Okay so first part we will continue with the
electronic switches and then uhh we will move to millimetre wave link budget and noise calculation
and in presence of noise how we characterise any millimetre wave system So let us continue
with electronic switches last day we were discussing about a given switch we can use
it in shunt configuration as well as in series configuration Now usually looking at value
typical capacitance or resistance values a few Ohms or capacitance a few pico farads
that we have seen that in shunt configuration the switch will give better isolation Now
let us say whatever isolation we are obtaining from a given switch it is not enough so in
that case how we can improve the isolation further So in that case we can use the same
switch uhh number 2 number in series shunt configuration so here is one example So you can see one antenna it is connected
to receiver side and transmitter transmitter side so now when the transmitter is connected
to antenna in that case the switch whatever we are using in series configuration it is
in ON condition and in shunt configuration it is in OFF condition So that means in equivalent
circuit in right hand side shown here so in ON condition the switch will replace by its
equivalent resistance R series and in OFF condition we will replace it by equivalent
capacitance which is c off and in this situation we call it a series shunt configuration Similarly
right hand side can be modelled right hand side together these 2 switches it forms a
series shunt switch and it is in OFF condition so the switch or a PIN diode which is OFF
we represent it by c off and the switch which is in ON condition we will represent it by
series resistance which is grounded So now how isolation improves in this case
So we see let us consider the switch which is in OFF condition that is the receiver side
so this PIN diode the first one it is open circuited so it is represented by a capacitor
so capacitor will give you some isolation For the second pin diode it is switched ON
so if there is any RF signal it will be further grounded so we have actually isolation improvement
from the contribution of these 2 switches but looking at the insertion loss value it
will not be that good if I consider now left hand side or the transmitter side so in that
case (isola) uhh insertion loss will increase because of the resistance and the capacitor
because in addition to the resister we have a capacitor which is also grounding some part
of the RF signal so insertion loss it will be poor But anyway for any given application the main
point is the difference between the insertion loss and isolation how we can improve this
difference Now let us say we plot a figure of merit versus frequency so left side it
shows insertion loss and right hand side it shows isolation versus frequency for different
figure of merits Now for the insertion loss plot what we expect that instead of single
switch if we use series shunt configuration insertion loss will be higher so let us compare
200 femtoseconds for this one at 140 gigahertz insertion loss is almost 6 dB Now if I compare
this one with previous case let us go back to previous one so here with r ON equal to
4 Ohms you can see insertion loss at 140 gigahertz it is 3 point 5 dB approximately so in series
shunt configuration insertion loss increases now look at the isolation values For 200 femtoseconds at 140 gigahertz we have
at least 15 dB isolation but if I go back in the previous case in this case for a (two)
200 femtoseconds you see isolation is less than 5 dB at 140 gigahertz So using series
shunt configuration this isolation has been improved from 5 dB at least 15 dB and the
same logic is true for other figure of merit values So what is the conclusion then we can
use series shunt configuration to improve isolation at the cost of little increased
insertion loss Now some cases we may face problem due to
input impedance matching problem why If I go back to this circuit once switch we are
using in series configuration another one in shunt configuration so as a whole this
switch is ON so now whatever impedance seen by the antenna if I change the state of the
switch then it will be OFF that means this resistor is open and the (capacity) and and
this OFF uhh this series resistance we have to replace by a capacitor and this shunt capacitor
we have to replace by a resister in that case input impedance seen by the antenna transmitting
side everything will change And not only that in series or shunt whatever we consider so
the capacitance or resistance whatever given by the switch it is it depends on that particular
device we do not have any control over that Now for a 50 ohms system we should match the
input impedance so sometimes it may happen if we simply by some switches and use in the
circuit so input impedance matching it may be very poor in that case how we can improve
the impedance matching We can use some inductor simply to improve the impedance matching so
let me show you one example So in this particular isolation and loss plot for a SPDT switch
with matching inductor so when we add matching inductor we add a small value of inductor
in series with the switch and the value of inductance L series it can be given as Z 0
square multiplied by c series plus c shunt so c series corresponds to the capacitance
of the switch in OFF condition whatever we are using in series combination here Similarly C shunt this is the capacitor of
that PIN diode which we are using in shunt configuration So then the matching inductance
value it depends on this capacitance values and the system impedance Z 0 Now if I look
at the source of losses when the switch is ON so we have mainly losses because of the
resistance so resistance it is defined by fabrication process and for resistance improvement
it can be done by changing the W by L where W represents the width of any MOSFET and L
represents the length same true for PIN diode W in that case represents the cross section
