Designing and Building

Transistor Linear Power Amplifiers

Part 2 — Apply techniques from Part 1 to single band HF
and 6 meter linear amplifiers.
Rick Campbell, KK7B

In 
Part 1 of this series, I 20 times (13 dB) more power,
described an experi- even in a low power (QRP)
mental method for contest. One more stage of
designing a linear amplifier amplification added to the
starting with a blank sheet of 0.25 W amplifier can overcome
paper, some basic test equip- that 13 dB difference. The
ment and an assortment of can- 5 to 10 W output level is a
didate transistors. In December common standard for portable
2006 QST I described a single radios with many commercially
band SSB exciter with 0 dBm available examples.
output.1 An output level of
0 dBm (1 mW) is very com- Putting Power in the
mon for signal interconnec- Antenna
tions between 50 Ω blocks in Figure 2 is the block diagram
radio systems. At this power of an experimental single-band
level, the SSB exciter output 36 dB gain 5 W linear amplifier
may be connected directly to that may be easily constructed
an antenna for very low power using whatever output device is
experiments, it may be ampli- available. Two noteworthy dif-
fied to any desired output level ferences between Figure 2 and
or it may be converted to a other commonly published cir-
different frequency using a cuits are the use of a resistive
mixer and VFO. It could also Figure 1 — attenuator and low-pass filter
be connected to a simple RF Some MicroT2 between the driver and final
clipper followed by a filter to applications. stage, and the floating ground
obtain higher average-to-peak at the final amplifier device.
ratio SSB, or it might even be These two features make it easy
converted back down to the to experiment with different
audio range with a second oscillator for a transmitted signal and noise and interfer- final amplifier transistors without mechani-
number of interesting analog signal process- ence are low at the receiver. I’ve played that cal headaches or oscillations.
ing applications. Several of these applica- game, and every contact entered into the log Figure 3 is the schematic of a 7 MHz
tions are illustrated in Figure 1. is cause for a little celebration. But the bands version of the amplifier.1 It was optimized to
are not always kind and high levels of noise use common, inexpensive ($0.79) switching
How Much Power do We Need? and interference are common. You may power supply transistors. Since the 2N5739
Once the signal has been moved around clearly hear the station on the other end of is not designed as an RF device, there are
and processed at the 0 dBm power level the contact but he is probably running at least no suggested RF operating conditions in
and is at the desired output frequency, most
applications will require more power. If
0.25 W is enough, then the amplifier
described in Experimental Methods in RF
Design, Figure 2.93 and included in the
December 2006 QST article is highly recom-
mended.2 Many on-air contacts have been
made at that power level over remarkable
distances when band conditions enhance the
1Notes appear at the end of the article.. Figure 2 — A 37 dB gain linear amplifier.
the data sheet. The operating conditions rying much about burning out the device. Give Me Power
were obtained experimentally by vary- I also varied the base bias, and changed The collector power supply for the output
ing the supply voltages while watching the drive level with a step attenuator. As stage is a common circuit, with a big capaci-
the output waveforms. The values of the expected, increased collector voltage made tor instead of the expected three terminal
π attenuators between stages were selected a big improvement in the linearity of strong regulator. That gives me about 18 V open-
experimentally for best gain and distor- signals. Increased base bias improved the circuit, and about 16 V at maximum output.
tion distribution among the three amplifier linearity of small signals. The big capacitor is split in two. The little
stages. The single-section low-pass filter on Since 100% modulated AM, SSB and box on the floor holds 2200 µF while another
the output of the driver transistor made a sig- two-tone outputs vary from some peak volt- 3500 µF is in the box with the speaker, vari-
nificant reduction in high-order intermodu- age all the way to zero, both collector supply able bias supply, TR relay and 12 V three
lation products, and seemed to improve the voltage and base bias determine the linearity terminal regulator. The regulator supplies
symmetry of the intermodulation distortion of the output. Each can also be used to destroy regulated voltage to the receiver, transmitter
products as well. the device. Too much collector voltage will and other amplifier stages.
burn out the transistor directly (remember The big capacitor provides the low
Tweaking it into Submission that the voltage at the collector will generally impedance at audio needed in a SSB linear
Figure 4 is the single-tone output spec- swing to significantly higher than twice the amplifier. By splitting it in two all of the
trum of the exciter driving the amplifier in supply voltage, even in a linear amplifier). components in the power supply and regula-
Figure 3 with the two-tone output spectrum Too much base bias will either destroy the tor circuitry are physically and electrically
shown in Figure 5. Excellent linearity was transistor quickly as it conducts too much close to a big reservoir capacitor. Keeping
obtained at a PEP output level of several collector current and overheats, or slowly as power supply lines clean is particularly
watts. the base-emitter junction warms up and the important around receivers. It is a simple
Since the amplifier is experimental and device goes into thermal runaway. I enjoyed power supply for an inexpensive transistor,
the parts are inexpensive, I adjusted the col- exploring these options in the design phase and any efficiency I would have gained by
lector supply voltage on the output stage of this amplifier, but have not burned out a using an expensive 13.8 V linear RF power
up and down and observed the impact on device since selecting the component values transistor in one of Granberg’s wonderfully
AM and two-tone waveforms without wor- and supply voltages shown in Figure 3. engineered circuits is more than offset by

