Lecture 45 - Mixer Design - IV

IIT Roorkee July 201
61 followers
Mar 1, 2024
Concepts Covered: Gilbert Multiplier, Active Mixer Operation and Characteristics, MOS Gilbert Cell Operation, Balanced Topology, Linearity ad IIP3 Considerations.
[Music]
[Music]
uh welcome to Arab transer design course
we are discussing about the mixer mixer
design and in last uh class we have seen
uh widely used to uh coupler based mixer
which will help us to uh design a
balance mixer by uh and image reject
mixer by using the coupler and
combination of the nonlinear components
now we'll try to see the one of the
widely used uh mixer that is a uh
Gilbert cell mixer and uh how it
evolved so this is a Gilbert multiplier
it is also uh used as a multiplier and
you can also name it as a mixer it is a
preze four quadrant analog multiplier
and it has application in a mixer so you
can have a uh both uh we can use it as a
both multiplier as well as a mixer
so you can see that it has a uh four so
the biasing is not shown here that is uh
tail basing at the bottom and at the top
however here uh there is how the mixing
happens that is uh explained so if you
see that it was Barry Gilbert uh this is
a very Landmark paper and highly cited
paper in 19
68 uh he has published this
groundbreaking paper uh introducing this
four quadrant analog multiplier uh he
has used the BJT so you can see there is
the npn transistor and uh by same thing
we can also discuss how we can realize
with the
mosad so here uh in this analysis uh it
is assumed that the device invols are
perfectly matched
and uh there is a process ideal uh
exponential
characteristics and have a high beta so
common emitor current gain which is
given by the beta it is a very
high so that we need to consider by
discussing this uh multiplier or a
mixer now if you apply a kvl
equation uh that what we will get is a
vbe1 minus vbe2 so this voltage equal to
vbe 3 minus vbe
4 equal to vbe 5 minus VB E6 so all are
bias such that all the base to emitter
voltage the difference is remain
same so the voltage vbe for each
transistor is
expressed as a uh we can use a vbe which
is a KT by Q lwn of IB by is if is is a
constant then we can write the ratio of
the current I E1 by
I2 so if IE current is written as a 1 +
x by 2 into I because it is a
differential signal and same because we
are applying the VX plus input we
consider V + x into I by 2 and here V -
x into I by 2 because we are applying
the negative voltage this is a VX plus
so there is a differential signal VX
plus and VX minus that is why here we
are considering 1 + x and 1 - x so the
current IE the ratio of current IE by
I2 we can write it down if we consider
vbe1 - VB E2 uh and vb3 - VB E4 or equal
to vb5 - vb4 and because this is a
logarithmic equation we can write it
down the i1 uh so it is i1 by is divide
by I2 by is so if is is is a constant uh
I E1 by I i1 / I2 equal to I3 by I4 = to
I5 by I6 which is 1 + x by 1 - x so the
ratio of the current between 1 to 2 uh 3
to 4 5 to 6 is given by 1 + x 1 -
x uh now uh if we see this is a tail uh
for both the tail if there is a this
current which is flowing through q1 and
Q2 is given by 1 - Y into i e by 2 and
the current which is flowing in Q3 and
Q4 which is given by 1 + y i by 2 so
this 1 - y i by 2 and 1 + y i e by 2 is
a differential Uh current which is a
generated through the uh other voltage
VY so here we have considered the
voltage
VX and uh VX plus and VX minus now this
voltage uh in q1 and Q2 is generated
through the VY means minus VY and this
is uh by by plus
VY now we'll try to find out the current
IO so now KCl is applied to the Circuit
we will get I E1 + I
E2 that is a 1 - y i by 2 so
here I E1 uh i1 +
I2 1 - y i e by 2 uh i
e i E3 and I E4 so here I I E3 and I E4
it is given 1 + y i e by
2 now the final expression of the
transistors which is I E12 so what will
happen
here that this is a this current which
is uh here is also multiplied with this
current which is a 1 + Y into 1 + x and
1 - Y
into 1
+ so X so 1 plus or - x is multip with 1
- y i e by 4 so IE 1 and 2 the total
current is 1 + X Plus or - x into 1 - y
i e by 2 and I3 and 4 we will get a 1 +
or- x and 1 + or 1 + y by 4 into i e so
what is your total current
IO so your total current IO it is a
combination of current uh so you can see
that the current from three and two are
added and current from uh four and one
are added and the difference of this is
here so that is two and three are added
so same thing two because this is a
connected at this node 2 and three are
added and 1 and four are added so this
is I E2 + i e 3 minus I E1 + I
E4 so if you if you write it down ie1
current here and plus I E3 what you will
get minus i1 + I3 if you use the if you
normalize the output current so so
