Lecture 38 - Low Noise Amplifier Design - III

IIT Roorkee July 201
61 followers
Feb 27, 2024
Classification of LNA based on Different Topologies
welcome to RF transer design course we
are discussing a lone noise amplifier
design
and up to now we have discussed the
lanoise amplifier design by using uh s
parameters and we have also seen that if
only the mosfet we try to concentrate
because that is a widely used technology
for RF receiver design then uh we have
also defined for mosat what is a noise
figure and how the noise figure is has a
impact on the Ft so that we have already
discussed in the last class
now in this class we will try to uh
develop few uh topologies and try to
find out the noise speaker of the
same so LNA design uh requirement
broadly we as uh I discussed in the last
class it should have a uh Broadband
input matching if we if we target the
higher band or if we have a narrow band
also input matching is important because
theow
amplifier is the first component in the
receiver
chain so that is why it is important to
have impedance matching it should have a
sufficient gain to suppress the noise of
the mixer because as we know the freeze
Formula
F which is F1 + F2 minus1 by gain as the
mixer has gain of a first stage so we
should have a very low noise figure of
the first stage and the higher gain of
the first stage so that it will nullify
the effect of the noise of the preceding
stages uh low noise figure to enhance
the receiver sensitivity as we reduce
the noise figure we can uh improve the
sensitivity we have already discussed
about the sensitivity in detail in the
previous modules and if we are targeting
the low power application like iot we
should uh also try to get the low uh
means lower power consumption to
increase the battery life and uh on
other side because uh there is a cost is
involved in the size of the chip uh we
need to have a minimum diara in order to
have the uh reduction in the cost so now
let's see what are the different kind of
topologies studied and available in the
literature so so if if we see
the mosat is
a so mostly we consider a three terminal
it is a four terminal device it is a
body
also but if we consider the basic
configuration of the mosfet it has a
three basic configuration one is a
common
source common gate and common dream it
is also called as a source follower but
this is mostly because it is only going
to give you the gain in the range of one
and it is widely used for the uh buffer
application we we are not going to see
this uh common drain but all the uh lowo
amplifier which is made from these two
topology common source and common gate
so if we try to have classify the LNA
based on the topology the first two is a
first is a common Source LNA second is a
common gate
LNA we also discuss the inductor L LNA
because it is
application the as uh we discussed in
the first slide that we need to reduce
the area and while discussing the
inductor uh we have already studied that
inductor is taking a more space on chip
so how you can make a design which is a
inductor less and there are noise
cancelling
circuits so uh in common source we will
try to see resistive uh resistive uh
feedback inductive Source degeneration
cast Code common source LNA with a
inductive degeneration we also try to
see the cast code uh common
gate uh then common gate with feedback
common gate with a ped one so all the
common gate so basic common gate
configuration and then inductorless uh
configuration uh the uh main advantage
is uh common gate Ln with GM boosting
technique this is a widely used noise
cancellation technique and uh there is a
inverter based uh inductor less LNA are
also possible uh in noise cancelling we
we will discuss a resistive feedback
noise cancelling and common gate common
source noise cancelling Ln so let's
start with the simple resistive
terminated uh common source Len so
whenever you see any
circuit as a uh RF designer you should
know that what are the noise sources in
the
circuit so basically the input impedance
means if we try to see the
input impedance at the gate side it is a
very high impedance
and what is the simplest way because we
are going to connect the our uh
amplifier lse amplifier with the antenna
so we need to have input impedance which
is a 50 ohm now if you try to see the
MOs pad only from the gate side the
input impedance is very high and that
cannot match with the antena impedance
of 50 ohm and there is a so much
reflection so we already have seen that
if there is no impedance match there is
a reflection will occur so what is the
uh simplest way to do the impedance
matching is to connect the RP
resistor in parallel to the uh input of
the uh moset so here when you connect
the resistance RP in parallel that will
decide your matching so here your input
impedance or input
resistance is equal to
the RP so whenever we try to uh solve
the any circuit for the low noise
amplifier we will uh the two is more
important uh parameters one is impedance
matching another is a noise speaker and
third is a
gain so for this circuit your input
impedance is are in which is is equal to
RP
and uh the if we want to get the gain so
if you just see the if there is a
