EE251 Lab Package: Winter 2025

Lab 1: Microstrip Structures

1.0 Motivation
In this assignment you will review the basics of microstrip transmission lines and a variety
of useful structures that can be readily made on FR4 circuit boards with adhesive copper
foil tape and BNCs. You will study the use of standard RF measurement equipment, such
as the vector network analyzer (VNA) and the time domain reflectometer (TDR), giving
you the bench time necessary to build up a proficiency with the instruments.

2.0 Background
Although the textbook and other materials contain numerous equations, here are a few
selected ones that you will likely find most useful for the lab exercises.
• The characteristic impedance of a microstrip line of width W separated from a ground
plane with a lossless dielectric of thickness H is reasonably well approximated by

1 - ----
L- = µ 0 ------------ H 1
Z0 = --- ⋅ - ⋅ -----------------------------------------------------------------
– 0.836
-, (EQ 1)
C εr ε0 W – 0.0724  W 
1 + 1.735 εr -----
 H
where εr is the relative dielectric constant of the board material, which generally varies
from batch to batch, and with frequency, as mentioned in class and in the text.
• The wavelength of a sinusoidal signal traveling along a microstrip line is

λ εr
λ = -------0- ⋅ -------------------------------------------------------------------------
- , (EQ 2)
εr 0.1255
1 + 0.63 ⋅ ( εr – 1 ) ⋅  ----
W
  H

where λ0 is the wavelength in vacuum. The formula is most accurate for W/H > 0.6.
That inequality is reasonably well satisfied for most practical lines.
• Fringing capacitance per width W of a wide open-circuited line is approximately

H
----
C 2 εrε
----- = εr ε0 ⋅ ---- = ---------0 . (EQ 3)
W H 2

Notice that H ultimately cancels out of the equation. We will later revisit this equation to
provide a first-order correction for fringing.

Lab 1: Microstrip Structures EE251 Winter 2025 1

 EE251 Lab Package: Winter 2025

3.0 Experiments
We advise that you carry out the following exercises in the order presented. Document
your steps carefully in a lab notebook, and save the completed circuit boards for use in
future experiments. Your documentation should enable you to explain your designs,
present key measurement results (including TDR/VNA plots), and otherwise demonstrate
your quantitative understanding of the observations including, most importantly, any dis-
crepancies.

3.1 50Ω lines

Mount two BNC bulkhead connectors on an FR4 board, as demonstrated during the lab
tutorial. You may have to abrade the connector surface with sandpaper before soldering, as
BNCs from some manufacturers have a solder-resistant coating on them. Cut out a strip of
copper tape of the width required to produce the impedance of 50Ω, and mount it on the
board. Useful information:
• FR4 boards that we use (this week, made by some unknown manufacturer) have a rela-
tive dielectric constant εr around 4.5 at low frequencies (but your mileage may vary).
The boards are 1/16” thick (1.6mm) and are plated with 1oz. of copper per square foot
(thickness t = 1.34 mils, or about 34µm, usually rounded to 35µm).
• You will probably discover that W = 3mm produces about the right Z0 (49.3Ω, assum-
ing εr = 4.4). Copper tape of 1/8” gets you close enough for most purposes. Care in this
step will reap big rewards later on, so try to cut as evenly as you can. [If we run out of
1/8” width Cu tape, manually slice -- carefully -- 1/4” tape.]

Examine your transmission line using the

TDR, with one of the line ends terminated in
a 50Ω BNC load. Observe and compensate
for the inductance associated with the con-
nector-board transition.

Before you characterize the line do not for-

get to calibrate the VNA (see the attached
notes). Measure S11 and S21 over a broad fre-
quency range. From an S21 plot (dB vs. fre-
quency) measure the loss of 50Ω lines on
FR4. Loss is typically ~0.1 dB/inch/GHz (the
plot on the right is for a 10” line), but again,
your mileage may vary. Note that this implies
a constant loss of about 0.5dB per wavelength. If your plot differs significantly from that
shown in the figure, double-check your work.

Explore the reflections produced when you load the line with an open and then a short. To
short the line you will have to extend the ground plane to the line-side of the board. Con-
structing a good ground connection is an instructive exercise in its own right. You’ll find

Lab 1: Microstrip Structures EE251 Winter 2025 2

 EE251 Lab Package: Winter 2025

that it isn’t easy (and you should be able to explain why). Making a good ground connec-
tion is a valuable skill to acquire.

