9808035.pdf

By Wes Hayward, W7ZOI, and Terry White, K7TAU

A Spectrum Analyzer for

the Radio Amateur

Good tools are priceless when you

need them. Here’s a piece of test

equipment you’ve always wanted

for your workbench. Now you can

have it—without spending a fortune.

A mong the many measurement

ments in the 50 kHz to 70 MHz region. The

design can be extended easily into the VHF

and UHF region with methods outlined

later. The instrument is configured to be

self-calibrating, or capable of calibration

with simple home-built test gear. 1

We often read and hear about “simple

designs.” Simplicity implies that some-

thing is eliminated to make the equipment

easier to build, use or afford. Unlike de-

signs that sacrifice performance for cost

and simplicity, this one sacrifices only con-

venience, while retaining the capabilities

needed for accurate measurements.

tools sought by the amateur

experimenter, the most de-

sired—but generally considered

the least accessible—is the radio-frequency

spectrum analyzer or SA. This need not be.

Simple and easily duplicated, this home-

built analyzer is capable of useful measure-

1 Notes appear on page 43.

Figure 1—Block diagram of the spectrum analyzer. The circuit is a double-conversion superheterodyne design with intermediate

frequencies of 110 and 10 MHz.

August 1998

Figure 2—Time base for the spectrum analyzer. Refer to the text for a discussion of the various circuit functions. Front-panel controls

include SWEEP RATE , SPAN and TUNE . Unless otherwise specified, resistors are 1 / 4 W, 5% tolerance carbon-composition or film units.

Equivalent parts can be substituted.

U401, U402, U403—LM358 op amp

D403, D406—6.2 V Zener diodes, 1 W

C401—Metal film or Mylar, 1.0 m F

capacitor

R420, R423—PC-mount trim pots, 5 k

control, 5 k

suitable. If a 10

turn pot is used for R3, R4 is not

needed.

or 10 k

10 k W suitable

R3, R4, R5, R6—Panel-mounted linear

Modern technology eases the construc-

tion of this spectrum analyzer. The loga-

rithmic amplifier uses an IF amplifier IC

found in cellular telephones and includes a

received signal strength indicator (RSSI)

function. Hybrid and monolithic IC build-

ing blocks are employed extensively. These

include mixers, amplifiers and VCOs—all

vital elements in an analyzer. Finally, it is

a rare devoted experimenter today who

does not own an oscilloscope. With good

basic ’scopes available for about the price

of a hand-held FM transceiver, every ex-

perimenter should have one. Our spectrum

analyzer uses a ’scope as the display. There

August 1998

are no special requirements for ’scope per-

formance other than an X-Y mode with dc

coupling in the X and Y axes.

dow, while preserving system dynamic

range. The proper distribution of gain, se-

lectivity and signal-handling capability (in-

tercepts) of the amplifiers and mixers is

vital to achieving good performance in a

spectrum analyzer, and indeed, any re-

ceiver. A proper design will have the same

number of stages as a poor one, but will

probably use different components and

consume more current.

The analyzer uses a

circuit commonly found in function gen-

erators. U401A operates as an integrator;

current is pulled from the inverting input

through a 56-kW resistor connected to the

SWEEP RATE control. This current must

flow through the capacitor (C401), creat-

ing a linearly changing op-amp output volt-

age. This ramp is applied to U401B, a re-

generative comparator, which provides a

reset signal to the integrator. The sawtooth

waveform (pin 1 of U401A) is asymmetri-

cal: The positive-going ramp grows with a

slope determined by the front-panel-

mounted SWEEP RATE pot, while the nega-

tive-going, faster reset ramp is determined

by fixed-value components.

The U401 ramp is used twice. U402A and

B process the ramp to generate a signal that

drives the ’scope’s X axis. The signal has a

0 V-centered range with just over a 10 V

total swing. Some of the “square wave” from

the basic time base (U401B, pin 7) is added

to the input of U402B to cause the sweep to

reset quickly, even though the return sweep

for the VCO occurs in a more stable, smooth

way. A slight overscan is generated for the X

axis, serving to hide an aberration occurring

near the sweep beginning.

The sweep also generates the signal

that controls the VCO. The sweep signal

(U401A pin 1) is applied to a SPAN control.

When the analyzer is set for maximum

span, the VCO voltage (about 2 to 10 V)

generates a sweep from 110 to 180 MHz.

