First off, let's remind everyone
that microwave electronics are by and large an analog science,
as opposed to most other electrical engineering, which has mostly
gone digital. We think of analog as real life, and digital
as the "reality TV" of electronics. No one really listens
to digital music or sees digital television, your senses are analog.
Digital communications must be carried on an analog radio signal.
Analog engineering will never go away.
If we had to summarize what sets
a microwave engineer apart from a "normal" electrical
engineer, we'd say that knowledge of just a few simple concepts
is required to fit in with microwave geeks. These are S-parameters,
the Smith chart, decibels, transmission
lines (including waveguide, which
really isn't a transmission line but performs the same function)
and skin depth. Notice that we didn't
mention antennas, because we consider
that almost a separate subject from microwave engineering! The funny
thing is that you can be expert on all of these without any advanced
math or even a college education, but without a college degree it
will be difficult to ever land a job as an engineer in this industry.
Here's a great list of colleges that
offer education in microwave engineering!
You'll learn about all of these
concepts and more, starting here on this page. Thankfully, there
is a ton of electronic design analysis software
out there that does all of the heavy lifting for you.
Here's a Microwaves101 piece
of advice: if you want to succeed in this field (or any other) the
most important thing is to love your work. Nothing trumps enthusiasm,
not even large brains. If you don't find microwaves particularly
interesting, go do something else. If you don't find any type of
work interesting and don't care what people think of you, become an investment banker and live off everyone else's hard work!
Here in the United States trade unions provide thousands of employees that perform much of the tedious work that engineers would not be good at (and perform it enthusiastically), and we thank them for their daily contributions and wish them luck on their next collective bargainization (if that is a word!) Be respectful of them and they will be nice to you. Thanks to Lyle for reminding us about the union label!
Look for the Union Label!
Will the U.S. microwave industry
be affected by outsourcing to developing nations, like the IT industry
was? We don't think so, for two reasons. A vast part of the microwave
industry is related to defense work and infrastructure (think
towers). Outsourcing these would make as much sense as outsourcing
your local fire department.
The second reason that microwave
engineering is a good field for US college students to consider
is that producing complex hardware takes a much higher level of
investment than running a call center: initial quality problems
can limit the worldwide microwave business of developing countries
even if they can offer lower pricing. Below is a picture of a simple
wrench the "Pittsburgh Forge" produced in India and sold
in the United States. One knuckle-bashing experience with hardware
like this will steer the consumer back to Craftsman (made in America)
tools. On the other hand, it is just a matter of time before emerging
countries develop microwave capabilities that they can be proud
of. When we wake up from the current recession, it is a safe bet
that many compound semiconductor fabs in the United States will
be gone, with Asia aggressively taking a large piece of that part
of the microwave industry. And in the not-too distant future you
may be driving a Tata!
Microwaves components and systems
are a multi-10-billion dollar industry, how's that for a vague data
point? The design community is small, perhaps only about 50,000
to 100,000 people in the US consider themselves in the microwave
field (this estimate is based on observed attendance at the annual
IEEE IMS symposium, factored for how many of us have to stay home
and do real work).
What are the "big three"
applications of microwaves in everyday life?
The most well-known remote sensing
systems are radars (radio direction
and ranging), which use a transmitter to illuminate an object, and
a receiver to detect its position or velocity (or both).
Another class of remote sensing
is radiometry. Radiometric
systems need no transmitter, they merely collect naturally-occurring
electromagnetic energy and process its to form images. Terahertz
radiometric receivers will soon be employed as security systems
in airports, provided that the ACLU will permit us all to be seen
in the nude by quarter-inch-brow security
guards. Another excellent example of remote sensing is the new "T-ray"
imaging being done at terahertz frequencies, by companies such as
Radio astronomy uses uses huge
dishes to capture incredibly weak RF signals from space to reconstruct
the origins of the universe starting with the big bang. We now have
a page on this topic!
Let's lump in global-positioning
systems into remote sensing, because a GPS unit "senses"
where it is.
Countermeasures to remote sensing
include all types of jamming equipment, usually associated with
military applications. Interested in electronic countermeasures?
Consider becoming an Old Crow!
We will also lump RFID in as
a use of microwaves to perform sensing.
Communications systems include
satellite, radio, television, wireless phone and data transmission
applications, and all combinations of these. We'll get into these
later... or sooner, if someone sends us some material!
Directed energy weapons will
eventually make up a new category of microwave applications. This
includes the Pentagon's new pain
ray, as well as high-power microwave (HPM) systems that can
be used to defeat weapons such as missiles and even disable ground
vehicles (with the exception of diesel engines which have no ignition
Here's some great info on the
pain ray from wired.com, in case you wanted to know about its effects
on dogs, sunblock, and drunks!
