Contents - setup.repair
Contents
12.1 Characterizing Receivers
12.6 Pulse Noise Reduction
12.2 Basics of Heterodyne Receivers
12.6.1 The Noise Limiter
12.2.1 The Direct Conversion Receiver
12.6.2 The Noise Blanker
Project: A Rock-Bending Receiver for 7 MHz
12.6.3 Operating Noise Limiters and Blankers
12.3 The Superheterodyne Receiver
12.3.1 Superhet Bandwidth
12.6.4 DSP Noise Reduction
12.7 VHF and UHF Receivers
12.3.2 Selection of IF Frequency for a Superhet
12.7.1 FM Receivers
12.3.3 Multiple-Conversion in the Digital Age
12.7.2 FM Receiver Weak-Signal Performance
Project: The MicroR2 — An Easy to Build SSB
or CW Receiver
12.7.3 FM Receiver ICs
12.4 Superhet Receiver Design Details
12.4.1 Receiver Sensitivity
12.4.2 Receiver Image Rejection
12.4.3 Receiver Dynamic Range
12.5 Control and Processing Outside the Primary
Signal Path
12.5.1 Automatic Gain Control (AGC)
12.7.4 VHF Receive Converters
12.8 UHF and Microwave Techniques
12.8.1 UHF Construction
12.8.2 UHF Design Aids
12.8.3 A 902 to 928 MHz (33 cm) Receiver
12.8.4 Microwave Receivers
12.9 References and Bibliography
12.5.2 Audio-Derived AGC
12.5.3 AGC Circuits
Chapter 12 —
CD-ROM Content
Supplemental Files
• “Amateur Radio Equipment
Development — An Historical
Perspective” by Joel Hallas, W1ZR
Project Files
• Rock Bending Receiver PCB
template by Randy Henderson,
WI5W
• 10 GHz preamp PCB template by
Zack Lau, W1VT
• Binaural Receiver project by Rick
Campbell, KK7B
Chapter
12
Receivers
Receivers are discussed in this
chapter from the standpoint of their
architecture, composed of functional
blocks such as mixers and oscillators
and amplifiers that are discussed in
the other chapters of this book. Other
functions such as AGC that control
the receiver from outside the primary
signal path are dealt with both
functionally and with examples at
the circuit level. Special techniques
associated with VHF, UHF and
microwave receivers are reviewed.
The performance of receivers is
described by standard terminology
and as a set of measurements that
allow a wide range of designs to be
evaluated in ways that reflect their
behavior in actual use.
This chapter, originally written
by Joel Hallas, W1ZR, includes an
update of existing material from
previous editions of this book.
Joel’s summary of receiver and
transmitter development, “Amateur
Radio Equipment Development —
An Historical Perspective,” provides
valuable insight as to why radios
are constructed the way they are
and is included on the CD-ROM
accompanying this book.
The major subsystems of a radio receiving system are the antenna, the receiver and the
information processor. The antenna’s task is to provide a transition from an electromagnetic
wave in space to an electrical signal that can be conducted on wires. The receiver has the job
of retrieving the information content from a particular ac signal coming from the antenna
and presenting it in a useful format to the processor for use.
The processor typically is an operator, but can also be an automated system. When you
consider that most “processors” require signals in the range of volts (to drive an operator’s
speaker or headphones, or even the input of an A/D converter), and the particular signal of
interest arrives from the antenna at a level of mere microvolts, the basic function of the receiver
is to amplify the desired signal by a factor of a million. It must do this, while in the presence
of signals many orders of magnitude greater and of completely different characteristics,
without distortion of the desired signal or loss of the information it carries.
12.1 Characterizing Receivers
As we discuss receivers we will need to characterize their performance, and often their
performance limitations, using certain key parameters. The most commonly encountered are
as follows:
Sensitivity — This parameter is a measure of how weak a signal the receiver can extract
information from. This generally is expressed at a particular signal-to-noise ratio (SNR)
since noise is generally the limiting factor. A typical specification might be: “Sensitivity:
1 µV for 10 dB SNR with 3 kHz bandwidth.” The bandwidth is stated because the amount (or
power) of the noise, the denominator of the SNR fraction, increases directly with bandwidth.
Generally the noise parameter refers to the noise generated within the receiver, often less
than the noise that arrives with the signal from the antenna.
Selectivity — Selectivity is just the bandwidth discussed above. This is important because
to a first order it identifies the receiver’s ability to separate stations. With a perfectly sharp
filter in an ideal receiver, stations within the bandwidth will be heard, while those outside it
won’t be detected. The selectivity thus describes how closely spaced adjacent channels can
be. With a perfect 3 kHz bandwidth selectivity, and signals restricted to a 3 kHz bandwidth
at the transmitter, a different station can be assigned every 3 kHz across the spectrum. In a
less than ideal situation, it is usually necessary to include a guard band between channels.
Note that the word channel is used here in its generic form, meaning the amount of spectrum
occupied by a signal, and not defining a fixed frequency such as an AM broadcast channel.
A CW channel is about 300 Hz wide, a SSB channel about 2.5-3 kHz wide, and so forth.
“Adjacent channel” refers to spectrum immediately higher or lower in frequency.
Dynamic Range — In the case of all real receivers, there is a range of signals that a receiver
can respond to. This is referred to as dynamic range and, as will be discussed in more detail,
can be established based on a number of different criteria. In its most basic form, dynamic
range is the range in amplitude of signals that can be usefully received, typically from as low
as the receiver’s noise level, or noise floor, to a level at which stages overload in some way.
The type and severity of the overload is often part of the specification. A straightforward
example might be a 130 dB dynamic range with less than 3% distortion. The nature of the
Receivers  12.1
distortion will determine the observed phenomenon. If the weakest and strongest signals
are both on the same channel, for example,
we would not expect to be able to process the
weaker of the two. However, the more interesting case would be with the strong signal
in an adjacent channel. In an ideal receiver,
we would never notice that the adjacent signal was there. In a real receiver with a finite
dynamic range or nonideal selectivity, there
will be some level of adjacent channel signal
or signals that will interfere with reception of
the weaker on-channel signal.
The parameters described above are often
the key performance parameters, but in many
cases there are others that are important to
specify. Examples are audio output power,
power consumption, size, weight, control
capabilities and so forth.
12.2 Basics of Heterodyne Receivers
The heterodyne receiver combines the
input signal with a signal from a local oscillator (LO) in a nonlinear device called a mixer
to result in the sum and difference frequencies
as shown in Fig 12.1. The receiver may be
designed so the output signal is anything from
dc (a so-called direct conversion receiver) to
any frequency above or below either of the
two frequencies. The major benefit is that
most of the gain, bandwidth setting and processing are performed at a single frequency.
By changing the frequency of the LO, the
operator shifts the input frequency that is
translated to the output, along with all its
modulated information. In most receivers the
mixer output frequency is designed to be an
RF signal, either the sum or difference — the
other being filtered out at this point. This
output frequency is called an intermediate
frequency or IF. The IF amplifier system
can be designed to provide the selectivity
and other desired characteristics centered at
a single fixed frequency.
When more than one frequency conversion process is used, the receiver becomes
a superВ­heterodyne or superhet. A block diagram of a typical superhet receiver is shown in
Fig 12.2. In traditional form, the RF filter
is used to limit the input frequency range
to those frequencies that include only the
desired sum or difference but not the other
— the so-called image frequency. The dotted line represents the fact that in receivers
with a wide tuning range, such as a simple
AM broadcast receiver that tunes from 500 to
1700 kHz, a more than 3:1 range, the input RF
amplifier and filter is often tracked along with
the local oscillator. The IF filter is used to establish operating selectivity — that required
by the information bandwidth. Circuits from
the antenna input through and including the
mixer (the first mixer if more than one mixing
stage is used) are generally referred to as the
receiver’s front-end.
For reception of suppressed carrier singlesideband voice (SSB) or on-off or frequencyshift keyed (FSK) signals, a second beat
frequency oscillator or BFO is employed
to provide an audible voice, an audio tone
or tones at the output for operator or FSK
Fig 12.1 — Basic architecture of a heterodyne direct conversion receiver.
Fig 12.2 — Elements of a traditional superheterodyne radio receiver
12.2   Chapter 12
processing. This is the same as a heterodyne
mixer with an output centered at dc, although
the IF filter is usually designed to remove one
of the output products.
Modern receivers using digital signal processing for operating bandwidth and information detection convert the incoming signal
to digital form at one of the intermediate
frequencies. Advancing speed and declining
costs of analog-to-digital converters (ADC)
and processors are moving the conversion
closer and closer to the incoming signal’s
frequency, in some cases converting directly
at RF, which is called direct sampling. See
the DSP and Software Radio Design chapter
for more information on these techniques.
12.2.1 The Direct Conversion
Receiver
The heterodyne process can occur at a
number of different points in the receiver.
The simplest form of heterodyne receiver is
called a direct conversion receiver because
it performs the translation directly from the
signal frequency to the audio output. It is,
in effect, just the BFO and detector of the
general superhet shown in Fig 12.2. In this
case, the detector is often preceded by an RF
amplifier with a typical complete receiver
shown in Fig 12.3. Such a receiver can be
very simple to construct, yet can be quite
Fig 12.3 — Block diagram of a direct conversion.
Nonlinear Signal Combinations
Although a mixer is often thought of as nonlinear, it is neither necessary nor desirable for a mixer to be nonlinear. An ideal mixer is one that linearly multiplies the LO
voltage by the signal voltage, creating two products at the sum and difference frequencies and only those two products. From the signal’s perspective, it is a perfectly
linear but time-varying device. Ideally a mixer should be as linear as possible.
If a signal is applied to a nonlinear device, however, the output will not be just
a copy of the input, but can be described as the following infinite series of output
signal products:
VOUT = K0 + K1 Г— VIN + K2 Г— VIN2 + K3 Г— VIN3 + . . . + KN Г— VINN(A)
What happens if the input VIN consists of two sinusoids at F1 and F2, or A Г— [sin
(2ПЂ1) Г— t] and B Г— [sin (2ПЂF2) Г— t]? Begin by simplifying the notation to use angular
frequency in radians/second (2πF = ω). Thus VIN becomes Asinω1t and Bsinω2t and
equation A becomes:
VOUT = K0 + K1 × (Asinω1t + Bsinω2t) + K2 × (Asinω1t + Bsinω2t)2 +
K3 × (Asinω1t + Bsinω2t)3 + … +KN × (Asinω1t + Bsinω2t)N(B)
The zero-order term, K0, represents a dc component and the first-order term, K1 Г—
(Asinω1t + Bsinω2t), is just a constant times the input signals. The second-order term
is the most interesting for our purposes. Performing the squaring operation, we end
up with:
Second order term = [K2A2sin2ω1t + 2K2AB(sinω1t × sinω2t) + K2B2sin2ω2t](C)
Using the trigonometric identity (see Reference 1):
sin á sin â = 1⁄2 {cos ( á – â ) – cos ( á + â )}
the product term becomes:
K2AB × [cos(ω1–ω2)t – cos(ω1+ω2)t](D)
Here are the products at the sum and difference frequency of the input signals!
The signals, originally sinusoids are now cosinusoids, signifying a phase shift. These
signals, however, are just two of the many products created by the nonlinear action
of the circuit, represented by the higher-order terms in the original series.
In the output of a mixer or amplifier, those unwanted signals create noise and interference and must be minimized or filtered out. This nonlinear process is responsible for the distortion and intermodulation products generated by amplifiers operated
nonlinearly in receivers and transmitters.
effective — especially for the ultra-compact
low-power consumer-oriented portable station known as the mobile telephone! In fact,
given the worldwide presence of the mobile
telephone, the direct conversion receiver is
the most widely used of all receivers.
The basic function of a mixer is to multiply two sinusoidal signals and generate two
new signals with frequencies that are the
sum and difference of the two original signals. This function can be performed by a
linear multiplier, a switch that turns one
input signal on and off at the frequency of
the other input signal, or a nonlinear circuit
such as a diode. (The output of a nonlinear
circuit is made up of an infinite series of
products, all different combinations of the
two input signals, as described in the sidebar
on Nonlinear Signal Combinations.) Much
more information about the theory, operation
and application of mixers may be found in the
Mixers, Modulators and Demodulators
chapter.
Fig 12.4 shows the progression of the spectrum of an on-off keyed CW signal through
such a receiver based on the relationships
described above. In 12.4C, we include an
undesired image signal on the other side of
the local oscillator that also shows up in the
output of the receiver. Note each of the desired
and undesired responses that occur as outputs
of the mixer.
Some mixers are designed to be balanced
in order to cancel one of the input signals at
the output while a double-balanced mixer
cancels both. A double-balanced mixer simplifies the output filtering job as shown in
Fig 12.4D.
Products generated by nonlinearities in the
mixing process (see the sidebar on Nonlinear
Signal Combinations) are heard as intermodulation distortion signals that we will discuss
later. Note that the nonlinearities also allow
mixing with unwanted signals near multiples
of local oscillator frequency. These signals,
such as those from TV or FM broadcast stations, must be eliminated in the filtering before the mixer since their audio output will
be right in the desired passband on the output
of the mixer.
Project: A Rock-Bending
Receiver for 7 MHz
There are many direct conversion receiver (DC) construction articles in the amateur lВ­iterature. Home builders considering
construction of a DC receiver should read
Chapter 8 in Experimental Methods in RF
Design, which outlines many of the pitfalls
and design limitations of such receivers, as
well as a providing suggestions for making
a successful DC receiver.
This DC receiver design by Randy
Receivers  12.3
Henderson, WI5W, is presented here as an
example. This receiver represents about as
simple a receiver as can be constructed that
will offer reasonable performance (see the
References entry). The user of such a DC
receiver will appreciate its simplicity and
the nice sounding response of CW and SSB
signals that it receives, but only if there aren’t
many signals sharing the band.
A major shortcoming of the DC receiver architecture will soon become evident — it has
no rejection of image signals. Looking again
at Fig 12.4B, while the 7049 kHz oscillator
will translate the desired 7050 kHz signal to
an audio tone of 1 kHz, it will equally well
translate a signal at 7048 kHz to the same frequency. This is called an image and results in
interference. The same kind of thing happens
while receiving SSB; the entire channel on
the other side of the BFO is received as well.
Building a stable oscillator is often the most
challenging part of a simple receiver. This one
uses a tunable crystal-controlled oscillator
that is both stable and easy to reproduce. All
of its parts are readily available from multiple
sources and the fixed value capacitors and
resistors are common components available
from many electronics parts suppliers.
THE CIRCUIT
This receiver works by mixing two radiofrequency signals together. One of them is
the signal you want to hear, and the other is
generated by an oscillator circuit (Q1 and
В­associated components) in the receiver. In
Fig 12.5, mixer U1 puts out sums and differences of these signals and their harmonics. We
don’t use the sum of the original frequencies,
which comes out of the mixer in the vicinity
of 14 MHz. Instead, we use the frequency difference between the incoming signal and the
receiver’s oscillator — a signal in the audio
range if the incoming signal and oscillator frequencies are close enough to each other. This
signal is filtered in U2, and amplified in U2
and U3. An audio transducer (a speaker or
headphones) converts U3’s electrical output
to audio.
Fig 12.4 — Frequency relationships in DC receiver. At (A), desired receive signal from
antenna, 7050 kHz on-off keyed carrier. At (B), internal local oscillator and receive frequency relationships. At (C), frequency relationships of mixer/detector products (not
to scale).At (D), sum and difference outputs from double balanced mixer (not to scale).
Note the balanced out mixer inputs that cancel at the output (dashed lines).
12.4   Chapter 12
How the Rock Bender Bends Rocks
The oscillator is a tunable crystal oscillator — a variable crystal oscillator, or VXO.
Moving the oscillation frequency of a crystal like this is often called pulling. Because
crystals consist of precisely sized pieces of
quartz, crystals have long been called rocks in
ham slang — and receivers, transmitters and
transceivers that can’t be tuned around due
to crystal frequency control have been said
to be rockbound. Widening this rockbound
receiver’s tuning range with crystal pulling
made rock bending seem just as appropriate!
L2’s value determines the degree of pulling
available. Using FT-243 style crystals and
larger L2 values, the oscillator reliably tunes
Fig 12.5 — An SBL-1 mixer (U1, which contains two small RF transformers and a Schottky-diode quad), a TL072 dual op-amp IC
(U2) and an LM386 low-voltage audio power amplifier IC (U3) do much of the Rock-Bending Receiver’s magic. Q1, a variable crystal
oscillator (VXO), generates a low-power radio signal that shifts incoming signals down to the audio range for amplification in U2
and U3. All of the circuit’s resistors are 1⁄4 W, 5% tolerance types; the circuit’s polarized capacitors are 16 V electrolytics, except C10,
which can be rated as low as 10 V. The 0.1 ВµF capacitors are monolithic or disc ceramics rated at 16 V or higher.
C1, C2 — Ceramic or mica, 10% tolerance.
C4, C5, and C6 — Polystyrene, dipped
silver mica, or C0G (formerly NP0) ceramic, 10% tolerance.
C7 — Dual gang broadcast variable capacitor (14-380 pF per section),
1
вЃ„4 inch dia shaft, available as #BC13380
from Ocean State Electronics. A rubber equipment foot serves as a knob.
(Any variable capacitor with a maximum
capacitance of 350 to 600 pF can be
substituted; the wider the capacitance
range, the better.)
C12, C13, C14 — 10% tolerance. For SSB,
change C12, C13 and C14 to 0.001 ВµF.
U2 — TL072CN or TL082CN dual JFET op
amp.
L1 — Four turns of AWG #18 wire on
3
вЃ„4 inch PVC pipe form. Actual pipe OD is
0.85 inch. The coil’s length is about 0.65
inch; adjust turns spacing for maximum
signal strength. Tack the turns in place
with cyanoacrylic adhesive, coil dope
or RTV sealant. (As a substitute, wind
8 turns of #18 wire around 75% of the
circumference of a T-50-2 powdered-iron
core. Once you’ve soldered the coil in
place and have the receiver working,
expand and compress the coil’s turns
to peak incoming signals, and then cement the winding in place.)