L represents the length so by fabrication process then we can tune the resistance value
Now look at this plot so insertion loss it increases but return loss it has a minimum
value and it is a function of figure of merit So for a given inductor then and for a given
capacitance value we have some optimum value of r for which return loss will be minimum
So for this particular example when we are using r series equal to 4 Ohms r shunt equal
to 7 point 2 Ohms at 60 gigahertz in that case we are having minimum return loss approximately
at 130 to 140 Femtoseconds Already we have discussed how to implement a switch by using
PIN diode now let us see how we can implement switch by using FET Now in NFET or NMOS or PMOS both are available
in market but usually when we go for high frequency applications NFET or NMOS is preferred
why Simply because electron mobility is higher than hole mobility So we can use NFET at even
millimetre wave frequencies and there are some alternatives also already we know the
devices for example we can use also MESFET as a switch we can also use HEMT as a switch
but in all the cases we will prefer electrons as the careers So in general we can represent
a FET MOSFET by this so we have this is a 3 terminal device if we neglect the body terminal
then we have source drain and gate the control voltages is applied at the gate terminal Now
if I look at the variation of drain current versus V DS so this first part of this graph
it looks like a straight line we call it the linear region or the triode region And after that for a given gate to source
voltage if we keep on increasing the gate to source voltage then this current component
it becomes constant or we call this is the saturation mode of the FET Now when we use
it in the switching application in that case we will be using this linear region and the
cut off region of the FET So linear region it occurs for V DS this is less than V GS
minus V TH where V TH this is the threshold voltage and V GS this is the gate to source
voltage then in that case drain current drain to source current this is Mu n into C dielectric
into W by L this ratio multiplied by V GS minus V TH into V DS minus VD V DS square
by 2 ampere where the C dielectric it represents the gate capacitance it depends on the whatever
insulation we have used for the gate and the thickness of that insulation So C dielectric it becomes a process parameter
not only that W by L it is also a process parameter which depends on the fabrication
procedure and here L is the channel length and W is the width of the gate
So the ON state resistance we can consider a V DS value sufficiently smaller then in
that case r ON that is equal to L by MU n C dielectric into W into V GS minus V TH Ohms
and the channel is OFF the switch is OFF then when V GS is less than V TH we do not have
any inversion layer inside FET So if L OV is overlapped length between the gate and
the source drain so OFF state capacitance we can we can represent like twice Epsilon
Epsilon dielectric into W L overlapped divided by T dielectric it is in Farad So for this
given FET then we can calculate what is the figure of merit it comes simply R into C which
is twice L L OV divided by Mu n into V GS minus V TH and the unit is second Now for input figure of merit what we have
to do we have to decrease the channel length Similarly we have to decrease the V TH so
W by L it is a process parameter in that case we have to use the minimum value so that you
can have the less figure of merit value for improved performance of the switch How we
can decrease V TH we can increase the dielectric constant of the insulator whatever we use
and the thickness of the insulator that should be as small as possible Another way we can
use uhh some group 3 group 5 material inside which carrier mobility is very high for example
gallium arsenide indium phosphide or gallium nitride so if the carrier mobility is higher
in that case we can expect that r ON will be much smaller Some hetero junction structure
like gallium arsenide aluminium gallium arsenide they also exhibit high carrier mobility Stress engineering also increases carrier
mobility so this stress engineering last day I was discussing that another crystal let
let us say germanium or maybe silicon so one crystal it is justů So if we place this crystal
on another one and if we use a very thin layer in that case the first crystal it will try
to follow the second crystal and sometimes the effective carrier mobility it depends
on the crystal structure so we can force the second layer to follow the first layers property
so this is called the stress engineering Now in CMOS technology usually the figure
of merit is very high and that is why it is not popular at millimetre wave frequency typically
they are used below 30 gigahertz so here is one example of I l R l and isolation plot
for a CMOS NFET designed using 65 nanometres technology typical R series value is given
as 5 ohms C off 36 Femtofarads and T (())(16:22) figure of merit is coming 182 Femtoseconds
Sometimes we can use parallel inductors to improve the performance of the switch so this
FET itself it will give you some capacitor and inductors together it forms a resonating
circuit and it behaves like a band pass filter So that is how sometimes we can improve the
performance of a FET switch So here is one example of NFET switch with
inductors so for this typical switch we are using 520 Pico Henry in series and 3 point
5 Nano Henry in parallel combination and that will give a band pass filter like response
the response is shown here So you can see insertion loss it is minimum at some the resonating
frequency approximately 41 gigahertz Not only that due to that resonant behaviour at this
frequency return loss is minimum sorry maximum so we have very less reflection from the input
of the switch and also isolation is maximum Now one problem with FET or CMOS switch is
that whenever we are using them we are considering only small signal equivalent circuit and switching
