[HBR] Another HBR Project -- Chapter 6

[hbr@mailman.qth.net]

Walt's first law of homebrewing:  Every project has at least one new 
form of weird unwanted behavior.  

But at least this one is amusing.   Got the front end wired, wound the 
coils according to the calculations, checking with a GDO and fixed 
caps of the right size -- should be close to decent tracking!    Put it 
all together, turned it on, and ... as I tuned down the band, I heard a 
nice stable whistle that varied slowly in frequency, occasionally 
above the range of the audio system, but mostly in the couple-kcs 
range?   Huh?   I'm used to assorted tunable whistles, but one that 
stays at a fairly constant frequency ... now that's a new one.

An hour later I knew what should have been obvious instantly:  the 
new RF stage was oscillating, tracking the oscillator mostly within a 
couple of kcs.   The good news is that's probably good enough 
tracking that no real effort at front end alignment (other than the 
oscillator) will be needed.  With a coil Q likely to be not much over 
50 (using the command set coil forms throughout, right now) I'll have 
80 kcs half-power bandwidth on 80 meters and its tracking within 
maybe 10% of that.  And even with toroids, 150 is about the 
maximum Q ... selected fixed caps will do fine.   

This result is one advantage of a ham-band only approach.  With 
frequency ranges of only 10-20%, tracking is a pretty trivial exercise; 
precision is not required.   The original command receivers -- up to 
3:1 frequency range! -- now *there* was a serious exercise in 
tracking.   They were good from the factory but could always be 
improved some, if you had the patience to adjust the coils and then 
finish up by bending the cap plates a tiny bit.

Of course the bad news is that I have to rip out most of the RF stage 
wiring and put a shield across the socket and over to the coil plugs.   
But it was still a pretty funny symptom -- at least, if you have my 
sense of humor.

By backing down the RF gain enough, I could kill the oscillation and 
with a clip lead on the RF tube grid, tuned in a sideband QSO.   It 
sounded fine, so everything else is indeed pretty much okay.   Still 
don't know if the total gain will be enough.   I can hear a few unstable 
warbly beat notes, suggesting a VHF parasitic someplace.

(The VHF always has a bit of 60 cps FM; the beats are with high 
harmonics of relatively powerful oscillators in the set so you hear the 
both the FM and rather rapid drift coming from the poor quality and 
low Q tank circuits that produce the parasitic.)

I did do a quick check of local osc. drift.   From a cold start, it drifts 
down at a slowly decreasing rate until it's down about 1.2 kcs after 
70 minutes or so.   It then stays within 20 cps indefinitely.   Not bad 
for this type of construction -- you'd cringe if you saw the length of 
some of the leads in that oscillator -- and should be pretty easy to 
reduce to the range of 100 cps total with a couple of compensating 
caps.   There is a random drift of up to 10 cps; that's typical of 
circuits having the cathode 'hot' for RF and is caused by the filament 
squirming inside the cathode.   I can probably improve that by 
selecting the best of a few tubes and yet more by moving the tap 
down some.   

The drift would likely be some higher in a cabinet -- I probably won't 
do that right away.

I wound up using the 6BH6.   0.15 x 6.3 = 1 watt but the total plate 
and screen current is around 1 ma at  50 volts -- i.e., 50 mW. Nearly 
all the heat from that stage is from the filament so (other things being 
equal and with no optimization to take advantage of a higher gain 
tube) there's a considerable advantage to using an 0.15 amp filament 
in that socket.

I've considered using something like a 1U4 for a local oscillator; the 
filament current (50 mA) would be easy to handle in a bifilar wound 
osc. coil and you'd get the heating down to near-nothing so the tube 
could be mounted right with the tuned circuit components.   Maybe 
someday ...

Walt's second law of homebrewing:  Every project develops early on 
some characteristic behavior, usually bad, which will be repeated in 
various stages as you go along, but with just enough variety to 
maintain your interest.   (Excessive drift, unwanted oscillation or no 
oscillation in an oscillator stage, intermittent shorts ...)    Surprise!  
The 1WHBR's characteristic is for most stuff to work pretty well the 
first time.   Electronically I believe this is the easiest project (of this 
size) that I ever attempted.   The mechanical stuff connected with the 
plug in coils is somewhat challenging however.

More tomorrow, probably.   

Next project (after wrapping up this and the HBR2K!!) will be to 
answer the question: What is the smallest possible good-quality 
multiband 'hollow state' SSB transceiver?   Always wanted to build 
something with a bunch of subminiature tubes and I'll bet such a 
project will cure me of that ...

A question that comes up in considering a 'Tiny HBX' is what's the 
best final tube?   Will operate class 'B' at a plate voltage not above 
480 (voltage tripler from the line) ... most likely a miniature tube.   I've 
thought about the 50C5, 16GK6, 12HL7, 27GB5 ... The 27GB5 (novar 
but a very small diameter envelope) will go to 25 watts PEP out at 
480 volts; the others are probably all less, even as pairs -- but of 
course they're smaller, and single-ended.   I'd really like to hit the 25 
watt range.  Filament current not to exceed 0.3 amps -- will be series 
string filaments.   Anybody got thoughts?

Walt 
KJ4KV