[HBR] What would W6TC do?

[Walt Hutchens]

About a month ago, I posted:
> The question comes from thinking about some of the questions and ideas
> posted here in the last few months. Specifically, what would Ted Cosby
> do differently if he were designing an HBR today?
> 
> In order to keep the question 'pure,' I rule out whatever thoughts he
> might have on integrated circuits, software defined receivers, and the
> like: Perhaps like me he got frozen on vacuum tubes a few decades back
> and wanted today to do a tube receiver along the old lines but using
> the parts available for vacuum tube construction today.
> 
> I'm assuming the goals of 'you can build it' and 'it will outperform
> most commercial sets of equivalent technology' would stay the same. I
> think that would mean keeping the plug in coils.

Naturally I had a couple of ideas about the answer and I have stolen
some more from the suggestions of others.

1. I think the first conversion to an IF in the 2 Mcs range and use of
plug-in coils in the front end would continue. Too much of the 'you
can build it' and 'it will be a top performer' lies in these
fundamental choices.

Solenoid wound coils with heavy wire and decent quality forms give
about the same Q as a toroid -- something like 200 -- and they're
easier to wind and adjust. They are not self-shielding as toroids are,
but that's a space/layout issue. PVC pipe forms -- 1" schedule 20
epoxy mounted on the bakelite bases from metal octal tubes -- work
fine and look like a commercial product. I think the solenoids would
remain.

The plug in coil design not only gives excellent front end selectivity
thus helping crossmodulation performance but allows taking one of the
most difficult parts of construction (front end coil winding) in
bite-size pieces. And with a first conversion to around 2 Mcs you get
a good combination of image rejection and local oscillator ease of
construction.

The alternative of using a fix-tuned 1st oscillator with a tunable 1st
IF (or equivalently a tunable single band receiver with converters)
sacrifices both low cost and ease of getting parts due to the need for
a set of 1st osc. crystals. Crossmod performance will be degraded if a
broadband RF stage is used.

2. WARC and SWL band coils would be part of the plan for those that
want them. The use of an AM-FM three gang capacitor with jumpers in
the coils to connect the larger sections would be an option for those
wanting general coverage -- say 2-6-18-30 Mcs.

3. The second conversion to a low fixed IF (100 kcs or 85 kcs
according to the variant) would be eliminated. In the 1960 time frame
this was an easy and relatively inexpensive way to obtain a 3 kcs (or
so) bandwidth as the necessary IF transformers were open stock or
available from junk 'command' receivers, but not anymore.

There are two easier ways to go now: A half lattice filter in this
frequency range, or use of a filter salvaged from one of the less
expensive ham sets of the 1960-70 time frame.  I suspect that a
popular salvage item might set the frequency with the option of a
half lattice design using a pair of custom crystals.  HOWEVER junk
vacuum tube ham sets are often cheaper than a pair of made-to-order
crystals and almost any ham set filter will have better skirt
performance than a half lattice filter.

Bonus: The second oscillator and conversion are eliminated giving
additional simplification AND reducing image and crossmodulation
issues. I suspect that these issues are the main reason for all those
stage gain controls.

The HBR 2010 design would have a single IF somewhere between 1.6-2.4
Mcs. IF transformers would be single tuned using home made slug-tuned
coils OR salvaged from the junk set that donated the filter.

Slightly better performance would result from using toroidal IF coils
(with the bonus of eliminating the need for shielding) but I can't
come up with a compact, stable, and readily available trimmer
capacitor for tuning. (The itty-bitty rotary ceramic trimmers like
those in the FT-101 series are too delicate.) There probably IS an
answer; I just haven't found it.

4. The mixer might be changed to a dual triode.

Pentode mixers generally are the worst performers in a strong signal
environment but injecting the signal and the oscillator voltage on the
same grid allows running the oscillator on on half the required
frequency with the mixer grid tuned circuit selecting the harmonic. A
dual triode mixer would be a better performer but separating signal
and LO injection would require operating the oscillator at the
injection frequency.