voltage here that is given by V in if
this is a v in this is a vs so
your because and we want RP should be
match with the RS and if RS and RP are
same so if RS
is equal to
RP that leads to
v s v in which is equal to v s by 2 so
the first disadvantage of this circuit
is as there is a
resistive matching what is happening is
that the your half of the power half of
the input voltage uh at the input side
itself it is a coming half so because of
that the your 50% of the gain is reduced
here itself now the gain we can write in
terms of RP and
rs if RS and RP are not equal we can
write
RP by rp+
RS into GM into Rd
so this is a gain of the circuit if you
consider RS is equal to RP it is equal
to GM Rd by
2 so your gain will be half because you
have use a resistive matching and there
is a resistive load so as I told now uh
noise speaker equation we already know
know so as discuss while discussing the
noise noise speaker is nothing but SNR
in by SNR
out so I can write it down it like this
SI by
ni into n o
by s
o that can again decompose in terms of
Si
by n i
into n o by g into
SI where SI is a cancel out so if you
try to uh see the input noise uh ni I in
this
case so mostly this
is 4 KTR
RS into n o
that is a total output noise and it can
if we try to represent in terms of the
mean Square voltage
noise power spectal
density divide by gain Square because
here we have consider the square we are
considering it it as a gain Square so
this is how we we can find out the noise
speaker so for this what we need to have
we need to find the VN Square
out so how how can we find out the VN
Square out so because there are rs and
RP which are parallel if there is a no
Source we can directly write the noise
due to RSN
RP so what what is the method to find
the VN out square first we need to find
that how many noise sources are there in
the circuit so uh can you think how many
noise sources are there but obvious
there are four noise sources one is due
to RS one is due to
RP another is a MOS pet and then load
resistance Rd so in order to find the V
and out square mean Square what we need
need to consider we need to use a
superposition theorem and for each noise
Source we will try to see the what is
the output noise and then we will add
all the noise to find out the V out VN
out square mean
Square then we we already know the gain
of the amplifier we divide it by gain
square and 1 by 4 KT RS so this is is a
methodology that we will try to
use so as RS and RP both are in parallel
we can directly write for RS and
RP that is noise due
to noise due
to RS and
RP it is given as a 4
KT RS parallel
RP and because the whatever noise Source
RS and RP uh that noise is going to
amplify by this amplifier so we need to
add the gain also that is
a GM
into r D whole
squ so this is a noise due to RS and RP
now noise due to
the
transistors
M so that is 4 KT so if we see the
current that
is 4kt gamma GM noise current that we
already discussed 4kt gamma GM current
that is a thermal noise and it
is given as a
4 KT and we are interested for the
voltage because the here the term is VN
Square out so we are multiplying with
the
Rd
Square so noise uh square root noise
current is Multiplied with the r d
square now noise due
to
Rd which is equal to
4 so here we are not adding the gain
because that is a at the output side
only it is not going to amplify so we
are just considering 4 KT into RT now uh
what what we need to do is we need to
put all this uh in this equation then
we'll get the noise figure
so which is equal
to 1 by 4 KT
RS then 4
KT
RS
RS parallel
RP
into
GM Rd whole
squ
plus
4 KT gamma
GM
Rd squ Plus
4
KT
RT 4 KT
RS is already there gain Square so when
we consider the gain the gain here
is GM into
Rd whole square and we also need to
consider RP
s RS +
RP
Square now we'll take each term one by
one so the for first term that
is if we if we consider the 4 KT is a
common in all we can cancel it out from
the top and bottom so 4 KT is
common in
all
so that is cancelled out then we can
write here
rsrp by RS +
RP into GM
s Rd
squ
by RP
squ RS +
RP
s
into GM s Rd
Square same we can also write for
here
gamma
GM Rd
s GM
squ Rd
Square this is Rd and RP don't
confuse RS
+
RP
Square then we will have
a and we also need to consider
everywhere there is a
RS because we just cancel out
the 4
KT so then
Rd
by GM into
Rd
squar into
RP Square by
RS +
RP
squ so here RS
is multiply so now we'll try to cancel
out the term that we can
cancel uh here it can be this can be
easily cancelled
out and then we can also cancel out the
RS we can also canel out one of the
RP and we can also cancel out the
RS RS plus RP so what remains uh that we
will write it down in the next
slide so then noise
figure is given by RS plus
RP by
RP so as RS + RP by RP this is the first
term we can write 1
+ RS
RP now in the second uh this
Rd is cancelled
out this Rd
square one of the GM is cancel out and
if we multiply RS at the bottom and rs
at the top this we can write it as RS
parallel RP
so another
term that is
a gamma gamma into GMA
by gamma
RS into
RS Square RP
s by RS + RP whole
squ
plus
here we can cancel out
the Rd so Rd is cancelling out so if
this is GM