L2 — Approximately 22.7 µH; consists of
one or more encapsulated RF chokes
in series (two 10-ВµH chokes [Mouser
#43HH105 suitable] and one 2.7-ВµH
choke [Mouser #43HH276 suitable] used
by author). See text.
L3 — 1 mH RF choke. As a substitute,
wind 34 turns of #30 enameled wire
around an FT-37-72 ferrite core.
Q1 — 2N2222, PN2222 or similar smallsignal, silicon NPN transistor.
R10 — 5 or 10 kW audio-taper control
(RadioShack No. 271-215 or 271-1721
suitable).
U1 — Mini-Circuits SBL-1 mixer.
Y1 — 7 MHz fundamental-mode quartz
crystal. Ocean State Electronics carries
7030, 7035, 7040, 7045, 7110 and 7125
kHz units.
PC boards for this project are available
from FAR Circuits.
from the frequency marked on the holder to
about 50 kHz below that point with larger
L2 values. (In the author’s receiver a 25 kHz
tuning range was achieved.) The oscillator’s
frequency stability is very good.
Inductor L2 and the crystal, Y1, have more
effect on the oscillator than any other components. Breaking up L2 into two or three
series-connected components often works
better than using one RF choke. (The author
used three molded RF chokes in series — two
10 ВµH chokes and one 2.7 ВµH unit.) Making
L2’s value too large makes the oscillator stop.
The author tested several crystals at Y1.
Those in FT-243 and HC-6-style holders
seemed more than happy to react to adjustment of C7 (TUNING). Crystals in the smaller
HC-18 metal holders need more inductance
at L2 to obtain the same tuning range. One
tiny HC-45 unit from International Crystals
needed 59 ВµH to eke out a mere 15 kHz of
tuning range.
Input Filter and Mixer
C1, L1, and C2 form the receiver’s input
filter. They act as a peaked low-pass network
to keep the mixer, U1, from responding to
signals higher in frequency than the 40 meter
band. (This is a good idea because it keeps
us from hearing video buzz from local television transmitters, and signals that might mix
with harmonics of the receiver’s VXO.) U1, a
Mini-Circuits SBL-1, is a passive diode-ring
Receivers  12.5
mixer. Diode-ring mixers usually perform
better if the output is terminated properly. R11
and C8 provide a resistive termination at RF
without disturbing U2A’s gain or noise figure.
Fig 12.6 — Two Q1-case
styles are shown because
plastic or metal transistors will work equally well
for Q1. If you build your
Rock-Bending Receiver
using a prefab PC board,
you should mount the ICs
in 8-pin mini-DIP sockets
rather than just soldering
the ICs to the board.
Audio Amplifier and Filter
U2B amplifies the audio signal from U1.
U2A serves as an active low-pass filter. The
values of C12, C13 and C14 are appropriate
for listening to CW signals. If you want SSB
stations to sound better, make the changes
shown in the parts list for Fig 12.5.
U3, an LM386 audio power amplifier IC,
serves as the receiver’s audio output stage.
The audio signal at U3’s output is much more
powerful than a weak signal at the receiver’s
input, so don’t run the speaker/earphone leads
near the circuit board. Doing so may cause a
squeal from audio oscillation at high volume
settings.
CONSTRUCTION
If you’re already an accomplished builder,
you know that this project can be built using a
number of construction techniques, so have at
it! If you’re new to building, you should consider building the Rock-Bending Receiver on
a printed circuit (PC) board. (The parts list
tells where you can buy one ready-made.) See
Fig 12.6 for details on the physical layout of
several important components used in the receiver. The receiver can be constructed either
using a PC board or by using ground-plane
(a.k.a “ugly”) construction. Fig 12.7 shows
a photo of the receiver built on a PC board,
but you can also build one on a ground-plane
formed by an unetched piece of PC board
material.
If you use the PC board available from
FAR Circuits or a homemade double-sided
circuit board based on the PC pattern on the
Handbook CD, you’ll notice that it has more
holes than it needs to. The extra holes (indicated in the part-placement diagram with
square pads) allow you to connect its ground
plane to the ground traces on its foil side.
(Doing so reduces the inductance of some of
the board’s ground paths.) Pass a short length
of bare wire (a clipped-off component lead is
fine) into each of these holes and solder on
both sides. Some of the circuit’s components
(C1, C2 and others) have grounded leads accessible on both sides of the board. Solder
these leads on both sides of the board.
Another important thing to do if you use
a homemade double-sided PC board is to
countersink the ground plane to clear all ungrounded holes. (Countersinking clears copper away from the holes so components won’t
short-circuit to the ground plane.) A 1вЃ„4-inch
diameter drill bit works well for this. Attach
a control knob to the bit’s shank and you can
safely use the bit as a manual countersinking
tool. If you countersink your board in a drill
press, set it to about 300 rpm or less, and use
12.6   Chapter 12
Fig 12.7 — Ground-plane construction or
PC-board construction as shown here —
either approach can produce the same
good Rock Bending Receiver performance.
very light pressure on the feed handle.
Mounting the receiver in a metal box or
cabinet is a good idea. Plastic enclosures can’t
shield the TUNING capacitor from the presence
of your hand, which may slightly affect the
receiver tuning. You don’t have to completely
enclose the receiver — a flat aluminum panel
screwed to a wooden base is an acceptable
alternative. The panel supports the tuning
capacitor, gain control and your choice of
audio connector. The base can support the
circuit board and antenna connector.
CHECKOUT
Before connecting the receiver to a power
source, thoroughly inspect your work to spot
obvious problems like solder bridges, incorrectly inserted components or incorrectly
wired connections. Using the schematic (and
PC-board layout if you built your receiver
on a PC board), recheck every component
and connection one at a time. If you have a
digital voltmeter (DVM), use it to measure
the resistance between ground and everything
that should be grounded. This includes things
like pin 4 of U2 and U3, pins 2, 5, 6 of U1,
and the rotor of C7.
If the grounded connections seem all right,
check some supply-side connections with the
meter. The connection between pin 6 of U3
and the positive power-supply lead should
show less than 1 W of resistance. The resistance between the supply lead and pin 8
of U1 should be about 47 W because of R1.
If everything seems okay, you can apply
power to the receiver. The receiver will work
with supply voltages as low as 6 V and as high
as 13.5 V, but it’s best to stay within the 9 to 12
V range. When first testing your receiver, use
a current-limited power supply (set its limiting
between 150 and 200 mA) or put a 150 mA
fuse in the connection between the receiver
and its power source. Once you’re sure that
everything is working as it should, you can
remove the fuse or turn off the current limiting.
If you don’t hear any signals with the antenna connected, you may have to do some
troubleshooting. Don’t worry; you can do it
with very little equipment.
TROUBLE?
The first clue to look for is noise. With the
GAIN control set to maximum, you should
hear a faint rushing sound in the speaker or
headphones. If not, you can use a small metallic tool and your body as a sort of test-signal
generator. (If you have any doubt about the
safety of your power supply, power the RockBending Receiver from a battery during this
test.) Turn the GAIN control to maximum.
Grasp the metallic part of a screwdriver,
needle or whatever in your fingers, and use
the tool to touch pin 3 of U3. If you hear
a loud scratchy popping sound, that stage
is working. If not, then something directly
related to U3 is the problem.
You can use this technique at U2 (pin 3,
then pin 6) and all the way to the antenna. If
you hear loud pops when touching either end
of L3 but not the antenna connector, the oscillator is probably not working. You can check
for oscillator activity by putting the receiver
near a friend’s transceiver (both must be in
the same room) and listening for the VXO.
Be sure to adjust the tuning control through
its range when checking the oscillator.
The dc voltage at Q1’s base (measured
without the RF probe) should be about half
the supply voltage. If Q1’s collector voltage
is about equal to the supply voltage, and Q1’s
base voltage is about half that value, Q1 is
probably okay. Reducing the value of L2 may
be necessary to make some crystals oscillate.
OPERATION
Although the Rock-Bending Receiver uses
only a handful of parts and its features are
limited, it performs surprisingly well. Based
on tests done with a Hewlett-Packard HP
606A signal generator, the receiver’s minimum discernible signal (by ear) appears to
be 0.3 ВµV. The author could easily copy 1 ВµV
signals with his version of the Rock-Bending
Receiver.
Although most HF-active hams use transceivers, there are advantages in using separate
receivers and transmitters. This is especially
true if you are trying to assemble a simple
home-built station.
12.3 The Superheterodyne Receiver
In many instances, it is not possible to
achieve all the receiver design goals with
a single-conversion receiver and multiple
conversion steps are used, creating the superheterodyne architecture. Traditionally, the
first conversion is tasked with removing the
RF image signals, while the second allows
processing of the IF signal to provide the
information based IF processing.
The superheterodyne receiver applies the
principles of the heterodyne receiver at least
twice. The concept was introduced by Major
Edwin Armstrong, a US Army artillery officer, just as WW I was coming to a close. He is
the same Armstrong who invented frequency
modulation (FM) some years later and who
held many radio patents between WW I and
WW II.
In a superhet, a local oscillator and mixer
are used to translate the received signal to an
intermediate frequency or IF rather than directly to audio. This provides an opportunity
for additional amplification and processing.
Then a second mixer is used as in the DC
receiver to detect the IF signal, translating it
to audio. The configuration was shown previously in Fig 12.2.
An example will illustrate how this works.
Let’s pick a common IF frequency used in a
simple AM broadcast radio, 455 kHz. If we
want to listen to a 600 kHz broadcast station,
the RF stage would be set to amplify the
600 kHz signal and the LO should be set to
600 + 455 kHz or 1055 kHz. The 600 kHz
signal, along with any audio information it
contains, is translated to the IF frequency and
is amplified and then detected.
Note that we could have also set the local
oscillator to 600 – 455 kHz or 145 kHz. By
setting it to the sum, we reduce the relative
range that the oscillator must tune. To cover
the 500 to 1700 kHz with the difference, our
LO would have to cover from 45 to 1245, a
28:1 range. Using the sum requires LO coverage from 955 to 2155 kHz, a range of about
2.5:1 — much easier to implement.
Note that to detect standard AM signals,
the receiver’s second oscillator, the beat fre-
quency oscillator (BFO), is turned off since
the AM station provides its own carrier signal over the air. Receivers designed only for
standard AM reception generally don’t have
a BFO at all.
It’s not clear yet that we’ve gained anything
by doing this; so let’s look at another example.
If we decide to change from listening to the
station at 600 kHz and want to listen to another station at, say, 1560 kHz, we can tune
the single dial of our superhet to 1560 kHz.
The RF stage is tuned to 1560 kHz, the LO
is set to 1560 + 455 or 2010 kHz, and now
the desired station is translated to our 455
kHz IF where the bulk of our amplification
can take place. Note also that with the superheterodyne configuration, selectivity (the
ability to separate stations) occurs primarily
in the intermediate-frequency (IF) stages and
is thus the same no matter what frequencies
we choose to listen to. This simplifies the
design of each stage considerably.
12.3.1 Superhet Bandwidth
Now we will discuss the bandwidth requirements of different operating modes and
how that affects superhet design. One advantage of a superhet is that the operating bandwidth can be established by the IF stages, and
further limited by the audio system. It is thus
independent of the RF frequency to which the
receiver is tuned. It should not be surprising
that the detailed design of a superhet receiver
is dependent on the nature of the signal being
received. We will briefly discuss the most
commonly received modulation types and the
bandwidth implications of each below. (Each
modulation type is discussed in more detail in
the Modulation or Digital Modes chapters.)
AMPLITUDE MODULATION (AM)
As shown in Fig 12.4, multiplying (in other
words, modulating) a carrier with a single
tone results in the tone being translated to
frequencies of the sum and difference of the
two. Thus, if a transmitter were to multiply a
600 Hz tone and a 600 kHz carrier signal, we
would generate additional new frequencies at
599.4 and 600.6 kHz. If instead we were to
modulate the 600 kHz carrier signal with a
band of frequencies corresponding to human
speech of 300 to 3300 Hz (the usual range
of communication quality voice signals), we
would have a pair of bands of information
carrying waveforms extending from 596.7
to 603.3 kHz, as shown in Fig 12.8. These
bands are called sidebands, and some form
of these is present in any AM signal that is
carrying information.
Note that the total bandwidth of this AM
voice signal is twice the highest frequency transmitted, or 6600 Hz. If we choose to
transmit speech and limited-range music,
we might allow modulating frequencies up
to 5000 Hz, resulting in a bandwidth of
10,000 Hz or 10 kHz. This is the standard
channel spacing that commercial AM broadcasters use in the US. (9 kHz is used in Europe)
In actual use, the adjacent channels are generally geographically separated, so broadcasters can extend some energy into the next
channels for improved fidelity. We would
refer to this as a narrow-bandwidth mode.
What does this say about the bandwidth
needed for our receiver? If we want to receive
the full information content transmitted by a
US AM broadcast station, then we need to set
the bandwidth to at least 10 kHz. What if our
receiver has a narrower bandwidth? Well, we
will lose the higher frequency components of
Fig 12.8 — Spectrum of sidebands of an
AM voice signal sent on a 600 kHz carrier.
Receivers  12.7
the transmitted signal — perhaps ending up
with a radio suitable for voice but not very
good at reproducing music.
On the other hand, what is the impact of
having too wide a bandwidth in our receiver?
In that case, we will be able to receive the full
transmitted spectrum but we will also receive
some of the adjacent channel information.
This will sound like interference and reduce
the quality of what we are receiving. If there
are no adjacent channel stations, we will
get any additional noise from the additional
bandwidth and minimal additional information. The general rule is that the received
bandwidth should be matched to the bandwidth of the signal we are trying to receive
to maximize SNR and minimize interference.
As the receiver bandwidth is reduced,
intelligibility suffers, although the SNR is
improved. With the carrier centered in the
receiver bandwidth, most voices are difficult
to understand at bandwidths less than around
4 kHz. In cases of heavy interference, full carrier AM can be received as if it were SSB, as
described below, with the carrier inserted at
the receiver, and the receiver tuned to whichever sideband has the least interference.
SINGLE-SIDEBAND SUPPRESSEDCARRIER MODULATION (SSB)
In looking at Fig 12.8, you might have
noticed that both sidebands carry the same
information, and are thus redundant. In addition, the carrier itself conveys no information.
It is thus possible to transmit a single sideband
and no carrier, as shown in Fig 12.9, relying
on the BFO (beat frequency oscillator) in
the receiver to provide a signal with which
to multiply the sideband in order to provide
demodulated audio output. The implications
in the receiver are that the bandwidth can be
slightly less than half that required for double
sideband AM (DSB). There must be an additional mechanism to carefully replace the
missing carrier within the receiver. This is the
function of the BFO, which must be at just
exactly the right frequency. If the frequency is
improperly set a baritone can come out sound-
Fig 12.9 — Spectrum of single sideband
AM voice signal sent adjacent to a suppressed 600 kHz carrier.
12.8   Chapter 12
ing like a soprano and vice versa! The effect is
quite audible even for small frequency errors
of a few tens of Hz.
This results in a requirement for a much
more stable receiver design with a much finer
tuning system — a more expensive proposition. An alternate is to transmit a reduced
level carrier and have the receiver lock on to
the weak carrier, usually called a pilot carrier. Note that the pilot carrier need not be
of sufficient amplitude to demodulate the
signal, just enough to allow a BFO to lock
to it. These alternatives are effective, but tend
to make SSB receivers expensive, complex
and most appropriate for the case in where a
small number of receivers are listening to a
single transmitter, as is the case of two-way
amateur communication.
Note that the bandwidth required to effectively demodulate an SSB signal is actually less than half that required for the
AM signal because the range centered on
the AM carrier need not be received. Thus
the communications-quality range of 300 to
3300 Hz can be received in a bandwidth of
3000 not 3300 Hz. Early SSB receivers typically used a bandwidth of around 3 kHz, but
with the heavy interference frequently found
in the amateur bands, it is more common for
amateurs to use bandwidths of 1.8 to 2.4 kHz
with the corresponding loss of some of the
higher consonant sounds.
RADIOTELEGRAPHY (CW)
We have described radiotelegraphy as
being transmitted by “on-off keying of a
carrier.” You might think that since a carrier
takes up just a single frequency, the receive
bandwidth needed should be almost zero. This
is only true if the carrier is never turned on
and off. In the case of CW, it will be turned
on and off quite rapidly. The rise and fall of
the carrier results in sidebands extending on
either side of the carrier, and they must be
received in order to reconstruct the signal in
the receiver.
A rule of thumb is to consider the rise and
fall time as about 10% of the pulse width
and the bandwidth as the reciprocal of the
quickest of rise or fall time. This results in a
bandwidth requirement of about 50 to 200 Hz
for the usual CW transmission rates. Another
way to visualize this is with the bandwidth
being set by a high-Q tuned circuit. Such a
circuit will continue to “ring” after the input
pulse is gone. Thus, too narrow a bandwidth
will actually “fill in” between the code ele­
ments and act like a “no bandwidth” full period carrier and this is exactly what is heard
if a very narrow crystal filter is used when
receiving CW.
DATA SIGNALS
(This is a short overview of amateur data
signals to establish receiver requirements
for the digital modes. See the Modulation
and Digital Modes chapters and the Digital
Communications supplement on the
Handbook CD for more in-depth treatment.)
The Baudot code (used for teletype communications) and ASCII code — two popu­lar digital communications codes used by
amateurs — are constructed with sequences
of elements or bits. The state of each bit
— ON or OFF — is represented by a signal at one of two distinct frequencies: one
designated mark and one designated space.
This is referred to as frequency shift keying
(FSK). The transmitter frequency shifts back
and forth with each character’s individual
elements.
Amateur Radio operators typically use
a 170 Hz separation between the mark and
space frequencies, depending on the data
rate and local convention, although 850 Hz
is sometimes used. The minimum bandwidth
required to recover the data is somewhat
greater than twice the spacing between the
tones. Note that the tones can be generated
by directly shifting the carrier frequency
(direct FSK), or by using a pair of 170 Hz
spaced audio tones applied to the audio input
of an SSB transmitter (audio FSK or AFSK).
Direct FSK and AFSK sound the same to a
receiver.
