[HBR] Another HBR Project -- Chapter 10

[hbr@mailman.qth.net]

Not too much more on this one!   The photos with cabinet are at:

http://www.collie-rescuers.com/ebay/september29-1.jpg
http://www.collie-rescuers.com/ebay/september29-2.jpg
http://www.collie-rescuers.com/ebay/september29-3.jpg

Making the cabinet has pretty well soaked up the last four days.  It's 
plywood -- sides, top, bottom are 15/32" and the rest is 9/32".   
Remember the days of plywood sized 3/8" and 1/4"?   Anyhow it 
looks okay, the speaker sounds a lot better with the panel edge 
clamped and the loading from the cabinet, and the coils have a 
home.  There's room to stow 8 coil sets -- it's not likely I'll ever make 
that many!   And behind the coils, a compartment for spare tubes 
and a larger one for miscellaneous items -- the (removable) line cord 
can be stowed there when the set's not in service.   

The panel is supported on 3/8" x 1/2" aluminum angle stock attached 
to the cabinet with 4-40 flat head machine screws.  

The main electronic work since the last chapter has been on VFO 
stability.  A week or two back it shifted from pretty stable after 1-1/2 
hours to a continuing slow downward drift of maybe 500 cps/hour for 
another three hours or so;  I've forgotten exactly what I did to provoke 
that, but it made good sense at the time.  Poking around with the 
digital thermometer I could see that the long term drift corresponded 
with heating of the coil set.   The coils never get above about 90F, 
but command set coils aren't noted for being rock-stable, either.   
And I had to use silver micas instead of NPO ceramics for the series 
padder caps on the 80 meter coils.   The padder is large compared to 
the tuning cap so its effect is a fraction, but likely still significant.

(On the higher bands the oscillator is below the signal freq. so the 
padder is on the RF/antenna coils -- that's one less drifty part in the 
oscillator circuit!)

The first step was to see if oscillator coil heating could be further 
reduced.   I had made the connections from the coil plug to the 
oscillator tube socket with regular insulated hookup wire, about #22 
size.  Well, that was wrong -- such connections are a pipeline for 
heat to get from the tube to the coil.   They should be made with bare 
or enamel wire of the smallest size that won't vibrate or flop around.  I 
use about #30 or 32 up to 1" length but in the 1MHBR these leads 
are so long that I had to use #26.   

(The tubes have to be far enough behind the tuning cap to clear the 
rotor gangs as it swings open.   The coils mount through the panel.   
I wasn't able to think of a way to get the coil plug to tube leads less 
than about 3" length.   Lead-length wise, the command set structure, 
in which the coils go in from the bottom of the set and can thus 
nestle *under* the tuning cap gangs, is better.)

(There was a semi-religious belief among the ham fraternity after WW-
II that oscillator circuits should be wired with the heaviest wire you 
could bend, for 'stability.'    This view is *incorrect* -- heavy wire not 
only more effectively pipes the heat from the tube to drifty parts, but 
also acts as a strut, moving other parts of the oscillator structure as 
it warms up.   The right size wire is the smallest that does not flop or 
vibrate under the conditions of use of the set.   I.e., a mobile rig 
might need one size larger wire than a home station unit.   

(I surmise that the 'heavy wire' idea started with the oscillators of the 
military sets that were then familiar -- but these sets were designed 
to continue functioning in any conditions that did not destroy the set, 
including nearby hits from enemy fire.   They were, however, mostly 
were designed with little consideration of warm up because the 
equipment was normally turned on well before it was needed.)  

Also I had mounted the grid coupling ceramic disk horizontally -- 
such caps should always be mounted on edge as the better air flow 
helps cool them.   Yes, stuff like this does matter -- go into a Collins 
PTO and rotate a cap or two and you'll see the effect big-time.

With these changes the drift dropped by half.   The oscillator now 
goes *up* a hundred cps or so, then is stable for 15 min or so, then 
drifts down 200 cps/hour or so.   The upward drift is the effect of the 3 
mmf N750 cap mounted on the tube socket.   The grounded end 
goes to the (cooler) chassis, and by shortening that lead I can delay 
and reduce its effect.

Another small self-inflicted wound:  The darn pilot light heats the coil 
area slightly.   There's very little effect on the oscillator coil because 
the pilot is located near the antenna coil end of the set, but it should 
have been all the way at the left side of the panel, where the phone 
jack is.   

I believe the remaining heat flow to the coils is mainly via the chassis 
which is heated (mainly) by the RF, mixer, and oscillator tubes that 
are located right behind the tuning cap.   (Can't be via the air, 
because the air flow there goes the other way -- coils to tubes.)  I put 
the mixer and oscillator on 1/4" spacers so air could flow up around 
the pins; in a day or so I'll do the same to the RF stage, then make a 
plate that mounts on the three socket mounting screws nearest the 
tuning cap and turns up to form a 3" shield between the cap and the 
tubes.   Such a shield isn't a bad idea in any case, to reduce 
radiation of heat to the cap (the tuning cap *does* affect stability!) but 
the main reason here is to sink the tube heat away from the chassis 
on the coil side.   

I'll also look for some unshielded tube sockets.   The shield base 
type that I used out of habit traps some tube heat and conducts it 
down to the screws from which it passes to the chassis.   Of course 
changing those sockets means re-wiring most of the front end, but 
what the heck ...

A receiver that doesn't drift during warm-up is first of all a mechanical 
and thermal design exercise.   Compensating caps for parts other 
than the tube itself should be the very last step in reducing drift.

Walt 
KJ4KV