RF lighting is a relatively new
topic for microwave engineering. The sulfur lamp uses a 2.45 GHz
magnetron to excite sulfur to give up an eye-pleasing spectrum of
light. We've started a page on this topic
versus commercial applications
We often divide microwave technology
based on commercial or military/aerospace applications. The mix
of people in microwaves is roughly half in commercial applications,
and half in military/aerospace. Everyone knows that people who work
in military/aerospace microwaves generally are more manly than their
Commercial applications of microwave
technology include the front-end of much of the wireless stuff you
use everyday, such as cell phones, pagers, wireless LANs, satellite
television, XM Radio, and that cool GPS playtoy you received on
Father's Day. Unfortunately the boom years of commercial microwave
technology seem to be behind us, as the telecom infrastructure was
overbuilt, while competition drove the price of wireless phone services
into unprofitable territory. Who knows, videophone and Bluetooth
tricks may eventually bring some real money back to this industry.
Doesn't everyone want to be able to buy a pack of gum from a vending
machine by clicking a few buttons on their cell phone? You can do
this in Finland right now thanks to Nokia! We're not holding our
breath for a lot of new gadgets to take hold here in the USA, the
Second Bush Recession still has a few more years to run out.
Military, aerospace applications
probably account for more research dollars than commercial stuff.
It's arguably a lot more fun to work in this arena, where cost is
often NOT as big a consideration as performance. How would you rather
spend your career, with a team of 100 engineers trying to shave
the cost of a $20 cell phone by one buck in six weeks, or with a
team of four engineers designing a million-dollar electronic warfare
pod in six years?
Perhaps the coolest microwave
development programs are sponsored by DARPA, the Defense Advanced
Research Projects Agency. Here's a page
that reviews some of their current work.
If you want to get into the U.
S. defense industry, chances are your employer will need to get
you a security clearance, granted from the Defense
Security Agency (DSS). This takes some time (perhaps six months),
and if you were born outside the country, or have been convicted
of a crime, or have declared bankruptcy, it might be better to rethink
your career choice. Although the DSS might ask you to sign something
that will permit them to use a polygraph to check out your background,
we've never heard of it being used. They will certainly ask you
if you have used illegal drugs, but chances are they will overlook
your use of weed during college, or the defense industry would lose
of all candidates. They don't care about your sexual orientation,
and won't ask about your religious preference.
So what's a microwave? There
is some controversy about the exact frequency limits. We define
it as an electromagnetic wave between 300 MHz and 300 GHz, in agreement
with Pozar's Microwave Engineering, which
allows microwave engineers as broad a stake as possible in the EM
spectrum. Below 300 MHz is called very high frequency (VHF,
thanks, Chris!), above 300 GHz you are into the sub-millimeter-wave
spectrum. Terahertz frequency means 1012 cycles per second,
approaching infrared radiation. Yikes!
Here's a separate Microwaves101
page that provides a table of frequencies used by different systems,
such as police radar, XM radio, automotive radar, etc. Check
Speaking of frequencies, you
will often encounter frequency-band letter designations within the
microwave field. Although the letter bands are considered obsolete,
you should memorize some of the more common designators (such as
the IEEE standards shown below) if
you ever want to be a Microwave Good-old Boy.
The following distinction between
millimeter-waves and microwaves is almost universally accepted:
frequencies with free-space wavelengths less than one centimeter
but greater than one millimeter are referred to as millimeter-waves.
Thus, the millimeter-wave spectrum starts at 30 GHz, and runs to
300 GHz, where the wavelength in free-space is less than one millimeter.
Welcome to the sub-millimeter-wave band, you are on your way to
infrared radiation and terahertz frequencies.
Some microwave engineers have
a fear of millimeter waves, thinking that solving problems is harder
at higher frequency. This is irrational thinking, millimeter-wave
hardware requires nothing new, the components are just smaller.
Let's illustrate the concept by comparing the rock group Kiss, versus
the midget tribute band Tiny Kiss:
Yes, they serve the same purpose,
but one is smaller. The details are all there, even the guy who
plays Chaim Witz (a.k.a, everyone's favorite band member Gene
Simmons) is the tallest. Got it?