Note that if the standard audio tones of
2125 Hz (mark) and 2295 Hz (space) are
used, they fit within the bandwidth of a
voice channel and thus a voice transmitter
and receiver can be employed without any
additional processing needed outside the
radio equipment. Alternately, the receiver
can employ detectors for each frequency and
provide an output directly to a computer.
If the receiver can shift its BFO frequency
appropriately, the two tones can be received
through a filter designed for CW reception
with a bandwidth of about 300 Hz or wider.
Some receivers provide such a narrow filter
with the center frequency shifted midway between the tones (2210 Hz) to avoid the need
for retuning. The most advanced receivers
provide a separate filter for mark and space
frequencies, thus maximizing interference
rejection and signal-to-noise ratio (SNR).
Using a pair of tones for FSK or AFSK results
in a maximum data rate of about 1200 bit/s
over a high-quality voice channel.
Phase shift keying (PSK) can also be used
to transmit bit sequences, requiring good frequency stability to maintain the required time
synchronization to detect shifts in phase. If
the channel has a high SNR, as is often the
case at VHF and higher, telephone network
data-modem techniques can be used.
At HF, the signal is subjected to phase
and amplitude distortion as it travels. Noise
is also substantially higher on the HF bands.
Under these conditions, modulation and demodulation techniques designed for “wire-
line” connections become unusable at bit
rates of more than a few hundred bps. As
a result, amateurs have begun adopting and
developing state of the art digital modulation
techniques. These include the use of multiple
carriers (MFSK, Clover, PACTOR III, etc.),
multiple amplitudes and phase shifts (QAM
and QPSK techniques), and advanced error
detection and correction methods to achieve
a data throughput as high as 3600 bits per
second (bps) over a voice-bandwidth channel. Newly developed coding methods for
digital voice using QPSK modulation result
in a signal bandwidth of less than 1200 Hz.
(See http://freedv.org for more information.) Spread-spectrum techniques are also
being adopted on the amateur UHF bands,
but are beyond the scope of this discussion.
The bandwidth required for data communications can be as low as 100 Hz for
PSK31 to 1 kHz or more for the faster speeds
of PACTOR III and Clover. Beyond having
sufficient bandwidth for the data signal, the
primary requirements for receivers used for
data communications are linear amplitude
and phase response over the bandwidth of
the data signal to avoid distorting these
В­critical signal characteristics. The receiver
must also have excellent frequency stability to avoid drift and frequency resolution
to enable the receiver filters to be set on frequency.
FREQUENCY MODULATION (FM)
Another popular voice mode is frequency
modulation or FM. FM can be found in a
number of variations depending on purpose.
In Amateur Radio and commercial mobile
communication use on the shortwave bands,
it is universally narrow band FM or NBFM.
In NBFM, the frequency deviation is limited to around the maximum modulating
frequency, typically 3 kHz. The bandwidth
requirements at the receiver can be approximated by Carson’s Rule of BW = 2 × (D + M),
where D is the deviation and M is the maxi-
mum modulating frequency. Thus 3 kHz
deviation and a maximum voice frequency
of 3 kHz results in a bandwidth of 12 kHz,
not far beyond the requirements for broadcast
AM. (Additional signal components extend
beyond this bandwidth, but are not required
for voice communications.)
In contrast, broadcast or wideband FM or
WBFM occupies a channel width of 150 kHz.
Originally, this provided for a higher modulation index, even with 15 kHz audio that
resulted in an improved SNR. However, with
multiple channel stereo and sub-channels all
in the same allocated bandwidth the deviation
is around the maximum transmitted signal
bandwidth.
In the US, FCC amateur rules limit wideband FM use to frequencies above 29 MHz.
Some, but not all, HF communication receivers provide for FM reception. For proper FM
reception, two changes are required in the
receiver architecture as shown within the
dashed line in Fig 12.10. The fundamental
change is that the detector must recover information from the frequency variations of
the input signal. The most common such
detector is called a discriminator. The discriminator does not require a BFO, so that is
turned off, or eliminated in a dedicated FM
receiver. Since amplitude variations convey
no information in FM, they are generally
eliminated by a limiter. The limiter is a highgain IF amplifier stage that clips the positive
and negative peaks of signals above a certain
threshold. Since most noise of natural origins
is amplitude modulated, the limiting process
also strips away noise from the signal.
SUMMARY OF RECEIVER
BANDWIDTH REQUIREMENTS
We now have briefly discussed the typical
operating modes expected to be encountered
by an HF communications receiver. These
are tabulated in Table 12.1 and will be used
as design requirements as we develop the
various receiver architectures.
Table 12.1
Typical Communications Bandwidths
for Various Operating Modes
Mode
Bandwidth (kHz)
FM Voice
15
AM Broadcast
10
AM Voice
4-6.6
SSB Voice
1.8-3
Digital Voice
1.0-1.2
RTTY (850 Hz shift)
0.3-1.0
CW0.1-0.5
12.3.2 Selection of IF
Frequency for a Superhet
Now that we have established the range of
bandwidths that our receiver will need to pass,
we are in a position to discuss the selection of
the IF frequency at which those bandwidths
will be established.
IF IMAGE RESPONSE
As noted earlier, a superhet with a single
local oscillator or LO and specified IF can
receive two frequencies, selected by the tuning of the RF stage. For example, using a
receiver with an IF of 455 kHz to listen to a
desired signal at 7000 kHz can use an LO of
7455 kHz. However, the receiver will also
receive a signal at 455 kHz above the LO
frequency, or 7910 kHz. This undesired signal
frequency, located at twice the IF frequency
from the desired signal, is called an image.
Images will be separated from the desired frequency by twice the IF and the filter
ahead of the associated mixer must reduce
the image signal by the amount of the required image rejection. For a given IF, this
gets more difficult as the received frequency
is increased. For example, with a 455 kHz
single conversion system tuned to 1 MHz,
the image will be at 1.91 MHz, almost a
2:1 frequency ratio and relatively easy to
reject with a filter. The same receiver tuned to
30 MHz, would have an image at 30.91 MHz,
Fig 12.10 — Block diagram of an FM ­superhet. Changes are in dashed box.
Receivers  12.9
a much more difficult filtering problem
While an image that falls on an occupied
channel is obviously a problem — it’s rarely
desirable to receive two signals at the same
time — problems occur even if the image
frequency is clear of signals. This is because
the atmospheric noise in the image bandwidth is added to the noise of the desired
channel, as well as any internally-generated
noise in the RF amplifier stage. If the image
response is at the same level as the desired
signal response, there will be a 3 dB reduction in SNR.
REDUCING IMAGES WITH A HIGHER
IF FREQUENCY
An obvious solution to the RF image response is to raise the IF frequency high enough
so that signals at twice the IF frequency from
the desired signal are sufficiently attenuated by filters ahead of the mixer. This can
easily be done with IF stages operating at 5
to 10% of the highest receiving frequency
(1.5 to 3 MHz for a receiver that covers the
3-30 MHz HF band). The concept is used at
higher frequencies as well. The FM broadcast band (150 kHz wide channels over 87.9
to 108 MHz in the US) is generally received
on superhet receivers with an IF of 10.7 MHz,
which places all image frequencies outside
the FM band, eliminating interference from
other FM stations.
The use of higher-frequency tuned cir-
Fig 12.11 — Schematic of representative superhet using crystal lattice filters to achieve selectivity with a high IF frequency.
12.10   Chapter 12
cuits for IF selectivity works well for the
150 kHz wide FM broadcast channels, but
not so well for the relatively narrow channels
encountered on HF or lower, or even for many
V/UHF narrowband services. Fortunately,
there are three solutions that were commonly
used to resolve this problem.
The first, double conversion, converted the
desired signal to a relatively high IF followed
by a second conversion to a lower IF to set
the selectivity. This was a popular technique
in the 1950s. Improvements on the double
conversion technique led to triple conversion
with a very low, highly selective third IF.
The Collins system of moving a single-range
VFO to the second mixer and using switchable crystal oscillators for the first mixer also
became popular. The pre-mixed arrangement,
a third approach to double conversion was a
combination of the two, used a single variable oscillator range, as with the Collins, but
mixes the VFO and LO before applying them
to the first mixer – outside of the signal path.
These methods are obsolete for current
receivers but are commonly encountered in
the vintage equipment popular with many
hams. They are discussed in previous editions
of the Handbook.
High Frequency Crystal
Lattice Filters
Commercial quartz-crystal filters with
bandwidths appropriate for CW and SSB
Receivers  12.11
became available in the 1970s with center
frequencies into the 10 MHz range. This allowed a single-conversion receiver (see Fig
12.2) with an IF in the HF range to provide
both high image rejection and needed channel selectivity. This single-conversion architecture remains popular among designers of
portable and low-power equipment. Crystal
filter design is discussed in the RF and AF
Filters chapter.
Fig 12.11 shows an example front-end and
IF sections of a single-conversion superhet
using simple filters centered 1500 kHz. While
the filters shown are actually buildable by
amateurs at low cost, multiple-section filters
with much better performance can be purchased or constructed. Other IF frequencies
can be used, depending on crystal or filter
availability.
The circuit shown demonstrates the concepts involved and can be reproduced at low
cost. Remaining receiver functional blocks
such as the AGC circuitry, detectors and BFO,
and audio amplifiers and filters can be found
elsewhere in the book.
The Image-Rejecting Mixer
Another technique for reduction of image
response in receivers is not as commonly encountered in HF receivers as the preceding
designs, but it deserves mention because it has
some very significant applications. The imagerejecting mixer requires phase-shift networks,
as shown in Fig 12.12. Frequency F1 represents the input frequency while F2 is that of
the local oscillator. Note that the two 90В° phase
shifts are applied at different frequencies.
The phase shift network following Mixer 1
is at a fixed center frequency corresponding
to the IF, while the phase shift network at F2
must provide the required phase shift as the
local oscillator tunes across the band.
If the local oscillator is required to tune
over a limited fractional frequency range,
this is a very feasible approach. On the other
hand, maintaining a 90В° phase shift over a
wide range can be tricky. The good news is
that this approach provides image reduction
that is independent of, and in addition to,
any other mechanisms such as filters that are
employed toward that end.
Additionally, with the ability of DSP
components to operate at higher and higher
frequencies, the necessary operations seen
in Fig 12.12 can be performed in software
which does not depend on precision hardware design to maintain nearly exact phase
relationships.
The image-rejecting mixing process has
several attractive features:
 It is the only way to provide “single signal” reception with a direct conversion receiver, effectively reducing the audio image.
This can make the DC receiver a very good
performer, although the added complexity
is not always warranted in typical amateur
DC applications.
 This option is frequently found in microwave receivers in which sufficiently selective RF filtering can be difficult to obtain.
Since they often operate on fixed frequencies,
maintaining the required phase shift can be
straightforward.
 It is found in advanced receivers that
are trying to achieve optimum performance.
Even with a high first IF frequency, additional
image rejection can be provided.
  In transmitters, the same system is called
the phasing method of SSB generation. The
same blocks run “backwards” — one of the
phase shift networks can be applied to the
speech band and used to cancel one sideband.
This is discussed in the Transmitters and
Transceivers chapter.
12.3.3 Multiple-Conversion in
the Digital Age
While advances in microminiaturization of
all circuit elements have had a radical change
in the dimensions of communication equipment, perhaps most significant technology
impacts on architecture come in two areas
— the application of digital signal processing
and direct digital synthesis frequency generation. Both are discussed in detail in the
chapter on DSP and Software Radio Design.
Digital signal processing provides a level
of bandwidth setting filter performance not
practical with other technologies. While
much better than most low frequency IF
LC bandwidth filters, the very good crystal
or mechanical bandwidth filters in amateur
gear are not very close to the rectangular
shaped frequency response of an ideal filter, but rather have skirts with a 6 to 60 dB
response of perhaps 1.4 to 1. That means if
we select an SSB filter with a nominal (6 dB)
bandwidth of 2400 Hz, the width at 60 dB
down will typically be 2400 Г— 1.4 or 3360
Hz. Thus a signal in the next channel that is
60 dB stronger than the signal we are trying
to copy (as often happens) will have energy
just as strong as our desired signal.
DSP filtering approaches the ideal response. Fig 12.13 shows the ARRL Lab
measured response of a DSP bandwidth filter with a 6 dB bandwidth of 2400 Hz. Note
how rapidly the skirts drop to the noise level.
In addition, while analog filtering generally
requires a separate filter assembly for each
desired bandwidth, DSP filtering is adjustable — often in steps as narrow as 50 Hz —
in both bandwidth and center frequency. In
addition to bandwidth filtering, the same DSP
can often provide digital noise reduction and
digital notch filtering to remove interference
from fixed frequency carriers.
The application of DSP filters at higher
and higher frequencies as analog-to-digital
conversion samples rates and processing
power increase, offers greatly improved per-
0
Reference Level: 0 dB
–10
–20
–30
–40
–50
–60
–70
–80
0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0
Audio Frequency: 0 to 4 kHz
Fig 12.12 — Image rejecting mixer, block diagram and signal relationships.
12.12   Chapter 12
Fig 12.13 — ARRL Lab measured response of an aftermarket 2400 Hz DSP
bandwidth filter.
formance for multiple-conversion receivers.
The state of the art for amateur receivers is
approaching direct sampling (digitization at
the signal frequency) which will eliminate
the need for separate IF filters, amplifiers,
and detectors as all of those functions will
be performed inside the DSP processor as
software operations. However, it is likely that
analog receivers will continue to be popular
in various applications, such as portable, lowpower, and home-built gear.
Project: The MicroR2 —
An Easy to Build SSB or
CW Receiver
The following is the description of a direct
conversion receiver that uses an image-rejecting mixer as a product detector for single
sideband or single-signal CW reception.
Throughout the 100-year history of
Amateur Radio, the receiver has been the
basic element of the amateur station. This
single PC board receiver, originally described
by Rick Campbell, KK7B, captures some of
the elegance of the classic all-analog projects
of the past while achieving the performance
needed for operation in 21st-century band
conditions across a selected segment of a
single band.
A major advantage of a station with a separate receiver and transmitter is that each can
be used to test the other. There is a particularly elegant simplicity to the combination
of a crystal controlled transmitter and simple
receiver. The receiver doesn’t need a wellcalibrated dial, because the transmitter is
used to spot the frequency. The transmitter
doesn’t need sidetone, VFO offset or other
circuitry because the crystal determines the
transmitted frequency and the receiver is
used to monitor the transmit note off the air.
Even the receiver frequency stability may
be relaxed, because there is no on-the-air
penalty for touching up receiver tuning during a contact.
Let’s dive into the design and construction
Fig 12.14 — Two versions of the MicroR2. For smooth tuning, it’s hard to beat a big
В­vernier dial!
Fig 12.15 — Block diagram of the MicroR2 SSB/CW receiver.
of a simple receiver (Fig 12.14) that may
be used as a companion for an SSB or CW
transmitter on 40 meters, or as a tunable IF for
higher bands. Fig 12.15 is the block diagram
of this little receiver, which can be built in a
few evenings. The block diagram is similar
to the simple direct conversion receivers of
the early 1970s, but there the similarity ends.
This radio takes advantage of 40 years of
evolution of direct conversion HF receivers
by some very talented designers.
“SINGLE SIGNAL” RECEPTION
MAKES ALL THE DIFFERENCE
One major difference from early direct
conversion receivers is the use of phasing to
eliminate the opposite sideband. The receiver
in Fig 12.15 is a “single-signal” receiver that
only responds to signals on one side of zero
beat. Basic receivers in the �50s and �60s
would hear every CW signal at two places on
the dial, on either side of zero beat. As you
tuned through a CW signal, first you heard
a high-pitched tone, and then progressively
lower audio frequencies until the individual
beats became audible, and then finally all the
way down to zero cycles per second: zero
beat. As you tuned further you heard the
pitch rise back up until the high frequency
CW signal became inaudible again at some
high pitch.
It’s not too terrible to hear the desired
signal at two places on the dial — but having twice as much interference and noise is
inconvenient at best. Most modern radios,
even simple ones, include a simple crystal or
mechanical filter for single-signal IF selectivity. Since a direct conversion receiver has no
IF, selectivity is obtained by a combination of
audio filtering and an image-rejecting mixer.
Chapter 9 of Experimental Methods in RF
Design (EMRFD) has a complete discussion
of image-reject or “phasing” receivers, and
an overview of both conventional superheterodyne and phasing receivers can be found
elsewhere in this chapter.
BLOCK BY BLOCK DESCRIPTION
A brief description of each block in
Fig 12.15 and the schematic in Fig 12.16
follows. A parts list is given in Table 12.2.
(Additional detail and rationale for the circuits is provided in EMRFD.)
The 50 W RF input connects to a narrow
tuned-RF amplifier. The first circuit elements
are a low-pass filter to prevent strong signals
above the receiver tuning range — primarily
local TV and FM broadcast stations — from
arriving at the mixers where they would be
down-converted to audio by harmonics of the
local oscillator. The common-gate JFET lownoise amplifier (LNA) sets the noise figure of
the receiver and provides some gain, but even
more importantly, it isolates the mixers from
the antenna. Isolation prevents the impedance
Receivers  12.13
at the RF input from affecting the opposite
sideband suppression, and reverse isolation
prevents local oscillator leakage from radiating out the antenna and causing interference
to others as well as tunable hum.
The tuned drain circuit provides additional
selectivity and greatly improves dynamic
range for signals more than a few hundred
kHz from the desired signal. Lifting the
source of the JFET allows a convenient mute
control — changing the LNA from a 10 dB
amplifier to a 40 dB attenuator.
Following the LNA are the I-Q (short for
“in-phase” and “quadrature,” the labels we
use for the two signal paths in a phasing
system) down-converter and local oscillator
(LO) I-Q hybrid. Many of the more subtle difficulties of phasing direct conversion reeivers
are avoided by a compact, symmetrical layout
and direct connection of all the mixer ports to
the appropriate circuit elements without the
use of transmission lines. The audio low-pass
filters ensure that only audio exits the downconverter block, and set the close-in dynamic
range of the receiver. The inductors in the
low-pass filters will pick up magnetic hum
from nearby power transformers — more on
this topic later.
