[HBR] Yet Another HBR Project -- Chapter 2

[hbr@mailman.qth.net]

Two days of testing and adjusting a push-pull oscillator using 1U4's, 
the worm drive and dial from an LM frequency meter, and a split-
stator tuning cap from an unknown piece of military UHF gear yielded 
an interesting collection of things that worked and one that hasn't -- 
yet.

The oscillator works -- at 60 V on the plate it draws around 1 mA and 
delivers 5V p-p across a 270 mmf capacitor -- actually, two such 
waveforms of opposite phase.   It's -- wow! -- really stable.   The lash-
up version does all its drift in a couple of minutes and may then be in 
the under 100 cps/hour range, indefinitely.  The drift is quite steady, 
so with compensation it should be better.  There are too many loose 
ends to really evaluate it -- a couple of poor-quality fixed ceramic 
caps in there, the turns on the coil aren't glued down, long 'haywire' 
leads, no cover on the oscillator box, filament isn't perfectly filtered or 
bypassed ... you get the picture.   But I'm ready to say that providing 
well-filtered DC to the filament of an oscillator tube (or tubes in this 
case) might be a worthwhile price if you need really low warm-up drift! 

(Of course it's only worthwhile because essentially no other heat will 
reach the oscillator except throught the air and with everything in a 
closed box, it'll all warm up at the same rate.   With oscillator parts 
strewn over several inches and multiple other nearby tubes and 
conduction heating paths as in the 1MHBR, a battery-tube oscillator 
would make no significant difference.)

Other than a couple of bad fixed capacitors, there really were none of 
the usual circuit-doesn't-do-what-you-expect kinds of problems.   
Everything just needs a good clean up, and it'll be ready to go.

Except for the darn tuning capacitor.  It tunes over a range of about 
240 degrees and that was one of its great attractions because it 
allowed picking a 180 degree range without end effects.  To tune over 
240 degrees, the stator has to be only 120 degrees -- otherwise the 
rotor would start to reenter it as the capacitance was decreasing.   
120 degree stator, 240 degree rotor ... but with this arrangement, as 
you turn in the increasing capacitance direction the rotor starts to 
*exit* the stator somewhere near mid-position.   Suddenly the 
capacitance is going down at the same time it's going up, meaning a 
step-decrease in the rate of increase.   The manufacturer dealt with 
that by placing a step increase of the rotor radius at just the right 
place to make the difference.   As the backside of the rotor starts to 
exit the stator, the radius of the rotor suddenly increases to 
compensate. 

I looked at that a dozen times while building the unit and wondered 
... this was done long before computers were used for such things so 
how would you design this feature?   Graphical electric field 
methods?   Wild guess?  Cut and try?   Did they get it right?

Well ... not quite.   In general I was able to linearize the tuning (i.e., 
track it to the linear calibration of the LM dial) within about 200 cps; I 
suspect that with some further work it could be gotten down to 100 
cps.   However just ahead of the step in the rotor radius -- covering 
say 50 kcs of the 500 kcs tuning range I'm using -- they ramped out 
the radius substantially, well ahead of the point where the rotor exit 
occurs.   There's a 5 kcs 'kink' in the tuning (a dip in frequency) at 
that point.

Later today I'll make one pass, removing the tuning cap, removing the 
shaft over a large mixing bowl so the 100 or so itty-bitty ball bearings 
(rat shot size) go someplace where I can find them, and grinding or 
filing the contour of the rotor plates to correct the problem. Working 
by uncalibrated eyeball and since I'm unwilling to do more than a one-
shot effort, I can probably get the error to the 1 kcs range and 
squeeze most of that into a smaller tuning range.

(Unwilling -- two hours of very fussy work and if I drop that ceramic 
shaft, the game's over.  Once is going to be plenty.)

This wasn't a problem in the original because these UHF sets used a 
bunch of ganged identical caps and switch selected crystals.  You 
peaked the capacitor knob (or knobs) according to the crystal, 
locked everything up, and an autopositioner did the job the next time 
that channel was selected.    

I think that capacitor may be from the RDZ -- an amazingly heavy 
RCA 200-390 Mcs shipboard receiver dating from about 1946-7.   The 
corresponding transmitter was the TDZ, at something like eight the 
all-time high water mark for number of Collins Autotune units in one 
radio.   The system had nearly a negative lifecycle, as it was based 
on the use of fundamental crystals in the 5 Mcs range multiplied up 
in conventional doublers and triplers.   But a task force of ships could 
easily need 100 or more channels and there was such severe mutual 
interference with all the spurious frequencies that the Navy rushed a 
series of single-channel transmitters and receivers based on overtone 
crystals into production within a few years.   

The units showed up in surplus but there was little you could do with 
them -- you got any ideas for a 400 pound 10-watt UHF transmitter?   
The receiver wasn't so bad but still required two men and a boy to lift.

Walt 
KJ4KV