Figure 3 — 7 MHz linear amplifier based on inexpensive active devices.

 QS0903-Campbell04 QS0903-Campbell05
0 0
-10 -10
-20 -20
-30 -30
-40 -40
dB -50 dB -50
-60 -60
-70 -70
-80 -80
-90 -90
-100 -100
0 2000 4000 6000 8000 10,000 12,000 0 2000 4000 6000 8000 10,000 12,000
Hz from Carrier Frequency Hz from Carrier Frequency

Figure 4 — 7.165 MHz LSB single-tone output spectrum at 1 W Figure 5 — 7.165 MHz LSB two-tone output spectrum at 1 W
(black) and 3 W (red) output. The desired LSB output tone is at (black) and 4 W (red) output. The carrier is at 5.0 kHz on this
3.7 kHz and the carrier is at 5.0 kHz on this display at about display at about 42 dB below the 1 W tone. Note that third
42 dB below the 1 W tone. Note suppressed upper sideband order products near the desired two-tone output increase by
tones at 6.3 kHz and unsuppressed modulator harmonic 3 dB for each 1 dB increase in output power level. Also note
distortion on upper sideband between about 9 kHz. the unsuppressed modulator harmonic distortion on the upper
sideband between 8 and 9 kHz.

eliminating the series regulator. For more

power, I’ll experiment with operating the
transistor closer to its breakdown limits. I
don’t mind burning out a few output transis-
tors during these experiments, because the
transistor is easy to change and costs less
than a cup of coffee.

Construction Techniques
Figure 6 is a photograph of the bias con-
trol supply and TR switching. The meter
reads current in the final transistor while
the knob below the meter is used to set the
quiescent (no signal) bias level. The power
transformer was mounted where the speaker
is now, but after a few experiments with
magnetic shielding to eliminate residual
Figure 6 — Classic 40 meter solid state QRP SSB station.
hum in the companion MicroR2 receiver, I
stopped fighting basic physics and cured the
problem by putting the power transformer, 16.7 MHz crystal oscillator and tripler, and the calculated using the formulas in QST and
rectifier and half of the capacitance in a box other with an 8.35 MHz oscillator, tripler and The 2009 ARRL Handbook.4,5
on the floor.3 This ancient cure, adopted doubler. An assortment of different crystals The ×3 and ×2 frequency multipliers were
long ago, was once common with sensitive was ordered from several different vendors, built on a scrap of unetched copperclad board,
receivers, and is still standard practice for including Peterson Radio and International. using ugly construction. If using ugly con-
sensitive audio and scientific instruments. The variable crystal oscillator (VXO) and struction at VHF, it is important to sketch the
After I removed the power transformer, ×6 multiplier schematic are shown in layout first and think about which leads need
I encountered another fringe benefit: any Figure 7. The VXO circuit is the original to be short and which ones can be longer.
source of 18 V dc will now power the whole circuitry in the MicroT2 circuit. The only The output of the doubler is about 1 mW
transceiver. Twelve AA cells in series will change from the values in the original 40 at 50 MHz, with all spurs suppressed at
run the receiver for days as well as power meter VXO described in QST is the value least 45 dB below the desired output. These
intermittent SSB transmitting for 4 hours of the low pass output transformer. This was spurs are further suppressed by the tuned
or so.