To the right of the I-Q hybrid is a basic
JFET Hartley VFO. In this application, it is
stripped of all the usual accoutrements like
buffer amplifiers and receiver incremental
tuning (RIT). The drain resistor and diodes
in the mixers set the operating level. This
simplified configuration, with a link on the
VFO inductor directly driving the twistedwire hybrid, works very well over a 5%
bandwidth (350 kHz at 7 MHz) when the
quadrature hybrid is optimized at the center
Fig 12.16 — Schematic diagram of the MicroR2 SSB/CW receiver. The parts are listed in Table 12.3.
12.14   Chapter 12
frequency and all voltages except the LO
are less than a few millivolts. (See R. Fisher,
W2CQH, “Twisted-Wire Quadrature Hybrid
Directional Couplers,” in QST for Jan 1978.)
The top half of the block diagram operates
entirely at audio. The left block is a matched
pair of audio LNAs with a noise figure of
about 5 dB and near 50 W input impedance,
to properly terminate the low-pass filters
at the I-Q mixer IF ports. These are directly
coupled into a circuit that performs the
В­mathematical operations needed to remove
amplitude and phase errors from the I and
Q signal paths. This is a classic use of operational amplifiers to do basic math.
The next block is a pair of second-order allpass networks. Use of second order networks
limits the opposite sideband suppression to
37 dB, which sounds exceptionally clean with
multiple CW signals and static crashes in the
passband.
It is straightforward to improve the opposite sideband suppression by 10 dB or so, but
doing so requires much more circuitry, not
just a third-order all-pass network. Relaxing
the opposite sideband suppression to a level
comparable with the Drake 2B permitted cutting the parts count in half, and the construction and alignment time by a factor of 10.
The final blocks are the summer, audio
low-pass filter and headphone amplifier.
These are conventional, near duplicates of
those used 10 years ago. When the board
is finished it is an operating receiver. The
completed circuitry is shown in Fig 12.17.
C34 — 2.2-22 pF, poly trimmer capacitor.
C35 — 68 pF, NP0 ceramic.
C36 — 2.7 pF 5% NP0 1000V ceramic.
C40 — Variable capacitor, see Note D.
L1, L2 — 33 mH inductor, see Note A.
L3 — 18t #28 T37-2, see Note E.
L4 — 40t #30 T37-2, see Note E.
L5, L6 — 3.9 mH inductor, see Note C.
L7 — 36t #28 T50-6, tap at 8 turns with a
2 turn link, see Note E.
Q1-Q3 — 2N3904.
Q4, Q5 — J310.
R1, R2 — 2.4 kΩ.
R3, R15-R18, R21, R22, R38, R43 —
10 kΩ.
R4 — 2.2 kΩ.
R5, R11, R12, R45 — 100 kΩ.
R6, R7 — 9.1 kΩ.
R8, R47, R52 — 100 Ω.
R9, R10, R13, R14, R41 — 27 kΩ.
R19, R23 — 10 kΩ trimpot.
R20 — 5.1 kΩ.
R24 — 2.32 kΩ, 1% see Note A.
R25 — 28.0 kΩ, 1% see Note A.
R26 — 9.53 kΩ, 1% see Note A.
R27 — 115 kΩ 1% see Note A.
R28-R37 — 10.0 kΩ 1%.
R39, R40, R42 — 470 Ω.
R44, R50 — 150 Ω.
R46, R48, R49, R53 — 51 Ω.
R51 — 1 MΩ.
R54 — 25 kΩ volume control.
RFC1,RFC2 — 15 µH molded RF choke.
T1 — 17 t two colors #28 bifilar T37-2, see
Note E.
U1-U5 — NE5532 or equivalent dual lownoise high-output op-amp.
U6 — LM7806 or equivalent 6 V three
terminal regulator.
U7, U8 — Mini-Circuits TUF-3 diode ring
mixer.
Table 12.2
MicroR2 Parts List
Parts for CW and SSB Version
C1-C4, C18, C39 — 33 µF, 16 V
electrolytic.
C5, C6 — 0.15 µF, 5% polyester.
C7 — 10 µF, 16 V electrolytic.
C8-C11 — 0.01 µF, polyester matched
to 1%.
C12 — 100 µF, 16 V electrolytic.
C13, C14 — 220 pF, NP0 ceramic.
C15 — 0.15 µF, 5% polyester, see Note A.
C16 — 0.22 µF, 5% polyester, see Note A.
C17 — 0.18 µF, 5% polyester, see Note A.
C19, C21, C23, C27, C37, C38 — 0.1 µF,
5% polyester.
C20, C22, C26 — 470 pF, NP0 ceramic.
C24 — 56 pF, NP0 ceramic.
C25 — 3-36 pF, poly trimmer capacitor.
C28 — 390 pF, NP0 ceramic.
C29 — 39 pF, NP0 ceramic on back of
board, see Note B.
C30-C33 — 0.82 µF, 5% polyester, see
Note C.
Note A: Make the following parts substitutions for a CW only
version:
R24 — 4.75 kΩ, 1%.
R25 — 41.2 kΩ, 1%.
R26 — 16.9 kΩ, 1%.
R27 — 147 kΩ, 1%.
C15 — 0.47 µF, 5% polyester.
C16 — 0.68 µF, 5% polyester.
C17 — 0.56 µF, 5% polyester.
L1, L2 — 100 mH inductor.
Note B: The total reactance of the parallel combination of
C28 and C29 plus the capacitance between the windings of
T1 is –j50 Ω at the center of the tuning range. Placing most
of the capacitance at one end is a different but equivalent
arrangement to the quadrature hybrid we often use with equal
capacitors. C29 is only needed if there is no standard value for
C28 within a few % of the required value. C29 is tack soldered
to the pads provided on the back of the PC board, and may be
a surface mount component if desired.
Note C: To sacrifice close-in dynamic range and selectivity for
reduced 60 Hz hum pickup, make the following substitutions, as
illustrated in Fig 12.18. This modification is recommended if the
MicroR2 must be used near 60 Hz power transformers.
C30, C32 — 0.10 µF 5% polyester.
C31, C33 — Not used.
L5, L6 — Not used. Place wire jumper between pads.
Note D: Capacitor C40 is the tuning capacitor for the receiver.
The MicroR2 was specifically designed to use an off-board
tuning capacitor to provide mounting flexibility, and to
encourage the substitution of whatever high-quality dual bearing
capacitor and reduction drive may be in the individual builder’s
junk box. Any variable capacitor may be used, but values
around 100 pF provide a little more than 100 kHz of tuning
range on the 40 meter band, which is a practical maximum.
The MicroR2 needs to be realigned (RF amplifier peaked and
the amplitude and phase trimpots readjusted) when making
frequency excursions of more than about В±50 kHz on the
40 meter band. To prevent tunable hum and other common
ills of direct conversion receivers, the MicroR2 PC board and
tuning capacitor should be in a shielded enclosure. See
Chapter 8 of EMRFD for a complete discussion.
Note E: L3, L4, L7 and T1 are listed as number of turns on
the specified core rather than a specific inductance. For those
who wish to study the design with a calculator, simulator and
inductance meter, L3 should be about +j100 Ω at about 8 MHz
(not particularly critical). L4 and L7 should be about +j250 Ω at
mid band. Each winding of T1 should be +j50 Ω at mid band.
Receivers  12.15
Fig 12.17 — Photo of the inside of the receiver. The PC board and parts are available
from Kanga US (www.kangaus.com).
Table 12.3
MicroR2 Receiver Performance
Parameter
Frequency coverage:
Power requirement:
Modes of operation:
SSB/CW sensitivity:
Opposite sideband rejection:
Two-tone, 3rd-order IMD dynamic range:
Third-order intercept: Third-order intercept points were determined
using equivalent S5 reference.
Measured in the ARRL Lab
7.0-7.1 or 7.2-7.3 MHz.
50 mA, tested at 13.8 V.
SSB, CW.
–131 dBm.
1000 Hz, 42 dB
20 kHz spacing, 81 dB
20 kHz spacing, –5 dBm.
ALIGNMENT
The receiver PC board may be aligned on
the bench without connecting the bandspread
capacitor. There are only four adjustments,
and the complete alignment takes just a few
minutes with just an antenna, the crystal oscillator in the companion transmitter (or a
test oscillator) and headphones. First, center
the trimming potentiometers using the dots.
Then slowly tune the VFO variable capacitor
until the crystal oscillator is heard. The signal
should be louder on one side of zero beat.
Peak the LNA tuning carefully for the loudest
signal. Then tune to the weaker side of zero
beat and null the signal with the amplitude
and phase trim pots. That’s all there is.
If the adjustments don’t work as expected
or there is no opposite sideband suppression,
look for a construction error. Don’t forget to
jumper the MUTE connection on the LNA —
it is a “ground- to-receive” terminal just like
simple receivers from the 1960s. Once the
receiver PC board is working on the bench,
mount it in a box, attach the bandspread
capacitor (with two wires — the ground
connection needs to be soldered to the PC
board too). Reset the VFO tuning capacitor for the desired tuning range, and do a
final alignment in the center of the desired
frequency range.
The simple on-board VFO is stable enough
for casual listening and tuning around the
band, but it can be greatly improved by eliminating the film trimmer capacitor that sets the
VFO tuning range. It may be replaced by an
air trimmer, a piston trimmer, a small compression trimmer, or a combination of small
fixed NP0 ceramic capacitors experimentally
selected to obtain the correct tuning range.
The holes in the PC board are large enough to
easily remove the film trimmer capacitor with
solder wick. The VFO stability may be improved still further by following the thermal
procedure described by W7ZOI in EMRFD.
12.16   Chapter 12
PERFORMANCE
Is it working? I don’t hear anything! Most
of us are familiar with receivers with a lot of
extra gain and noise, and AGC systems that
turn the gain all the way up when there are no
signals in the passband. Imagine instead a perfect receiver with no internal noise. With no
antenna connected, you would hear absolutely
nothing, just like a CD player with no disk.
The MicroR2 is far from perfect — but it may
be much quieter than your usual 40 meter receiver, and it probably has less gain. It is designed to drive good headphones. Weak signals
will be weak and clear and strong signals strong
and clear over the entire 70 dB in-channel
dynamic range — just like a CD player. This
receiver rewards listening skill and a good
antenna — don’t expect to hear much on a
couple feet of wire in a noisy room. The measured performance is shown in Table 12.3.
Now, about that hum. One reason this receiver sounds so good is that the selectivity is
established right at the mixer IF ports, before
any audio gain. Then, after much of the system
gain, a second low-pass filter provides additional selectivity. Distributed selectivity is
a classic technique. Earlier designs included
aggressive high-pass filtering to suppress frequencies below 300 Hz. That works well and
has other advantages — but the audio sounds
a bit thin. K7XNK suggested trying a receiver
with the low frequency response wide open.
It sounds great, but 60 Hz hum is audible
if the receiver is within about a meter of a
power transformer. Any closer and the hum
gets loud. This makes it difficult to measure
receiver performance in the lab, as there is
a power transformer in every piece of test
equipment.
Hum pickup may be eliminated by modifying the circuit as shown in Fig 12.18. This
hurts performance in the presence of strong
off-channel signals, but in some applications
— an instrumentation receiver on the bench
— it is a good trade-off. The receiver sounds
the same either way. For battery operation in
the hills, or on a picnic table in the backyard,
hum pickup isn’t a problem.
Fig 12.18 — Modified circuit to minimize hum.
12.4 Superhet Receiver Design Details
The previous sections have provided an
overview of the major architectural issues involved in superheterodyne receiver design. In
this section, we will cover more of the details
for the superhet constructed as a hybrid of
analog electronics and DSP functions applied
in an IF stage, the primary receiver design
architecture used in Amateur Radio today.
12.4.1 Receiver Sensitivity
The sensitivity of a receiver is a measure of
the lowest power input signal that can be received with a specified signal-to-noise ratio.
In the early days of radio, this was a very
important parameter and designers tried to
achieve the maximum practical sensitivity.
In recent years, device and design technology have improved to the point that other
parameters may be of higher importance,
particularly in the HF region and below.
THE RELATIONSHIP OF NOISE
TO SENSITIVITY
Noise level is as important as signal level
in determining sensitivity. This section
builds on the discussion of noise in the RF
Techniques chapter. The most important
noise parameters affecting receiver sensitivity are noise bandwidth, noise figure, noise
factor, and noise temperature.
Since received noise power is directly proportional to receiver bandwidth, any specification of sensitivity must be made for a
particular noise bandwidth. For DSP receivers with extremely steep filter skirts, receiver
bandwidth is approximately the same as the
filter or operating bandwidth. For other filter
types, noise bandwidth is somewhat larger
than the filter’s 6 dB response bandwidth.
The relationship of noise bandwidth to
noise power is one of the reasons that nВ­ arrow
bandwidth modes, such as CW, have a significant signal-to-noise advantage over modes
with wider bandwidth, such as voice, assuming the receiver bandwidth is the minimum
necessary to receive the signal. For example,
compared to a 2400-Hz SSB filter bandwidth,
a CW signal received in a 200 Hz bandwidth
will have a 2400/200 = 12 = 10.8 dB advantage in received noise power. That is the same
difference as an increase in transmitter from
100 to 1200 W.
Sources of Noise
Any electrical component will generate
a certain amount of noise due to random
electron motion. Any gain stages after the
internal noise source will amplify the noise
along with the signal. Thus a receiver with no
signal input source will have a certain amount
of noise generated and amplified within the
receiver itself.
Table 12.4
Typical Noise Levels (Into the Receiver) and Their Source, by Frequency
Frequency range
LF 30 to 300 kHz
MF 300 to 3000 kHz
Low HF 3 to 10 MHz
High HF 10 to 30 MHz
VHF 30 to 300 MHz
UHF 300 to 3000 MHz
Dominant noise sources
atmospheric
atmospheric/man-made
man-made/atmospheric
man-made/thermal
thermal/galactic
galactic/ thermal
Typical level (mv/m)*
150
70
20
10
0.3
0.2
*The level assumes a 10 kHz bandwidth. Data from Reference Data for Engineers,
4th Ed, p 273, Fig 1.
Upon connecting an antenna to a receiver,
there will be introduction of any noise external to the receiver that is on the received
frequency. The usual sources and their properties are described below. Table 12.5 presents typical levels of external noise in a 10
kHz bandwidth present in the environment
from different sources.
Atmospheric noise. This is noise generated within our atmosphere due to natural
phenomena. The principal cause is lightning
which sends wideband signals great distances. All points on the Earth receive this
noise, but it is much stronger in some regions
than others depending on the amount of local
lightning activity. This source is usually the
strongest noise source in the LF range and
may dominate well into the HF region, depending on the other noises in the region. The
level of atmospheric noise tends to drop off
by around 50 dB every time the frequency
is increased by a factor of 10. This source
usually drops in importance by the top of the
HF range (30 MHz).
Man-made noise. This source acts in a similar manner to atmospheric noise, although it
is more dependent on local activity rather than
geography and weather. The sources tend
to be sparks from rotating and other kinds
of electrical machinery as well as gasoline
engine ignition systems and some types of
lighting. In recent years, noise from computing and network equipment, switchmode
power supplies, and appliances has increased
significantly in urban and suburban environments. All things being equal, this source, on
average, drops off by about 20 dB every time
the frequency is increased by a factor of ten.
The slower decrease at higher frequencies is
due to the sparks having faster rise times than
lightning. The effect tends to be comparable
to atmospheric noise in the broadcast band,
less at lower frequencies and a bit more at HF.
Galactic Noise. This is noise generated by
the radiation from heavenly bodies outside
our atmosphere. Of course, while this is noise
to communicators, it is the desired signal
for radio-astronomers. This noise source is
a major factor at VHF and UHF and is quite
dependent on exactly where you point an
antenna (antennas for those ranges tend to be
small and are often pointable). It also happens
that the Earth turns and sometimes moves an
antenna into a position where it inadvertently
is aimed at a noisy area of the galaxy. If the
Sun, not surprisingly the strongest signal in
our solar system, appears behind a communications satellite, communications is generally
disrupted until the Sun is out of the antenna’s
receiving pattern. Galactic noise occurs on
HF, as well: Noise from the planet Jupiter can
be heard on the 15 meter band under quiet
conditions, for example.
Thermal Noise. Unlike the previous noise
sources, this one comes from our equipment.
All atomic structures have electrons that
move within their structures. This motion
results in very small currents that generate
small amounts of wideband signals. While
each particle’s radiation is small, the cumulative effect of all particles becomes significant
as the previous sources roll off with increasing frequency. The reason that this effect is
called thermal noise is because the electron
motion increases with the particle’s temperature. In fact the noise strength is directly
proportional to the temperature, if measured
in terms of absolute zero (0 K). For example,
if we increase the temperature from 270 to
280 K, that represents an increase in noise
power of 10/270, 0.037, or about 0.16 dB.
Some extremely sensitive microwave receivers use cryogenically-cooled front-end
amplifiers to provide large reductions in
thermal noise.
Oscillator Phase Noise. As noted in the
Oscillators and Synthesizers chapter, real
oscillators will have phase-noise sidebands
that extend out on either side of the nominal
carrier frequency at low amplitudes. Any such
noise will be transferred to the received signal
and through reciprocal mixing create noise
products from signals on adjacent channels
through the mixer. A good receiver will be designed to have phase noise that is well below
the level of other internally generated noise.
Receivers  12.17
Noise Power and Sensitivity
There are a number of related measures that
can be used to specify the amount of noise that
is generated within a receiver. If that noise
approaches, or is within perhaps 10 dB of
the amount of external noise received, then
it must be carefully considered and becomes
a major design parameter. If the internally
generated noise is less than perhaps 10 dB
below that expected from the environment,
efforts to minimize internal noise are generally not beneficial and can, in some cases, be
counterproductive.
While the total noise in a receiving system
is, as discussed, proportional to bandwidth,
the noise generating elements are generally
not. Thus it is useful to be able to specify
the internal noise of a system in a way that
is independent of bandwidth. It is important
to note that even though such a specification is useful, the actual noise is still directly
proportional to bandwidth and any bandwidth
beyond that needed to receive signal information will result in reduced SNR.
