Lecture 37 - Low Noise Amplifier Design - II

IIT Roorkee July 201
61 followers
Feb 27, 2024
MOSFET 2 Port Noise Parameters, LNA Design Parameters and Requirement.
welcome to RF trans design course we are
discussing about uh lanoise
amplifier uh design and in last class we
have seen if s parameter is given to us
and based on the S parameter how we can
proceed to design a uh Lano amplifier so
in order to get the S parameter of the
amplifier you need to bias your
transistor and based on that you will
get the S parameter it it is provided
might be from The Foundry or the device
manufacturer or you can also use a
uh simulator to simulate the S parameter
based on your design it might consist a
single transistor or more than one
transistor or transistor network but it
will be represented as a two port and
you should know the prior s parameter of
that for Designing the luno amplifier by
using the S parameter we have also seen
the uh how to make the noise circles
gain circles and how we can identify
the input reflection coefficient gamma s
by which you can make the input matching
Network so that you will get a low noise
amplifier as per your desired noise
figure so now uh we will specifically
Target the MOs pet because uh it is in
rfic it is a widely used uh components
and how we will
make a low noise amplifier design by
using a MOS ped so let's discuss the
noise in the
moset so if a moset if we consider a two
Port noise parameter so moset is
comprised of a two
Source uh if we consider the thermal
noise uh then it has a drain current
noise so when is a drain current noise
and another is a gate current noise
because there is a gate current as well
as a drain current and gate current is
happens because here we are considering
a very high frequency and when we
consider the high frequency we cannot
avoid the capacitance CGS so if you are
working at low frequency you can
generally consider this is as a open
circuit but at high frequency there is a
if this is
a gate if this is a source and if this
is a
dream so GM into
vgs and there is a
arrow so
if we are working at high frequency we
cannot award this capacitor CGS and
while discussing the Active Components
we have also defined the ft and F Max by
using this
CGS so that that is why there is a gate
current noise and it is given by iing
square this is a mean Square value uh of
the noise current and it is given by 4
KT gamma GG into Delta F uh we already
have seen in the while we discuss the
noise how this Delta f is coming that is
a per unit bandwidth so we we have
defined the noise power is KTB and we
need to consider the band
width similarly for the drain current i
d uh mean Square current noise current
is given by 4 KT gamma GD KN into Delta
F so this is for the gate and this is
for the
drain so here GG is given by Omega squ
CGA sare 5 into GD KN so gate noise is a
correlated with the drain noise with a
correlation coefficient C that is I into
i d conjugate divide by whole Square in
uh square root of in square into in d
square mean Square so if we consider the
correlation and if we find out the two
fundamental moset noise sources if we
back to the input Port we already have
done this for uh while discussing the uh
LNA design and in that we have defined
the input referred means if you see the
voltage and current at the input side
you can write it down a drain current
noise back to the input as a noise
voltage so uh in uh last class we have
also seen that if we want to define the
voltage and current at the input side uh
so here there is a current that is
voltage is defined as a e n square and
similarly there is
a current can be defined as a i N1 squ
this is a mean Square so this we are
considering at the input side and if
mosat is a two Port Network in the
previous
uh example also when we have calculated
the noise figure minimum noise figure uh
at that time also we have consider the
input referred noise current and noise
voltage so en Square can be written as i
n d square by GM
squ because uh we are considering at the
input side and then you can write 4 KT
GMA GD Delta F by GM s so RN so
Capital RN that is defined in that noise
figure equation can be given by E n² by
4 KT Delta F so this is a voltage by
current we can get gamma GD by GM squ so
this is a resistance RN now noisy drain
current also flows when the input is
open circuit so uh this we we are also
considering the
noisy uh drain current that is I N1
square mean Square so i n d square into
J Omega CGS by GM squ so we we should
consider the impedance J Omega
CGS so here this is
a uh this is given by I and D squar that
is e n sare into GM Square so we can
write Square n square into uh GM Square
GM Square will be cancel out so your I
in square is nothing but e n sare into J
Omega CGS whole Square so total input
current noise is equal
to reflected drain noise plus induced
gate current noise so this is a total
input current noise so induc gate noise
current it is a correlated with the
drain current noise and uncorrelated
with the drain current noise so the way
we have defined in the correlation
admittance YC to find the correlation
admittance YC equal to i n 1 + iing GC
so this is a correlated dra drain
current and then this is a uncorrelated
drain current so i1 + IC by n we'll get
J Omega CGS plus I GC by n so fi finally
we get J Omega CGS plus GM into i n GC
by i n D so the YC is equal to uh J
Omega C GS plus GM i n g i n d conjugate