Bandwidth is a measure of how
much spectrum your microwave system can respond to. Bandwidth is
often given in megahertz or gigahertz, calculated from from a low
frequency FL to an high frequency FH, the
bandwidth is given by (FU-FL). Bandwidth is
expressed a number of other ways, which we will define here:
Three-dB bandwidth: for
a network that has a non-ideal frequency response (which includes
all physical networks), the three-dB bandwidth is where the
transmission coefficient S21 falls off from its highest peak by
three dB. Similarly, you could describe a network by its two-dB
or one-dB bandwidths.
for a system that works from a low frequency FL to an
high frequency FH, the percentage bandwidth is given
by 100%x(FH-FL)/FC. FC is the
center frequency, equal to (FH+FL)/2. Note
that it is possible to have more than 100% bandwidth by this definition;
an amplifier that works from 100 MHz to 10 GHz has a bandwidth of
200%. There is additional explanation about percentage bandwidth
on this page.
this is a measure of how wide a spectrum a system can respond to,
without any type tuning. Using the analogy of radio, the IF bandwidth
in an American FM receiver is about 200 kHz, which is necessary
to pass the full spectrum of a broadcast FM signal. The demodulator
processes this bandwidth to obtain the approximately 18 kHz baseband
bandwidth. The "despreading" effect of this processing
results in the superior signal to noise ratio enjoyed by FM transmission.
(Thanks for the correction, Miles!)
Tunable bandwidth: tunable
bandwidth is a measure of how wide a spectrum a system can respond
to with the user allowed to change settings such as local oscillator
frequency. For a receiver, the tunable bandwidth is almost always
more than the instantaneous bandwidth. An AM radio has a tunable
bandwidth of 540 kHz to 1600 kHz, or over one MHz of bandwidth.
This is about 100X its instantaneous bandwidth.
What does octave bandwidth
mean? It implies that the the upper frequency of operation is double
the lower frequency of operation, for example, an amplifier that
works from 2 to 4 GHz has one octave bandwidth. The origin of the
word octave goes back to music theory, where an octave is an interval
of eight notes in the major scale. For reference, the interval from
middle C to high C on a piano is an octave; high C is double the
audio frequency of middle C.
A device with an octave bandwidth
always has 67% bandwidth (do the math for homework!)
A fundamental problem in electromagnetics
is that for a signal to be radiated into free space, an antenna
must be on the order of 1/10 or more of a wavelength. Thus transmitting
voice without some type of upconversion would require a 30 kilometer
antenna for a 10 kHz signal! Thus, baseband signals need to ride
on carrier waves, which are at RF and microwave frequencies. Mixers
are the devices that are used to convert from one frequency to another.
Upconversion means you are increasing the frequency of your signal,
and downconversion means you are decreasing it.
A harmonic frequency is 2X, 3X,
4X, etc. the frequency of a signal. Why is it called a harmonic?
Because in music, harmonic frequencies of 2X, 3X, 4X sound good
together (they are harmonious, like the Del
Vikings). 2X and 4X frequencies are octaves, 3X is an octave
plus a perfect fifth.
A subharmonic frequency
is one that is 1/2, 1/3, 1/4 of a signal.
What is a transmission line?
Here's our definition: it's any conducting structure that supports
an electromagnetic wave "in captivity". Most transmission
lines use two conductors, where one is considered ground. This includes
coax (the outer conductor is ground), microstrip and stripline.
The transmission line that does not use a pair of conductors is
waveguide. By the way, we are talking about lossless transmission
lines here, or at least near-lossless. We have an entire chapter
devoted to transmission lines, click
here and we'll take you there.
What's a "substrate?"
It is the insulating material that support the the transmission
lines. In microstrip and stripline, the substrate is the dielectric
slab onto which the strip conductors and groundplanes are plated
When microwave engineers talk
about a "fifty-ohm system", what does that mean? A common
misconception is that if you placed an ohmmeter across the ground
and conductor of a fifty-ohm coax cable, you would always read 50
ohms. This is not the case, here's what we're talking about: transmission
lines have two important properties that depend on their geometry,
their inductance per unit length, and their capacitance per unit
length. The "characteristic impedance" of a system is
calculated from the ratio of these two:
where L' is the inductance per
unit length and C' is the capacitance per unit length. Note that
higher inductance translates to higher impedance, and higher capacitance
translates to lower impedance. Notice also that the units of length
don't matter, since they are "lost in the sauce". The
units of inductance and capacitance must be self-consistent, such
as pico-henries/foot and pico-farads/foot.
How do you know the inductance
and capacitance per unit length of a particular transmission line?
Who cares, when this has all been calculated for you about a million
times already and plenty of software exists that will calculate
it for you. The thing you should care about is what parameters within
a transmission line geometry control the relative capacitance and
inductance per unit length, so you get a feeling for what controls
Let's start with coax cable.