3.2 90o corners and mitered bends

The world rarely allows you to lay out
lines without requiring a turn some- l
where. To explore how these behave,
construct four consecutive 90o turns
as shown in the picture. Using the
VNA and/or TDR determine whether
W
the corners look capacitive or induc-
tive.

Then, try to minimize impedance dis-

continuities by trimming the corners as illustrated. Trim the corners one by one, beginning
with the one closest to the input, since filtering by a discontinuity reduces your ability to
resolve the behavior of downstream discontinuities.

Measure the length, l, of the cut-out section (see diagram). Compare your measured l to
the textbook value of 1.8W.

3.3 Inductive notches

A notch in a transmission line is a convenient way to obtain inductance (if that’s your
intent). Make a small notch and measure the resulting inductance using the TDR. You will
probably observe something like the TDR trace shown here:

l TDR:
a

w a = 50mV
s
s = 8V/ns
L= 0.63nH

Methods that rely on edge-speed measurements are usually too sensitive to various
sources of error, so use the integral method, rather than the derivative method, to estimate
reactance. This advice holds for both inductive and capacitive transients. To validate this
advice, cross-check your computations using the derivative method. Comment on the
likely sources and magnitudes of error for both approaches, and explain why the integral
method should yield more reliable values. (Or convincingly refute our advice!)

Lab 1: Microstrip Structures EE251 Winter 2025 3

 EE251 Lab Package: Winter 2025

3.4 Capacitive bulges

Explore the effect of a modest widening in the microstrip line.
First estimate the additional capacitance using the parallel
plate capacitance formula. Then use the TDR to determine the
capacitance experimentally (again, the area method is gener-
ally more accurate, but feel free to explore alternatives).
Determine whether a correction for fringing capacitance is
needed. Does the capacitance arise from the flanges above and
below the line, or from the entire rectangular piece in the center of the picture?

Hint: Do not forget that the capacitor is shunted by two 50Ω resistors in parallel when the
line is properly terminated. You should be able to explain why this is so.

3.5 Halfwave filter

The previous exercise deals with one particular manifestation of a more general effect: A
discontinuity can in fact behave as a bandpass filter, passing signals when the length of the
discontinuity is an integer multiple of a half-wavelength:

2l
λ p a ss = ----- , m = 1, 2, 3… . (EQ 4)
m

The short widening that we examined in the

previous section is a bandpass filter for very
high frequencies and, as you know, a resona-
tor approximates a pure inductance or capaci-
tance far from center frequency.

The text expands on this idea, and describes

how a periodically loaded structure can be
used as a lowpass and bandpass filter. Con-
struct a 1GHz bandpass filter of this type.
Keep in mind that the half-wavelength of a
1GHz signal on FR4 is very roughly of the
order of 75-80mm, or about 3”, so you don’t
have space for too many sections on our stan-
dard boards. Do your best.

Measure the –3dB bandwidth, maximum stopband rejection, and insertion loss. Explain
why the insertion loss is as large as it is.

Although we will study filters in much greater detail later, it should seem intuitively rea-
sonable that the wider the wide section, the greater the mismatch at frequencies where the
filter is to reject signals and, hence, the better the filter action. In future labs you will con-
struct a receiver front end, consisting of an antenna, filter and LNA, so try to begin acquir-
ing some insight about filters now.

Lab 1: Microstrip Structures EE251 Winter 2025 4

 EE251 Lab Package: Winter 2025

4.0 Appendix: Network Analyzer Calibration

Depending on how abused a given VNA has been, calibration can be a lot like a visit to
Monte Carlo (or any of a number of vending machines around the campus). You should
get in the habit of checking calibration quite often (“If you get S21 > 0 dB for a passive
network, it’s time to recalibrate the VNA!”). Inability to hold calibration over a short
period of time (e.g., minutes) is most often a sign of inattention to connectors, including
pilot error and poor connector maintenance. A common noob error is to tighten both
halves of an APC-7 connector. As emphasized in class and in the lab tutorial, ONLY
ONE half should be tightened. The other half should rotate freely. Tightening both will
force the center conductors away from each other, resulting in a poor contact.