Putting the MicroT2 on 6 Meters Figure 7 — Prototype

6 meter VXO tripler
The fundamental crystal oscillator in the and doubler. Note
MicroT2 won’t go above about 25 MHz with the short leads,
gimmick capacitors
common crystals, so another approach must and symmetry in the
be used for the bands above 12 meters. The layout. Such circuitry
ancient lore suggests frequency multipliers, is common on the low
and that approach was chosen for simplicity. VHF bands, and works
exceptionally well.
Two 6 meter exciters were built, one with a The schematic is on
the gollowing page.
 scraps left on bench

��� ���

�� ��
�����
��� ����� � ��� ������
� � �� �� ��� �� ��� ���
��� ����� � �� ���
�� ����
��� �� �� ��
��
��� �� ��
�� ���
��
�����
��� ��� ���� �� ��
�� ��
�����
�� ��
�� ���
�����
������ ��� ��� ��� ������� �������
�� ��

���� �� ������� �� ��� �� ����� �� �������� ������ ���� � ��� ������ ������ ������� ������ ���
������ ������� ��������� ���������� � ����������� ��������
��� �� �� ������� ����� ����� ��� ����� �������� �� �� ������� �����
�
����� �� �� ������� ����� ��� ���� ��� �� �������� �� �������� ����������� ���� ��
�

Prototype ��������
version with VXO on same PC board.
There
������ �� are some minor circuit
� differences.
�
������ �

��� ���

�� ��
�����
��� �����
��� Rick Campbell
�
KK7B � ��
��� ���� November 2007
����� ��� ��� �
������� ���
��� �� ���
��
��
�� �����
��� � ���� ��
��� ���
��� ���
��� �� ��
��� ��
����
����� ��

��� � ����������� ������ �� ���������� ���� ��������� �� ���� �� ������� �� ��� �� ����� �
�� �� �������� �� �������
�� ��� ���������� ����� ���������� �� ��� ����� �������� �� �� ������� ����� ����� ��� ����� �������
����� ���� �� ������� ����� �� ������ ����� �������� �� �� ������� ����� ��� ���� ��� �� ������

Prototype ve
There are so
Figure 8 — Schematic of 6 meter amplifier.

RF amplifier in the MicroT2, and by the

narrow-band interstage tuning networks
in the linear amplifier. The tuned doubler
output drives the original MicroT2 buffer
amplifier circuit. The only other change
to the MicroT2 is retuning the RF ampli-
fier output to 50 MHz. That may be easily
done by changing MicroT2 L3 to 16 turns
on a T25-6 ferrite toroid, changing C21 to
a 20 pF trimmer, and leaving C20 out of the
Figure 9 —
Photograph of
circuit. If the MicroT2 RF amplifier stage is
the 6 meter 5 W built using ugly construction, short leads are
SSB transmitter. necessary, particularly for connections to the
The MicroT2 with gate of Q6.
×6 exciter is in
the black diecast Experiments with running the TUF-3
box in the lower mixers as third harmonic mixers with direct
right. The VXO IQ LO drive at 16.7 MHz were initially
knob is in the
middle, and the encouraging, with very good carrier sup-
MIC connector pression at 50 MHz, but distortion was high.
on the right. The A very simple rig could be built with that
crystal socket on approach.
the front panel
is a mistake The 8 MHz crystal from International
— the crystal in the large can provides a very stable fre-
belongs inside quency tuning range of about 50 kHz on
the box where
it is thermally 6 meters, over the useful range from
and electrically 50.115 to 50.165 MHz. The 16.7 MHz crys-
shielded. tal and tripler provides a narrower range,
 QS0903-Campbell11 QS0903-Campbell12
0 0
-10 -10