between the cut off state and linear region But let us say one FET switch already it is
an cut off region we are expecting that switch will be in OFF condition but the applied electromagnetic
signal magnitude is so high sometimes it can turn on the FET so we cannot use it for very
high power applications so this is the limitation for FET switches So there is an alternative
that we can use stacked FET configuration so here is one example stack FET device So
instead of one we are using 3 FETs in parallel shunt configuration so whatever voltage we
are applying it is now being divided into 3 components so that is how we can improve
linearity of the given FET switch but only one problem is that conventional CMOS fabrication
procedure they cannot fabricate this type of stacked FET so we have to use some non
standard procedure and fabrication cost will increase Another important point when we go for high
power applications so what is the 1dB compression point of the switch So you remember that we
are assuming the switch as a linear device so here how we obtain 1dB compression point
this is first we have to plot the output power versus input power it should follow a straight
line if it is a linear device but for high power it starts to deviate from the straight
line and when this deviation is 1dB the corresponding input power we call the 1dB compression point
of the switch Here this is the plot for a typical FET switch and we see for this particular
switch 1dB compression point is that 40 gigahertz and it is coming for 3 point 6 dBm of input
power So here is performance comparison we have
different types of FET switches in HEMT technology in travelling wave HEMT technology so what
is travelling wave HEMT technology Here multiple HEMT switches they are used in Cascade connection
by using separate uhh by separate transmission line section so we use some transmission line
among the FET switches to connect them and that is why we call it travelling wave type
HEMT So if we compare these switches (per) performances we see the insertion loss is
lower for reference 3 and the blue ne reference 4 so mainly the HEMT and PHEMT they are better
performance compared to conventional FET switches Look at the isolation values again PHEMT switches
they are providing let us say 30dB isolation over uhh 60 to 70 gigahertz Typically switches
using standard 90 nanometre or 130 nanometres CMOS technology we use below 30 gigahertz
because of their very high figure of merit values Now we know different technologies we have
FET switches HEMT switches PIN diode switches so let us compare their performances So these
are the major performance comparison of some switches which are available commercially
in market So measured values for FET switches typically at 77 gigahertz is providing insertion
loss 2 point 7dB isolation 29 dB and we see figure of merit is 264 femtoseconds For gallium
arsenide HEMT at 60 gigahertz we see figure of merit is 324 femtoseconds but for PIN diode
it is just 30 femtoseconds so PIN diode among these 3 is a better performer Here are some
more examples in all the almost all the cases we see that PIN diode is providing lower insertion
loss as well as better isolation Now what are the challenges associated with
electronic switches So the first example is the maximum power handling compatibility P
max it can be given by E g to the power 4 into V s square by f square where E g this
is the band gap energy so higher is the band gap energy maximum power handling capability
is higher V s is the saturation velocity if the saturation velocity increases in that
case also we can increase power handling capability and it decreases with increasing frequency
so different materials will provide different maximum power handling capability Now if we
use PIN diode although in previous chart we have seen that PIN diode is providing minimum
insertion loss and maximum isolation even at 100 gigahertz and beyond but the main problem
with PIN diode is the biasing current For PIN diode typical bias current it can
be as high as 10 mili ampere or even more than that Now consider we are using a phase
carry antenna or similar type of applications where we have hundreds of switches so it will
consume very high DC power so it is a problem for PIN diode Next for FET FET the problem
is its low power handling capability for high power applications we cannot use FET switches
but the advantage of FET switch is its low power consumption compared to other switches
Similarly gallium nitride or gallium arsenide switches although they provide better figure
of merit compared to FET switches but we face integration problem high contact resistance
so finally we have to provide some contact switch by using some sort of metal then third
order intermodulation so it can generate high frequency components and high negatives pinch
off voltage typically for gallium nitride based switches So here are some examples of some fabricated
switches these are some photographs left hand side shows millimetre wave CMOS switches so
this top layer this spiral they are actually inductors we cannot replicate inductors inside
chip we have to fabricate inductors on top of the chip so you can see this spiral they
represents the inductor So not only inductor we need also capacitors so inductors and capacitors
they use for RF and DC blocking Right hand side picture it shows several PIN diodes with
their packages so they are based in rectangular waveguide technology inside the rectangular
waveguide PIN diode is housed you can see for a given component this small aperture
it shows the input line we have output line on the other side and the semiconductors they
are actually using for the biasing and for control signal These are some name of the
companies different types of switches are available from these companies so we will
take a break then we will start millimetre wave propagation

Leave a Reply

Your email address will not be published. Required fields are marked *