Calibrate the instrument with the calibration kit, which includes reference shorts, opens,
and a 50Ω transmission standard. Calibration standards and connectors are quite
expensive (thousands of dollars!), so please be careful and treat them with respect. In
particular, don’t leave them on the benchtops, where they will roll off onto the floor,
get dented and dusty, and disappear into oblivion. Put them back in their cases as
soon as you are finished with them.

We were once forced to perform calibrations using only SMA standards (because these are
much more common), even though all of the experiments use BNC connectors. Happily
we now have BNC standards, one of which is from a trusted vendor, and another that I
have made by hand. Be aware that the random BNC 50Ω terminators floating around the
lab are NOT to be used in calibration, as they almost always contain a through-hole resis-
tor with long leads. If, for some reason, the correct BNC 50Ω standard has been mis-
placed, you may use an N-type 50Ω standard (which we do have) for the particular step in
the calibration suite calling for a reflection-free termination. The result is more than good
enough over the frequency range that one would use a BNC.

Since the phase of a signal changes as it propagates through the cables/connectors, you
might need to perform a mathematical port extension if you are using fixturing that
extends beyond the plane where the calibrations were performed. Without dialing in a port
extension value an open will have Γ = ejβ(f)x, where x is the cable length, rather than Γ = 1.
One performs port extensions after the main calibration, using the built-in function of the
VNA. Large port extension values are to be avoided, because such purely mathematical
corrections necessarily fail to take into account the actual physical structure of the setup.
During the calibration step, you should include as much of the final setup as possible, and
use a post-calibration port extension to perform only the tiniest of tweaks. Typically, port
extensions are used to compensate for slight differences in the length of connectors used in
the DUT, versus those for the calibration standards. If you are using this capability to com-
pensate for meters of extra cable, etc., STOP. Instead, perform a calibration with those
meters of cable in place, then use small port extensions if necessary to fix up what should
be 2nd-order remaining discrepancies.

Lab 1: Microstrip Structures EE251 Winter 2025 5

 EE251 Lab Package: Winter 2025

Lab 2: Narrow- and Broadband Amplifiers

1.0 Motivation
Having mastered the art of microstrip construction and measurement in Lab 1, you are
now ready to tackle the design of active circuits, specifically, two different types of ampli-
fiers. One is a narrowband amplifier that might be suitable as the first stage of a commer-
cial FM broadcast receiver, and the other is a highly useful general-purpose broadband
amplifier whose basic topology is found in numerous commercial products.

2.0 Narrowband (single-tuned) amplifier

In this first part of the lab, you’ll design a tuned (bandpass) amplifier centered at 98MHz,
with a total –3dB bandwidth of at least 20MHz. The corresponding 88-108MHz passband
covers the commercial FM radio band. The input and output impedances should provide a
sufficiently good match to 50Ω to present a return loss (|S11| and |S22|) in excess of 15dB at
the center frequency, and in excess of 10dB at the band edges.

To constrain the design space to something manageable, you’ll use just one transistor, the
2SC2753 (see datasheets provided in the lab package). Within this constraint, maximize
the gain (|S21|), above a minimum value of 15dB.

If you elect to wind your own inductor, you might find it useful to recall Harold Wheeler’s
formula for single-layer cylindrical coils which, in SI units, is
2 2
µ 0 n πr
L ≈ -------------------- . (EQ 1)
l + 0.9r

As discussed in class, impedance transformations as part of matching may be provided by

tapping the inductor appropriately to produce the desired real part. Addition of an appro-
priate reactive element will remove any remanent reactive part.

Important reminder: Be sure that your amplifier is AC-coupled at both ports, as dis-
cussed in class. Otherwise the instruments could (will) upset your amplifier’s bias, and
your amplifier might damage the instruments.

3.0 Broadband amplifier

As demonstrated in the previous part of this exercise, a narrowband amplifier is capable of
high gain at high frequencies; the gain-bandwidth product “limit” applies to bandwidth,
not center frequency. For applications that may be served by narrowband circuits, tuned
amplifiers provide higher gain (or consume less power for the same gain). They also
reduce problems caused by out-of-band sources of interference and noise.