The only way to make this decision is to build the receiver both ways.
It could be that with better layout, reduced power dissipation and
some simple trickery, warm-up drift can be sufficiently reduced that
the higher oscillator frequencies are okay.

5. The RF and IF stages would use 6EH7's. These tubes are capable of
gain that is enough higher than those in the original designs as to
cause stability problems for the novice constructor but you don't have
to use all the possible gain. With cathode resistors of 220 ohms or so
and 100 ohm resistors at the grid connection (to suppress VHF
parasitics) they're as tame as 6SK7s and you still have the
spectacular crossmodulation performance.

6. The audio stages would be a 'high fi' design using feedback from the
output transformer secondary to the cathode of the 1st audio to
achieve excellent linearity. Until you've listened to a receiver built
this way you've no idea of the resulting improvement in
intelligibility.

7. A simple fixed-frequency audio tuned circuit at the 1st audio input
would give selectivity in the 100 cps range for CW.

8. I think the physical layout could be improved and the elimination
of home-made shielding metalwork is both possible and a design plus.
The National 5:1 dials are no longer useful (even if you can find one)
but the JB 36/6:1 drive with a home made or National escutcheon would
work for those who can't find an Eddystone dial. It seems to me those
two would be the standard approaches.

The two-section tubes of the later W6TC sets would remain.
Compactrons and such offer no real advantages in this design and
increase the congestion around each socket.

9. Instead of the conventional multi-winding power transformer, a 15
or so watt isolation transformer would be used with a solid state full
wave bridge rectifier to deliver 130 volts for plate service and a 25
watt 12.6 VAC transformer would power the filaments. Almost nobody
makes multi-winding transformers for vacuum tube equipment anymore,
the usually-available plate voltages are much higher than desirable
(higher voltage means more heat and thus more thermal drift), and what
you can buy is almighty expensive. Two small competitive items are
cheaper than one that's nearly custom made.

The small Radio Shark 12.6 volt transformers are bobbin-wound with the
windings side-by side. A couple of times I've taken two of these
transformers apart, cut off the secondary bobbins, and reassembled the
two primaries on the same core for a plate transformer. But buying the
proper units is cheaper and less work.

In concept at least there is another way to go for power: Series
filaments and 'transformerless' high voltage. This further reduces the
total dissipation (small transformers aren't very efficient), expense,
and weight at the cost of some additional design and construction
wrinkles. If all the circuits in a large color TV can be made to work
without a power transformer, certainly a short wave receiver can.

Design changes are required: 18 watts of filament power is plenty for
a short wave receiver of the HBR genre but not all tube types have 150
mA versions. (For example as far as I can tell there is no 14EH7 or
12GM6.) The more difficult questions come from heater-to-cathode
leakage when a high impedance cathode circuit is necessary as in an
infinite impedance, plate, or diode detector, or Hartley oscillator.

Every technique or design error appears as either instability or hum
modulation of strong carriers. Hum isn't directly an issue -- you can
eliminate it in the audio section -- but it means that 60 cps is being
modulated onto the signal and the same will occur with voice signals,
appearing as distortion and loss of intelligibility. It has to be
eliminated at the source. The bag of tricks includes eliminating where
possible circuits in which the cathode isn't at signal ground, placing
the problem stage(s) near the neutral end of the filament string, and
(sometimes) biasing the cathode more negative than the negative peak
filament voltage on that stage.

In addition, wiring technique has to change to deal with the fact that
the chassis can't be used for the common negative. It's best to wire
all stages in star fashion from a central common negative AND
positive. Antenna, audio, and perhaps standby connections aren't
isolated from the power line by a power transformer so that has to be
done in other ways.  The use of a filtered IERC connector to bring in
the power line is helpful in reducing RF trash between the neutral and
ground wires.

All the issues can be resolved and the techniques can be learned but
this happens one circuit at a time and I'm still working on it. You
don't build a ship in a bottle because you want a model ship and I
wouldn't recommend a transformerless HBR as the way to go because you
want a receiver.

Walt
KJ4KV