by i n d whole Square so that can be
written as a j Omega CGS plus GM because
this is the same as a correlation
coefficient C uh square root of i n g
Square by i n d
squ now we add both in and I N D from
the previous equation that is iing and i
n
d so this is in D and in G one is a 4kt
gamma GD not Delta F and in G 4kt gamma
GG Delta F and where GG can be
considered as this term so we will get
the YC in terms of Delta J Omega CGS GM
by GD
correlation coefficient Delta by 5 gamma
Omega into CGS so this is both are the
process dependent parameter Delta and
Gamma so we can get the YC which is G
Omega CG is 1 minus Alpha into C square
root of Delta by 5 gamma so this is this
term YC is depend upon
CGS the
alpha then uh correlation coefficient C
Delta by 5
GMA so here Alpha is equal to GM by GD
KN and the parameter Alpha exhibits
value of one for long Channel device so
mostly for long Channel device it will
be one whereas it gradually di
diminishes as the channel size decreases
so here we get all the parameter G T
which is
zero BT which is uh this value that
is uh
BT Omega CGS into 1us Alpha Omega CG is
1 - Alpha C square root of
Delta by pi
gamma RN we have already find out the
value of RN and then gu can be given
given by this equation so we got all GT
BT RN and gu then we will try to get the
G Optimum based on the GU so G Optimum
is given by square root gu by RN plus CC
squ so very if you try to keep value of
gu and GC so we have already defined
this equation uh in the previous slides
for how to get the f mean = to Alpha
Omega CGS Delta by 5 GMA 1 -
c² so
here if
we want to find the f mean which is
equal to 1 + 2 RN G op plus GC equal to
1 + RN means you can replace the value
of
RN and uh G of T GC you will get the
final value in this
term so now for different value of Omega
by Omega T so here Omega by Omega T is a
20 you will get f mean which is equal to
0.5 now for in order to find the f mean
here gamma is considered a 2 Delta is
equal to 4 and the based on the value of
Omega by Omega T we will get the uh for
Omega by Omega T of value of 20 we'll
get f mean which
is5 and C is a correlation coefficient
so how much correlation is there between
I N means gate and drain current noise
current that is given by the correlation
coefficient C So based on the value of
Omega by Omega T you will get the f mean
so here the things that we need to
observe that
as your noise minimum noise speaker f
mean is inversely proportional to Omega
T so while designing the low noise
amplifier if you are targeting high
frequency then your Omega T if you want
to have a low noise figure for higher
frequencies you need to have a higher
Omega T so your device should have a it
means you you need to use very fast
device where your ft is higher so ft and
Omega T both should be high in order to
get the minimum noise figer so here
while considering the minimum noise
figure we have only calculated the uh
based on the source impedance now if
there is a source impedance is match
with the optimize impedance now if we
want to consider the power match so
there are two things one is the
impedance match and another is the power
match so in low noise emplify design we
should consider both impedance and power
match how can we try to optimize the
power and have a battery impedance
matching so noise optimization with a
power constraint we all we all already
have find out the equation of noise
speaker which is f mean plus RN by GS
gs- T op square plus bs- P op Square so
here if we assume BS is equal to B
opt if uh the source susceptance means
if we consider the Y so y
s is equal
to if we want to have
a conjugate match with the Y
optimization so we should have BS is
equal to minus uh BS is equal to B
opt so your f is equal to f mean can
directly be written as a uh f is equal
to f mean plus RN by GS GS minus G opt
so your G op here can be rearrange in
terms of quality
Factor so quality Factor Q is given by
uh
imaginary part so Q op which is equal to
G op by Omega
CGS so here this is a real part so we
already have defined that Q is nothing
but imaginary
part of Y by real part of
Y so here imaginary part is nothing but
a G opt and real part is given by the
Omega into
CGS
so Q op is given by Alpha square root of
Delta 5 gamma so if we try to replace
the G opt from this equation if we try
to replace the G opt from this equation
we'll get a q opt which is C by Omega
CGS Alpha root Delta 5 GMA 1 - c² and Qs
is given by for S side 1 by Omega CGS
into RS so because there is a series uh
there is a capacitance here Q is
considered as a 1 by Omega CGS into RS
so your quality
Factor Q op and Qs if you try to replace
this in this
equation GS and G opt we'll get
noise speaker in terms of Q op and Qs so
f is equal to f mean plus comma by Alpha
GM into RS so that is what we have done
we have replace RN and
GS gamma by Alpha GM into RS 1 - Q out
by Qs whole
s now we already know the drain equation
drain current equation is given by
the complete drain equation is given by
this mu CX W by 2 vgs minus VT and here
vgs minus VT uh is in relation with the
uh electric field in saturation and L is
a length of moset so ID you can we can
rewrite as a WL
coox uh V saturation then e saturation
and there is another ter row which is
depend depend upon o Drive voltage which
is vgs minus VT by L into e saturation
so the this is the another way of
writing the ID
equation now we'll find the power
disspation that is a PD which is equal
