[HBR] HB-65, anyone, anyone?

[Walt Hutchens]

Jim said:
> The HB-65 has some good ideas but it could be a lot better.

Yep.  And I generally agree with Jim about the details.

When you get past using single conversion, 455 kcs IF, on all bands --
prewar and some early postwar designs (HROs, etc.) -- there were four
recognized ways to handle the IF and conversion scheme for ham receivers:

1. Dual conversion with two fixed IFs as in the HB-65.  This offers both
selectivity without a crystal filter and adequate image rejection,
getting rid of the two main problems of using 455 kcs.  These designs
were pretty common -- the whole W6TC series of HBRs are examples.

 From a home constructor's point of view this is the simplest approach. 
  Q's are modest, meaning minimal issues with feedback.  Using any 
established design there's almost no way to go wrong if you use a 
crystal controlled oscillator for the 2nd conversion and for anyone 
without serious RX building experience this is an excellent way to go. 
Keep the IF leads short, observe the designer's cautions about layout, 
wire backward from the AF output, and it'll work the first time.

(Well, give or take outright mistakes.  I'm proud to say that I now get
filaments wired correctly the first time on about 9 of 10 stages.)

The main disadvantage is that getting optimum selectivity for an SSB
receiver is likely to require extra tuned circuits at the low IF and
once you get there, there's no easy way to broaden things out for good
AM reception.  An audio filter (Select-O-Ject, etc.) can be used for CW.
In some designs a Q-multiplier was used, sometimes at the 1st IF,
sometimes at the second, and sometimes at the signal frequency.

The elaborate DCS-500 (1960 Handbook) shows how the variable bandwidth
problem can be confronted the hard way:  A single IF stage at a high
frequency sets a 6 kcs bandwidth using two half-duplex crystal filters;
this is backed up by two stages at 50 kcs using conventional LC circuits
with switched Q's.

It's tough to get outstanding dynamic range, too.  'Dynamic range' is
what it takes to make a receiver sound sharp and clean on any signal you
can hear, regardless of how many high power signals are on the band.  If
you want to listen to DX, or even follow a round table on 40 in the
evening, it's critical.  The usual problem is that second mixer, which
must (like the first) deal with most of the signals on the band, but
AFTER they've been amplified by the first mixer.

The W6TC designs represent about the apex of this approach.  By using
manual gain controls on every radio frequency stage and a Q-multiplier
at the 1st IF they allow optimization of the gain distribution for
varying conditions and reduction of unwanted signals before they hit the
2nd mixer, thus improving the dynamic range.  The selectivity given by
the 100 kcs 2nd IF can be optimized for AM and the Q-multiplier can be
used to narrow it down enough for SSB, although care must be taken with
the stability of the coil/capacitor.

Of course the W6TC designs aren't great SSB performers anyway,
because of stability issues with plug-in coils but the compromise --
most all-round RX performance in a buildable package -- is outstanding.

2. Single conversion with a high fixed IF -- say 1700 kcs, 4550 kcs, 9
Mcs, or something like that.

This demands an IF crystal filter because at such high IFs you can't
get good enough selectivity any other way.  However even an inexpensive
crystal filter (two crystal 'half duplex') gives you a good bet for a
well-performing all-round receiver.  The filter can be placed right at
the start of the IF (mixer plate circuit) so all the unwanted signals
are knocked down right at the start.

Receivers with a crystal filter at the start of the IF require a bit
more care with layout and construction.  Because of the extremely high
Q, ANY feedback that gets to either side of that crystal filter will
absolutely trash the passband shape and I usually wind up shielding the
whole area from the mixer plate through the filter to the first IF grid.
  Filtering and bypassing must be done perfectly, single-point grounding
must be used, and all precautions applicable to the tubes you're using
-- grounding of internal shields, bypassing two cathode leads, and so on
-- observed.  But all this stuff is pretty mechanical.

I think many home constructors would say that this is the way to go,
once you get a bit of experience.  With a hard-wired front end you have
a single band receiver; using plug-in coils or a bandswitching RF stage
and mixer gives you an all-band set.  With a suitable choice of IF and
local oscillator frequency, you can tune two bands in RF or mixer input
-- the 'band imaging' idea is an excellent choice for a well-performing
receiver with a bit of extra capability for very little more effort.

(My everyday-use receiver -- it's in the kitchen -- is the five tube
(+35W4) series filament R8040A band imaging design of a year or two
back.  Front end tunes 80/40, no RF stage, push-pull mixer, pp
oscillator 1400-1900 kcs, half-lattice crystal filter at 5400 kcs.  I
copy everything that anyone else does, except for the guys using
three-element 80 meter rotatable dipoles and Sterba curtains, and that's
about all you can expect using a 160M full wave loop at 10-15' altitude
and fed with 50' of twisted pair.  No, I am NOT kidding.  The antenna is
500-some feet of #22 magnet wire fed with twisted #18 hookup wire.

(Use of a mixer combining a push-pull antenna signal with a push-pull
local oscillator injection has the advantage that both inputs (and all
odd-order products) commit suicide in the single-ended output, thus
taking out IF feed through and any LO noise.  Moreover, the curvatures
of the two mixer sections tend to cancel, giving you a mixer that's more
nearly linear on the signal path and thus improving dynamic range.  And
the thing is dirt-simple.

(The gain's not outstanding but on the low HF, gain isn't an issue.
It's usual to double tune the input to such a mixer but using one very
high Q coil (a toroid, but a miniductor would be close) both increases
gain and eliminates coupling and tracking issues.