To evaluate the effect of noise power on
sensitivity, we will use the following equations from the discussion of noise in the RF
Techniques chapter:
Fig 12.19 — The effect of adding a low noise preamplifier in front of a noisy receiver
system.
F = 1 + TE/T0(1)
where
F is the noise factor,
TE is the equivalent noise temperature
and
T0 is the standard noise temperature,
usually 290 K.
Ni = k Г— B Г— TE
(2)
where
Ni is the equivalent noise power in watts
at the input of a perfect receiver
that would result in the same noise
output,
k is Boltzman’s constant, 1.38 × 10-23
joules/kelvin, and
B is the system noise bandwidth in
hertz.
Expressing Ni in dBm is more convenient
and is calculated as:
dBmi = –198.6 + (10 × log10 B) +
(10 Г— log10 TE)(3)
If Ni is greater than the noise generated
internally by the receiver, the receiver’s sensitivity is limited by the external noise. This
is usually the case for HF receivers where
atmospheric and man-made noise are much
stronger than the receiver’s internal noise
floor. If Ni is within, perhaps, 10 dB greater
than the receiver’s internal noise, then the
12.18   Chapter 12
Fig 12.20 — Illustration of adding a VHF low noise preamplifier at the antenna (A) compared to one at the receiver system input (B).
effect of the receiver’s internal circuits on
overall system sensitivity must be taken into
account.
Noise Figure of Cascaded Stages
Often we are faced with the requirement of
determining the noise figure of a system of
multiple stages. In general adding an amplification stage between the antenna and the rest
of the system will reduce the equivalent noise
figure of the system by the amount of gain of
the stage but adds in the noise of the added
stage directly. The formula for determining
the resultant noise factor is:
F= F1 +
F2 в€’ 1
(4)
G1
where
F1 is the noise factor of the stage closest
to the input,
F2 is the noise factor of the balance of
the system, and
G1 is the gain of the stage closest to the
input.
Note that these are noise factors not noise
figures. The change to noise figure is easily
computed at the end, if desired.
This is commonly encountered through the
addition of a low noise preamplifier ahead of
a noisy receiver as shown in Fig 12.19. As
shown in the figure for fairly typical values,
while the addition of a low noise preamplifier does reduce the noise figure, as can be
observed, the amplifier gain and noise figure
of the rest of the receiver can make a big
difference.
In many cases elements of a receiving system exhibit loss rather than gain. Since they
reduce the desired signal, while not changing
the noise of following stages, they increase
the noise figure at their input by an amount
equal to the loss. This is the reason that
VHF low noise preamps are often antenna
mounted. If the coax between the antenna
and radio has a loss of 1 dB, and the preamp
has a noise figure of 1 dB, the resulting noise
figure with the preamp at the radio will be
2 dB, while if at the antenna, the noise added
by the coax will be reduced by the gain of the
preamp, resulting in a significant improvement in received SNR at VHF and above.
Fig 12.20 shows a typical example.
To determine the system noise factor of
multiple cascaded stages, Equation 4 can be
extended as follows:
FN в€’ 1
F в€’ 1 F3 в€’ 1
+
+ ... +
F = F1 + 2
G
G
G
G
G
1
1
2
1
2 ...G N в€’1
(5)
where FN and GN are the noise factor and
gain of the Nth stage.
This can proceed from the antenna input
to the last linear stage, stopping at the point
of detection or analog-to-digital conversion,
whichever occurs first.
Why Do We Care?
A receiver designer needs to know how
strong the signals are to establish the range of
signals the receiver will be required to handle.
One may compare the equivalent noise power
determined in Equation 3 with the expected
external noise to determine whether the overall receiver SNR will be determined by external or internal noise. A reasonable design
objective is the have the internal noise be less
than perhaps 10 to 20 dB below the expected
noise. As noted above, this is related closely
to the frequency of signals we want to receive.
Any additional sensitivity will not provide a
noticeable benefit to SNR, and may result in
reduced dynamic range, as will be discussed
in the next section.
For frequencies at which external noise
sources are strongest, the noise power (and
signal power) will also be a function of the
antenna design. In such cases, the signalto-noise ratio can be improved by using an
antenna that picks up more signal and less
noise, such as with a directional antenna that
can reject noise from directions other than
that of the signal. Some antennas improve
SNR simply by rejecting noise, such as the
Beverage antenna. Signals from a Beverage
antenna are usually much weaker than from
a conventional antenna but their ability to
reject noise from undesired directions cre-
ates a net improvement in the SNR at the
receiver output.
12.4.2 Receiver Image
Rejection
The major design choice in determining
image response is selection of the first and
any following IF frequencies. This is important in receiving weak signals on bands where
strong signals are present. If the desired signal is near the noise, a signal at an image
frequency could easily be 100 dB stronger,
and thus to avoid interference, an image rejection of 110 dB would be needed. While
some receivers meet that target, the receiver
sections of most current amateur transceivers
are in the 70 to 100 dB range.
Receiver Architectures for Image
Rejection
As noted earlier, improved crystal filter
technology allows down-conversion HF receivers to use an IF in the 4-10 MHz range.
With a 10 MHz IF and an LO above the signal frequency, a 30 MHz signal would have
an image at 50 MHz. This makes imagerejection filtering relatively straightforward,
although many receiver IF frequencies tend to
be at the lower end of the above range. Still,
as will be discussed in the next section, they
have other advantages.
Many current HF receivers (or receiver sections of transceivers) have elected to employ
an up-converting architecture. They typically
have an IF in the VHF range, perhaps 60 to
70 MHz, making HF image rejection easy. A
30 MHz signal with a 60 MHz IF will have an
image at 150 MHz. Not only is it five times
the signal frequency, but signals in this range
(other than perhaps the occasional taxicab)
tend to be weaker than some undesired HF
signals. Receivers with this architecture have
image responses at the upper end of the above
range, often with the image rejected by a
relatively simple low-pass filter with a cut
off at the top of the receiver range.
Another advantage of this architecture
is that the local oscillator can cover a wide
continuous range, making it convenient for
a general coverage receiver. For example,
with a 60 MHz IF, a receiver designed for
LF through HF would need an LO covering
60.03 to 90 MHz, a 1.5 to 1 range, easily
provided by a number of synthesizer technologies, as described in the Oscillators and
Synthesizers chapter.
The typical upconverting receiver uses
multiple conversions to move signals to frequencies at which operating bandwidth can be
established. While crystal filters in the VHF
range used by receivers with upconverting
IFs have become available with bandwidths
as narrow as around 3 kHz, they do not yet
achieve the shape factor of similar bandwidth
filters at MF and HF. Thus, these are commonly used as roofing filters, discussed in
the next section, prior to a conversion to one
or more lower IF frequencies at which the
operating bandwidth is established. Fig 12.21
is a block diagram of a typical upconverting
receiver using DSP for setting the operating
bandwidth.
12.4.3 Receiver Dynamic
Range
Dynamic range can be defined as the ratio
of the strongest to the weakest signal that a
system, in this case our receiver, can respond
to linearly. Table 12.5 gives us an idea of
how small a signal we might want to receive.
The designer must create a receiver that will
handle signals from below the noise floor to
as strong as the closest nearby transmitter can
generate. Most receivers have a specified (or
sometimes not) highest input power that can
be tolerated, representing the other end of
the spectrum. Usually the maximum power
specified is the power at which the receiver
will not be damaged, while a somewhat lower
power level is generally the highest that the
receiver can operate at without overload and
the accompanying degradation of quality of
reception of the desired signal.
ON-CHANNEL DYNAMIC RANGE
The signal you wish to listen to can range
Fig 12.21 — Block diagram of a typical upconverting receiver using DSP for operating
bandwidth (BW) determination. Receivers applying DSP filtering at the second IF and at
higher frequencies are common.
Receivers  12.19
from the strongest to the weakest, sometimes
changing rapidly with conditions, or in a situation with multiple stations such as a net.
While a slow change in signal level can be
handled with manual gain controls, rapid
changes require automatic systems to avoid
overload and operator discomfort.
This is a problem that has been long solved
with automatic gain control (AGC) systems.
These systems are described in a further section of the chapter, but it is worth pointing
out that the measurement and gain control
points need to be applied carefully to the
most appropriate portions of the receiver to
maintain optimum performance. If all control
is applied to early stages, the SNR for strong
stations may suffer, while if applied in later
stages, overload of early stages may occur
in the presence of strong stations. Thus, gain
control has to be designed into the receiver
distributed from the input to the detector.
The next two sections illustrate a frequent
limitation of receiver performance — dynamic range between the reception of a weak
signal in the presence of one or more strong
signals outside of the channel.
BLOCKING GAIN COMPRESSION
A very strong signal outside the channel
bandwidth can cause a number of problems
that limit receiver performance. Blocking
gain compression (or “blocking”) occurs
when strong signals overload the receiver’s
high gain amplifiers and reduce its ability to
amplify weak signals. (Note that the term
“blocking” is often used outside Amateur
Radio when referring to reciprocal mixing
of oscillator noise with strong local signals.)
While listening to a weak signal, all stages
operate at maximum gain. If the weak signal
were at a level of S0, a strong signal could
be at S9 + 60 dB. Using the standard of S9
representing a 50 ВµV input signal, and each
S unit reflecting a change of 6 dB, the receiver’s front-end stages would be receiving
a 0.1 ВµV signal and a 50,000 ВµV into the front
end at the same time. A perfectly linear receiver would amplify each signal equally
until the undesired signal is eliminated at
the operating bandwidth setting stage. However, in practical receivers, after a few stages
of full gain amplification, the stronger signal
causes amplifier clipping, which reduces the
gain available to the strong signal. This is
seen as a gradual reduction in gain as the
input signal amplitude increases. Gain reduction also reduces the amplitude of the weaker
signal which is perceived to fade as the strong
signal increases in amplitude. Eventually, the
weaker signal is no longer receivable and is
said to have been “blocked”, thus the name
for the effect.
The ratio in dB between the strongest
signal that a receiver can amplify linearly,
with no more than 1 dB of gain reduction,
12.20   Chapter 12
and the receiver’s noise floor in a specified
bandwidth is called the receiver’s blocking
dynamic range or BDR or the compressionfree dynamic range or CFDR. In an analog
superhet, BDR is established by the linear
regions of the IF amplifiers and mixers. If
the receiver employs DSP, the range of the
analog-to-digital converter usually establishes the receiver’s BDR.
A related term is “near-far interference”
which is used primarily in the commercial
environment to refer to a strong signal causing
a receiver to reduce its gain and along with it
the strength of weak received signals.
INTERMODULATION DYNAMIC
RANGE
Blocking dynamic range is the straightforward response of a receiver to a single strong
interfering signal outside the operating passband. In amateur operation, we often have
more than one interferer. While such signals
contribute to the blocking gain compression
in the same manner as a single signal described above, multiple signals also result in
a potentially more serious problem resulting
from intermodulation products.
If we look again at Equation B in the earlier
sidebar on Nonlinear Signal Combinations,
we note that there are an infinite number of
higher order terms. In general, the coefficients of these terms are progressively lower
in amplitude, but they are still greater than
zero. Of primary interest is third-order term,
K3 Г— VIN3, when considering VIN as the sum
of two interfering signals (f1 and f2) near our
desired signal (f0) and within the first IF passband, but outside the operating bandwidth.
VOUT = K3 Г— [Asin(f1)t+Bsin(f2)t]3 (6)
= K3 Г— {A3sin3(f1)t + 3A2B [sin2(f1)t
Г— sin(f2)t]
+ 3AB2 [sin(f1)t Г— sin2(f2)t] + B3sin3(f2)t }
The cubic terms in Equation 6 (the first and
last terms) result in products at three times the
frequency and can be ignored in this discussion. Using trigonometric identities to reduce
the remaining sin2 terms and the subsequent
cos( )sin( ) products reveal individual intermodulation (IM) products, recognizing that
the signals have cross-modulated each other
due to the nonlinear action of the circuit.
(Math handbooks such as the CRC Standard
Mathematical Tables and Formulae have all
the necessary trigonometry information.)
IM products have frequencies that are
linear combinations of the input signal frequencies, written as n(f1) В± m(f2), where n
and m are integer values. The entire group
of products that result from intermodulation
are broadly referred to as intermodulation
distortion or IMD. The ratio in dB between
the amplitude of the interfering signals, f1 and
f2, and the resulting IM products is called the
intermodulation ratio.
If all of the higher-order terms in the original equation are considered, n and m can take
on any integer value. If the sum of n and m
is odd, (2 and 1, or 3 and 2, or 3 and 4, etc.)
the result is products that have frequencies
near our desired signal, for example, 2(f1) –
1(f2). Those are called odd-order products.
Odd-order products have frequencies close
enough to those of the original signals that
they can cause interference to the desired
signal. If the sum of n and m is three, those
are third-order IM products or third-order
IMD. For fifth-order IMD, the sum of n and
m is five, and so forth. The higher the order of
the IM products, the smaller their amplitude,
so our main concern is with third-order IMD.
If the two interfering signals have frequencies of f0 + D, and f0 + 2D, where D is some
offset frequency, we have for the third-order
term:
VOUT = K3 Г— [A sin(f0 +D)t +
B sin (f0 + 2D)t]3
(7)
A good example would be interfering signals with offsets of 2 kHz and 2 Г— 2 kHz or
4 kHz from the desired frequency, a common
situation on the amateur bands.
Discarding the cubic terms and applying
the necessary trigonometric identities shows
that a product can be produced from this combination of interfering frequencies that has a
frequency of exactly f0 — the same frequency
as the desired signal! (The higher-order terms
of Equation B can also produce products at
f0, but their amplitude is usually well below
those of the third-order products.)
Thus we have two interfering signals that
are not within our operating bandwidth so
we don’t hear either by themselves. Yet they
combine in a nonlinear circuit and produce a
signal exactly on top of our desired signal. If
the interfering signals are within the passband
of our first IF and are strong enough the IM
product will be heard.
As the strength of the interfering signals
increases, so does that of the resulting intermodulation products. For every dB of increase
in the interfering signals, the third-order IM
products increase by approximately 3 dB.
Fifth-order IM increases by 5 dB for every
dB increase in the interfering signals, and so
forth. Our primary concern, however, is with
the third-order products because they are the
strongest and cause the most interference.
INTERCEPT POINT
Intercept point describes the IMD performance of an individual stage or a complete
receiver. For example, third-order IM prod-
ucts increase at the rate of 3 dB for every
1-dB increase in the level of each of the interfering input signals (ideally, but not always
exactly true). As the input levels increase,
the distortion products seen at the output on
a spectrum analyzer could catch up to, and
equal, the level of the two desired signals if
the receiver did not begin to exhibit blocking
as discussed earlier.
The input level at which this occurs is the
input intercept point. Fig 12.22 shows the
concept graphically, and also derives from
the geometry an equation that relates signal
level, distortion and intercept point. The intercept point of the most interest in receiver
evaluation is that for third-order IM products
and is called the third-order intercept point
or IP3. A similar process is used to get a
second-order intercept point for second-order
IMD. A higher IP3 means that third-order IM
products will be weaker for specific input
signal strengths and the operator will experience less interference from IM products from
strong adjacent signals.
These formulas are very useful in designing radio systems and circuits. If the input
intercept point (dBm) and the gain of the
stage (dB) are added the result is an output intercept point (dBm). Receivers are specified
by input intercept point, referring distortion
back to the receive antenna input. Intercept
point is a major performance limitation of
receivers used in high density contest or DX
operations. Keep in mind that we have been
discussing this as an effect of two signals,
one that is D away from our operating frequency and another at twice D. In real life,
we may be trying to copy a weak signal at
f0, and have other signals at f0 В± 500, 750,
1000, 1250…5000 Hz. There will be many
combinations that produce products at or near
our weak signal’s frequency.
Note that the products don’t need to end up
exactly on top of the desired signal to cause
a problem; they just need to be within the
operating bandwidth. So far we have been
talking about steady carriers, such as would
be encountered during CW operation with
interference from nearby CW stations. SSB or
other wider bandwidth modes with spectrum
distributed across a few kHz will have signal
components that go in and out of a relationship that results in on-channel interference
from IMD. This manifests itself as a timevarying synthetic noise floor, composed of
all the resulting products across the channel.
The difference in this low level “noise” can
be dramatic between different receivers, especially when added to phase noise received
from other stations and reciprocal mixing
inside the receiver!
SPURIOUS-FREE DYNAMIC RANGE
IM products increase with the amplitude
of the interfering signals that cause them and
at some point become detectable above the
receiver’s noise floor. The ratio of the strength
of the interfering signals to the noise floor,
in dB, is the receiver’s spurious-free dynamic
range or SFDR. This is the range of signal
strengths over which the receiver does not
produce any detectable spurious products.
SFDR can be specified for a specific order
of IM products; for example, SFDR3 is the
SFDR measured for third-order IM products
only. The bandwidth for which the receiver’s
noise floor is measured must also be speci-
Fig 12.22 — Graphical representation of
the third-order intercept concept.
fied, since smaller bandwidths will result in
a lower noise floor.
RECEIVER DESIGN FOR DYNAMIC
RANGE
As noted previously, there are a number of
possible architectural choices for an amateur
receiver. In the past, the receivers with the best
close-in third order intermodulation distortion and maximum blocking dynamic range
were amateur-band-only receivers, such as
the primary receiver in the TEN-TEC Orion
series, the receiver in the Elecraft K2 , the
earlier TEN-TEC Omni VI and both receivers
in the Elecraft K3. A look at a typical block
diagram, as shown in Fig 12.23, makes it
easy to see why. The problems resulting from
strong unwanted signals near a desired one are
minimized if the unwanted signals are kept out
of the places in the receiver where they can be
amplified even more and cause the nonlinear
effects that we try to avoid.
Note that in Fig 12.23, the only place where
the desired and undesired signals all coexist
is before the first mixer. If the first mixer
and any RF preamp stages have sufficient
strong-signal handling capability, the undesired signals will be eliminated in the filter
immediately behind the first mixer. This HF
crystal filter is generally switchable to support desired bandwidths as narrow as 200 Hz.
The later amplifier, mixer and DSP circuits
only have to deal with the signal we want.
