This set did a lot better than last year, new ferrite rod
coils and a 10k pot for regen control. The circuit is the very
simple mpf102 jfets regen from Owen Pooles website, with a
separate antenna coil loosely coupled to the detector coil
and a 10K pot between the tickler coil and the jfet.

Both coils are 210 uH, the tickler coil is 5 turns added to
the ferrite rod and secured with electrical tape. All coils
44/40 litz. I tried a few different jfet's and voltages in
the circuit and all were about the same so I stuck with the
mpf102 and 9 volt battery although the circuit works very
well at 4.5 volts. The beauty of this circuit is decent
performance in a circuit that just can't get any simpler.

Description. My 2007 1-AD Contest set was the “Modern Homodyne,”
designed and described by G.W. Short in a highly recommended article:

http://www.vintageradio.me.uk/radconnav/transtrf/

I made several modifications to, and omissions from,
Short's original circuit:

1. ferrite rod coil L1 replaced by basketwound 660/46 litz coil

2. no gimmick capacitor across gate resistor R1
-- not necessary at BCB frequencies

3. no shunt detector diode D3 in the half-wave doubler
– simple, single diode detector D4 only

4. no low-pass audio filter -- it eliminates highs,
but it is a real detriment to DX listening

5. no final audio amp stage -- not necessary
when using sound powered phones

6. no ground plane under circuit board or other shielding
(actually, I forgot these, and the circuit seems
to work just fine without them).

Performance.

Antenna. The Homodyne uses no outside antenna. Its tuning coil
serves as the antenna -- in this case it's the 6" diameter
(that's right, 6 INCH) litz basketwound coil L1, mounted as
a rotatable, tuned loop on top of the set. This coil is all
the antenna there is. I'm reasonably sure of no coupling
going on to my disconnected outside antenna. The lead-in is
several feet away; the set works well in other rooms far
away; and the loop exhibits nice directionality. But, larger
coils tried, including a 1-meter diameter loop antenna,
tended to overload the set and ruin its selectivity.

Selectivity. Feedback works like none I've ever experienced.
There is no popping in and out of regeneration. As feedback
is increased, the squeal-free tuning range about the desired
station narrows, and the volume increases. Further increasing
feedback, the squeal-free region closes out, and tuning the
station is very touchy -- just like a BFO’ed exalted-carrier
set which is always in oscillation. In essence, feedback
control R7 can vary the selectivity from broad as a barn
to razor sharp, depending upon one’s need..

Noise. One drawback to the Homodyne is it is very sensitive
to line noise – perhaps because it is so sensitive to
everything else as well. During evening listening sessions
I can hear the ambient line noise level steadily increase
as neighbors switch on SCR light dimmers, cable TV’s,
fluorescents, etc. This din usually increases to a point
when some idiot switches on something defective, and the
noise becomes as loud as an SCR operating in my own house.
Unfortunately, the power conduit to the house is right
outside the radio room wall. Switching off the main power
provides little relief. By then, the noise has been induced
into the phone lines, and the cat’s out of the bag. I get a
good measure of relief by retreating with the set to an
outpost on the electrically quiet lanai (enclosed sun porch)
in back. Having only a 6-inch loop antenna makes this feasible.

Wave-Trapping. The directional pattern of the rotatable loop
antenna reduces the need for wave traps for all but the strongest
interfering signals. I got by in the contest fairly nicely
without using any wave traps. Since the loop antenna pattern
has sharp nulls off the faces and broad peaks off the ends,
directing the antenna to exactly null an interfering station
usually leaves enough gain for the weak DX station, whatever
direction it may lie in. One really weird effect I haven’t had
time to investigate is, on strong local stations and in-house
line noise, the loop antenna shows a sharp null off only one
face but not the other. This means the loop is showing a
cardiod (heart-shaped) pattern instead of the usual figure-eight.
One of the real beauties of a loop antenna is that, unlike
other traps, it lets you ‘trap’ interfering signals on the
same frequency as the desired DX signal.

Without a doubt, the Homodyne is the finest broadcast band
DX set I’ve ever operated -- bar none. That includes the likes
of the Hammarlund HQ-180A, Collins 51S-1, and others.

How It Works. Drawing extensively from G.W. Short’s
article, here is how the set works:
Amplifier. The set is basically a two-stage RF
amplifier using a “direct-coupled, complementary-cascade”
transistor pair TR1-TR2. All tuning is done by L1-C2,
the 6-inch basket coil-variable capacitor tank. Little
amplification is done by the 2N3819 JFET TR1 – it serves
mainly to match the high impedance input from the tuning
tank, preserving the tank’s high Q. Most of the actual
gain comes from the PNP germanium
AF 239 UHF-type transistor TR2.

Oscillator. Positive feedback* derived at the collector of the
AF 239 (‘collector follower’) is applied to the 2N3819 gate
(input) to generate strong oscillation at approximately the
same frequency as the incoming carrier. The 10 k-ohm
pot R7 controls feedback level.

(* In going from 2N3819 gate to AF 239 collector, the phase
is inverted twice, for a 360 degree ‘total phase shift,’ thus
leaving the feedback in-phase with the input. At least,
that’s true at the lower frequencies including the broadcast
band. At short waves, phase shift creeps in
and must be compensated for.)

Locking. An important feature of this local oscillation is
that it will lock onto the frequency of the incoming signal’s
carrier, even if the tank is not tuned exactly to that carrier.
Ideally when the oscillator tank is tuned to the ‘vicinity’
of a weak signal, it will lock on that signal and not on much
stronger signals that may lie in adjacent channels.

Limiter. The front-to-back diode pair D1-D2 serves as a limiter
that suppresses the modulation of the fed-back carrier and
prevents the level of oscillation from becoming too large.
Ideally the feedback consists of un-modulated carrier of fixed
amplitude. The 10 k-ohm fixed resistor R6 feeding the limiter
allows good clipping and at the same time prevents the limiter
from loading down the detector input.

Demodulator / Mixer / Detector. Demodulation (detection) occurs
at diode D4 that goes to the output. Incoming signal
(carrier-plus-sidebands = RF input) is mixed with the
locally-generated oscillation (LO) at the carrier’s frequency.
The difference frequency product of the two mixed signals
emerges as the so-called ‘intermediate frequency’ (IF) which
in our case is simply the audio frequency. Any mixing products
of the LO with adjacent channel signals are much higher in
frequency and thus out of the audible range. This is how
‘direct conversion’ receivers of this type
get their excellent selectivity.

Acknowledgments: Special thanks to Macrohenry for originally
recommending the Homodyne circuit to me, and to Brian Hawes for
tracking down the special AF 239 transistor
as well as the 2N3819 FET for me.