[HBR] Revamping a HB RX

[Ian Wilson]

[Posted on Glowbugs. Kees suggested that I report here. So here you are].

A while ago, a list member generously made available to me a homebrew RX
that he'd built a few decades ago. It bubbled up to the top of my projects
pile a couple of weeks ago. Thought I'd share some of my experiences getting
it back on the air again.

The basic architecture is 'SuperX' - that is, a superhet, 1700kHz IF with LO
around 5MHz. That allows both 80m and 40m to be covered by selecting one or
other with front end tuning.

This particular RX was very similar to the 2X4 superhet that can be found on
the Glowbugs web site or in the 1963 ARRL Handbook (and others). It uses a
pair of 1700kHz xtals in a half-lattice (I think) arrangement. The parts
list calls for an "RF choke resonant around 1700 kc" but my receiver
contains a parallel inductor/capacitor combination of unknown values.

Original lineup used a 6U8A triode/pentode as oscillator+mixer; 6BA6 pentode
as IF amplifier; 6U8A as BFO osc/grid leak detector; 6CG7 dual triode audio;
voltage stabilizer.

I modified the LO to use the pentode as oscillator and used the triode as a
common-grid amplifier. This turned out not to be a very happy combination,
so I recast it as a Hartley oscillator (using the pentode screen & grid; fed
from stabilized HT; output taken from resistor in the plate lead) followed
by the triode as cathode follower. This has plenty of output, the level is
easily changed by the value of the plate resistor, and has a low impedance
output.

I tried a couple of mixers: 6CB6 pentode with cathode LO injection; and a
6J6 dual triode. In both cases I had severe problems with the mixer wanting
to oscillate at the frequency of its input tuned circuit. (They also had a
tendency to oscillate at VHF; a grid stopper resistor of 180 ohms cured
that). The only way I could overcome this was by moving the mixer to be
closer to the tuned circuits. A point to bear in mind when planning the
layout!

When I replaced the 6CB6 with the 6J6 mixer, there was quite a drop (10dB or
more) in sensitivity. Part of this is probably due to the lower plate
resistance of the triode (discussed earlier). The LO level was also rather
low; increasing this to 3-4V pk-pk helped noticeably. I considered using the
arrangement proposed by Chris Trask (taking the output from the plates tied
together) but since I am using dissimilar biasing on the 2 triodes, this
wouldn't produce much LO cancellation. My present arrangement is a Pullen
mixer, with the mixer triode being biased according to the GE tube data
under 'Mixer Service'.

The preselector that I wound [ha!] up using consists of 2 tuned circuits,
each with a 21uH toroid tuned by half of a 5-100pF variable capacitor. The
antenna is coupled with a few turns on one of the toroids. The two tuned
circuits are "top-coupled" with a 2.5pF capacitor. This is not what's used
in the various Handbook designs; I may experiment with this later. This ends
up being very sharp, needing to be retuned every 10kc or so, but does get
rid of out-of-band noise quite effectively. I can't assess 80m performance,
since there are no signals on the band here much of the time.

Replacing the 6BA6 IF amp with a 6BZ6 provided a worthwhile increase in gain
(needed to add a grid stopper here too). The 6BZ6 has about 2x the
transconductance of the 6BA6 without being untamable.

The end result (in its current incarnation) is working well with about 4' of
wire as an "antenna". The LO is reasonably stable (there are some thermal
issues in the original design that I can't do much about). I might buffer
the BFO, and use a different detector scheme. I'm not sure what the
selectivity of the xtal filter is. Will see what happens with SSB to get
some idea of that.

This is a fun radio, I've learnt a lot, and the fact that I've been able to
make changes without destroying it is a tribute to the original builder. If
I get an hour or two I will try to put a schematic together, but it may be a
while.

73, ian K3IMW