For additional discussion of these issues, see
“International Radio Roofing Filters for the
Yaesu FT-1000 MP Series Transceivers,” by
Joel Hallas, W1ZR, in QST Product Review
for February 2005.
Now look at a typical modern general
coverage receiver as shown previously in
Fig 12.21. In this arrangement, a single digital synthesizer, perhaps covering from 70
to 100 MHz, shifts the incoming signal(s)
to a VHF IF, often near 70 MHz. A roofing
filter at 70 MHz follows the first mixer.
This arrangement offers simplified local
oscillator (LO) design and the possibility
of excellent image rejection. Unfortunately,
crystal filter technology has only recently
been able to produce narrow filters for
Fig 12.23 — Block diagram of a downconverting amateur band receiver with HF IF filtering to eliminate near channel interference.
Receivers  12.21
70 MHz, and so far they have much
wider skirts than the crystal filters used in
Fig 12.23.
Many receivers and transceivers set this
filter bandwidth wider than any operating
bandwidth and use DSP filtering much later
in the signal chain to set the final operating
bandwidth. For a receiver that will receive
FM and AM, as well as SSB and CW, that
usually means a roofing filter with a bandwidth of around 20 kHz. With this arrangement, all signals in that 20 kHz bandwidth
pass all the way through IF amplifiers and
mixers and into the A/D converter before we
attempt to eliminate them using DSP filters.
By that time they have had an opportunity to
generate intermodulation products and cause
the blocking and IMD problems that we are
trying to eliminate. This situation is changing
with the introduction of the top of the line radios that feature both general coverage at HF
and VHF roofing filters, such as the ICOM
IC-7800 and Yaesu FTdx9000 transceivers.
(See the Transceiver Survey by W1ZR on the
Handbook’s CD-ROM for more information
on the latest models.)
A hybrid architecture has recently appeared in the TEN-TEC Omni VII transВ­ceiver that effectively combines the two technologies. The first IF has a 20 kHz wide roofing filter at 70 MHz, followed by selectable
steep skirted 455 kHz Collins mechanical filters at the second IF and then DSP filters at the
third IF. The topology is shown in Fig 12.24.
Careful attention to gain distribution
among the stages between the filters maintains desired sensitivity, but not so high that
the undesired products have a chance to become a serious problem. With bandwidths
of 20, 6 and 2.5 kHz supplied, and 500 and
300 Hz as accessories, the undesired close-in
signals are eliminated before they have an
opportunity to cause serious trouble in the
DSP stages that follow.
Another variation is found in the Kenwood
TS-590S which switches between downconversion on the more crowded “contest
bands” (160, 80, 40, 20 and 15 meters) and
up-conversion on the remaining bands. In
effect, trading sensitivity for dynamic range.
WHAT KIND OF PERFORMANCE
CAN WE EXPECT?
As we would expect, the blocking and IMD
dynamic range (IMD DR) performance of a
receiver will depend on a combination of the
early stage filtering, the linearity of the mixers
and amplifiers, and the dynamic range of any
ADC used for DSP. Product reviews of receivers and the receiver sections of transceivers
published in QST now provide the measured
dynamic range in the presence of interfering
signals with spacings of 2, 5 and 20 kHz.
(Details of the test procedures used are given
in the Test Equipment and Measurements
chapter.) At 20 kHz spacing, the interfering
signal is usually outside of the roofing filter
bandwidth of any of the above architectures.
Spacing of 2 and 5 kHz represents likely conditions on a crowded band.
A look at recent receiver measurements
indicates that receivers have IMD dynamic
range with 2 kHz spacing results in the following ranges:
Upconverting with VHF IF (Fig 12.21):
60 to 80 dB
Downconverting with HF IF (Fig 12.23):
75 to 105 dB
Hybrid distributed architecture (Fig 12.24):
Omni VII, 82 dB
Let’s take an example of what this would
mean. If we are listening to a signal at S3, for
signals to generate a third-order IMD product
at the same level in a receiver with a dynamic
range of 60 dB, the f0 + D and f0 + 2D signals
would have to have a combined power equal
to S9 +27 dB, or each at S9 +24 dB. This is
not unusual on today’s amateur bands. On
the other hand, if we had an IMD dynamic
range of 102 dB, the interfering signals would
have to at S9 +66 dB, much less likely. How
much dynamic range you need depends in
large measure on the kind of operating you
do, how much gain your receiving antennas
have and the closeness of the nearest station
that operates on the same bands as you.
Keep in mind also that it is often difficult to tell whether or not you are suffering
from IMD — it just sounds like there are
many more signals than are really present.
A good test to assess the source of interference is first switch off any preamplifiers and
noise-blankers or noise-reduction systems
that affect the receiver’s linearity. Observe
the level of the interference (if it’s still there)
and then switch in some attenuation at the
front-end of the receiver. If the level of the
interference goes down by more than the level
of attenuation (estimate 6 dB per S unit),
then the interference is being generated (or
at least aggravated) by non-linearity inside
the receiver. Continue to increase attenuation
until the interference either goes away or goes
down at the same rate as the attenuation is
increased. You might be surprised at how
much better the band “sounds” when your
receiver is operating in its linear region!
Fig 12.24 — Block diagram of a upconverting multiple conversion receiver with distributed roofing filter architecture.
12.22   Chapter 12
12.5 Control and Processing Outside the
Primary Signal Path
Discussion thus far has been about processes that occur within the primary path
for signals between antenna input and transducer or system output. There are a number
of control and processing subsystems that
occur outside of that path that contribute to
the features and performance of receiving
systems.
12.5.1 Automatic Gain
Control (AGC)
The amplitude of the desired signal at each
point in the receiver is generally controlled
by the AGC system, although manual control
is usually provided as well. Each stage has
a distortion versus signal level characteristic
that must be known, and the stage input level
must not become excessive. The signal being
received has a certain signal-to-distortion
ratio that must not be degraded too much by
the receiver. For example, if an SSB signal
has –30 dB distortion products the receiver
should create additional distortion no greater
than -40 dB with respect to the desired signal.
The correct AGC design ensures that each
stage gets the right input level. It is often
necessary to redesign some stages in order
to accomplish this task.
While this chapter deals exclusively with
AGC in the guise of analog circuits, the same
function is also implemented digitally in DSP
and SDR receivers. The goal of both is the
same — to maintain a signal level at all stages
of the receiver that is neither too large nor too
small so that the various processing systems
operate properly. Whether or not the AGC
offset and time constant are implemented by
an analog component or by a microprocessor
output is immaterial. The point is to manage the RF amplifier gain so that the overall
receiver behavior is satisfactory.
The effects of an improperly operating
AGC system can be quite subtle or nearly
disabling to a receiver and vary with how
Fig 12.25 — AGC principles. At A: typical superhet receiver with AGC applied to multiple stages of RF and IF. At B: audio output as a
function of antenna signal level.
Receivers  12.23
the AGC system is constructed. This chapter attempts to describe the requirements for
proper operation and provides some examples of implementation and common AGC
failures in terms of analog circuitry which is
somewhat easier to describe than software
algorithms, noting that similar behaviors
exist even in purely software receivers. The
interested student should consider studying
the AGC systems of commercial receivers to
understand how professional design teams
deal with the problem of managing so much
gain with such stringent requirements for
linearity and distortion.
THE AGC LOOP
Fig 12.25A shows a typical AGC loop that
is often used in amateur superhet receivers.
The AGC is applied to the stages through RF
decoupling circuits that prevent the stages
from interacting with each other. The AGC
amplifier helps to provide enough AGC loop
gain so that the gain-control characteristic of
Fig 12.25B is achieved. If effect, the AGC
system causes the receiver to act as a compression amplifier with lower overall gain for
stronger signals.
The AGC action does not begin until a
certain level, called the AGC threshold, is
reached. The THRESHOLD VOLTS input in
Fig 12.25A serves this purpose. After that
level is exceeded, the audio level increases
more slowly than for weaker signals. The
audio rise beyond the threshold value is usually in the 5 to 10 dB range. Too much or
too little audio rise are both undesirable for
most operators.
As an option, the AGC signal to the RF
amplifier is offset by the 0.6 V forward drop
of the diode so that the RF gain does not start
to decrease until larger signals appear. This
prevents a premature increase of the receiver
noise figure. Also, a time constant of one or
two seconds after this diode helps keep the
RF gain steady for the short term.
Fig 12.26 is a typical plot of the signal levels at the various stages of a certain ham band
receiver using analog circuitry. Each stage
has the proper level and a 115 dB change in
input level produces a 10 dB change in audio
level. A manual gain control could produce
the same effect.
AGC TIME CONSTANTS
There are two primary AGC time constants. AGC attack time describes the time it
takes the AGC system to respond to the presence of a signal. AGC decay time describes
the response of the AGC system to changes
in a signal that is present. The optimum time
constants for the AGC system depends on
the type of signal being received, the type of
operation being conducted, and the operator’s
preference.
In Fig 12.25, following the precision rectifier, R1 and C1 set an attack time, to prevent
excessively fast application of AGC. One or
two milliseconds is a good value for the R1
Г— C1 product. If the antenna signal suddenly
disappears, the AGC loop is opened because
the precision rectifier stops conducting. C1
then discharges through R2 and the C1 Г— R2
product can be in the range of 100 to 200 ms.
At some point the rectifier again becomes
active, and the loop is closed again.
An optional modification of this behavior
is the hang AGC circuit. If we make R2 Г—
C1 much longer, say 3 seconds or more, the
AGC voltage remains almost constant until
the R5, C2 circuit decays with a switch selectable time constant of 100 to 1000 ms. At
that time R3 quickly discharges C1 and full
Fig 12.26 — Gain control of a ham-band receiver using AGC. A manual gain control
could produce the same result.
12.24   Chapter 12
receiver gain is quickly restored. This type
of control is appreciated by many operators
because of the lack of AGC pumping due to
modulation, rapid fading and other sudden
signal level changes.
AGC PUMPING
AGC pumping can occur in receivers in
which the AGC measurement point is located
ahead of the stages that determine operating
bandwidth, such as when an audio filter is
added to a receiver externally and outside the
reach of the AGC system. If the weak signal
is the only signal within the first IF passband,
the AGC will cause the receiver to be at maximum gain and optimum SNR. If an interfering signal is within the first IF passband, but
outside the audio DSP filter’s passband, we
won’t hear the interfering signal, but it will
enter the AGC system and reduce the gain so
we might not hear our desired weak signal.
AGC pumping is audible as sudden reductions in signal strength without a strong signal
in the passband of the receiver
AGC LOOP RESPONSE PROBLEMS
If the various stages have the property that
each 1 V change in AGC voltage changes the
gain by a constant amount (in dB), the AGC
loop is said to be log-linear and regular feedback principles can be used to analyze and
design the loop. But there are some difficulties
that complicate this textbook model. One has
already been mentioned, that when the signal
is rapidly decreasing the loop becomes open
and the various capacitors discharge in an open
loop manner. As the signal is increasing beyond the threshold, or if it is decreasing slowly
enough, the feedback theory applies more accurately.
In SSB and CW receivers rapid changes are
the rule and not the exception. It is important
that the AGC loop not overshoot or ring when
the signal level rises past the threshold. The
idea is to design the ALC loop to be stable
when the loop is closed. It obviously won’t
oscillate when open (during decay time). But
the loop must have smooth and consistent
transient response when the loop goes from
open to closed state.
Another problem involves the narrow bandpass IF filter. The group delay of these filters
constitutes a time lag in the loop that can make
loop stabilization difficult. Moreover, these
filters nearly always have much greater group
delay at the edges of the passband, so that loop
problems are aggravated at these frequencies.
Overshoots and undershoots, called gulping,
are very common. Compensation networks
that advance the phase of the feedback help to
offset these group delays. The design problem
arises because some of the AGC is applied
before the filter and some after the filter. It is
a good idea to put as much fast AGC as possible after the filter and use a slower decaying
Fig 12.27 — Some gain controllable amplifiers and a rectifier suitable for audio derived AGC.
Receivers  12.25
AGC ahead of the filter. The delay diode and
RC in Fig 12.25A are helpful in that respect.
Complex AGC designs using two or more
compensated loops are also in the literature.
If a second cascaded narrow filter is used in
the IF it is usually a lot easier to leave the
second or downstream filter out of the AGC
loop at the risk of allowing AGC pumping as
described in the preceding section.
Another problem is that the control characteristic is often not log-linear. For example,
dual-gate MOSFETs tend to have much larger
dB/V at large values of gain reduction. Many
IC amplifiers have the same problem. The
result is that large signals cause instability
because of excessive loop gain. Variable gain
op amps and other similar ICs are available
that are intended for gain control loops.
Audio frequency components on the AGC
bus can cause problems because the amplifier gains are modulated by the audio and
distort the desired signal. A hang AGC circuit
(essentially a low-pass filter) can reduce or
eliminate this problem.
Finally, if we try to reduce the change in
audio levels to a very low value, the required
loop gain becomes very large, and stability
problems become very difficult. It is much
better to accept a 5 to 10 dB variation of
audio output.
Because many parameters are involved and
many of them are not strictly log-linear, it
is best to achieve good AGC performance
through an initial design effort and finalize
the design experimentally. Use a signal generator, attenuator and a signal pulser (2 ms
rise and fall times, adjustable pulse rate and
duration) at the antenna and a synchronized
oscilloscope to look at the IF envelope. Tweak
the time constants and AGC distribution by
means of resistor and capacitor decade substitution boxes. Be sure to test throughout
the passband of each filter. The final result
should be a smooth and pleasant sounding
SSB/CW response, even with maximum RF
gain and strong signals. Patience and experience are helpful.
12.5.2 Audio-Derived AGC
Some receivers, especially direct-conversion types, use audio-derived AGC. There are
problems with this approach as well. At low
audio frequencies the AGC control action can
be slow to develop. That is, low-frequency
audio sine waves take longer to reach their
peaks than the AGC time constants. During
this time the RF/IF/AF stages can be overdriven. If the RF and IF gains are kept at
a low level this problem can be reduced.
Also, attenuating low audio frequencies
prior to the first audio amplifier should help.
With audio AGC, it is important to avoid
so-called “charge pump” rectifiers or other
slow-responding circuits that require multiple cycles of audio to pump up the AGC
voltage. Instead, use a peak-detecting circuit
that responds accurately on the first positive
or negative half-cycle.
12.5.3 AGC Circuits
Fig 12.27 shows some gain-controllable
circuits. Fig 12.27A shows a two-stage
455-kHz IF amplifier with PIN diode gain
control. This circuit is a simplified adaptation
from a production receiver, the Collins 651S.
The IF amplifier section shown is preceded
and followed by selectivity circuits and additional gain stages with AGC. The 1.0 ВµF
capacitors aid in loop compensation. The favorable thing about this approach is that the
transistors remain biased at their optimum
operating point. Right at the point at which
the diodes start to conduct, a small increase
in IMD may be noticed, but that goes away
as diode current increases slightly. Two or
more diodes can be used in series, if this is a
problem (it very seldom is). Another solution
is to use a PIN diode that is more suitable for
such a low-frequency IF. Look for a device
with t > 10 / (2pf) where t is the minority
carrier lifetime in ms and f is the frequency
in MHz.
Fig 12.27B is an audio derived AGC circuit
using a full-wave rectifier that responds to
positive or negative excursions of the audio
signal. The RC circuit follows the audio
closely.
Fig 12.27C shows a typical circuit for the
MC1350P RF/IF amplifier. The graph of gain
control versus AGC voltage shows the change
in dB/V. If the control is limited to the first
20 dB of gain reduction this chip should be
favorable for good AGC transient response
and good IMD performance. Use multiple
low-gain stages rather than a single highgain stage for these reasons. The gain control
within the MC1350P is accomplished by diverting signal current from the first amplifier
stage into a current sink. This is also known
as the Gilbert cell multiplier architecture.
Another chip of this type is the NE/SA5209.
This type of approach is simpler to implement than discrete circuit approaches, such
as dual-gate MOSFETs that are now being
replaced by IC designs.
Fig 12.27D shows the high performance
National Semiconductor LMH6502MA
(14-pin DIP plastic package) voltage controlled amplifier. It is specially designed
for accurate log-linear AGC from 0 to 40 dB
with respect to a preset maximum voltage
gain from 6 to 40 dB. Its В±3 dB bandwidth
is 130 MHz. It is an excellent IF amplifier
for high performance receiver or transmitter
projects.
Additional info on voltage-controlled
amplifier ICs can be found on the Analog
Devices web site (www.analog.com). Search
the site for Tutorial MT-073, which describes
the operation of various types of gain-controlled amplifiers with numerous product
examples.
12.6 Pulse Noise Reduction
A major problem for those listening to
receivers has historically been local impulse
noise. For HF and VHF receivers it is often
from the sparks of internal combustion engine spark plugs, electric fence chargers, light
dimmers, faulty power-line insulators and
many other similar devices that put out short
duration wide band signals. In the UHF and
microwave region, radar systems can cause
similar problems. There have been three general methods of attempting to deal with such
noise over the years, some more successful
12.26   Chapter 12
than others. We will briefly describe the approaches.
12.6.1 The Noise Limiter
The first device used in an early (1930s)
attempt to limit impulse noise was called a
noise limiter or clipper circuit as originally
described by H. Robinson, W3LW. (see references) This circuit would clip or limit noise
(or signal) peaks that exceeded a preset limit.
The idea was to have the limit set to about as
loud as you wanted to hear anything and nothing louder would get through. This was helpful in eliminating the loudest part of impulse
noise or even nearby lightning crashes, but
it had two problems. First it didn’t eliminate
the noise, it just reduced the peak loudness;
second, it also reduced the loudness of loud
non-noise signals and in the process distorted
them considerably.
The second problem was fixed shortly
thereafter, with the advent of the automatic noise limiter or ANL as described by
J. Dickert (see references). The ANL automatically set the clipping threshold to that
of a loud signal. It thus would adjust itself
as the loudness of signals you listened to
changed with time. An ANL was fairly easy to
implement and became standard equipment
on amateur receivers from the late 1930s on.
While ANL circuits are no longer common,
simple receivers used today do sometimes
incorporate passive clipping circuits to account for their limited AGC ability.