Lab 2: Narrow- and Broadband Amplifiers EE251 Winter 2025 1

 EE251 Lab Package: Winter 2025

Some applications, though, are truly broadband. This part of the exercise explores the
design of a versatile broadband gain block that uses negative feedback to stabilize gain
and port impedances (and also reduce distortion at the same time):

FIGURE 1. Shunt-series amplifier (biasing details and DC-blocking capacitors omitted)

Rout

vOUT
RF
RL
RS
vIN

Rin RE

As discussed in the book, this topology employs two types of negative feedback together.
Emitter degeneration provides one level of negative feedback, which acts with another,
provided by RF. In the usual incarnation of this gain block, the amount of emitter degener-
ation is sufficient to exert first-order control of the gain (thus stabilizing the gain against
variations). The feedback resistance, RF, then acts primarily to reduce the input and output
resistances. If the source and load resistances are equal, it happens that the input and out-
put impedances will also be approximately equal. This property is in addition to the usual
improvement in linearity provided by negative feedback. These qualities explain why so
many commercial products are based on this topology.

Design generally starts by choosing RE to set the gain, and then selecting RF to meet input
and output matching objectives. Here, you are to design for a gain of at least 15dB and a
return loss of at least 10dB at both ports. The objective is then to maximize the bandwidth
over which both sets of specifications are satisfied. You are encouraged to explore shunt-
peaking and capacitive bypass ideas to extend bandwidth beyond what would otherwise
be the case. Take note of the tradeoffs between achieving good broadband gain and match-
ing.

You should provide an open-circuit time constants hand-estimate of bandwidth for the
unpeaked amplifier. Consider only the capacitances inherent in the transistor itself (the
base-emitter and collector-base capacitances).

As a final recommendation, save even your unsuccessful prototype amplifiers. They may
prove useful in providing additional gain when assembling your complete system. Stu-
dents in prior years have often commented that doing so helps quite a bit.

Lab 2: Narrow- and Broadband Amplifiers EE251 Winter 2025 2

 EE251 Lab Package: Winter 2025

4.0 Appendix: Summary of bipolar transistor model

The textbook section on bipolar transistor modeling is useful if you haven’t had much
exposure to the bipolar transistor. Excerpted material is provided here for quick reference:

FIGURE 2. Basic small-signal bipolar transistor model

rb cµ
base collector
cπ rπ gm v π

emitter

The model is very similar to that for a MOSFET, as both are fundamentally transconduc-
tances (indeed, every active device that has driven the electronics age—vacuum tubes,
bipolar transistors, JFETs, MOSFETs, MESFETs, etc.—has been a voltage-controlled cur-
rent source). However the bias and geometry dependencies of the elements differ a bit.
Unlike a MOSFET, a bipolar transistor’s transconductance is solely a function—and a lin-
ear one, at that—of bias (collector) current. At room temperature, a 1mA collector current
always produces about a 40mS gm, independent of the dimensions of the transistor.

Other differences arise in the behavior of the parasitics. The base-to-emitter capacitance
results from two different contributions. One is a traditional junction capacitance, and is
thus solely junction-voltage dependent. In forward bias, the base-emitter voltage varies lit-
tle, implying a roughly constant junction capacitance. The other (and generally dominant)
component of base-emitter capacitance behaves quite differently; it is linearly propor-
tional to collector current and has only a second-order sensitivity to junction voltages.

The bipolar transistor has a resistor, rπ, in shunt with the base-emitter capacitance. This
resistance has a value β/gm. At 1mA collector current and a β = 100, rπ = 2.5kΩ. At fre-
quencies well above 1/rπcπ, the capacitance cπ dominates the base-emitter impedance.

The collector-base capacitance cµ is well-modeled as a pure junction capacitance. In nor-

mal operation, the collector-base junction is reverse-biased.

The undesired, but unavoidable, extrinsic base resistance rb is a weak function of bias. At
high collector currents, the base current is also large. The voltage thus drops as one moves
away from the base contact, assuring that the first moment of the base current distribution
moves closer to the base contact as current increases. The consequent reduction in average
distance between the contact and the current distribution causes a diminution in rb. It is
not uncommon to see a halving in base resistance under high-level injection conditions.
The extrinsic base resistance of the 2SC2753 transistor we use is around 7Ω under nomi-
nal conditions, and drops below 5Ω at high currents. Such modest changes are by them-
selves generally inconsequential except, perhaps, in the design of low-noise amplifiers
(about which we’ll say more later in the course).