normalizing the output current iio by
dividing it by IE so if you divide it by
IE what you will get output which is a x
into y so the our final aim is to
multiply the VX and v y and that we
achieved by using this
configuration now here you can use the X
signal as RF signal and Y signal as a Lo
signal and what you will get J which is
a if you use a low pass filter it is a
if
signal so if you try to see this
configuration if you try to see this
configuration this configuration
something look like a uh we can also uh
draw differently so if you try to draw
this
q1 Q2
[Music]
then Q2 and Q4 has the
same
voltage so this is
a
Q4 and
then
Q3 so
uh this q1 Q2 what is the current here
that is 1 -
y i
e by
2 this is 1 +
y i e
by
2 and how it is stepped out so the Q3
and
Q4 so there is a resistor so so Q3 and
Q2 are connected so you can connect this
two and q1 and Q4 are
connected so this q1
and Q4 are
connected and there is a resistor
between these
two this is a r
o and
this is widely because this circuit is
uh uh if you try to redraw the circuit
this is you are seeing in multiple
papers and this is nothing but a Gilbert
cell mixer so uh here in uh this is
drawn something differently to explain
but however the this is the same circuit
that is widely uh used and uh available
on on the literature so and now you can
use uh here uh one more transistor at
this
side and one more transistor at this
side and you can also use one more
transistor to bias all so this is a v so
what is this this is a VX so at q1 there
is a VX
plus and this is VX Plus at q1 and Q 3
and this is VX minus and here what is
coming 1 - y so this is a v v y minus
and this is a
vy+ so you this is a complete uh mixer
and you can also use a VX as a l and v y
as a RF and there are other application
also where uh V uh Y is used as a l and
VX is used as a uh RF so this this is
how this Gilbert cell mixer um is
evolved so it is based on the um BJT
same thing you can also use for mosad so
whatever we have drawn it is the same
thing so the what is the active mixer
operation active mixers are based on the
concept of current switching and Gilbert
multiplayer this is a rearrangement of
the same whatever I explain how it is a
rearrange q1 Q2 and Q3 Q Q4 and this
output current if you get the VX this is
how you can IO which is equal to I 10
hyperbolic minus VX by VT and if you are
if you are completely switching this a
square wave like signal is applied to
the core transistor VX plus and VX minus
to steer the ametric current abate ly
between the device at each zoc
Crossing so your I out which is equal to
i e e 10 hyperbolic minus VX by 2 VT so
this is your uh 2 VT mean you at least
need to have a amplitude which is high
to have the complete switching so this
is how this mixer can be
used so the em couple pair IV
characteristics so as it is shown in the
figure uh so illustrate an emitter
couple pair with the I characteristics
and when your VX is greater than 2 VT
most of the current is rapidly switch
between the device due to the sharp 10
hyperbolic transition and resulting in
uh almost the square wave so uh that
will if we try to use the same thing uh
what what we have discussed uh ideal
switch and that is uh applied the square
wave and what is what you need is a your
outward current IO can be expressed as a
series s terms with the fundamental
component being 4 by pi * the RF input
modulated by the L frequency Omega l so
this is the io current that we have seen
4 by pi I sin Omega l - 1 by 3 sin 3
omega L + 1 byk sin Omega L there there
we have consider cos here you can
consider sin as well and your emiter
current consist of a DC component and
small AC component due to the RF input
voltage what you will get is a i e which
is 1 - Y into i e which is 2 which is
equal 1 - a COS Omega RF T into i e by 2
so here in that your X is a s and uh
here Y is your cos if you multiply this
you will get a sin Omega l o
t IO current which is multiply with 4 pi
and if
you uh if you multiply this it is a uh
if final is means this lellow and if it
is Multiplied you will get 2 by pi
uh
into your fully differential circuit uh
the second pair Q3 and Q4 is added to
create a fully differential circuit and
this result in the characteristics of
allot terms and the addition of RF input
terms in the
output so your IO by i e that is a 4 by
pi Y into sin Omega L and then what you
will get is a COS Omega RF minus Omega l
t so this is a this is called double
balance mixer because you are applying
the RF and lo differential and the mixer
is a considered that is why it is a
double balance and
uh so here you uh your
l l ped through is um canel out and
there is a RF F
through so there is a no neither uh L
feed through nor the RF input feed
through to the output in the double