-20 -20

-30 -30
-40 -40

dB -50 dB -50
-60 -60

-70 -70
-80 -80

-90 -90
-100 -100
0 2000 4000 6000 8000 10,000 12,000 0 2000 4000 6000 8000 10,000 12,000
Hz from Carrier Frequency Hz from Carrier Frequency
MicroT2 1 tone Icom 502 1 tone MicroT2 2 tone Icom 502 2 tone

Figure 10 — MicroT2 and PA 50.125 MHz single-tone USB Figure 11 — MicroT2 and PA 50.125 MHz two-tone USB output
output spectrum compared with ICOM IC-502 single-tone output spectrum (black) compared with IC-502 two-tone output
spectrum. In each case the suppressed carrier is at 5500 Hz on spectrum (red). In each case the suppressed carrier is at 5500 Hz
this plot. The '502 output is about 0.5 W and the MicroT2 and PA is on this plot. The '502 output is about 1 W PEP and the MicroT2
at about 5 W output. Carrier suppression and opposite sideband and PA is at about 5 W PEP output. Carrier suppression and
suppression are not as good with the MicroT2, but in-channel opposite sideband suppression are not as good with the MicroT2,
distortion is lower, even if the ’502 is backed off 8 dB from its but in-channel distortion is lower, even when the '502 is backed
rated output. off 5 dB from its rated output.

about 35 kHz. These are observations with ure 8, which is used on both the driver and Figure 10 is the single-tone output spec-
a sample size of one, so I hesitate to draw final. The MRF 134 and 136 have enough trum at the 5 W level, compared with the
any firm conclusions. One observation is gain at 50 MHz that two stages are sufficient output of a commercial radio in the same
that the carrier suppression improves when to amplify the output of the MicroT2 exciter power class. Note that the carrier level is
the cans of the TUF-3 mixers are soldered to to the 5 W level. Figure 9 is a photograph higher than we would like, at only about
the circuit board all the way around. This is of the 6 meter amplifier connected to the 33 dB below the peak output power. The
prevented by solder mask on the current pro- 6 meter exciter with MicroT2 exciter and ×6 two-tone spectrum shown in Figure 11 is
duction circuit boards, but it is a worthwhilemultiplier. This amplifier is simply the pack- certainly good enough to use on the air. A
modification. For much better carrier sup- aged up version of the amplifier I tested on comparison between the spectra of these
pression, wider tuning range and transceive the bench. It is still a work in progress. Note two rigs on 6 meters illustrates a common
operation, a heterodyne approach from a how easily the final output transistor may be difference between phasing and filter SSB
lower frequency is recommended. changed to a different FET, or even a bipolar transmitters. Phasing rigs have lower distor-
transistor. tion in-channel, so they sound very good, but
The 6 Meter Amplifier Each modular section of the 6 meter filter rigs have fewer off-channel spurious
The 2SC5739 transistor has low gain at transmitter is separate, including the bias Ts outputs.
50 MHz, so different driver and final devices described in Part 1 and the output networks. This 6 meter SSB signal sounds excep-
were chosen for the 6 meter linear amplifier. The bias networks and low-pass filter are on tional on the air, and the carrier could be
In keeping with the experimental nature of separate boards. This makes it very easy to acquired and locked by a phase locked loop
the project, both of these devices came from measure, adjust and modify any part of the (PLL) for a signal with the fidelity of the best
the junk box, in fact they were unsoldered circuit, or insert additional gain or attenu- AM signal, at a fraction of the bandwidth
from a printed circuit (PC) board obtained ation anywhere. All of the pieces except and total radiated power. There is room for
at a flea market. The driver transistor is an the black diecast box are from the author’s experiments with such modes on the VHF
MRF134, and the final is a MRF136 power junk box, including the thick aluminum and UHF bands.
FET unsoldered from an old PC board. The plates used for heat sinks. Even the MicroT2 My personal choice is to use phasing for
MRF136 is capable of much more power exciter was pieced together from prototype QRP SSB rigs and VHF-UHF SSB, where
output than the 2SC5739, so a larger heat modules from the MicroT2 project. the off-channel products don’t generally
sink was used. The bias connection is differ- The plethora of 2200 µF, 50 V electro- bother anyone. For high power operation
ent as well, since the FETs require forward lytics will continue until that bag is empty. on crowded HF bands, I follow the phasing
bias of about 4 V on the gate. These small Because this transmitter was assembled at exciter with a crystal filter and heterodyne
variations are easily seen in the schematic low cost from the available ingredients, I system.
of Figure 8. refer to it as the “Blue Plate Special.” The
The performance of the broadband 4PDT relay in the lower right corner is the Checking it Twice
4:1 bias T described in Part 1 degrades TR relay, and includes an RF output and The procedure I use for a quick check
above 35 MHz, and with a 28 V supply, a muting to the receiver and 6 meter converter. of amplifier performance before putting a
12.5 Ω load impedance is too low for a 5 W The set of contacts next to the ground plane signal on the air is very different from the
class amplifier. A quick search of the junk are used for the 6 meter RF, an additional measurements and experiments I use to
box and a few minutes with a Smith Chart shield was added, and RF connections to the design an amplifier. With these amplifiers,
resulted in the output network seen in Fig- relay are made with small diameter coax. I connect a dummy load and wattmeter to
the output of the low-pass filter, and drive ment, it’s a choice, and in many cases it will or the ARRL Bookstore, ARRL order no.
the amplifier chain to saturation with a get through when nothing else will. SSB is 8799. Telephone 860-594-0355, or toll-free in
the US 888-277-5289; www.arrl.org/shop;
CW signal. I observe the current drain at this nice for casual conversations, digital modes [email protected].
saturated output level, and then set the rest- are wonderful if you don’t mind sharing the 3R. Campbell, KK7B, “The MicroR2 — An Easy