to vdd into ID so your PD is given by
vdd wlc Ox V saturation into e
saturation R square by 1 + r s and your
GM is equal to Alpha GD now what we want
is a minimize this power dissipation so
how can we minimize the power
dissipation if we try to find out the
value means the we if we want to
minimize we need to minimize with
respect to the value of row and uh that
uh if we try to work for this
we will get the optimiz
width for which we will get the minimum
power and that is given by the 3x2 1 by
WL coox into RS so this is this will
provide as the optimum
width for power
consideration now when you try to use
the optimum width for power
consideration your f mean
is can be rederived and it can be
mentioned it has a two term 2 by 2.4
gamma into Alpha Omega by Omega T and
here the f mean that we have derived in
the previous slides so that is 1 1 + 2.3
Omega by Omega T so here again your F
me is inversely proportional to Omega T
however if let's say you have omega T by
Omega is equal to T 20 if your Omega T
by Omega is equal to 20 if you consider
gamma is equal to 2 Delta isal 4 and
Alpha is equal
85 then you you will get a f mean which
is equal to5 DB f mean power is equal to
1.1 DB so this is uh considering only
the uh Source uh matching so and this is
considering means this is considering
only the impedance matching and this is
considering by uh Power match so when
there is a power and impedance both is
match at that time your noise figer is
increased by around5 to 6 DB so for
Omega T by Omega if it is 15 it is
increased 2.6 and 1.4 if Omega T by
Omega is 10 it is .9 and it is increased
1.9 and here if Omega T by Omega is 5
then your noise figure is 1.6 in this
case and 3.3 in this case if you
consider the Omega Optimum P this value
this is a power minimum noise figure
that you will achieve so what it says
for the
mosfet if you want to design a low noise
amplifier and if your Omega T by
Omega is equal to
20 that means if it is Omega T if it is
200 gahz and if Omega is
a 10
gahz then you will get a
20 so if you are designing a low noise
amplifier in the technology which is
providing you the Ft of 200 uh Omega T
of 200
GHz and uh your Omega is if your desired
Omega is 10 so if your ft is 200 GHz and
your desired frequency F not so Omega T
by we can also write ft by F both are
same because 2 pi will be cancel out so
ft by F which is equal to 200 by 10 so
at designed frequency of 10 uh you will
get a minimum noise
figure5 DB so you cannot get less than
this noise figure if you consider the
power match then your noise figure is
1.1 DB so this is uh how how means you
should you will get a 10 GHz now if you
try to increase the frequency your Omega
T by Omega is uh means Omega T by Omega
is reducing and that
will uh that will increase the noise
speaker so this is what is a uh Power
match and the impedance match constraint
and while calculating this chart the
gamma is considered as a two Delta is a
four and Alpha is
a85 so this is what is uh the impedance
match and power match so the uh I think
this is uh about the moset noise now we
will try to uh consider the
various uh various topology of the Lono
amplifier and how that various topology
we can design so we have already
discussed while discussing the uh
receiver architecture that your noise
figure f is equal to SNR in by SNR out
and it is given by 10 so this is a noise
factor F and noise speaker is 10 log
into F your gain voltage gain is given
by this that's uh that is a 20 log of V
out by V in and we already have seen the
free formula that your first component
in the receiver will uh identify the uh
noise figure of the complete receiver
chain because the if we see the freeze
formula the first term is fub1 + FS2
minus1 by the gain of the uh first stage
so the gain of the low noise amplifier
should be high and the noise figer
should be less in order to minimize the
noise figure of the complete receiver so
this is uh by this way we can able to
get the minimum noise in Our receiver
design so that is why low no amplifier
is important uh in the transer design
and uh we are going to see the many
different kind of topologies for
Designing the low noise
amplifier so linearity also we have
already seen uh in linearity if you uh
want to achieve the we have discussed
the two kinds of linearity one is a
intercept input intercept point we have
also discuss the
iip2 uh while discussing the receiver
architecture so both linearity is also
important when we are designing the low
noise amplifier and stability uh also we
have discussed so if we want to design
LNA design requirement it should have a
Broadband input matching to minimize the
return loss it should have a sufficient
gain to suppress the noise of the mixer
and low noise speaker to enhance the
receiver sensitivity and low power
consumption to increase the battery life
and small di area to reduce the cost so
this is very important for uh designing
the low noise amplifier we have already
discussed one of the way if s parameter
is given to us how to find the how to
design the low noise amplifier now we
will try to design the low noise
amplifier by using the circuit
methodology where we try to uh see the
we'll try to discuss the step by step
what is the input impedance what is the
output impedance how we can calculate
the noise and what is a gain so this is
about the low noise amplifier thank you
we'll discuss uh more in detail about
the all topology of low noise
amplifier
a