balance because you are cancelling out
the common mode signal for both hello
and
RF so in Practical situation a small pit
through may occur because there is a
transition there is not a perfect
transition ition happen at one point of
time it acts as a differential amplifier
and it will amplify the signal for a
very small duration of time but still
you will get a l or rpid through because
it is not completely switching for the
entire cycle so that is why there is a p
through will always be there in
Practical
situation so for Moss same thing you can
realize by using a moss Gilbert cell
mixer so what is this uh as we have seen
this uh will convert voltage into the
current and then this uh moset will
switch the current so this is many times
it is called the trans conductance
stage this is called a trans conductance
stage it is also called a GM stage and
this is a switching stage so that is
many times written as a s w
stage and then this resistor is
converting current into voltage you can
use any component which can convert
current into voltage or you can also use
a uh like OPM which will generate the uh
which is like a OT operational
transconductance amplifier so only the
uh trans means current which is a output
and then later on you can use a circuit
which will give you the uh current to
voltage conversion so this is uh here it
is a GM into switch uh so many times it
is also called a transconductance switch
uh GM into s w mixer so and uh this is a
moss version and your I out IO current
in this case is a one of M mu coox W by
L
VX into 4 is uh mu coox by WL minus
VX so uh this is going to square the
input so your VX so current steering
condition complete current steering from
one device to other device occurs when
your VX which is equal to square root of
2 I by mu C so it is a you can consider
mu coox W by L you can also write < tk2
into effective voltage V
effective so the conversion gain is
given by giv given condition 2 by pi GM
into RL where VX is greater than < tk2 V
effective and where GM represent the
trans conductance of the input pair M1
and M2 and RL is a load resistance so
same thing whatever we have discussed
for the BJT you can use for the uh moset
or AP
mixer so mixer function the mixer
operation can broken down into three
function as I told the input voltage to
current uh converter often implemented
using a shudo differential pair for
Simplicity and a balance between head
room noise and
linearity uh it is a switching cord that
is a device M3 and M6 that commutes the
RF current and uh load is responsible
for current to voltage conversion so we
have already seen that trans conductance
stage switching stage and a current to
voltage conversion stage so this is
almost there in all the mixer and the
trans conductance is Switched by the RF
and loo means switching is done by loo
and then you will have a voltage current
to voltage
conversion there is one uh important
thing that is a uh your because this a
differential uh uh differential signal
this is a singlehanded antenna which
uh mostly we have antenna which is a
single-ended and in that case you need
to have a single balance mixer so before
that I just want to add one thing that
here in this
uh
configuration there is a uh constructive
interference for RF and destructive for
loo that is why the
L2 if F through is a minimum in a
Gilbert cell mixer and that is why it is
widely and both RFP 3 is also minimum so
that is why it is also widely uh
considered mixer in multiple
application but it has a uh one
disadvantage that it consume a more
power and it has a multiple strike
transistors are there so uh and it
require a differential input so for your
antenna if uh because your LNA is a
single-ended now if from Ln single ended
input is coming then you need to convert
into the differential in order to get
the higher uh in order to use this uh
Gilbert cell mixer or otherwise you can
use the input as a means you can give
the input at one side and other is a
ground so there's a differential uh the
way we use in the uh differential
amplifier but it will have a higher
noise so that is why there is one more
configuration that is widely used is a
single balance
mixer so in single balance mixer uh that
is a uh single Ed ended RF input which
is so how the single balance mixer will
be
done
so this is
a RF
input there is a v
l+ v v l
minus and this RF if you consider the DC
current
IC plus
irf where irf is given by GM into VF
where because this is going to convert
voltage into the current and that
current is nothing but GM into vgs
whatever voltage you apply here GM into
vgs and that vgs is nothing but V RF cos
Omega RF
T and differential loo you can represent
differential L as a 4 by pi cos Omega
loo same we have used the same uh