The inductance per unit length is mainly attributed to the diameter
of the center conductor. Decrease this diameter (keeping everything
else the same) and you will increase the inductance. This also raises
the characteristic impedance, referring to the equation above. Filling
the cable with a material of higher relative dielectric raises the
unit capacitance, and lowers the line impedance.
Another example: microstrip.
Here unit capacitance and inductance are inexorably linked together;
widening the microstrip line decreases its inductance while it increases
it capacitance. Hence, wide lines are always lower in impedance
than narrow lines for a given substrate height. As with coax, the
dielectric constant of the substrate has a big effect on capacitance;
using a higher dielectric substrate will yield a lower impedance
line, all other things being equal. So it is important not to mix
up your Rogers Duroid materials, once your circuit is etched it
is pretty hard to judge the dielectric constant from color and texture
The exact characteristic impedance
of free space is 120
ohms, which is approximately 377 ohms. Why? This is explained (or
should be) on our page on characteristic impedance.
Impedance matching of source
and load is important to get maximum power transfer. If you have
a 75 ohm load, you don't want to drive it with a 50 ohm source,
because it is inefficient. You can learn more about the simple math
behind maximum power transfer by clicking
Simple impedance transformation
can be done using quarterwave transformers. Click
here to go to our main page on quarter-wave tricks!
constant and effective dielectric constant
is another way to say "relative permittivity". Check out
our separate page on permittivity
for more info on this subject. Although some people use the phrase
"relative dielectric constant", this is incorrect, akin
to saying "deja vu again".
Remember back to your physics
class, when you learned that dielectric constant is used to calculate
the value of a capacitor? The higher
the dielectric constant, the higher the capacitor value. For an
ideal parallel plate capacitor, the capacitance is calculated by:
is the permittivity of free space (thanks, Maarten!), R
is the relative permittivity (the dielectric constant) of
the material between the plates, A is the area of the parallel plates,
and D is the distance they are separated. Technically for this expression
to be 100% accurate, the material surrounding the plates must be
of the same relative dielectric constant R,
but this induces only a small error in the calculation under most
is equal to 8.854x10-12 Farads per meter (you should
commit this to memory). Most often it is the dielectric constant
that is most important in microwaves.
For electromagnetic radiation,
the permittivity of the medium that the wave is propagating in is
equal to R0.
In a vacuum or in dry air, R
is equal to unity, and the signal travels at the speed of light.
All electromagnetic energy, from 60 Hertz power that your electric
company sells you, to signals that the latest Mars satellite returns
to earth, travels really, really fast. In a vacuum, the speed of
light, denoted "c" in textbooks, is 2.998 x 1010
centimeters/second (thanks, Jared!) , or 2.998 x 108
meters per second, or about 186,000 miles per second, which puts
the moon about 1.5 seconds away by radio.
The dielectric constant of a
material can be used to quantify how much a material "slows"
an electromagnetic signal. The velocity of the signal within any
transmission line that is 100% filled with a material of dielectric
is computed by:
So if your stripline or coax
transmission line is fabricated on a material with dielectric constant
2.2, the velocity of propagation is only 67% of the speed of light
in free space. Similarly, because wavelength is proportional to
velocity, the length of a quarter-wave transformer is also 67% of
what it would be in free space. Thus one of the tricks of reducing
the size of microwave components is revealed; by using materials
of higher dielectric constant, distributed structures can be made
smaller. One of the advantages of using GaAs
for microwave ICs (known in the industry as MMICs)
is its dielectric constant of 12.9, which is appreciably higher
than ceramics such as alumina, and most soft substrates.
very good rule of thumb is that electromagnetic radiation in free
space travels about one foot in one nanosecond; a more exact value
is 0.983571 feet per nanosecond. This slows to about 8 inches per
nanosecond for coax cables filled with PTFE
(almost all coax cables are filled with PTFE, or a combination of
PTFE and air.) For more information please see our discussion of
This brings us to the subject
of "effective dielectric constant". In transmission lines
realized in microstrip media, most of the electric fields are constrained
within the substrate, but a fraction of the total energy exists
within the air above the board. The effective dielectric constant
takes this into account. The effective dielectric constant of a
fifty-ohm transmission line on ten mil alumina is a number somewhere
around 7, which is less than the dielectric constant of the substrate
bulk material (9.8). Another example of an effective dielectric
constant is if you were to create a stripline circuit using two
sheets of substrates with different dielectric constants. To a first
order, the effective dielectric constant would be the average of
the two materials' dielectric constants. A third example is coplanar
waveguide transmission lines with air above the substrate. Here
the effective dielectric constant is very nearly the average of
the substrate dielectric constant and one (the dielectric constant
of air=1). Thus the effective dielectric constant of CPW circuits
on GaAs (R=12.9)
is approximately 6.5.
