit does and at low
Supply voltage cascode output stages
offer only a slight voltage swing
advantage or their common source count
counterpart but at the cost of
efficiency and linearity so linearity
degradation is happening because of the
stack transistor and that will uh
affect the linearity as well as
efficiency only advantage of this
circuit is stability so the stability of
this uh circuit is uh better compared to
the uh what we have seen a simple uh
common source amplifier because we are
shielding the output with the input
however the uh efficiency and linearity
is a
poor now
what we have done up to all uh when we
have discussed the all um the model
means all the amplifier Class A B C D
all the amplifier what uh we uh have
considered that it's uh
transistor which is uh input impedance
is a means its impedance is a constant
and its capacitor and everything is not
changing as the voltage swing is
changing so uh in reality it is not true
because if we consider the full Voltage
swing the other uh the capacitor
associated with the M1 transistor like
cdb and the output resistance r o is
also varying with the uh voltage
swing now this simply uh Simplicity
model assume that out output matching
Network simply transform RL to a lower
value so this matching Network what we
have done it just transform the RL to
the lower value that we have simplified
but in practice the situation is more
complex a nonlinear complex output
impedance must be matched to the linear
Lo load because the uh there are
capacitor and the impedance is also
Associated so if we consider the large
signal impedance matching very simple
case if we consider the output
resistance
R1 which is this instead in this we are
not considering the capacitor uh but we
are only considering the R1 and uh the
there is a capacitor cdb one which which
needs to be considered while making
through the AR so if we compute the
power delivered by M12 RL which is PRL
and that consumed by the transistor
output resistance P R1 what we have a p
RL output power which is a IP s by 2 RL
uh and the there is a vdd which is R1
means the voltage division rule R1 by RL
+ R1 whole
squ this is a p RL and P R1 which is
given by IP s by 2 uh RL uh into R1
Square by RL + R1 squ so in this in the
both the case if you write a p RL in one
of the case there is RL which is
considered and in another case uh r o
means in one case RL Square will be
there and in another case R
square so the maximum power transfer RL
is chosen equal to R1 it yields to P RL
equal to P R1 so so P RL by P R1 equal
to so if we write P RL by P R1 which is
given by R1 by RL Square because here
there is RL into R1 square and here
there is a RL square into
R1 so P RL by P R1 is given by r o by RL
so what the relation above show that
reducing the RL minimizes the relative
power consumed by the uh
transistor he here there is a trade off
come between the maximum power transfer
and the efficiency that we want to
achieve so uh that uh we need to uh make
sure that there is a maximum power
transfer will also happen and also we
will get a maximum efficiency so in
order to do that we need to uh design a
matching Network such that it will be
properly tuned with its output and input
impedance so that for for that we need a
a load pull matching that is a load pull
measurement and uh this is widely uh
used method in the power amplifier
design in order to do the uh in order to
have the better matching for getting
maximum output power and the efficiency
for for Designing the power
amplifier so here how it is it has been
done we have already discussed that if
there is
a uh any two Port
Network there is a output matching
Network and input matching
Network
this is input matching Network and then
we have considered the
RL and uh there is
a input RS
and while designing the amplifier we
have
discussed now we also discuss that if
there is a change in the uh gamma l
means your gamma in is depend upon how
uh we have designed the output matching
Network and your gamma out is depend
upon how we have designed the input
matching Network so what what we need to
do we need to put a tuner so this is a
two Port Network so this is a input uh
means TW Port Network there is a tuner
which will change this input impedence
matching network uh reflection
coefficient and that is a tuner ad the
output which control so here it is shown
the uh input resistance means RS which
is uh RS is a constant but here it is
considered as a impedance but we can
also consider as a reflection
coefficient gamma in and Gamma out so as
we are changing this control
voltage if we change the control voltage
such that initially we got some power
which is a
P1 so what we do we vary the J 1 such
that the power delivered to the RL
remain constant and equal to P1 that is
this and th obtaining a control depicted
so we make a contour or we can also plot
the power on the SMI chart we already
seen the so how how we can do that so if
we have
a uh if we write a again
GP which is equal to P
out by P
in so we can write a p
out because p in already we know
GP into p
in and based on different gain Circle we
already have uh seen that GP uh we can
plot a different gain circles for GP
value and that that can be put on the
SMI
chart so similarly we can also
plot the power P out for the
different this is
a uh if it is the maximum power that
that will be shown as a DOT and for
other impedance means uh other gains we
we we have a different Contours and this
is called the power Contour that is P1
and P2 so the same thing has been shown
here that is this is a power P1 and P2
and when we have a Optimum impedance at
that time it will be have a DOT so for
that what we need is a load pool
measurement we need a repeated
measurement at the input side and output
side which increasing the power level
and eventually we want to arrive at the
optimum impedance set out opt to have a
maximum output power so this uh this
kind of measurement technique is widely