equation for all the square loo
now you multiply this uh with
differential output what you will get is
a 2 by pi GM into RL because you have a
load
RL cos Omega RF is multip with Omega L
cos Omega RF minus Omega L by because it
is by 2 this 4 by pi is converted into 2
by pi so plus cos Omega r + Omega T you
if you consider 3 omega it is a three
Omega F and it has a multiple component
but we considered it as a filtered out
so that you will get a COS Omega RF
minus Omega l o t and if you filtered
out this as well your conversion
gain so this is active mixer gain and we
should write as a conversion gain which
is given by 2 by pi GM into RL and that
we have already seen in the previous uh
cases also which is a 2 by pi GM into RL
or for uh Gilbert cell also it has the
same and your I3 as we have discussed in
the uh while discussing the low noise
amplifier it is a uh proportional to the
overdrive voltage of the input
stage now the single balance mixer it is
the same we have drawn so IDC plus irf
it is replaced with the me uh BJ moset
and uh that is a uh M RF mosat and this
is the same configuration it is a
single-ended input because here the
input is a singlehanded that is why it
is called a single
balance the loo is a
differential it is a differential output
and uh that RF transistor provide and
gain is a 2 by pi GM into
RL it has a disadvantage that because
there is a single means ended input so
there is a high L to RF L to if or RF to
if head through but the advantage is it
consume a low power and it
has a single-ended input you don't need
to convert into the differential so in
many cases this mixer is a widely used
where there is a constraint in terms of
the power and you have a advantage in
this case you don't have a multiple St
transistor so you you can have a uh
better linearity as well so this is uh
about the single balance
mixer and single balance mixer operation
this is already we have discussed uh and
one more thing I want to add is a noise
and interference so in a single balance
uh any noise or interference at the lput
appears at the output making it
important to consider uh potentially
potential noise
source and
linearity uh and ip3 of active mixer so
as I told it is proportional to over
drive voltage uh in this context the
load that is a
chedel uh load uh
chedel uh if designed properly and
especially if it is a
resistive uh typically does not
contribute significantly to
nonlinearity now the switching component
the switches used in a mixer also have a
limited impact on on overall
nonlinearity so it is a crucial to
select the appropriate D DC biases for
both the output and the allo signal to
ensure that switches remain in
saturation this switches must remain
into the
saturation during the half cycle they
are active
then this helps in isolating the lower
transconductance device from the mixer
output and this uh capacitor which is a
CP it um should be
minimized this should be this remain in
the constant and this capacitor here
that should be minimized then you will
have a uh better better uh linear
output so minimizing the parasitic
capacitance the parasitic capacitance at
the common source of each switching pair
that we have seen in the
CP must be minimized and in modern
nanometer simos processes this condition
is generally met comfortably for most
gigs are
application so your
GM by
GM 3 and four so that is a GM of three
and 4 by CP should be greater than
greater than of so
GM 3 and 4 by CP so this here this CP
consist of multiple capacitors from the
M3 M4 and the M1 and that
parasitic CP if we try to reduce the CP
and increase the GM if it is greater
than it should be much greater than
Omega so we need to minimize the TM 3x4
and we should uh maximize uh we should
minimize the CP and we should increase
the gm3 comma 4 so
uh and impact of transconductance stage
the transconductance stage appears to be
a critical Factor influencing the
linearity and uh as I told the O Drive
voltage should be as high as possible
and what is the challenges in
multi-stage mixer active mixer usually
consist of multiple stages stti together
which can pause the challenges in
maintaining Headroom and that is why
your o Drive voltage is is reducing and
preventing unwanted Distortion so
achieving I ap3 despite the challenges
achieving a ip3 better than 1 Volt or
let's say if we consider 10 dbm is
possible and some linearity some
literature reports success in achieving
this level of linearity by active mixer
operating at Supply voltage of 1.2 volt
so uh aim is to have the voltage
Headroom which is a high enough so now
we'll discuss about the more
architecture of the mixer and how can we
design the mixer uh by based on the
parameters available in the next class
thank
[Music]
you
[Music]
oh