12.6.2 The Noise Blanker
It turned out that improvements in receiver
selectivity over the 1950s and beyond, while
improving the ability to reduce random noise,
actually made receiver response to impulse
noise worse. The reason for this is that a very
short duration pulse will actually be lengthened while going through a narrow filter. This
is due to the filter’s different delay times for
the pulse’s wide spectrum of components,
resulting in the components arriving at the
filter output at different times. You can demonstrate this in your superhet receiver if it
has a narrow crystal filter. Find a frequency
with heavy impulse noise and switch between
wide and narrow filters. If your narrow filter
is 500 Hz or less, the noise pulses will likely
be more prominent with the narrow filter.
DSP filters with their superior group delay
performance exhibit less smearing than their
analog counterparts.
The noise limiters described previously
were all connected at the output of the IF
amplifiers and thus the noise had passed most
of the selectivity before the limiter and had
been widened by the receiver filters. In addition, modern receivers include automatic
gain control (AGC), a system that reduces
the receiver gain in the presence of strong
signals to avoid overload of both receiver circuits and ears. In SSB receivers, since signals
vary in strength as someone talks, the usual
AGC responds quickly to reduce the gain of
a strong signal and then slowly increases it if
the signal is no longer there. This means that
a strong noise pulse may reduce the receiver
gain for much longer than it lasts.
The solution — a noise blanker. A noise
blanker is almost a separate wideband receiver. It takes its input from an early stage in
the receiver before much selectivity or AGC
has been applied. It amplifies the wideband
signal and detects the narrow noise pulses
without lengthening them. The still-narrow
noise pulses are used to shut off the receiver at
a point ahead of the selectivity and AGC, thus
keeping the noise from getting to the parts
of the receiver at which the pulses would be
extended. In other words, the receiver is shut
off or gated during the noise pulse.
A well-designed noise blanker can be very
effective. Instead of just keeping the noise
at the level of the signal as a noise limiter
does, the noise blanker can actually eliminate
the noise. If the pulses are narrow enough,
the loss of desired signal during the time the
receiver is disabled is not noticeable and the
noise may seem to disappear entirely.
In addition to an ON/OFF switch, many
noise blanker designs include a control labeled THRESHOLD. The THRESHOLD control
adjusts the level of noise that will blank the
receiver. If it is set for too low a level, it
will blank on signal peaks as well as noise,
resulting in distortion of the signal. The usual
approach is to turn on the blanker, then adjust
the THRESHOLD control until the noise is just
blanked. Don’t forget to turn it off when the
noise goes away.
Noise blankers can also create problems.
The wide-band receiver circuit that detects
the noise pulses detects any signals in that
bandwidth. If such a signal is strong and has
sharp peaks (as voice and CW signals do), the
noise blanker will treat them as noise pulses
and shut down the receiver accordingly. This
causes tremendous distortion and can make
it sound as if the strong signal to which the
noise blanker is responding is generating spurious signals that cause the distortion. Before
you assume that the strong signal is causing
problems, turn the noise blanker on and off to
check. When the band is full of strong signals,
a noise blanker may cause more problems
than it solves.
12.6.3 Operating Noise
Limiters and Blankers
Many current receivers include both a noise
limiter and a noise blanker. If your receiver
has both, they will have separate controls
and it is worthwhile to try them both. There
are times at which one will work better than
the other, and other times when it goes the
other way, depending on the characteristics
of the noise. There are other times when both
work better than either. In any case, they can
make listening a lot more pleasant — just
remember to turn them off when you don’t
need them since either type can cause some
distortion, especially on strong signals that
should otherwise be easy to listen to.
Recognizing that it is difficult for a single noise blanker to work properly with the
wide variations of noise pulses, it is common
for late-model receivers to have two noise
blankers with different characteristics that
are optimized for the different pulse types.
One noise blanker is typically optimized for
very short pulses and the other for longer
pulses. The operator can switch between
the blankers to see which works best on the
noise at hand.
OTHER TECHNIQUES
The previous techniques represent the
most commonly available techniques to reduce impulse noise. There are a few other
solutions as well. Note that we haven’t been
talking about reducing interference here.
By interference, we mean another intended
signal encroaching on the channel to which
we want to listen. There are a number of
techniques to reduce interference, and some
also can help with impulse noise.
Many times impulse noise is coming
from a particular direction. If so, by using
a directional antenna, we can adjust the direction for minimum noise. When we think
about directional antennas, the giant HF Yagi
springs to mind. For receiving purposes, especially on the lower bands such as 160, 80
and 40 meters (where the impulse noise often
seems the worst), a small indoor or outdoor
receiving loop antenna as described in the
ARRL Antenna Book can be very effective
at eliminating either interfering stations or
noise (both if they happen to be in the same
direction).
Another technique that can be used to
eliminate either interference or noise is to
obtain a copy of the noise (or interference)
that is 180В° out of phase from the one you
are receiving. By adjusting the amplitude to
match the incoming signal, the signal can be
cancelled at the input to the receiver. Several
available commercial units perform this task.
Digital signal processing, described in the
next section, is another multifunction system
that can help with all kinds of noise.
12.6.4 DSP Noise Reduction
DSP noise reduction can actually look at
the statistics of the signal and noise and figure
out which is which and then reduce the noise
significantly. These adaptive filters can’t
quite eliminate the noise, and need enough
of the desired signal to figure out what’s happening, so they won’t work if the signal is far
below the noise. Many DSP systems “color”
the resulting audio to a degree. Nonetheless,
they do improve the SNR of a signal in random or impulse noise. As with noise blankers,
receivers frequently offer two or more noise
reduction settings that apply different noise
reduction algorithms optimized for different
conditions. It’s always worth experimenting
with the radio’s features to find out which
work better. The DSP and Software Radio
Design chapter discusses these features in
more detail.
Receivers  12.27
12.7 VHF and UHF Receivers
Most of the basic ideas presented in previous sections apply equally well in receivers
that are intended for the VHF and UHF bands.
This section will focus on the difference between VHF/UHF and HF receivers.
12.7.1 FM Receivers
Narrow-band frequency modulation
(NBFM) is the most common mode used
on VHF and UHF. Fig 12.28A is a block
diagram of an FM receiver for the VHF/UHF
amateur bands.
FRONT END
A low-noise front end is desirable because
of the decreasing atmospheric noise level at
these frequencies and also because portable
gear often uses short rod antennas at ground
level. Nonetheless, the possibilities for gain
compression and harmonic IMD, multi-tone
IMD and cross modulation are also substantial. Therefore dynamic range is an important design consideration, especially if large,
high-gain antennas are used. FM limiting
should not occur until after the crystal filter.
Because of the high occupancy of the VHF/
UHF spectrum by powerful broadcast transmitters and nearby two-way radio services,
front-end preselection is desirable, so that a
Fig 12.28 — At A, block diagram of a typical VHF FM receiver. At B, a 2 meter to 10 meter receive converter (partial schematic; some
power supply connections omitted.)
12.28   Chapter 12
low noise figure can be achieved economically within the amateur band.
12.7.2 FM Receiver WeakSignal Performance
DOWN-CONVERSION
Down-conversion to the final IF can occur
in one or two stages. Favorite IFs are in the
5 to 10 MHz region, but at the higher freВ­
quencies rejection of the image 10 to 20 MHz
away can be difficult, requiring considerable
preselection. At the higher frequencies an
intermediate IF in the 30 to 50 MHz region
is a better choice. Fig 12.28A shows dual
down-conversion.
The noise bandwidth of the IF filter is not
much greater than twice the audio bandwidth
of the speech modulation, less than it would
be in wideband FM. Therefore such things
as capture effect, the threshold effect and the
noise quieting effect so familiar to wideband
FM are still operational, but somewhat less
so, in FM. For FM receivers, sensitivity is
specified in terms of a SINAD (see the Test
Equipment and Measurements chapter)
ratio of 12 dB. Typical values are –110 to
–125 dBm, depending on the low-noise RF
pre-amplification that often can be selected
or deselected (in strong signal environments).
IF FILTERS
The customary peak frequency deviation
in amateur FM on frequencies above 29 MHz
is about 5 kHz and the audio speech band extends to 3 kHz. This defines a maximum modulation index (defined as the deviation ratio)
of 5/3 = 1.67. An inspection of the Bessel
functions that describe the resulting FM signal shows that this condition confines most
of the 300 to 3000 Hz speech information
sidebands within a 15 kHz or so bandwidth.
Using filters of this bandwidth, channel separations of 20 or 25 kHz are achievable.
Many amateur FM transceivers are
channelВ­ized in steps that can vary from 1 to
25 kHz. For low distortion of the audio output
(after FM detection), this filter should have
good phase linearity across the bandwidth.
This would seem to preclude filters with very
steep descent outside the passband, which
tend to have very nonlinear phase near the
band edges. But since the amount of energy
in the higher speech frequencies is naturally
less, the actual distortion due to this effect
may be acceptable for speech purposes. The
normal practice is to apply pre-emphasis to
the higher speech frequencies at the transmitter and de-emphasis compensates at the
receiver.
LO PHASE NOISE
In an FM receiver, LO phase noise superimposes phase modulation, and therefore
frequency modulation, onto the desired signal. This reduces the ultimate signal-to-noise
ratio within the passband. This effect is called
“incidental FM (IFM).” The power density
of IFM (W/Hz) is proportional to the phase
noise power density (W/Hz) multiplied by
the square of the modulating frequency (the
familiar parabolic effect in FM). If the receiver uses high-frequency de-emphasis at
the audio output (–6 dB per octave from 300
to 3000 Hz, a common practice), the IFM
level at higher audio frequencies can be reduced. Ordinarily, as the signal increases the
Fig 12.29 — The NE/SA5204A,
NE/SA602A, NE/SA604A NBFM
ICs and the LM386 audio amplifier in a typical amateur application for 50 MHz.
LIMITING
After the filter, hard limiting of the IF is
needed to remove any amplitude modulation components. In a high-quality receiver,
special attention is given to any nonlinear
phase shift that might result from the limiter
circuit design. This is especially important
in data receivers in which phase response
must be controlled. In amateur receivers for
speech it may be less important. Also, the
ratio detector (see the Mixers, Modulators
and Demodulators chapter) largely eliminates the need for a limiter stage, although the
limiter approach is probably still preferred.
FM DETECTION
The discussion of this subject is deferred to
the Mixers, Modulators and Demodulators
chapter. Quadrature detection is used in some
popular FM multistage ICs. An example receiver IC will be presented later.
Receivers  12.29
noise would be “quieted” (that is, “captured”)
in an FM receiver, but in this case the signal and the phase noise riding “piggy back”
on the signal increase in the same proportion as described in the Oscillators and
Synthesizers chapter’s discuss of reciprocal mixing. IFM is not a significant problem
in modern FM radios, but phase noise can
become a concern for adjacent-channel interference.
As the signal becomes large the signal-tonoise ratio therefore approaches some final
value. A similar ultimate SNR effect occurs
in SSB receivers. On the other hand, a perfect
AM receiver tends to suppress LO phase
noise. (See the reference entry for Sabin.)
12.7.3 FM Receiver ICs
A wide variety of special ICs for communications-bandwidth FM receivers are
available. Many of these were designed for
“cordless” or mobile telephone applications
and are widely used. Fig 12.29 shows some
popular versions for a 50 MHz FM receiver.
One is an RF amplifier chip (NE/SA5204A)
for 50 W input to 50 W output with 20 dB
of gain. The second chip (NE/SA602A) is a
front-end device with an RF amplifier, mixer
and LO. The third is an IF amplifier, limiter
and quadrature FM detector (NE/SA604A)
that also has a very useful RSSI (logarithmic
Received Signal Strength Indicator) output
and also a “mute” function. The fourth is the
LM386, a widely used audio-amplifier chip.
Another FM receiver chip, complete in one
package, is the MC3371P.
The NE/SA5204A plus the two tuned circuits help to improve image rejection. An
alternative would be a single double-tuned
filter with some loss of noise figure. The
Mini-Circuits MAR/ERA series of MMIC
amplifiers are excellent devices also. The
crystal filters restrict the noise bandwidth
as well as the signal bandwidth. A cascade
of two low-cost filters is suggested by the
vendors. Half-lattice filters at 10 MHz are
shown, but a wide variety of alternatives, such
as ladder networks, are possible.
Another recent IC is the MC13135, which
features double conversion and two IF amplifier frequencies. This allows more gain on a
single chip with less of the cross coupling that
can degrade stability. This desirable feature
of multiple down-conversion was mentioned
previously in this chapter.
The diagram in Fig 12.29 is (intentionally)
only a general outline that shows how chips
can be combined to build complete equipment.
The design details and specific parts values
can be learned from a careful study of the data
sheets and application notes provided by the IC
vendors. Amateur designers should learn how
to use these data sheets and other information
such as application notes available (usually for
free) from the manufacturers or on the web.
12.7.4 VHF Receive
Converters
Rather than building an entire transceiver
for VHF SSB and CW, one approach is to use
a receive converter. A receive converter (also
called a downconverter) takes VHF signals
and converts them to an HF band for reception using existing receiver or transceiver as
a tunable IF.
Although many commercial transceivers cover the VHF bands (either multiband,
multimode VHF/UHF transceivers, or
HF+VHF transceivers), receive converters
are sometimes preferred for demanding applications because they may be used with
high-performance HF transceivers. Receive
converters are often packaged with a comВ­
panion transmit converter and control cir-
cuitry to make a transverter.
A typical 2 meter downconverter uses an IF
of 28-30 MHz. Signals on 2 meters are amplified by a low-noise front-end before mixing
with a 116 MHz LO. Fig 12.28B shows the
schematic for a high-performance converter.
The front-end design was contributed by
Ulrich Rohde, N1UL, who recommends a
triple-balanced mixer such as the Synergy
CVP206 or SLD-K5M.
The diplexer filter at the mixer output sВ­ elects
the difference product: (144 to 146 MHz)
– 116 = (28 to 30 MHz). A common-base
buffer amВ­plifier (the 2N5432 FET) and tuned
filter form the input to the 10 meter receiver.
(N1UL suggests that using an IF of 21 MHz
and an LO at 165 MHz would avoid interВ­
ference problems with 222 MHz band signals.) For additional oscillator designs, refer
to the papers on oscillators by N1UL in the
supplemental CD-ROM files for the Mixers,
Modulators and Demodulators chapter
В­accompanying this Handbook.
Based on the Philips BFG198 8 GHz transistor, the 20 dB gain front-end amplifier is
optimized for noise figure (NF is approximately 2.6 dB), not for input impedance. The
output circuit is optimized for best selectivity.
The transistor bias is designed for dc stability
at IC = 30 mA and VC = 6 V. Both of the transistor’s emitter terminals should be grounded
to prevent oscillation. NF might be improved
with a higher performance transistor, such
as a GaAs FET, but stability problems are
often encountered with FET designs in this
application.
If a mast-mounted preamplifier is used to
improve the system noise figure, an attenuator should be available to prevent overload.
Simulation predicts the circuit to have an IP3
figure of at least +25 dBm at 145 MHz with
an IC of 30 mA and a terminating impedance
of 50 W.
12.8 UHF and Microwave Techniques
The ultra high frequency spectrum comprises the range from 300 MHz to 3 GHz. All
of the basic principles of radio system design
and circuit design that have been discussed so
far apply as well in this range, but the higher
frequencies require some special thinking
about the methods of circuit design and the
devices that are used. Additional material
on construction for microwave circuits can
be found in the Construction Techniques
chapter and in the series of QST columns,
“Microwavelengths” by Paul Wade, W1GHZ.
12.8.1 UHF Construction
Modern receiver designs make use of
12.30   Chapter 12
highly miniaturized monolithic microwave
ICs (MMICs). Among these are the Avago
MODAMP and the Mini Circuits MAR and
MAV/ERA lines. They come in a wide variety of gains, intercepts and noise figures
for frequency ranges from dc to well into
the GHz range. (See the Component Data
and References chapter for information on
available parts.)
Fig 12.30 shows the schematic diagram
and the physical construction of a typical
RF circuit at 430 MHz. It is a GaAsFET
preamplifier intended for low noise SSB/
CW, moonbounce or satellite reception. The
construction uses ceramic chip capacitors,
small helical inductors and a stripline surface-
mount GaAsFET, all mounted on a G10 (two
layers of copper) glass-epoxy PC board. The
very short length of interconnection leads
is typical. The bottom of the PC board is a
ground plane. At this frequency, lumped components are still feasible, while microstrip
circuitry tends to be rather large.
At higher frequencies, microstrip methods
become more desirable in most cases because
of their smaller dimensions. However, the
advent of tiny chip capacitors and chip resistors has extended the frequency range of discrete components. For example, the literature
shows methods of building LC filters at as
high as 2 GHz or more, using chip capacitors and tiny helical inductors. Amplifier and
Fig 12.30 — GaAsFET preamplifier
schematic and construction details
for 430 MHz. Illustrates circuit, parts
layout and construction techniques
suitable for 430-MHz frequency
range.
C1 — 5.6 pF silver-mica or same as
C2.
C2 — 0.6 to 6 pF ceramic piston
trimmer (Johanson 5700 series or
equiv).
C3, C4, C5 — 200 pF ceramic chip.
C6, C7 — 0.1 µF disc ceramic, 50 V or
greater.
C8 — 15 pF silver-mica.
C9 — 500 to 1000 pF feedthrough.
D1 — 16 to 30 V, 500 mW Zener
(1N966B or equiv).
D2 — 1N914, 1N4148 or any diode
with ratings of at least 25 PIV at
50 mA or greater.
J1, J2 — Female chassis-mount
Type-N connectors, PTFE dielectric
(UG-58 or equiv).
L1, L2 — 3t, #24 tinned wire,
0.110-inch ID spaced 1 wire dia.
L3 — 5t, #24 tinned wire, 3⁄16-inch ID,
spaced 1 wire dia. or closer. Slightly
larger diameter (0.010 inch) may be
required with some FETs.
L4, L6 — 1t #24 tinned wire, 1⁄8-inch
ID.
L5 — 4t #24 tinned wire, 1⁄8-inch ID,
spaced 1 wire dia.
