[HBR] Another HBR Project -- Chapter 8

[hbr@mailman.qth.net]

Subtitled ... "And it looked so easy ..."

Yesterday I wired the 100 kcs calibration oscillator.   There are 
several things to be said about this:

1.  It was probably unnecessary for a reasonably stable receiver 
that's only going to be calibrated every 50 kcs or so and that already 
has a marker at 3500 kcs courtesy of the 1750 kcs 2nd oscillator.

2.  It *seemed* really easy ... after all, I was copying the circuit 
'verbatum' from another set, and,

3.  It turned into a real can of worms, because ...

4.  It was a dumb idea trying to put it in the same envelope (other half 
of a 6U8) as the 85 kcs BFO.

It wasn't more than the usual amount of trouble getting the oscillator 
to work -- there are a lot of parts and (Walt's third rule of 
homebrewing) I seem to wire about 1/4 of all parts wrong the first 
time.   But then ... beat notes every few kcs (like, less 5) all up and 
down the dial, and close enough to the same strength that picking 
the right ones is impossible.   Gosh ... I sure didn't expect that from 
a 100 kcs crystal.   Same thing happened with another crystal, too -- 
and both of them believed to be good.

Took me a while to figure out, but the problem was that the 85 kcs 
BFO was modulating the 100 kcs calibration oscillator.   So the 
output frequencies of the latter are every multiple of the common 
divisor -- since the two frequencies aren't precise, there isn't actually 
an exact common divisor so there are an infinite number of 
frequencies but fortunately you can't hear all of them.   I spent half an 
hour looking for oddball coupling mechanisms but finally decided that 
there was (a) too much stuff too close together, since both 
oscillators are in the same tube, and (b) way too much BFO voltage 
floating around.   How much is too much?   Try 120 volts p-p on the 
plate of the oscillator tube!    Since the calibrator is the pentode half 
of a 6U8 and the BFO is the triode, coupling is unavoidable.   And 
that's with just enough BFO plate voltage to get good injection.  

The reason the BFO operates that way is that I stole the command 
set circuit along with the BFO coil.   "It seemed like a good idea at 
the time" but the command sets run a high-output oscillator very 
loosely coupled to the detector so they get a good waveform -- no 
spurious signals.   I need to put the BFO into the cathode circuit of a 
plate detector -- that's a low impedance circuit.  And getting 6V p-p 
at the detector cathode with loose coupling requires 120V p-p on the 
BFO plate.   Increasing the coupling to get a greater fraction of the 
plate signal killed the oscillator.

Obvious answer -- redesign the BFO to deliver output at a lower 
impedance and voltage -- and BTW, to minimize coupling within the 
6U8, why not put the plate at RF ground?   In other words use a 
tickler to feed back from the cathode to the grid circuit.   That took 
most of a day of thrashing about because I had to go into the 
oscillator coil assembly, reverse the connections to one of the coils 
and replace the internal cap with an external one so I could get the 
frequency right.   And of course there's Rule 3.

But it worked.   I moved the tuned winding from the oscillator plate to 
the grid and the tickler from the grid to the cathode.  The largest 
signal is now 20 volts p-p  on the grid and the 6 volt cathode signal 
can be coupled through a large cap to the detector cathode.   The 
(cathode) output waveform is pretty poor, but the voltage is low 
enough that pickup of harmonics in other parts of the receiver 
shouldn't be a problem.

With this change modulation of the 100 kcs calibrator signal by the 
BFO is no longer visible and although with the antenna disconnected 
you can hear lots of beat notes, the 'real' ones are S-8 or so and the 
others barely move the needle.   Not great, but good enough.

The change solved a second problem as well.   The detector circuit 
I'm using is a plate detector with the BFO injected on the cathode for 
SSB/CW.   But for AM, the cathode should be bypassed to ground -- 
'plate detector,' right?   Turning the BFO off by shorting its cathode to 
ground (shorting the tickler winding) turns the coupling cap into the 
needed bypass for the detector and for the first time in my use of that 
detector circuit, the output volume is the same on AM as on SSB.   
The tube is protected by the large plate resistor.

Other progress -- I extracted the 3 mmf N750 temp compensating 
cap from the command receiver that donated the coil sockets and 
installed it on the 1WHBR; it helps but it's not enough.   Well, it 
wasn't enough in the command receiver, either ... they generally are 
much improved if you add a second cap, and in fact the late-
manufacture ARC-5 receivers have two caps factory installed.  (Look 
for an 'S' inside a circle stamped next to the tuning knob; the second 
cap is on the coil socket inside the tuning cap shield.)  I'll either add 
a second one or go to 10 mmf N750.   

As Bart Simpson would say ... The deep fat fryer is not a toy.   I will 
not call her Mrs. Dumbface.  And I will *never again* put two 
oscillators at nearly the same frequency in the same envelope.   

Still have to hook up the STDBY switch, finish the alignment of the 
80 meter coils and wind some for another band or two.   Also do the 
basic performance measurements.   Another data point on 
sensitivity: in the evening it picks up 75 meter phone signals with no 
antenna.   

Walt 
KJ4KV