When the behavior of a resistor,
capacitor, or inductor can be fully described by a simple linear
equation, microwave engineers refer to it as a lumped element. For
example, a 50-ohm resistor at low frequencies will obey Ohm's law
(V=IxR). Put five volts across it and it will draw 100 milliamps
of current. "Lumped elementhood" is restricted to components
that are operate at frequencies where they are physically much smaller
than a quarter-wavelength. For example, axial-leaded components
perform well up to 10s of MHz, but at one GHz, chances are that
an axial-leaded resistor is closer to an open circuit, or a lousy
inductor, rather than an ideal resistor. This is why you will rarely
be asked the resistor color code as a microwave engineer!
At microwave frequencies, other
factors must be considered. To accurately calculate the behavior
of that same 50-ohm resistor, you need to consider its length, width,
and thickness of metal (due to the skin effect), and its proximity
to the ground plane. This is when we must consider it as a distributed
By designing really tiny parts,
you can often consider them lumped elements, even at microwave frequencies.
You must keep the critical dimensions (such as length and width
of a thin-film resistor) small compared to an electrical quarter
wavelength. For example, if you are designing a 50 ohm microstrip
load resistor at X-band, on an alumina substrate (dielectric constant
9.8), a quarter wavelength is approximately 120 mils. You'd better
keep both the length and width of the resistor to less than 40 mils,
or you else you have to spend some time with a EDA
simulation tool such as Agilent ADS or Eagleware Genesis evaluating
the performance. Where else but microwave engineering can you make
a project out of designing a stupid fifty-ohm resistor?!
Yet another rule of thumb: to be considered a "lumped element",
no feature of the structure can exceed 1/10 of a wavelength at the
maximum frequency of it usage.
At low frequencies, the metal
that connects components together is treated as an ideal connection,
with no loss, no characteristic impedance, and no transmission phase
angle. When interconnects become an appreciable fraction of the
signal wavelength, these interconnections themselves must be treated
as distributed elements or transmission lines. An extreme example
of the need to consider the distributed properties of transmission
lines is when we are dealing with a quarter-wavelength. At this
electrical length (90 degrees), an open circuit is transformed to
a short circuit, and a short-circuit is transformed to an open circuit!
Think about this: a short-circuited 90 degree "stub" hanging
in shunt off of a transmission line will be invisible to signals
propagating down the the transmission line, while an open circuited
90 degree stub shunting a transmission line will cause a short circuit
and the propagating signal will get hosed! A whole lot of microwave
engineering exploits this concept, so you'd better understand it.
One "classic" distributed
element is the quarter-wave transformer (we've written an entire
chapter on this and other quarterwave
tricks! The quarterwave transformer is used to shift the impedance
of a circuit by the following simple formula:
where Z2 is the characteristic
impedance of the transformer, ZL is the load impedance,
and Z0 is the characteristic impedance of the system
you are trying to maintain. Do you detect a pattern? Most of the
equations on this page use the square-root function... perhaps they
put that button on your Casio calculator for a reason!
VSWR stands for voltage standing
wave ratio. It is a measure of how well a network is matched to
it's intended characteristic impedance (Z0), which is
almost always 50 ohms in microwave engineering. Return loss is just
another way to express the same thing. Both are used in microwave
engineering, that's just to keep you on your toes.
VSWR dates back to the days when
a "standing wave meter" was an important piece of lab
equipment. Long before you could buy a network analyzer for measuring
how well a part is impedance matched, the standing wave meter was
used by engineers to evaluate the same problem. A small probe was
inserted into a waveguide, the output of which was rectified, producing
a current or voltage proportional to the electric field with the
waveguide. The engineer would pull the probe longitudinally along
the waveguide, in search of local maxima and minima readings. These
are due to the standing wave within the transmission line. The ratio
of the maximum to the minimum voltage recorded was known as the
voltage standing wave ratio (VSWR). To this day VSWR is often used
to quantify how well a part is impedance matched. Always expressed
as a ratio to unity, a VSWR of 1.0:1 indicates perfection (there
is no standing wave). A VSWR of 2:1 means the maxima are twice the
voltage of the minima. A high VSWR such as 10:1 usually indicates
you have a problem, such as a near open or near short circuit.