[HBR] HBR - BFO progress

[Walt Hutchens]

B. Smith said:

> I would like to post these comments on the BFO page.

Anything I post can be used in any way anyone wants.

> I guess I use a lot of RF chokes, ferrite beads are also some of my
> favorite decoupling devices.

Modern 'vintage' builders are hurt by the fact that the common
oscillator circuits were developed and popularized when standards of
stability were much lower than those needed for SSB. So the references
on those circuits too are mostly older and don't include the
considerations needed for precision tunable oscillators. Good stuff is
out there but it isn't found in the most common vintage sources.

If you want to receive SSB on 80M that's 50 cps or so at 4 mcs: 12
ppm. You can hardly buy a nominally zero temperature coefficient cap
that's much better than 50 ppm/degree/F. A 2.5mH choke will be MUCH,
MUCH worse. Plus of course gradual drift due to changes in the varnish
and moisture effects.

The first line of defense is eliminating (so far as possible) reactive
components. In other words, you want the tank circuit to be the tank
circuit AND NOTHING ELSE SHOULD BE IN THAT GAME. Other inductors and
capacitors should be eliminated.

There are some surprises in doing this. The cathode-to-filament
junction (inside the cathode sleeve) is an unstable capacitor (the
filament wires squirm in there) shunted by a poor quality diode. This
makes the circuits with an RF-hot cathode problematic. Ground the
cathode if possible.

(And the damn diode may cause hum modulation if the filament is AC.
This doesn't matter at a Mc or less but on the higher bands it can be
limiting unless designed around. Bias the filament positive or put the
cathode at RF ground.)

Try to feed the DC at points of zero RF voltage and use a resistor to
decouple. I like the Colpitts circuit with a tapped inductor for plate
voltage coupling. You let the coil tap float for RF.

This is easier to get working than it sounds: Wind the first coil,
hook up the plate voltage through a resistor to the plate of the tube,
and check it out. Then touch the coil with a pencil point looking for
the point at which there's no effect on frequency -- suitable HV
precautions required when doing this. Rewind the coil -- I never get
them right the first time anyway -- with a tap at the zero effect
point and connect the plate resistor there.

You now have an oscillator for which -- pretty much -- the parts that
you think are the tank circuit ARE the tank circuit. Focus on
stability of those parts and you'll have a stable oscillator.

One part that you might not think of as part of the tank is the tube
itself. Transit time and gain limits cause some phase shift and
because the loop phase shift must be exactly 360 degrees for
oscillation, any change in the effects from the tube will shift the
frequency.

(Suppose the tube phase shift INCREASES by 5 degrees due to a plate
voltage change. Then the frequency must change so that the tank
circuit phase shift DECREASES by 5 degrees to maintain the 360 degree
condition.)

This can be reduced with higher gain and UHF-design tubes but those
bring their own problems.

Another approach is to decouple the tube as much as possible and there
are several 'high stability' oscillator circuits (Clapp, Vackar ...)
that do this. THESE CIRCUITS ONLY HELP WITH TUBE-RELATED FREQUENCY
CHANGES. A 'drifty' tank circuit will cause just as much drift as for
any other circuit.

Usually the practical answer is to use a fairly conventional circuit,
minimize the number of parts with RF on them and then stabilize those
parts.

When stabilizing, look at the big picture: In my Eddystone dial HBR I
found that the aluminum chassis bulged upward as it warmed and thus
rotated the tuning cap ever so slightly. This effect was far too
small to be visible, but it led to a couple of kcs drift on 80M. I
took steps to minimize the heating of that part of the chassis but in
future designs I'll change how the cap is mounted so this can't
happen.

(Diagnosis by using a digital thermometer on the chassis and noting
that the frequency changed as it warmed, then gently applying the
corner of an ice cube there and watching it drift rapidly back the
other way ...)

For a BFO at 85kcs SSB stability isn't a big job. But for a local
oscillator on 20 or 10M it will be a very big deal indeed. Even a BFO
at 1700 kcs (as for the Heath HR-10 to HBR project) requires care.

> I've had a lot of fun with the receiver works great. Now working on home
> brew IF coils.
> I  recommend the HBR project to anyone. But start off simple.
> Bread board that BFO before building it.  :-)

Yep. I'd also recommend following an established design for the first
project. Every change from a straight copy will require debugging and
the first time 'round you want as little of that as possible.

Walt Hutchens
KJ4KV