used when uh we have a transistor
network available and we want to
identify the matching Network which is
giving the maximum power
for the uh and uh Optimum output
impendence said op we then we can fix
the impedance matching Network once we
know this J
one uh that uh for the tuner that will
be used to design the output matching
Network as well as input matching
Network so this is how load pull will
work and uh there are various
measurement technique uh is uh provided
in the
literature now we will discuss the basic
linearization techniques uh that is uh
just uh I will give the brief idea you
can go in detail and understand more
about it uh if you are interested in uh
Power Amplifier
design so one of the technique that is a
feed forward technique now in feed
forward technique
what we do so the there is a two diagram
which is drawn here this is a main power
amplifier and uh whatever nonlinearity
which is added by this power
amplifier uh that will come at the point
M and this is the actual signal V in so
what we what we will try to find out
that how much non nonlinearity will be
added
uh we will get from uh by subtracting
from uh m point and we will put error
amplifier here which will substract the
nonlinear nonlinearity from the main
power amplifier and there is a
substraction needs to be
done so
here in this pit proper architecture
what it commutes the error and with
proper scaling substract it from the
output
waveform what will happen there is a
second figure which is drawn because
when you directly take this output and
if you uh get after the main power
amplifier M it will also observe the
delay in the signal and that delay we
need to consider by using this Delta 1
there is also delay means when we use
this error
amplifier also
when we use the error amplifier and that
error amplifier Del delay also needs to
be incorporated here Delta 2 now in
order to reduce uh the uh nonlinearity
in order to improve the
linearity here there are how many extra
component that we are adding that is uh
uh this amplifier and after this
amplifier your v in input at V VN is
equal to V which is this plus VD by
voltage gain
a so uh you need this uh the error
amplifier uh to adder and subtractor now
this uh additional term will also
consume a power and it will reduce the
efficiency so feed forward uh suffers
from a several uh
disadvantage that made its use in
integrated PA design difficult the
analog delay elements introduce a loss
if they are passive or Distortion if
they are active so it will also add a
loss and Distortion second the loss of
the output substract degrades the
efficiency and the linearity Improvement
depends on the gain and phase matching
of the signals that sense by each
subtractor so that uh gain and phase
matching also is important otherwise it
will not substract the uh nonlinearity
and we cannot get the uh completely
linear output so these are the three uh
shortcoming of this effect now we will
see the another method that is called
the cartisian
feedback so in the cartisian feedback
what has been done uh
we take output means this is a input
which is uh coming to the power
amplifier and then with sign Omega it is
will be up converted then power
amplifier will amplify the signal so
this we have already seen when we have
discussed the transmitter
architecture as we are
taking from the output some of the
portion
vpa that we take to the uh
we need to down convert that signal with
sin Omega L plus Theta we do the low
pass filtering and at the we subtract
input with this uh whatever down
converted signal that we are getting and
we can see that how much error it is
introducing and we again multiply the
signal and
amplify the power by using power
amplifier so uh if the power amplifier
output is down conver
and compared with the basement signal
this an error term proportional to the
Nority of the transmitter chain can be
created and with a quadr down conversion
this method that is why it is called a
cartisian
feedback now cartisian
feedback avoids the output subtractor
and much less sensitive to the path
mismatch but required some linearity in
the
PA it can also be designed
uh That Base band we can also make I and
Q in phas and quadrature component and
separately we can do the uh subtraction
that is shown in the second uh figure
now in this uh method of uh linearity
Improvement we need to make sure that
this uh L
signal that is a
properly uh delayed so that the amount
of down when we are down converting the
signal it is not experiencing the delay
which is added by the power amplifier so
this is uh why you you need to be extra
careful for Designing the uh L and this
uh additional two uh parts that we are
adding in the
feedback we'll also be uh add adding a
loss and mismatch that also needs to be
considered another method which is the
predistortion uh this is widely used
method because uh in this method simply
what we are trying to do is we know we
have designed a power amplifier and that
power amplifier how much nonlinearity it
is adding so we we can plot this V out
by V in curve v out by V in we have
already Pro uh known the power amplifier
we have measured the uh is
characteristics nonlinear
characteristics and it is defined by the
function
GX if some function
GX if we try to make the inverse
function of that at the input side means
the input signal whichever is coming if
we give if we pass to the G inverse X
means which is a a uh inverted function
then what will happen that if we combine
this uh output so what we are doing we
are adding the
predistortion uh in in the incoming
signal and that pre-distortion when it
is adding with the Distortion the we are
getting the
linearity so if the power amplifier
nonlinear characteristics are known so
that is very important power amplifier
nonlinear characteristics are known it
is a possible to pre- distort the input
waveform in a such a manner that After
experiencing the PA nonlinearity it