Lab 2: Narrow- and Broadband Amplifiers EE251 Winter 2025 3

 EE251 Lab Package: Winter 2025

Lab 3: Downconversion Module

1.0 Motivation
By now you have developed familiarity with the design and characterization of active
bipolar stages. In this laboratory exercise, you will hone those skills further in the course
of designing an oscillator-mixer combination. The output will eventually feed an interme-
diate-frequency amplifier/filter that you will construct in a future laboratory experiment.

2.0 Mixer
This block accepts the amplified 88-108MHz RF signal and downconverts it to a
10.7MHz (nominal) intermediate frequency. Of the many possible mixer topologies one
could imagine employing, you will use a single transistor circuit that exploits the inciden-
tal mixing one obtains from a nonlinearity (specifically, the second-order nonlinear term
in the series expansion of the iC-vBE relationship). It is by no means the world’s best
mixer, but in terms of performance obtained per transistor, it is hard to beat. This type of
mixer dates back to the earliest years of the vacuum tube era, and it is still widely used in
consumer radios today.

As a suggestion, feed the RF into, say, the base, and the local oscillator (LO) into the emit-
ter circuit. You are certainly free to use other feed arrangements if you wish. Just consider
the pros and cons of each option and be prepared to explain the rationale behind your
choice. A simple LC tank in the collector circuit should be used to pass only the desired
difference-frequency IF component. The bandwidth of this tank just needs to exceed
200kHz in order to pass broadcast FM signals. In this case, we actually want a somewhat
large bandwidth to avoid distortion due to nonlinear phase, so a bandwidth of ~300kHz is
recommended. There is, of course, a tradeoff between achieving a large enough bandwidth
to avoid phase distortion, and having a low enough bandwidth to attenuate the undesired
sum-frequency IF component. That’s engineering.

3.0 Local Oscillator

You may select any oscillator topology you wish, but just to point you in a direction, con-
sider the Colpitts configuration. It has relatively few failure modes, and well-known fixes
for those modes.

We wish to tune over the 88-108MHz band, so your oscillator will also have to tune over a
20MHz range. Theoretically, you have the choice of placing that range above or below the
RF range (both will produce a difference frequency equal to the IF). However, placing it
above reduces the percentage frequency change represented by that fixed 20MHz range.
Such “high-side” injection is favored because design difficulty is a strong function of the
normalized tuning range you seek. In our case, with a 10.7MHz IF, the LO would need to

Lab 3: Downconversion Module EE251 Winter 2025 1

 EE251 Lab Package: Winter 2025

tune from 77.3MHz to 97.3MHz (if using low-side injection), or from 98.7MHz to
118.7MHz. The low-side choice obligates us to tune 87.3MHz +/– 11.5%, which implies a
total capacitance tuning range of about 46%. High-side injection reduces the tuning
requirement to something like a 37% change in capacitance. That difference may not seem
dramatic, but you will come to appreciate that the design difficulty is not a linear function
of that percentage. [Consider, for example, that making a power converter with 105% effi-
ciency is not a little bit harder than making one with 95% efficiency.]

The MMBV105G/MV105G varactors we stock in the lab (see the collection of datasheets)
can accommodate the tuning range requirement without resort to external tuning voltages.
We’ll make that a requirement: The tuning voltage should be derived from the supply volt-
age for the rest of the circuit.

4.0 Specs to be achieved

In addition to achieving the tuning range specified, your conversion gain (measured as the
IF power delivered by you, divided by the RF power supplied to you) should exceed
10dB. Both ports should exhibit a return loss of at least 10dB (yes, this is much easier than
15dB, by design). It may be difficult with the instrumentation in the laboratory to make
this measurement, so please stand by for additional announcements.

Within the constraints of a 12V supply, you are to minimize the supply current consumed
to achieve these specifications.

You are to measure the relative suppression of the undesired sum-frequency IF compo-
nent, compared to the desired difference-frequency component. You need not meet any
particular value of this suppression, but you should show that the measured value is expli-
cable from computations carried out on your circuit.