squ and if this is Rd Square One of the
Rd is cancel
out and we will have
a RS
by GM
Square R
is parallel
RP square
into
Rd so same thing is written here GM
square and uh RS so RS we have multiply
at the top and bottom so it is RS
parallel
RP so the finally we can write
the noise speaker which is
comma RS by
par RP
S
Plus
RS G
m² RS parallel
RP square
into r
d now we'll take if because we want to
have impedance matching I RS is equal to
RP the noise
figure can be written as a this RS is
equal to RP so 1 + 1
plus because RS is equal to RP we can
write RS by 2 and that will come as RS
Square uh by 2 so we can
write that is
4 gamma
GM into RS because there is a RS Square
so because RS square is there RS RS can
be cancel
out
plus we will have
a uh RS
Pi TM s
RS Square by
4 into
Rd so this four will go up so we can
write it as
a this is a 2 + 4
gamma GM
RS plus 4
by G
m²
RS r
d so this is very important equation it
will tells us
many that
this
configuration minimum noise bigger if we
try to neglect this term also then noise
figer is a two and that is a uh around 3
DB so that will be very high noise
figure and the minimum noise figure is a
3db will go more than this also so for
low noise amp uh application we cannot
use this uh structure but it has
uh let's say if you uh have a uh not
that much stranged noise speaker
requirement you can use this amplifier
it has one advantage that is
because
so the disadvantage is noise figure is
too
high greater than
3db
and Advantage is
uh Broadband
matching because it is a resistive
matching and mostly the half of the
power is reduced the gain is also
less
gain
is less than
3db uh less than 3db of the
desired uh let's say if he because at
the input side itself there is a half of
the voltage so this is about the uh
common source amplifier with resistive
matching so there is one more advanced
circuit that we can try to uh look after
that that is a if we do the resistive
feedback from drain to gate that is also
widely used circuit so
this uh if we try to draw the
circuit that is the RF
feedback and if there is a
capacitor
that is a DC blocking
capacitor
RS there is a
vs this is a
ground so this in this Advantage is in
this structure that you are uh in the
previous uh topology we are our voltage
is half at the input side here the
voltage is not half and if you if you
see as this is uh considered as a diode
connected your input impedance that is r
in is equal to 1 by GM of
this
m so it is a straightforward R in is
equal to 1 by GM and if you try to match
RN is equal
to r s which is 1 by GM and if you are
targeting the 50 ohm impedance matching
that is RS then GM you need is around 20
M
cements so if we try to if you want to
find out the gain of this
circuit it is a
simple we just take this as a
RF if there is a voltage VX at this
side
then we will
get
RF GM into
VX
so this is a VX
here and then this is a v out so we can
get directly VX if this is a
VX
minus GM
into
GM
VX into RF equal to V
out so V
out by
VX is given by
1
minus and if we try to put the GM is
equal to 1 by RS it is RF
by
RS it can also be uh because the same
here the input impedance is uh 50 50 ohm
so same we can also write the V out by v
as
is equal to
half 1 - RF by RS we can also mention it
as a minus because if this term is a
higher minus RF by RS into
half so this is a gain of the amplifier
so
as as we uh discussed in the previous
example what we need is the input
impedance gain and uh we also need a
output impedance because in this case it
is not
straightforward so we also need output
impedance so this is uh this is a for
input
impedance this is for the gain
then input impedance now we'll try to
find out the output impedance
so uh for small signal if this is a
RS and the input is grounded there is a
vgs and here there is
RF
we will have
a
current GM into
vgs and V out vgs we can also mention as
a VX because we have already mentioned
here as a
VX so this uh here we can directly write
the kclv out
minus
vgs
by
RF
plus because we are interested to find
out the output impedance V out by I out
GM into
VDS is equal
to I
out so vgs we can mention it as
RS by RS +
RF
into V out because there is only one
potential V output voltage V out is
there so we can write V GS which is RS
by RS + RF into V
out what we will directly get uh then
this
is if we try to put the vgs in this
equation we'll
get V
out 1
minus RS upon rs+
RF then
GM
RS RF into V
out into I
out so if V out we take it out and take
on this
side we'll get 1
by R out is equal
to because if this term will go up that
is 1
by RS + RF
because there is RS
also in the denominator 1 by RS +
RF
plus 1
by 1 by
GM into 1
+
RS RF
+ RF by RS
so if we try to make RS in this case
denominator so that is 1 + RS by RF if
GM also in denominator so we can
directly write R
out is equal to 1 by
GM 1
+ RF by
RS parallel
to RS +
R and this
term implies if if we consider GM is
equal to 1 by
RS the r out will come out as
a RF + RS by
2 so now we got everything we have a
input impedance we have output
impedance
and we have again we will try to use the
same formula uh of the noise figure and
we'll find out the noise figure in the
next class thank
you
a
