Wadley loops and similar matters (was 'Re: [HBR] Anybody got a rotary ...')

[hbr@mailman.qth.net]

Bob asks:
> Wonder if you considered Wadley loop instead of synth and
> if so, why synth was chosen instead?

This project -- the third in about three years since joining the list -- 
started out as an attempt to get the max dynamic range in a 
straightforward receiver.   That led to a push-pull front end (as in the 
G2DAF Mk II) and the desire to limit the number of mixers in the 
signal path to one.   Once I got thinking about it, I decided to look at 
general coverage, rather than ham band only.   

In the Wadley loop design (made famous by the Racal RA-17) a 
tunable oscillator of only moderate stability is used to up-convert.  
For example (I think the following numbers are probably the RA-17 
ones) the oscillator might cover 41.5-69.5 Mcs to upconvert to a 1st 
IF of 40Mcs. That IF is 1 Mcs wide so that gives coverage 1-30 Mcs.

The same oscillator is mixed with a harmonic of a 1 Mcs crystal.  A 
bandpass filter 1 Mcs wide selects whatever combination will go 
through a 37.5 Mcs filter that's 1 Mcs wide.  This signal is at 37.5 
Mcs nominally (when the band is peaked) but is 2.5 Mcs lower in 
frequency than the signal IF *with drift exactly matching that of the IF 
signals*.
   
The 37.5 Mcs signal is amplified and fed to a mixer that has the 1 
Mcs wide chunk of signals centered at 40 Mcs.   The result is a 1 
Mcs band covering 2-3 Mcs; assuming the crystal has no drift we 
have  *no drift*, because the tunable oscillator drift was cancelled out.

Now we follow with a third conversion from the range 2-3 Mcs to a 
suitable IF where a crystal (or other high quality filter) selects the 
desired signal.   The tunable oscillator for this conversion operates at 
such a low frequency that stability is not a problem.   

The advantages are (1) wide tuning range with outstanding stability, 
and, (2) only one crystal is required. The disadvantages are (1) three 
mixers in the signal path.  (2) Two high quality bandpass filters *plus* 
the usual crystal (etc.) filter are needed, and (3) a rather unpleasant 
tracking problem if the band selection oscillator tuning is ganged to 
the front end as it logically should be -- perhaps not much of a 
problem if you can specify a custom-made capacitor, but definitely 
awkward for the ham.   

Basically the Wadley design made sense for moderately high quality 
commercial receivers back when crystals were very expensive.   It 
traded numbers of crystals for some very difficult design issues.   
You might possibly get to the top of the dynamic range numbers by 
using beam mixers on the signal path but they have their own 
(difficult) design issues.  Today, I think the Wadley approach is a 
curiosity -- certainly interesting, could be fun to build, but it wasn't a 
way to go where I wanted to go.

Another thread that merged into the design approach I chose was 
looking at the ARC-27 carcasses I have.   Anything that produces 
1750 crystal-synthesized channels at 100 kcs intervals ought to have 
an assembly or two or at least a set of crystals that could be used in 
a multiband receiver design.   When I dug in and found that the 20-30 
Mcs IF assembly generated crystal signals every 1 Mcs from 25.7-
34.7 Mcs and some study convinced me that the tunable IF stages 
might be rebuilt as mixers so the assembly would function as the 
complete synthesizer for a general coverage receiver (the tunable 
oscillator is external), I was hooked.   

I got ten crystals (the 25.7-34.7 ones) in the assembly.   For 
convenience I'm moving that range down 1 Mcs so I have to buy a 
custom 24.7 Mcs crystal.   10 and 20 Mcs crystals are clock 
crystals and I already have lots of 10's.   I also need 8.05 and 7.55 
Mcs crystals -- those must be custom.   And that's all -- maybe $50-
75 in new crystals?

Why not aluminum shields ... well, using tin cans (tin plated low 
carbon steel) I can cut with snips, clean up with a file, solder 
anyplace without problems.    I do this only for non-structural 
shielding (like across a socket or in the current case between 
sections of a rotary switch); when support, bracing or whatever is 
required I use aluminum.   Most often that's a piece cut from a dead 
chassis, with a nice flange already formed.  

Coffee cans ... didn't think of that.   I'll check into it.   They're 
generally painted on one side I think, but let me see what's in the 
stores.   No better way to get good play on the Wal-Mart security 
cams than by picking up and replacing a bunch of items ... been 
there, done that.

6061-T6 ... definitely a real man.    I just use the dead soft chassies 
as from the 60's and 5xxx-series for the panel.

Walt
KJ4KV