Note: Some students in the past have encountered difficulty by only simulating with val-
ues that correspond to an ideal circuit that they did not actually build. If you make this
error, you should not be puzzled at the existence of discrepancies! Common errors include
assuming that signal sources have zero Thevenin impedance, and failing to take into
account various parasitics (e.g., don’t treat wires as ideal, impedance-free interconnect --
take the lessons of Lab 1 to heart). Be sure to incorporate knowledge that some parasitics
may matter a great deal. Get into the habit of estimating parasitics and using those esti-
mates to inform your design and simulation methodology.

Lab 3: Downconversion Module EE251 Winter 2025 2

 EE251 Lab Package: Winter 2025

Lab 4: IF Chain and Demodulator

1.0 Intro
After having engaged in life-and-death combat with the intricacies of single-transistor
mixers and oscillators, you are probably ready for a little relaxation. In this penultimate
exercise, you will design an IF amplifier that employs a double-tuned filter, followed by
an FM demodulator. [In the final exercise of the course—system integration—you will
connect everything together and to an antenna and audio amplifier, to see if your radio
actually works.]

2.0 IF amplifier
This block is tuned to the same 10.7MHz intermediate frequency produced by the down-
converter you built in the previous exercise. To pass the full bandwidth required (as given
by Carson’s rule, to avoid excessive distortion) you will want to have a 3dB bandwidth of
at least 180kHz, and a monotonic (half–) passband. At the same time, the filter needs to
attenuate alternate adjacent channels (i.e., the ones 400kHz away) by a “significant”
amount. For this exercise, “significant” = 20dB, relative to the center frequency. [It should
be relatively straightforward to achieve 25dB rejection and, with care and nontrivial effort,
values approaching 30dB should be attainable.] In a commercial FM receiver, two or three
stages of such filtering would be employed, to provide alternate channel rejections of 50-
65dB, typically. We will not be quite so ambitious.

You are to wind your own IF transformer; that will probably be your biggest challenge,
time-wise. That will require you to determine what inductances you should target, as well
as their mutual coupling. Then you can use the formulas presented earlier to figure out
how closely spaced the coils should be. You should not have to iterate too many times to
converge on a workable design, because the specifications are sufficiently loose not to
require absolute perfection. In lecture, we will review how to determine the coupling coef-
ficient experimentally.

Your IF amplifier input should present a >10dB return loss over the passband in order to
connect gracefully to your downconverter. The output of the filter should similarly match
to 50Ω to that same level of quality, but largely for the purpose of easy, accurate character-
ization with our equipment. You will almost certainly want to use a transistor at the input
and output (e.g., a common-base input transistor, and a follower output transistor), and
embed your transformer in between.

The suggested demodulator is a slope detector/leaky peak detector, consisting of a diode,

resistor and capacitor. You will need to design this block to filter out the 10.7MHz IF,
while not significantly attenuating the baseband modulation (which can go to as high as
15kHz). Given the several orders of magnitude separating those two frequencies, you
should not encounter undue difficulty satisfying those requirements.

Lab 4: IF Chain and Demodulator EE251 Winter 2025 1

 EE251 Lab Package: Winter 2025

To avoid loading the filter too much, you can drive this demodulator with the same fol-
lower assumed to exist at the output to drive the instrumentation. The effect on S22 should
not be large. You may, if you wish, also replace the diode with the base-emitter junction of
a transistor connected as a follower. This connection implements something that is often
called an “infinite impedance” leaky peak detector because it combines the input imped-
ance-boosting properties of a follower with the nonlinearity of a diode.

The combined IF amplifier and filter should provide a minimum of 60dB of insertion gain
when terminated in 50Ω. Students in the past advise even greater gain, which is why we
advised that you save your leftover amplifier prototypes from previous labs (or borrowed
from fellow students). Commercial radio receivers provide overall end-to-end gains well
in excess of 100dB, so you see that achieving the bare minimum gains specified will give
you a radio that is quite insensitive.

Other hints: Given the impossibility of fabricating perfect coils while simultaneously
matching the parasitics and intended loading, achieving correct tuning by construction
alone is not a reasonable expectation. Plan on using trimmer capacitors where it makes
sense.

Lab 4: IF Chain and Demodulator EE251 Winter 2025 2