The VCO uses only positive sweep volt-

ages, so the output of U403B is diode-

clamped to prevent negative output. The

center frequency TUNE , FINE TUNE and a

MAX SPAN calibration pot set up the proper

sweep for maximum span. As the span is

reduced with the SPAN control, the sweep

expands on (or zeroes in on) whatever ap-

pears at the center of the screen, determined

by the tuning. The center frequency must

be set for 35 MHz at maximum span, which

coincides with having the zero signal, or

Some Spectrum-Analysis Basics

The RF spectrum analyzer is essentially

a swept receiver with a visual display. The

display shows the strength of all signals

within a user-defined frequency span. Each

signal is represented by a line or blip that

rises out of a background noise, much like

the action of an S meter. Commercial ana-

lyzers are calibrated for signal power, with

all signals referred to a reference level

at the top of the screen. Our analyzer is

designed for a basic reference level of

–30 dBm, a common value in commercial

analyzers. 2

Signal levels are read from the display

by noting that power drops by 10 dB for

each major division on the ’scope. You can

change the reference level. Adding gain to

the analyzer moves the reference to lower

levels; introducing attenuation ahead

of the instrument moves the reference to

higher power levels.

15 V power supply.

The positive supply delivers about 0.5 A.

The negative supply current drain is under

50 mA.

Following sections present the circuit

blocks in greater detail, in the order that

they should be built. The partial but grow-

ing system can then be used to test the other

sections as they are built, turned on and

integrated. We strongly discourage build-

ing the entire analyzer before testing spe-

cific sections. Such an approach may work

for casual kits, but is not suitable when

careful control of signal levels is required.

That approach also robs you (the builder)

of the excitement of the process: the learn-

ing that comes from detailed examination.

Before jumping into the circuit details,

we reemphasize that this analyzer—al-

though simple—is intended for serious

measurements. This means that a normal

maximum span display contains no spuri-

ous signals. When clean (well-filtered, har-

monic-free) signals are applied to the ana-

lyzer, there should be no extra products as

long as the signal level is kept on screen.

This performance goal applies for a single

tone, or for two equal signals at the top of

the screen.

Circuit Overview

Figure 1 is a block diagram of our spec-

trum analyzer. A double-conversion super-

heterodyne, it begins with a step attenua-

tor, followed by a low-pass filter and the

first mixer, where incoming signals are

upconverted to a 110 MHz first IF. After

some gain and band-pass filtering, a sec-

ond conversion moves the signals to a

10 MHz IF. The resolution bandwidths

available are 30 kHz and 300 kHz. A video

filter smooths or averages noise. The avail-

able frequency spans range from a per-

division maximum of 7 MHz to about

50 kHz. The center frequency can be ad-

justed over the entire 70 MHz range. An

uncalibrated SPAN control allows expan-

sion of the display about the screen center.

An uncalibrated SWEEP RATE control

allows the sweep to be controlled and

matched to a given span while avoiding

excessively fast scans that could introduce

errors.

Ideally, a receiver’s first IF should be

greater than twice the highest input fre-

quency, a design rule that we bend in this

application. The input tuning range in-

cludes all HF amateur bands and 6 meters.

(We’ll discuss higher tuning ranges later.)

We picked the 10 MHz second IF because

surplus-crystal filters and LC filters for this

frequency are easily built. You can easily

adapt the design’s IF to 10.7 MHz, or other

close, convenient values.

The swept LO tunes from 110 to

180 MHz with a commercial VCO module.

The VCO output is amplified to drive a

high-level-input mixer. The commercial

VCO is a recent modification to a design

that started with a homebrew oscillator. 3

Amplifiers are included at the 10 and

110 MHz IFs. These establish signal levels

that properly match the log-amplifier win-

Time Base

Figure 2 shows the analyzer time base,

designed for basic functionality without

frills; the result is a circuit using only a

handful of op amps. 4 U401A and U401B

form a free-running sawtooth generator, a

Figure 3—An experimental logarithmic amplifier breadboarded to evaluate performance

prior to analyzer construction. You may want to duplicate this circuit and analyze its

performance if you decide to use other log-amp ICs.

August 1998

transforming networks. The bandwidth of

the AD8307 is about 500 MHz, so care is

required in its use.

Our analyzer uses the inexpensive and

readily available MC3356 log amp shown

in Figure 5. 8 An op amp, U303, used to in-

crease the signal output to 0.5 V per divi-

sion, follows the log chip, U301. The 0 V

level corresponds to the bottom of the

screen; a signal of 4 V brings the response

to the top of the screen. The op-amp output

is slightly higher than this, but is then at-

tenuated with a LOG AMP CAL control, R2.

This pot should be accessible from the out-

side of the instrument.

The log amp is preceded by an IF ampli-

fier, Q301 through Q303. These stages are

biased for relatively high-current operation

to preserve linearity. Gain is controlled

through variable emitter degeneration in

the form of a PIN diode, D301. Most com-

mon 1N4000-series power rectifiers work

well for gain control. The IF GAIN ADJ con-

trol (R1) should be available from the exte-

rior of the RF-tight amplifier box. We have

placed it on the front panel of our analyzers.

Calibration of the IF and log amplifier is

straightforward. First, set the ’scope’s Y

axis to 0.5 V/division and short it. Set the

now-working time base to drive the X axis

and adjust the ’scope’s vertical position

control to place the horizontal line at the

bottom of the screen. Inject a

Figure 4—Transfer characteristics for three different logarithmic amplifier ICs. Although

the MC3356 is used in our analyzers, use of the AD8307, shown in the lower curve, is

recommended. Some curves have been linearly scaled to ease comparison.

“zero spur” at the left edge of the screen.

Setting up the time-base function is gen-

erally straightforward. The ’scope can be

used to debug, check and study the circuits.

The X-axis signal is a ramp ranging from

–6 to +6 V with a reset to –15 V during the

retrace. A similar ramp appears (without a

reset pulse) at the VCO output, but with an

amplitude dependent on the SPAN control

setting.

Although the op amps are carefully by-

passed, and the signal that tunes the VCO is

shielded, most circuits are noncritical.

Normal op-amp circuit precautions are

taken with resistors injecting signals into

inverting inputs positioned close to the

op-amps. 5

A 10-turn front-panel-mounted pot is

used for the TUNE control (any value from

5 k

rithmic processing is used.

The circuit element we use for this pro-

cessing is the log amplifier. 7 The term is a

misnomer, for the usual log-amp IC is both

a logarithmic processor (amplifier) and a

detector. The chips provide a dc output

voltage that increases in proportion to the

logarithm of the input amplitude. The cen-

tral sensitivity specification for a log amp

is a voltage slope that is equal to the volt-

age change (per decade or per decibel) of

input-voltage-amplitude change.

An experimental log amp is shown in

Figure 3. We breadboarded and tested this

circuit to evaluate the log IC. To produce

the MC3356 curve shown in Figure 4, the

10 MHz output of an HP-8654 signal gen-

erator was applied through HP-355 step

attenuators. Exact dc output levels are in-

significant, for they can be adjusted with

dc voltage gain in a following amplifier.

The salient detail that we observe is the dy-

namic input window. The MC3356, with a

50-

10 dBm sig-

nal from a signal generator into the log

amplifier input, remove the short circuit

and adjust R2, LOG AMP CAL , for a full-

screen (reference level) response. The in-

put level is next reduced in 10 dB steps.

The horizontal sweep line should drop

down one major division for each 10 dB

reduction over a 60 dB range. If this does

not happen, repeat the procedure with a

slightly different drive level. In our analyz-

ers, a typical drive level of –13 dBm pro-

duced good accuracy.

Now, attach the IF amplifier to the log

amplifier and drive them with an input

level of –23 dBm. Peak the IF output filter

for maximum response and set R1, IF GAIN

ADJ , for a full-screen response. A true fil-

ter peak can be confirmed by varying the

generator frequency. There is considerable

extra range in the IF GAIN ADJ , providing

extra flexibility during use.

is suitable). A single-turn

pot can be substituted if a 10-turn pot is not

available. A fine-tuning function is in-

cluded in this design, but may be omitted if

a 10-turn pot is used for the main tuning.

to 50 k

input termination, produces a nearly

straight-line output voltage versus input

power for inputs in the –80 to –10 dBm

range. Hence, the analyzer log amp should

operate with an input signal of –10 dBm for

signals at the top of the screen.

We evaluated two other ICs. One, the

commonly available NE/SA604, shows

considerable ripple. The best performance

offered came from a recently introduced

chip from Analog Devices: the AD8307.

This IC is designed specifically for mea-

surement applications and offers outstand-

ing logarithmic accuracy, a dynamic range

exceeding 90 dB and better temperature

stability than found with the usual cellular-

receiver chips. The AD8307 requires a high

drive level, so it must be preceded with

higher-power amplifiers or impedance-

Log Amplifier and Detector

Central to any spectrum analyzer is a

logarithmic amplifier . The need for loga-

rithmic processing becomes clear if we

consider the range of signals we want

to measure: At the low end, we may want

to look at submicrovolt levels: under

–107 dBm in a 50-

Resolution Filters

Continuing the backward progression

through the system, we encounter the

resolution-bandwidth-determining filters.

Our analyzer uses bandwidths of 30 and

300 kHz, provided by crystal and LC fil-

ters, respectively. The 300 kHz LC filter,

the crystal filter and the relay circuitry for

bandwidth switching are shown in Figure

6. Although shown as individual modules,

they can be incorporated into one. The PC

board for the filter includes the LC filter

and switching relays with room for a user-

selected crystal filter. Builders may want

to implement their own scheme here. We

system. At the other

extreme, we may want to measure the out-

put of small transmitters, perhaps up to a

power of 1 W, or +30 dBm. The difference

between the two levels is 137 dB. The hu-

man ear is capable of handling linear

ranges well over 60 dB. 6 This is a wide

dynamic range world and linear displays,

such as our screen, are inadequate unless

some form of data compression or loga-

August 1998

Figure 5—The 10 MHz IF amplifier and log amplifier used in the analyzer. Refer to the text for adjustment details. Unless otherwise

specified, resistors are 1 / 4 W, 5% tolerance carbon-composition or film units. Equivalent parts can be substituted.

D301—PIN diode; 1N4007 used

L301—1.35 uH, 18 turns #24 enameled

wire on T-44-6 core, Q >150

Q301, Q302, Q303—2N3904

C309—Plastic dielectric trim cap

(Sprague-Goodman GYD65000)

C307, C308, C310—Silver mica or NP0

ceramic capacitors, 10% tolerance

C316—0.22 m F ceramic

R1—Panel-mount, 1 k W linear

R2—Panel-mount, 5 k

linear

U301—Motorola MC3356

U302—78L05 +5 V regulator

U303—CA3140 op amp

reasoned that builders would want to imple-

ment their own ideas. Maintain reasonable

shielding for this part of the system. Addi-

tional attenuator pads can be inserted in line

with one filter or the other to approximately

equalize filter loss in the two paths.

You may want to build crystal filters for

your analyzer. 9 The VCO stability in this

analyzer will support resolution band-

widths as narrow as 3 to 5 kHz. For a sim-

plified beginning, a very practical analyzer

can be built with only one resolution band-

width of 300 kHz.

Temporarily replace the crystal, Y201,

with a 51

front end. Accordingly, the second mixer

should have an intercept similar to that of

the first mixer. This is the usual weak point

in all too many homebrew spectrum analyz-

ers—as well as more than a few receivers!

The second mixer, U202, uses a +17 dBm

level Mini-Circuits TUF-1H. This is not

the place for a current-starved telephone

component! The second mixer is terminated

in a high-pass/low-pass diplexer followed

by an IF amplifier (Q202) biased at 50 mA.

This is a critical stage for dynamic range:

Don’t replace it with a monolithic substi-

tute of reduced gain or intercept.

The second LO begins with a 100 MHz,

fifth-overtone crystal oscillator (Q201),

followed by a pad and a power amplifier.

The oscillator inductor, L201, in Q201’s

collector is made of five turns of #22 wire

wound on a 6-32 machine screw. (Remove

the screw before installing the coil.) Here’s

an excellent way to align the oscillator:

resistor. Adjust the tuned cir-

cuit until oscillation occurs at the desired

100 MHz frequency. Then, replace the

resistor with the 100 MHz crystal; no

further tuning is required. Measure the

oscillator’s output with a power meter be-

fore applying it to U202. Adjust the pad

attenuation (R205, R206, R207) to realize

the specified LO drive level.

After the second LO is operating, attach

it to the second mixer and the rest of the

analyzer. With a second mixer input of

–35 dBm at 110 MHz, you should obtain a

reference-level response.

Second Mixer and Second Local

Oscillator

Figure 7 shows the second mixer and

related LO. The heart of this module—and

to some extent that of the entire analyzer—

is U202, a high-level second mixer. This

mixer is bombarded by large signals that

are as strong or stronger than those at the

Voltage-Controlled Local Oscillator

and First Mixer

Figure 8 shows the analyzer’s swept LO.

The foundation for this module is a Mini-

Circuits POS-200 VCO module, U101.

Similar VCOs are available from many

August 1998

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: