[HBR] Yet ANOTHER HBR Project -- Preface

[hbr@mailman.qth.net]

Yeah, I gotta do it.   A premixed receiver but with a push-pull front 
end per G2DAF Mk II.  No, of course I'm not addicted ... I can quit 
ANY time.   Really.   I might decide to quit next Mon..., umm, 
Monday after next.

I got a copy of the SSB for the Radio Amateur book dated 1970 so I 
could read the article on the W5OMX receiver on paper rather than 
with the only microfiche viewer I could afford ten years ago.   It's an 
interesting job and certainly represents a bold approach.   The 
author's technical skills and experience seem to have been modest, 
but he charged ahead, got help from friends, fixed and redid as 
necessary and seems to have come up with a receiver that looks like 
one of the better performers built on this side of the Atlantic.   

There are a few points where it could have been improved.   He's right 
that a significant 'birdy' in every band (and 6 on 10M) is excessive;  a 
good hamband-only design ought to have essentially nothing in-band 
below 10 and not much, there.   The pentode half of a 6AZ8 isn't 
close to 1st choice as an RF stage -- in fact, with a 7360 mixer and 
3 IF stages, he didn't need an RF stage (at least not one with net 
gain) and would have gotten even better performance without it.   Like 
everyone else he used single-ended rather than balanced oscillator 
drive to the 7360 and single-ended output, as well.  (Making the 
birdies stronger and increasing IF feedthrough, tho he did not remark 
a problem with the latter.)  The Miller MD-7 (36:1 and 6:1 ratios, ball 
planetary) is about the minimum acceptable dial.  But overall this is a 
fine job, probably second only to the G2DAF Mk II design in overall 
performance.

Didn't the AF use SS-1Rs?   Wonder if that was where he got the 
idea?   

As the first step toward actual construction of Yet Another ... (and 
recollecting Jim's comments about the BC-221) I pulled apart an LM-
20 frequency meter parts unit and extracted the reduction gear and 
dial.  This is a 100:1 anti-backlash worm drive with a cast frame and 
shafts coming out both ends.   (Frame is sand cast from a two piece 
pattern and machined -- gee ...)   Four sealed ball bearings, preload 
adjustable with shims.  The left end has a drum type dial, 0-50 
revolutions of the knob shaft.   The right end has a 1/4" shaft and four 
posts for coupling and mounting of the capacitor.  

The knob shaft has a dial graduated 0-100 with a vernier reading in 
tenths.  If used to drive a linear oscillator tuning a 500 kcs range, this 
would give a dial graduated in units of 100 cps with 10 cps steps 
readable on the vernier. The tuning rate would be 10 kcs/knob 
revolution which is about ideal for SSB, in my opinion.   

The linear oscillator is of course rather a challenge -- it's tough to get 
better than about 200 cps linearity.   However, with careful work you 
can make the error change fairly slowly, meaning that if the dial is 
calibrated somewhere near the frequency in use you can use the 100 
cps units, and the 10's are available for relative measurements.    I 
never had a receiver that would do that ... 

It will require a 10 kcs calibrator to get the full value of the 
mechanism.

This is about the only commonly available dial mechanism that has 
significantly better resolution than the FT-101.  The -101 dial reads 
directly to 1 kcs and has a reasonable tuning rate, but there are 
several gears that run in sleeve bearings and a ball drive is used for 
the knob reduction, so you get way too much backlash (100-200 cps 
minimum) for serious precision.    

I don't relish the thought of trying to use a randomly chosen tuning 
cap (even one of excellent quality) to get linear tuning.   For one 
thing, most military units are not straight-line capacitance -- they're 
SLF for a fairly large frequency range.   Unless I find a surprise in the 
archives, the FT-101 tuning cap will get the nod.   It's small, dual 
section (allowing push-pull operation), and face mounting (rather than 
base mounting) is an option so it will go right on a plate mounted on 
the dial drive posts -- mechanically it's a good bet for the job.

Building an oscillator that's up to that dial should be interesting.  
Tube heat has to be kept completely away so the tube has to go ... 
where?   The dial mechanism is tall, so I'm not sure the tube can go 
above.   Then there's the problem of where to put the premixer 
('nuther heatmaking tube) and how to get the signal there.  But we'll 
see.   This front end clearly has to be done as a subassembly.   

A serious problem is that the dial is 0-50 over 180 degrees.   But end 
effects will make any 180 degree tuning cap non-linear near the 
ends.   Either I'll have to accept that -- it probably would only affect 
the 1st and last 25 kcs of the band, or so -- or offset the dial slightly 
and cover only about the first 450 kcs of each band.  That would only 
affect 80 and 10.

An alternative would be to use only 250 kcs of the tuning range and 
jigger the 1.75 Mcs oscillator to take care of it.  That probably 
requires another tube or two, another bandswitch section and more 
positions, and another set of coils -- or going to individual overtone 
crystals for each band and pitching the Hahnel oscillator.   

You could do using something like the regenerative frequency divider 
in the ART-13.   In that case a 200 kcs crystal was used to control a 
50 kcs oscillator, but output also was available at 150 kcs, i.e., (n-
1)/n x f.   Two tubes are used ... this was going to be simple, right?

Probably the first thing should be to see how bad the non-linearity 
problem is.   If it's not off by more than a kc or two within 25 kcs of 
the ends, I'd probably swallow it -- I rarely tune there anyway.   A 
frequency counter on the output of an FT-101 VFO should do for 
a test.

I remember my first LM -- in 1961 I was delighted to get a unit that 
was complete except for the dial, calibration book, case, and power 
supply for only $35.   Actually, it must have been a BC-221 ...   I got 
the dial somewhere else for $5 and built a case (another plywood job) 
and a power supply.  Sometime '64-'67 I wrote my first higher 
language computer program -- in ALGOL -- and used it to print a 
calibration book from calibration points on an IBM 7094.   I believe I 
still have that unit, somewhere in the archives ... 

Going prices for complete units with a calibration book matching 
serial number (but you had to check under the s/n plate to be sure it 
really *did* match) were $85 up, as I recall.   What would that be -- 
$400 now?   It was about 10 days pay for me, at the time.   Now you 
find them under hamfest tables, $5 max.

Walt 
KJ4KV