(It has excellent IF-derived AGC -- the mixer's on the AGC line -- and a
useful S-meter.  The 'back end' is pretty much stolen from the
Tempo-ONE/FT-200 -- separate triode plate detectors for the audio and
AGC with a crystal controlled oscillator injected on the cathode of the
audio detector for SSB/CW.  Audio output is the pentode half of a 19JN8
driving half of a 12AU7, lots of feedback.  The 35W4 could be replaced
with a two tube converter and silicon diode for all-band operation.)

By adding a crystal-controlled converter ahead of the mixer, this
approach gives you an all-band receiver using ...

3.  Dual conversion, tunable first IF and high second (fixed) IF.

This approach was very common in commercial ham designs of the early SSB
era -- Tempo-ONE, FT-101, and MANY more.  It's probably the most common
choice for the few builders of more elaborate vacuum tube designs,
today.  A great advantage is that you can build up a complete working RX
(for a chosen ham band if you like) and then add a crystal controlled
converter ahead of it.  A number of Handbook designs took this approach.
  W6BD's wonderful GCR-100A communications receiver is another example
-- single band on 80 meters, with a built-in crystal-controlled
converter to cover every thing else, 4.0-30 Mcs.

A variation is dropping back to a 455 kcs fixed IF to build a single
band receiver for 80M (about the highest band for which 455 kcs will
allow control of images) and then using a converter to get other bands.
  That's the 'Advanced 6 tube receiver' of the 1969 Handbook, said to be
based on the W1DX HB 67.

My current GC-HBR project uses a slightly adapted version of the R8040A
as an IF with a push-pull mixer driven by a signal mixed from two
crystal oscillators to cover 1.5-30.5 Mcs in bands of 250 kcs, using the
120 crystals from the WW-II BC-1335 FM set.  300 mA filaments on this
one.  Unfortunately 'slightly adapted' is taking a bit longer than
planned ... I am nowhere near working on the 'interesting' parts.

The one remaining sort-of-standard way to go is ...

4. Single conversion to a high IF using a partially synthesized mixer
injection signal, generally made up of a crystal frequency that varies
by band, and a tunable local osc.  This has the great advantage of
giving all band coverage with a single mixer but delivers in exchange
the major issue of avoiding 'birdies' coming from unwanted mixing
products on the injection signal.  The W5OMX communications receiver is
one of the best of this uncommon genre; he counted one or two birdies
on each ham band and six on 15M.

I don't think the full story has been written on these 'premixed'
receivers yet; I had a project of this type going until one of the dogs 
'marked' it.  But unless you copy one of the very few established
designs -- the Drake TR-4's (perhaps other TR-x's?) are likely to be be
the best engineered -- the process of choosing the various frequencies
and engineering the front end that generates and combines them is a very
serious challenge.

There are many other ways to handle the conversions and IF section of a
receiver but they most often come up in the context of commercial design
requirements.  For example, most ham high-fixed IF designs don't have
front end coverage of the band containing the IF; a common solution for
a general coverage set is to switch to another IF when covering that
range.  Some designs switch sidebands by changing the conversion scheme,
using a final fixed IF and crystal filter with the signal shifted in the
preceding IF so that either the upper or lower sideband is in the
passband -- this approach demands a separate 'roofing' crystal filter to
dump most of the trash at a much earlier point in the receiver.
Manufacturers have tended to standardize filters at just a few
frequencies and there has been a tendency in recent years to put at
least a final IF at 455 kcs in order to sell more of these very
profitable accessories.

If you are using separate converter crystals for each 500 kcs band, a
general coverage design can run into some money; there have been designs
that used two different IFs so arranged to cut the number of crystals in
half.

The Wadley drift canceling loop receiver beloved of users of older
Racal products converted all signals to a broadband very high fixed IF
using a tunable oscillator, then converted them back down to a much
lower fixed IF using a signal that contains crystal controlled steps
plus a difference frequency containing all the drift of the original
tunable oscillator.  POOF -- the drift is gone.  The low IF is a single
band tunable receiver.  The first tunable oscillator selects the band;
the second selects the particular signal.  This was a good way to get
around the need for lots of crystals but demanded an excellent IF filter
in exchange, plus very careful design to avoid spurious signal and
dynamic range issues.  Crystals (and in current designs, synthesizers)
are so cheap now that the scheme was very little used.

Designs I'd give first consideration to building, according to skill 
level and desire for adventure:

1. A two tube regen receiver.  Just to get the feel of vacuum tube RX
construction.

2. The 2x4+1 four tube band imaging design from the 1960 handbook.  This
has a half-lattice filter at the 1700 kcs IF and a plate detector.
Manual gain control only, no frills, but also no tricks; will get you
into superhet construction and should work correctly 'out of the box.'

Within its goals, there's nothing to fault in this design.

3. The 'Advanced 6 tube receiver' of the 1969 handbook.  AGC, S-meter,
455 kcs IF with mechanical filter, front it with the two-tube converter
described for all-band use.  AM reception is only by zero-beat method
and I HATE audio AGC but these circuits could be substituted with an
IF-derived one.  A combination AM plate detector/AGC circuit (the only
essential difference on the two paths is filtering) could replace the
audio amp AGC driver, for example.

It's surprising that there doesn't seem to be a really good widely
published design at this (6 tube full function single band/8 tube all
band) level of complexity.  The little R8040A might eventually get
there, but must be regarded as still under development until I actually
PRINT a circuit diagram somewhere.

4. More complex designs according to taste.  The HBR-series is one way
to go and as a 1960 plug-in coil set, unbeatable.  Among later/more 
sophisticated approaches: the G2DAF Mk II would be worth a look as well 
as W6BD's fine GC-100A, perhaps simplified to ham band only operation.

Walt Hutchens
KJ4KV