This first attempt at a totally homebrew project was inspired by the
Norcal 2N2222 building contest.

The circuits in this rig were patched together from the following sources:

FINAL AMPLIFIER
I used the Class A linear power amplifier for the WA7MLH dsb
transceiver, described on p. 202 of Hayward and DeMaw's SOLID
STATE DESIGN (1986), but simulating the 44C6 transistor with
8 paralleled 2n2222s. The WA7MLH amplifier is designed for a
standing current of 250 ma and an output of about 1 W PEP DSB.
With a 5 ohm collector resistor, this amp draws about 500 ma,
and has about 1 W PEP DSB output, as guesstimated from watching
the wiggle of a WM-2 needle while hollering 'Hola!' Greater
efficiency might be attained by using the recommended
1 ohm ballasting resistors in the emitter circuits of the
2n2222's, instead of the 10 ohm resistors found in my junkbox.

The output matching circuit is a controlled-Q L network with
values calculated using formulas from p. 53 of SOLID STATE
DESIGN. The low pass filter is a kit for 40 m from Hands
Electronics.

MIKE AMP
I used a 2n2222 mike preamp circuit from p. 60 of Rusgrove and
DeMaw's UNDERSTANDING AMATEUR RADIO (1977), and soldered an
electret mike element to the input.

BALANCED MODULATOR
This circuit comes from p. 202 of SOLID STATE DESIGN (the
WA7MLH dsb transceiver again). Unlike prior failed balanced
modulators, this was built on a separate circuit board from the
mike amp and the vfo. I measure less than 1 mW output from the
XCVR on transmit, with no audio input, using a WM-2 wattmeter on
the 100 mW scale.

RF AMPS
This circuit comes from DeMaw's W1FB QRP NOTEBOOK (Second
Edition). There, on p. 135,DeMaw daisy-chains four of these
broadband amplifiers to form an all-purpose rf amplifier.
To avoid overdriving the final amplifier, I measured output from
the final amp with first one, then two, and finally three of
these driver stages ahead of the final. A 10 mW crystal
oscillator output (as measured in a WM-2 wattmeter) was used as
input for the drivers. Extensive use of feedback, while
reducing output from each stage, helps ensure stability. The
broadband matching transformers used between stages in DeMaw's
original circuit have been omitted. One of these amps is also
used to amplify the vfo output for the receive mixer.

VFO/BUFFER
A 2n2222 VFO was built using the design for a Colpitts VFO given
\ on p. 34 of SOLID STATE DESIGN, together with the buffer stage
from p. 45 of the second edition of DeMaw's W1FB QRP NOTEBOOK
(Second Edition). It's plenty stable enough for casual phone
operation, even though I used silver-dipped mica caps instead of
polystyrene or NPO caps. Also, I'm running the VFO at too high a
voltage (about 11 V), with a seat-of-the-pants 12 V Zener
regulator circuit. I should have bought a 9.1 V Zener and
followed the Zener regulator design information given in SOLID
STATE DESIGN (p. 156). Instead, I slapped a 12 V Zener from
VFO collector to ground, put resistors in series with the
collector to drop my power supply's 14 V down closer to, but
still above, 12 V, until the Zener was no longer running hot to
the touch. I don't know how much regulating it does, but the
VFO works.

RCVR
For the receiver, I built a simple direct conversion circuit,
using the 2 - 2n2222 audio amplifier shown on p. 81 of Rusgrove
and DeMaw's UNDERSTANDING AMATEUR RADIO. The mixer amp
is shown in SOLID STATE DESIGN on p. 75. The simple preselector
circuit comes from DeMaw's design for a Universal Direct
Conversion Receiver, on p. 79 of the W1FB QRP NOTEBOOK (Second
Edition). The preselector often functions as an rf gain control,
detuning the input to prevent overload.