Q1 — Mitsubishi MGF1402.
R1 — 200 or 500-W Cermet potentiometer (initially set to midrange).
R2 — 62 W, 1⁄4 W.
R3 — 51 W, 1⁄8 W carbon composition
resistor, 5% tolerance.
RFC1 — 5t #26 enameled wire on a
ferrite bead.
U1 — 5 V, 100-mA 3 terminal regulator
(LM78L05 or equiv. TO-92 package).
Receivers  12.31
mixer circuits operate at well into the GHz
range using these types of components on
controlled-dielectric PC board material such
as Duroid or on ceramic substrates.
Current designs emphasize simplicity of
construction and adjustment, leading to “no
tune” designs. The use of printed-circuit
microstrip filters that require little or no adjustment, along with IC or MMIC devices,
or discrete transistors, in precise PC-board
layouts that have been carefully worked out,
make it much easier to “get going” on the
higher frequencies.
12.8.2 UHF Design Aids
Circuit design and evaluation at the higher
frequencies usually require some kind of minimal lab facilities, such as a signal generator,
a calibrated noise generator and, hopefully,
some kind of simple (or surplus) spectrum
analyzer. This is true because circuit behavior
and stability depend on a number of factors
that are difficult to “guess at,” and intuition
is often unreliable. The ideal instrument is
a vector network analyzer with all of the attachments (such as an S parameter measuring
setup), an instrument that has become surprisingly affordable in recent years. (See the Test
Equipment and Measurements chapter.)
Another very desirable thing would be
a circuit design and analysis program for
the personal computer. Software packages
created especially for UHF and microwave
circuit design are available. They tend to be
somewhat expensive, but worthwhile for a serious designer. Inexpensive SPICE programs
are a good compromise but have significant
limitations at VHF and above. See the chapter on Computer-Aided Circuit Design for
information on these tools.
12.8.3 A 902 to 928 MHz
(33 cm) Receiver
This 902 MHz downconverter is a fairly
typical example of receiver design methods
for the 500 to 3000 MHz range, in which
down-conversion to an existing HF receiver
(or 2 meter multimode receiver) is the most
convenient and cost-effective approach for
most amateurs. At higher frequencies a
double down-conversion with a first IF of
200 MHz or so, to improve image rejection,
might be necessary. Usually, though, the presence of strong signals at image frequencies
is less likely. Image-reducing mixers plus
down-conversion to 28 MHz is also coming
into use, when strong interfering signals are
not likely at the image frequency.
Fig 12.31A is the block diagram of the
902 MHz down-converting receiver. A cavity
resonator at the antenna input provides high
selectivity with low loss. The first RF amplifier is a GaAsFET. Two additional 902 MHz
band-pass microstrip filters and a second RF
amplifier transistor provide more gain and
image rejection (at RF – 56 MHz) for the
mixer. The output is at 28.0 MHz so that
an HF receiver can be used as a tunable IF/
demodulator stage.
CUMULATIVE NOISE FIGURE
Fig 12.31B shows the cumulative noise
figure (NF) of the signal path, including the
28 MHz receiver. The 1.5 dB cumulative NF
of the input cavity and first RF-amplifier combination, considered by itself, is degraded to
1.9 dB by the rest of the system following
the first RF amplifier. The NF values of the
various components for this example are reasonable, but may vary somewhat for actual
hardware. Also, losses prior to the input such
as transmission line losses (very important)
are not included. They would be part of the
complete receive system analysis, however.
It is common practice to place a low noise
preamp outdoors, right at the antenna, to
overcome coax loss (and to permit use of
less expensive coax).
Fig 12.31 — A downconverter for the 902 to 928 MHz band. At A: block diagram; At B: cumulative noise figure of the signal path; At C:
alternative LO multiplier using a phase locked loop.
12.32   Chapter 12
LOCAL OSCILLATOR (LO) DESIGN
The +7-dBm LO at 874 to 900 MHz is derived from a set of crystal oscillators and frequency multipliers, separated by band-pass
filters. These filters prevent a wide assortment
of spurious frequencies from appearing at the
mixer LO port. They also enhance the ability
of the doubler stage to generate the second
harmonic. That is, they have very low impedance at the input frequency, thereby causing
a large current to flow at the fundamental
frequency. This increases the nonlinearity
of the circuit, which increases the secondharmonic component. The higher filter impedance at the second harmonic produces a
large harmonic output.
For very narrow-bandwidth use, such as
EME, the crystal oscillators are often oven
controlled or otherwise temperature compensated. The entire LO chain must be of
low-noise design and the mixer should have
good isolation from LO port to RF port (to
minimize noise transfer from LO to RF).
A phase-locked loop using GHz range
prescalers (as shown in Fig 12.31C) is an
alternative to the multiplier chain. The di-
vide-by-N block is a simplification; in practice, an auxiliary dual-modulus divider (see
the Oscillators and Synthesizers chapter)
would be involved in this segment. The cascaded 902 MHz band-pass filters in the signal
path should attenuate any image frequency
noise (at RF–56 MHz) that might degrade the
mixer noise figure.
12.8.4 Microwave Receivers
The world above 3 GHz is a vast territory
with a special and complex technology well
beyond the scope of this chapter. We will
scratch the surface by describing a specific
receiver for the 10 GHz frequency range and
point out some of the important special features that are unique to this frequency range.
A 10 GHZ PREAMPLIFIER
Fig 12.32B is a schematic and parts list,
Fig 12.32C is a PC board parts layout and
Fig 12.32A is a photograph of a 10 GHz
preamp, designed by Senior ARRL Lab
Engineer Zack Lau, W1VT. With very careful design and packaging techniques a noise
figure approaching the 1 to 1.5 dB range was
achieved. This depends on an accurate 50-W
generator impedance and noise matching
the input using a microwave circuit-design
program such as Touchstone or Harmonica.
Note that microstrip capacitors, inductors
and transmission-line segments are used
almost exclusively. The circuit is built on
a 15-mil Duroid PC board. In general, this
kind of performance requires some elegant
measurement equipment that few amateurs
have. On the other hand, preamp noise figures
in the 2 to 4-dB range are much easier to get
(with simple test equipment) and are often
satisfactory for amateur terrestrial communication.
Articles written by those with expertise
and the necessary lab facilities almost always
include PC board patterns, parts lists and detailed instructions that are easily duplicated
by readers. Microwave ham clubs and their
publications are a good way to get started in
microwave amateur technology.
Because of the frequencies involved, dimensions of microstrip circuitry must be very
accurate. Dimensional stability and dielectric
constant reliability of the boards must be very
good.
Fig 12.32 — At A, a low-noise preamplifier for 10 GHz, illustrating the methods used at microwaves. At B: schematic. At C: PC board
layout. Use 15-mil 5880 Duroid, dielectric constant of 2.2 and a dissipation factor of 0.0011. A template of the PC board is available on
the CD-ROM included with this book.
C1, C4 — 1 pF ATC 100 A chip capacitors.
C1 must be very low loss.
C2, C3 — 1000 pF chip capacitors. (Not
critical.) The ones from Mini Circuits
work fine.
F1, F2 — Pieces of copper foil used to
tune the preamp.
J1, J2 — SMA jacks. Ideally these should
be microstrip launchers. The pin should
be flush against the board.
L1, L2 — The 15 mil lead length going
through the board to the ground plane.
R1, R2 — 51 W chip resistors.
Z1-Z15 — Microstrip lines etched on the
PC board.
Receivers  12.33
Analysis of a 10.368 GHz communication link with SSB modulation:
Free space path loss (FSPL) over a 50-mile line-of-sight path (S) at F = 10.368 GHz:
FSPL = 36.6 (dB) + 20 log F (MHz) + 20 log S (Mi) = 36.6 + 80.3 + 34 = 150.9 dB.
Effective isotropic radiated power (EIRP) from transmitter:
EIRP (dBm) = PXMIT (dBm) + Antenna Gain (dBi)
The antenna is a 2-ft diameter (D) dish whose gain GA (dBi) is:
GA = 7.0 + 20 log D (ft) + 20 log F (GHz) = 7.0 + 6.0 + 20.32 = 33.3 dBi
Assume a transmission-line loss LT, of 3 dB
The transmitter power PT = 0.5 (mW PEP) = –3 (dBm PEP)
PXMIT = PT (dBm PEP) – LT (dB) = (–3) – (3) = –6 (dBm PEP)
EIRP = PXMIT + GA = –6 + 33.3 = 27.3 (dBm PEP)
Using these numbers the received signal level is:
PRCVD = EIRP (dBm) – Path loss (dB) = 27.3 (dBm PEP) – 150.9 (dB) = –123.6 (dBm PEP)
Add to this a receive antenna gain of 17 dB. The received signal is then PRCVD = –123.6 +17 = –106.6 dBm
Now find the receiver’s ability to receive the signal:
The antenna noise temperature TA is 200 K. The receiver noise figure NFR is 6 dB (FR=3.98, noise temperature
TR = 864.5 K) and its noise bandwidth (B) is 2400 Hz. The feedline loss LL is 3 dB (F = 2.00, noise temperature TL
= 288.6 K). The system noise temperature is:
TS =TA + TL + (LL) (TR)
TS = 200 + 288.6 + (2.0) (864.5) = 2217.6 K
NS = kTSB = 1.38 × 10–23 × 2217.6 × 2400 = 7.34 × 10–17 W = –131.3 dBm
This indicates that the PEP signal is –106.6 –( –131.3) = 24.7 dB above the noise level. However, because the
average power of speech, using a speech processor, is about 8 dB less than PEP, the average signal power is
about 16.7 dB above the noise level.
To find the system noise factor FS we note that the system noise is proportional to the system temperature TS and
the “generator” (antenna) noise is proportional to the antenna temperature TA. Using the idea of a “system noise
factor”:
FS = TS / TA = 2217.6 / 200 = 11.09 = 10.45 dB.
If the antenna temperature were 290 K the system noise figure would be 9.0 dB, which is precisely the sum of
receiver and receiver coax noise figures (6.0 + 3.0).
Fig
system performance
performancecalculation.
calculation.
Noise
temperature
noise
factor
of receiver
the receiver
are
Fig 14.37—Example
12.33 — Example of
of aa 10-GHz system
Noise
temperature
andand
noise
factor
of the
are considconsidered
in detail.
ered in detail.
14.40
Chapter 14
System Performance
At microwaves, an estimation of system
performance can often be performed using
known data about the signal path terrain, atmosphere, transmitter and receivers systems.
In the present context of receiver design we
wish to establish an approximate goal for the
receiver system, including the antenna and
transmission line. Fig 12.33 shows a simplified example of how this works.
A more detailed analysis includes terrain
variations, refraction effects, the Earth’s curvature, diffraction effects and interactions
with the atmosphere’s chemical constituents
and temperature gradients.
In microwave work, where very low noise
levels and low noise figures are encountered,
experimenters like to use the “effective noise
temperature” concept, rather than noise factor. The relationship between the two is given
by
TE = 290 (F – 1)
(8)
where the noise factor F = 10NF/10 and NF is
the noise figure in dB.
TE is a measure, in terms of temperature,
of the “excess noise” of a component (such
12.34   Chapter 12
as an amplifier). A resistor at TE would have
the same available noise power as the device
(referred to the device’s input) specified by
TE. For a passive device (such as a lossy transmission line or filter) that introduces no noise
of its own, TE is zero and G is a number less
than one equal to the power loss of the device.
The cascade of noise temperatures is similar
to the formula for cascaded noise factors.
the ratio of total system output noise to that
system output noise attributed to the “generator” alone, regardless of the temperature of
the equipment or the nature of the generator,
which may be an antenna at some arbitrary
temperature, for example. This ratio is, in
fact, a special “system noise factor (or figure),
FS” that need not be tied to any particular
temperature such as 290 K. (Note that regular
noise factor (or figure) does depend on referTE1 TE3
TE 4
TS =TG + TE1 +
+
+
+ ... ence temperature.) The use of the FS notaG1 G1G 2 G1G 2 G3
tion avoids any confusion. As the example of
(9)
Fig 12.33 shows, the value of this system
where TS is the system noise temperature noise factor FS is just the ratio of the total sys(including the generator, which may be an tem temperature to the antenna temperature.
Having calculated a system noise temperaantenna) and TG is the noise temperature of
ture,
the receive system noise floor (that is,
the generator or the field of view of the antenna, usually assumed 290 В°K for terrestrial the antenna input level of a signal that would
exactly equal system noise, both observed
communications.
The number 290 in the formulas for TE at the receiver output) associated with that
is the standard ambient temperature (in kel- temperature is:
vins) at which the noise factor of a two-port
transducer is defined and measured, accord- N = k TS BN(10)
ing to an IEEE recommendation. So those
formulas relate a noise factor F, measured where
k = 1.38 × 10-23 (Stefan-Boltzmann’s
at 290 K, to the temperature TE. In general,
constant) and
though, it is perfectly correct to say that the
BN= noise bandwidth
ratio (SI/NI)/(SO/NO) can be thought of as
The system noise figure FS is indicated in
the example also. It is higher than the sum of
the receiver and coax noise figures.
The example includes a loss of 3 dB in the
receiver transmission line. The formula for
TS in the example shows that this loss has a
double effect on the system noise temperature,
once in the second term (288.6) and again in
the third term (2.0). If the receiver (or highgain preamp with a 6 dB NF) were mounted
at the antenna, the receive-system noise temperature would be reduced to 1064.5 K and
a system noise figure, FS, of 7.26 dB, a very
substantial improvement. Thus, it is the common practice to mount a preamp at the antenna,
although transmission line loss must still be
included in system noise figure calculations.
MICROWAVE RECEIVER
FOR 10 GHZ
Here is a good example of amateur techniques for the 10 GHz band. The intended
use for the radio is narrowband CW and
SSB work, which requires extremely good
frequency stability in the LO. Here, we will
discuss the receiver circuit.
Block Diagram
Fig 12.34 is a block diagram of the receiver.
Here are some important facets of the design.
1) The antenna should have sufficient gain.
At 10 GHz, gains of 30 dBi are not difficult
to get, as the example of Fig 12.33 demonstrates. A 4-foot dish might be difficult to
aim, however.
2) For best results a very low-noise preamp at the antenna reduces loss of system
sensitivity when antenna temperature is low.
For example, if the antenna temperature at
a quiet direction of the sky is 50 K and the
receiver noise figure is 4 dB (due in part to
transmission-line loss), the system temperature is 488 K for a system noise figure of
4.3 dB. If the receiver noise figure is reduced
to 1.5 dB by adding a preamp at the antenna
the system temperature is reduced to 170 K
for a system noise figure of 2.0 dB, which is
a very big improvement.
3) After two stages of RF amplification
using GaAsFETs, a probe-coupled cavity
resonator attenuates noise at the mixer’s
image frequency, which is 10.368 – 0.288
Fig 12.34 — A block diagram of the microwave receiver discussed in the text.
Receivers  12.35
= 10.080 GHz. An image reduction of 15 to
20 dB is enough to prevent image frequency
noise generated by the RF amplifiers from
affecting the mixer’s noise figure.
4) The single-balanced diode mixer uses a
“rat-race” 180° hybrid. Each terminal of the
ring is 1вЃ„4 wavelength (90В°) from its closest
neighbors. So the anodes of the two diodes
are 180В° (1вЃ„2 wavelength) apart with respect
to the LO port, but in-phase with respect to
the RF port. The inductors (L1, L2) connected to ground present a low impedance at the
IF frequency. The mixer microstrip circuit
is carefully “tweaked” to improve system
performance. Use the better mixer in the
transmitter.
5) The crystal oscillator is a fifth-overtone
Butler circuit that is capable of high stabil-
ity. The crystal frequency error and drift are
multiplied 96 times (10.224/0.1065), so for
narrowband SSB or CW work it may be difficult to get on (and stay on) the “calling
frequency” at 10.368 GHz. One acceptable
(not perfect) solution might be to count the
106.5 MHz with a frequency counter whose
internal clock is constantly compared with
WWV. Adjust to 106.5 MHz as required.
At times there may be a small Doppler shift
on the WWV signal. It may be necessary
to switch to a different WWV frequency, or
WWV’s signals may not be strong enough.
Surplus frequency standards of high quality
are sometimes available. Many operators just
“tune” over the expected range of uncertainty.
6) The frequency multiplier chain has
numerous band-pass filters to “purify” the
harmonics by reducing various frequency
components that might affect the signal path
and cause spurious responses. The final filter
is a tuned cavity resonator that reduces spurs
from previous stages. Oscillator phase noise
amplitude is multiplied by 96 also, so the
oscillator must have very good short-term
stability to prevent contamination of the desired signal.
7) A second hybrid splitter provides an
LO output for the transmitter section of the
radio. The 50-Ω resistor improves isolation
between the two output ports. The original
two-part QST article (see the references) is
recommended reading for this very interesting project, which provides a fairly straightforward (but not extremely simple) way to
get started on 10 GHz.
12.9 References and Bibliography
J. Dickert, “A New Automatic Noise
Limiter,” QST, Nov 1938, pp 19-21.
W. Hayward, W7ZOI, R. Campbell,
KK7B, and B. Larkin, W7PUA,
Experimental Methods in RF Design.
(ARRL, Newington, 2009).
R. Henderson, WI5W, “A Rock Bending
Receiver for 7 MHz,” QST, Aug 1995,
pp 22-25.
12.36   Chapter 12
McClanning and Vito, Radio Receiver
Design, Noble Publishing Corporation,
2001.
H. Robinson, W3LW, “Audio Output
Limiters for Improving the Signal-toNoise Ratio in Reception,” QST, Feb
1936, pp 27-29.
Rohde, Whittaker, and Bucher,
Communications Receivers, 2nd Edition,
McGraw-Hill, 1997.
Rohde and Whittaker, Communications
Receivers: DSP, Software Radios, and
Design, 3rd Edition, McGraw-Hill
Professional, 2000.
W. Sabin, “Envelope Detection and AM
Noise-Figure Measurement,” RF Design,
Nov 1988, p 29.
Z. Lau, W1VT, “Home-Brewing a 10-GHz
SSB/CW Transverter,” QST, May and
Jun 1993.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement