at the end of the cable, and you can see

Table 2
that the other end is shorted. The re-
flected pulse is inverted, as predicted Velocity Factor Tabulation: Practical
by the wave equation. The oscilloscope Transmission Line Velocity Factor Round Trip m / ns Round Trip ft / ns
time base is set to 10 ns per division, Space (vacuum) 1.00 0.15 0.49
and the time between the leading edges Air, 600 Ω open wire 0.98 0.15 0.48
of the incident and reflected pulses is
450 Ω twinlead 0.88 0.13 0.43
45 ns. Multiplying this by 0.39 ns/ft, the
length of this cable was calculated to 300 Ω twinlead 0.82 0.12 0.40
be 17.55 feet long (a tape measure read Foam coax, hardline 0.80 0.12 0.39
17.5 feet). Figure 10 is the same ar- Polyethylene coax 0.66 0.10 0.32
rangement, but with an open end.
Figure 11 is also an open end, but with
a 20 ns pulse; in Figure 12, with a
50 ns pulse, you see the superposition
that occurs if the reflected pulse returns
before the incident pulse is over. Use
whatever pulse width produces the best
reflected pulse. As Figure 13 demon-
strates, there is no reflected pulse when
the cable is terminated in its charac-
teristic impedance.
One Practical Application
A practical application of the
Pizzicato is shown in Figure 14. This is
my Butternut HF6V vertical antenna
at the end of 110 feet of direct-bury feed
line. Note that the oscilloscope time
base has been changed to 100 ns per
division, and that the pulse width is
100 ns. The step at 100 feet is the be-
ginning of a length of 75 Ω cable that
matches the vertical’s radiation resis-
tance of 35 Ω to the feed line. Figure 8 — Front view with violin.

Figure 9 — Shorted 10 ns pulse. Figure 10 — Open 10 ns pulse.

Figure 11 — Open 20 ns pulse. Figure 12 — Open 50 ns pulse.

Mar/Apr 2006 7
Figure 13 — Terminated with resistor. Figure 14 — Vertical antenna.

An S-shape is characteristic of an an- Figure 8) and a length of line with radar,’’’ Electronic Design, October 1,
tenna, and the additional ornamenta- PL-259 connectors. Be careful to sup- 1998.
tion is produced by the HF6V’s port the transmission line so its
bandswitching capacitors and coils. If weight does not overstress the ABS Continuously licensed since 1964, Gary
the feed line ever develops a flaw, I will enclosure. Happy plucking with your Steinbaugh, AF8L, is an ARRL Life Mem-
know within a foot or so where to start Pizzicato! ber and a licensed Professional Engineer.
digging! He holds a BSEE from Case Institute of
If your transmission lines have Notes
Technology, plays (using the word advis-
UHF connectors, a UG-255 BNC-to- 1Feynman, Richard Phillips. The Feynman edly) many different musical instruments,
SO-239 adapter will be a handy ac- Lectures on Physics. Reading, MA: and is a Certified Flight Instructor.
cessory, as will an extender made of Addison-Wesley, 1965. He may be reached at gsteinbaugh@
a PL-258 double-female adapter (see 2Englund, Gunnar. “Build your own ‘cable yahoo.com. ��

hams (www.arrl.org/ard/). aren’t enough example problem-and-

Volume 1 starts with the basics, sur- solution sets to really learn a sub-
Book Review veys common radio architectures, lays
out the Smith chart and s-parameters
ject exclusively from these books.
However, each section has an exten-
and then dives into impedance-match- sive set of references for further
Practical RF Circuit Design ing techniques. A section on the tech- study. In numerous areas, the au-
for Modern Wireless Systems, niques used to simulate and optimize thors present a useful set of design
Vol. 1 and 2
RF circuits is followed by a full 70 pages equations and principles, but avail-
on component models, an often-ne- able space prevents them from
by Rowan Gilmore and Les Besser, glected topic. The compromise is that covering the interactions and sensi-
Artech House Publishers, ISBN the next section, on filters and resonant tivities that inevitably occur in ac-
158053-522-4, $95 each volume. circuits, must cover much ground at a tual designs. For these, readers
Browsing one of my electrical engi- medium to high level. There is so much must do additional research. Never-
neering trade magazines, my eye caught literature on filters that it would be un- theless, there is enough solid foun-
on the publication notice for this two- reasonable to expect that depth to be dation material for an engineer to
volume set. In recent years, the RF reproduced here. The volume concludes create a basic design and enough
designer’s world has expanded well be- with a welcome discussion of the RF references to deal with some second-
yond what can be captured in a single characteristics of high-speed digital de- ary considerations. My only com-
handbook, leading to a blizzard of books signs, something that often comes as a plaint is that the graphic style is
on various specialties, but few that cap- surprise to digital designers. erratic, as is typical with heavy use
ture “the big picture.” Would this be the Volume 2 gets into the serious details of reprinted material. Many of the
answer? I borrowed two copies from the of RF amplification. It starts with lin- graphics would benefit as well from
library to find out. The two weeks I had ear RF amplifiers and continues to a more refined use of line width,
the books was obviously not enough time optimization and comparisons of the style, and density, but these are mi-
for an exhaustive review, but I believe I different designs. Modeling of active nor distractions, at worst.
was able to assess the books’ coverage. devices and nonlinear circuits is then These texts are available through
The two volumes divide the RF world covered as a detailed survey. Special con- inter-library loan, so you can evalu-
into passive circuits and systems (Vol- cerns of high-power amplifier design are ate them before making the invest-
ume 1) and active circuits and systems presented along with a design example. ment. At nearly $100 per volume,
(Volume 2). Computer-aided techniques Oscillators, mixers and multipliers each these are professional tools. Yet, if
get extensive coverage in each volume get an overview. The final section covers you work in the RF field and want a
and go hand-in-hand with each topic. several system-level examples, such as solid handbook to unify your niche ref-
The books assume that the reader mobile telephony, software-defined ra- erences and theoretical textbooks, the
understands the fundamentals of RF dios and radio chipsets. set would be a good use of your book
design and has access to RF computer- Even at 500 pages per volume, budget or a good addition to the com-
aided design, such as the limited-edi- there still isn’t enough space to pany library. — H. Ward Silver, NØAX,
tion of Super-Compact from Compact reach a detailed examination of 22916 107th Ave SW, Vashon, WA
Software that can be downloaded by many topics. For example, there 98070; [email protected] ��

8 Mar/Apr 2006
 A 100-W Class-D Power
Amplifier for LF and MF
Get high efficiency from a complementary
pair of inexpensive MOSFETs.

By Frederick H. Raab, W1FR; Mike Gladu, N1FBZ, and Dan Rupp

Introduction tion. The finished project is shown U101 is a standard 4-A gate driver
This paper describes a simple, high- in Figure 1. that converts the input signal into
efficiency transmitter for LF and MF Note: This is a project for experi- square-wave drive for the final. It is bi-
experimental operation. The power enced experimenters. The 165-V dc- ased near the threshold by adjustment
amplifier (PA) is a class-D design supply voltage can be lethal. Treat this of R103; hence, any input signal pushes
using a complementary pair of circuit with the same respect you the output of U101 into either a high or
MOSFETs.1 It operates from 165 V would give a vacuum-tube amplifier. low state. To adjust the threshold, re-
and delivers 100 W with an efficiency move the signal input and look at the
Driver drive output with a scope or voltmeter.
of 89%. The PA is configured for op-
eration at 500 kHz, but can be The driver (Figure 2) operates from Adjust R103 to increase the bias volt-
adapted to any frequency from +12 V and requires about 50 mA. The age until a change is observed in the
100 kHz to 3 MHz by changing the signal input can be a sine wave of 3 V drive output, then back off R101 slightly
output filter. The driver is a single (peak) or more (20 dBm), TTL, or a so the output of U101 rests in the low
gate-driver IC, and a one-transistor square wave of about 1 V peak or state. The resultant bias voltage is
circuit keys the drive for CW opera- more. Resistor R101 provides a load about 1.5 V. (Alternately, R101 can be
for the signal source. It should be re- adjusted so that U101 rests in the high
1Notes appear on page 13. moved for TTL or if necessary to get a state.)
sufficient output from the signal Keying for CW operation is accom-
Frederick H. Raab, W1FR source. An insufficient input signal plished by turning the drive on and
240 Stamford Rd can lead to chaotic oscillations at high off. This is done by Q101 and the re-
Burlington, VT 05401 power due to transients from the fi- lated components. An open circuit or
[email protected] nal entering the input to the driver. greater than +9 V at the key input

Mar/Apr 2006 9
J102 results in no drive, while a short
circuit or 0 V turns the drive on. This
allows operation from either key or
a programmable waveform genera-
tor with a range of 0 to 10 V. Capaci-
tor C110 and resistors R105 and
R106 control the shape of the wave-
form to avoid key clicks. The value
of 10 µF produces a rise/fall time of
about 10 ms.
Final Amplifier
The final amplifier (Figure 3) con-
sists of a complementary pair of
200-V power MOSFETs Q201 and
Q202. The MOSFETs are biased near
the threshold of conduction by R205 and
R206 so that any RF drive pushes one
into an on state and the other into an
off state. Figure 1 — The LF/MF complementary class-D transmitter.
Gate biases are supplied by two
Zener-regulated supplies. The bias for
n-channel MOSFET Q202 is derived
directly from the +12 V supply. The
source of p-channel MOSFET Q201 is
connected to the main supply VDD,
which may be variable. Consequently,
the bias supply for Q201 floats against
VDD. This is accomplished by coupling
the square-wave drive signal through
C205 to the floating supply consisting
of D203, D204 and C206, which pro-
vides a voltage of approximately
VDD – 12 V to operate the bias circuit.
Turning both Q201 and Q202 on si-
multaneously produces a short circuit.
Therefore, it is important to set the bi-
ases via the following procedure. Begin
by disconnecting C210 and C211 and
setting VDD to 0. Connect the +12-V sup-
ply, then adjust R205 to produce 0 V.
Apply drive, verify that the floating bias
supply is working, and set R206 to pro-
duce VDD – 0 V. Increase VDD and make
sure no drain current flows, then return
to zero. Temporarily connect a resistor
of approximately 100 Ω from the two
drains to ground. Set VDD to about 25 V.
Slowly adjust R206 until approximately
5 mA is flowing in Q201. The resultant
bias voltage is about VDD – 2.7 V for the
VP1220N5, but may be different for
other MOSFETs. Turn off VDD, remove
the 100-Ω resistor from the drains, and
Figure 2 — Circuit of driver.
connect a voltmeter in its place. Return
VDD to 25 V and slowly adjust R205 C101 — 20-µ µF, 250-V electrolytic. J104 — European-style binding post,
until the drain voltage is about 12.5 V, C102 — 3300-pF mica. black, E.F. Johnson 111-0103-001.
which implies equal quiescent currents C103 — 0.1-µ µF, 50-WV chip, Mount vertically.
in the two MOSFETs. The resultant bias ATC 200B104NP50X. R101 — 51-Ω Ω RC07 (omit for TTL input).
C104 — 0.1-µ µF disk. R102 — 4.7-kΩ Ω RC07.
voltage is about 3.5 V for the IRF610. C105-C109 — 1-µ µF, 50-WV X7R chip, R103 — 5-kΩ Ω trimpot, Bournes
The drain voltage will drift and vary with Kemet 1210C105K5RATCU. 3006P-1-502.
temperature, but the MOSFETs will re- C110 — 4.7-µ µF, 25-V electrolytic. R104 — 620-Ω Ω RC07.
main sufficiently close to their thresh- D101 — 1N751A, 5.1-V, 0.25-W Zener. R105, R106 — 2.2-kΩΩ RC07.
olds of conduction for proper operation D102 — 1N746A, 3.3-V, 0.25-W Zener. Q101 — PNP BJT, 2N2907.
when drive is applied. Finally, reconnect J101 — BNC jack. U101 — 4-A gate driver, TI UCC32325P or
J102 — Key jack, shorting. equivalent.
C210 and C211. J103 — European-style binding post,
The output network provides a red, E.F. Johnson 111-0102-001. Mount
50-Ω load to the drains, high imped- vertically.