ing bias to about one tenth that level. These fun with a computer, but CW is simple and to Build ‘Single Signal’ SSB or CW Receiver,”
QST, Oct 2006, pp 28-33.
5 W amplifiers idle at around 100 mA, so the power advantage on transmit is only part 4The ARRL Handbook for Radio Communica-

that is 1.8 W for the 7 MHz amplifier and of the equation. tions, 2009 Edition. Available from your ARRL
dealer or the ARRL Bookstore, ARRL order
2.7 W for the 6 meter version. Then I speak The linear amplifiers described in this no. 0261 (Hardcover 0292). Telephone
into the microphone and observe that the article are remarkable in two ways: 860-594-0355, or toll-free in the US
output peaks are about 3 dB below the satu-  The experimental design procedure 888-277-5289; www.arrl.org/shop;
[email protected].
rated CW output level. That results in a very provides a real education in linear ampli- 5This circuitry is nearly identical to the circuit
nice sounding signal on the air. fier design, measurement, adjustment and used for the 40 meter version, and can use
If you have been doing the math, you construction. the same circuit traces. The output may be
taken from MicroT2 C41 by leaving R59, 60
can quickly estimate that my average output  They were designed around the devices and 61 off the circuit board. The VHF ugly
power on SSB ends up being about one tenth on hand, and built at nearly zero cost. constructed tripler and doubler circuitry is
the saturated CW output power. PEP output In that sense, they follow the best tradi- inserted in the MicroT2 circuit in place of
R60. Short small diameter coax or twisted
capability is considerably higher than the tions of the Amateur Radio service and inno- pair should be used for the interconnections.
average, but I prefer natural sounding SSB vative design engineering. This circuitry should be shielded from the
to the highly processed sound that results in RF output amplifier to improve carrier sup-
pression.
an average power output very close to the 1Notes
R. Campbell, KK7B, “The MicroT2 — A
PEP output capability of the amplifier. This Compact Single-Band SSB Transmitter,”
is personal preference, and directly related QST, Dec 2006, pp 28-33. See Part 1 for Rick’s biography. You can con-
to my willingness to switch to CW when sig-
2W. Hayward, W7ZOI, R. Campbell, KK7B, and tact Rick at 4105 NW Carlton Ct, Portland,
B. Larkin, W7PUA, Experimental Methods in OR 97229, or at [email protected].
nals are marginal. CW is no longer a require- RF Design. Available from your ARRL dealer