Remains the ideal waveform this
predistortion you can also bring to the
digital side before the means at I and Q
we make it pre-distortion at the digital
side we use a digital to analog
converted deck and then there is a
multiplier that can be used uh to
upconvert the signal and it will be
added and applied to the power
amplifier now this method uh has
Advantage if you know the power
amplifier characteristics so there are
three drawbacks of this method uh the
performance degrades if the power
amplifier non
nonlinearity varies with the process
temperature and the load impedance while
the
predistortion does not track the these
changes
so if there is some change in the power
amplifier uh with due to the temperature
or mismatch all the process then it
cannot we we have already predistorted
and then uh the signal which is at the
output will also be predistorted the
power amplifier cannot be arbitrarily
nonlinear as a no amount of pred
predistortion can correct for abrupt
nonlinearity it should not have any
abrupt nonlinearity otherwise there is a
uh predistorted signal will uh will not
mean this function G inverse into GX
will not be a
linear and variation in the anten
impedance because we are going to
connect antenna at the output uh
somewhat affect the power amplifier non
linearity but predistortion provide a
fix correction what is a uh uh solution
for this in order to solve this problem
we need to add some feedback to the
predistorted uh so that this uh variable
that we have identified in the preist
predistorter should automatically change
the way the power amplifier output
linearity is is changing uh so that
needs a additional feedback circuit and
uh the mechanism in the uh predistorted
so that this uh can be linearized so
these kind of uh predistortion digital
predistortion are widely used and you
can go through the literature which uh
will discuss about the predistortion you
can improve the
nonlinearity of the the power
amplifier so and another method that is
envelope
feedback in the anal of feedback what we
uh do is generally the anual of
experience uh
nonlinearity so what we try to do is we
detect the
envelope uh means there is a input
voltage and then there is envelope
detector and at the output also there is
a whatever anvelope with attenuation
Alpha we detect the envelope by envelope
detector and there is a amplifier error
amplifier AB which is amp uh supply to
the variable gain amplifier so if there
is a
some Distortion in the amplitude which
is uh taken care by the envelop detector
if it is happening this variable gain
amplifier will change its gain
accordingly that how much is a change in
the uh output after the power amplifier
and in the difference of the input which
is coming from the uh base band so that
difference will be Amplified by using a
variable gain
amplifier so here in order to reduce the
envelope nonlinearity in a PA it is
possible to apply a negative feedback
only to the envelope of the signal
instead of considering the pH so only we
will play with the
amplitude and how does the Distortion of
envelope detector affect the performance
of above
system if the two
detectors remain identical their
Distortion does not affect the
performance because the feedback loop
will yield VA is equal to VB and hence
if VD which is uh equal to V in this
property proves greatly helpful here as
a typical envelope detector suffers from
a nonlinearity so we need to ensure that
envelope detector is identical if it is
identical it will have and we need to
attenuate Alpha then this V in should be
equal to
VT there are different kind of method by
which we can design envelope detector
the simple method is we give the mixer
input will be input and allow signal
will be shorted it will uh use a square
so squaring of the signal so what we are
doing this is a v int which is V
envelope cos Omega + PT if V out T which
is a square because two signal are
multiplied beta with uh beta V envelope
Square t cos Square
Omega t + 5 T so what you will get
envelope in the DC plus cos 2 Omega t +
25 so you need one low pass
filter which will passing out the second
and other harmonics then you will get a
envelope at the
output you can also use a simple Source
follower in order to get only the
envelope so uh this will uh track the
envelope uh not the
carrier you can also use a limited
circuit this is a limited Circuit by
using a differential amplifier and then
this is a simple multiplier Gilbert cell
multiplier that we have discussed and
that will uh V in means when it is
limiter is applied and VA and VB which
is a multiply you will get the envelope
output that also needs to be lowas
filter so limiter and mixer can also be
combined and used as a an
detector so in practice the limiter may
require two or more cascaded
differential
pairs there is another method uh by
which we can improve the nonlinear uh
nonlinearity in the power
amplifier uh what we can do is a polar
modulation where we uh change the phase
and
uh amplitude
envelope and then we try to change the
amplitude of the power
amplifier with the envelope so any bath
signal can be represented in terms of V
envelope cos Omega t + 5T and we
decompose the input into an envelope
signal and a phase signal amplify each
separately and combine the results at
the
end so this approach is called a polar
modulation because it possesses the
signal in the form of magnitude is
possesses a signal in the form of
magnitude envelope and component and
phase component so this is how this is
input where envelope detector uh it is
applied to the switching PA and this is
a phase which will be Amplified so this
is how we can
design the uh polar uh
modulator and the final is a method
which is a
outphasing now in outphasing what we can
do is a uh this is a basic uh idea uh we
will discuss about more uh about uh
outphasing and one of the widely used uh
uh Power Amplifier that is the doti
power amplif in the next class thank
[Music]
you