10 Mar/Apr 2006
ances to the harmonics, adequate efficiency to ensure the PA is oper- drain capacitance through the induc-
suppression of the harmonics in the ating correctly. Then increase VDD to tance of a transformer.2,3 The small
output, and single-knob tuning to the 100 V and tweak C214 slightly to al- droops are the result of drain cur-
frequency of operation. The T configu- low for any variation of drain capaci- rent flowing through the on-state
ration may be regarded as two back- tance with voltage. Then increase resistances of the MOSFETs. The
to-back L networks that match VDD to its full value of about 165 V output vO is a pure sine wave.
upward to about 150 Ω at the center. without readjustment. The dc-input The output and efficiency charac-
Inductor L201 has a Q of 3, which is current should be about 0.72 A. teristics for operation at 500 kHz are
sufficient to ensure that third- shown in Figure 5. Output voltage
harmonic current circulating in the Performance Vom varies linearly with supply volt-
drains is no more than 10% of the The waveforms are shown in age V DD. The transfer curve fits a
fundamental-frequency current. Figure 4 for operation at 500 kHz straight line with an rms error of only
Adjustment is quite simple, since from VDD = 100 V. In this case, the 0.2%, indicating excellent amplitude-
maximum RF output, maximum dc input vi is a TTL signal. The drive modulation linearity (IMDs less than
input, and maximum efficiency occur signal v DR produced by U101 is a –50 dBc). The efficiency is about 89%
so close to each other that the settings square wave that ranges from 0 to over most of the output range. It
are indistinguishable. Begin with VDD 12 V. The drain voltage vD produced drops slightly at high levels because
set to 25 to 50 V. Adjust C214 for ei- by the two MOSFETs is a clean of increased MOSFET resistance
ther the maximum dc power or maxi- square wave. The complementary and slightly at low levels because of
mum RF output. Make a quick check configuration eliminates most of the increased drain capacitance.
on the RF output (Figure 4) and the transients associated with charging The PA can be used at frequen-

Figure 3 — Circuit of final amplifier.

C201 — 1-µµF, 50-WV X7R chip, Kemet D201, D202 — 7.5-V 0.25-W Zener, L203 — 30 turns #24 AWG on Ferroxcube
1210C105K5RATCU. 1N755A. 768XT188 3E2A toroid.
C202, C207, C111 — 0.1-µ µF, 50-WV chip, D203, D204 — MUR120 60-V 1-A ultra-fast Q201 — Supertex VP1220N5 200-V, 2-A
ATC 200B104NP50X. switching diode. P-channel MOSFET.
C203, C208, C215, C217, C219 — 3300-pF J201 — BNC jack. Q202 — IRF610 200-V, 2-A, N-channel
mica. J202 — European-style binding post, MOSFET.
C204, C209 — 5-µ µF 25-V electrolytic. green, E.F. Johnson 111-0104-001. R201, R202 — 1-kΩ Ω RC07.
C205, C210, C211 — 0.022-µ µF disk. J203 — European-style binding post, R203, R204 — 10-kΩ Ω RC07.
C206 — 10-µµF 25-V electrolytic. red, E.F. Johnson 111-0102-001. Mount R205, R206 — 10-kΩ Ω trimpot, top-
C212, C213, C218 — 0.1-µ µF, 200-WV chip vertically. adjustable, Bournes 3299X-1-103.
capacitor (American Technical Ceramics J204 — European-style binding post, Heat sink — Extruded aluminum, 3.75-in
900C104NP200). black, E.F. Johnson 111-0103-001. wide, Aavid 60675.
C214 — Three-gang breadslicer, 15-525 Mount vertically. Insulating pads for Q201, Q202 —
pF per section, 70% closed at 500 kHz. L201, L202 — 47.7 µH. 61 turns #24 AWG Thermalloy 43-77-2.
C216, C220 — 20-µ µF, 250-V electrolytic. enameled wire on Micrometals T200-2
toroid.

Mar/Apr 2006 11