[HBR] Need advise: BFO coil not tapped

[Walt Hutchens]

> I have a Miller adjustable coil, 12-25 mH, that doesn't have a tap.  I would
> like to use the BFO circuit shown in the HBR schematic (take your
> pick...they all seem to have that same tapped coil configuration).  Is there
> any way I can formulate something equivalent without having to resort to
> removing windings and adding a tap?

This is actually simple enough that it's how I do most oscillators
now. This is a variation on the usual Colpitts circuit: It is the same
for RF purposes but places the ground more advantageously, on the
cathode.

From plate to B+, a resistor of 10k or so.  NO CAP bypassing this.

One end of your coil to the plate.  Other end to a capacitor to the
grid, 100 mmf or so.

Capacitors from both ends of the coil to ground; generally the one on
the plate end should be ~2x the other one.  The series combination
should tune the desired frequency with the coil at midpoint.

A tuning cap can be used from either end of the coil to ground, if
desired -- or use a double section cap.  This is a good application
for one of those BC receiver caps with a smaller section for the
oscillator.

Cathode of tube is grounded; adjust power input by changing the plate
resistor. A plate voltage of 25 or so is probably fine for most tubes
and BFO service.

1. First variation: Instead of a cap from the grid end of the coil to
ground, put one from the grid to ground.  The value of the three caps
that are now in series (coil to grid, grid to ground, and plate end of
coil to ground) should tune the desired frequency.  The two caps on
the grid end form a voltage divider for the grid.  This is the Vackar
circuit; with high gain (almost all later tubes) this allows more
'swamping' of tube grid capacitance, contributing to warm-up
stability.  Also this circuit will give higher output voltage.

2. Second variation: If you are winding the coil, tap it at the zero
RF voltage point.  This point can be estimated from the inverse ratio
of the capacitance from each end of the coil to ground.  This will
slightly raise the Q of the tank circuit which will improve stability
against some kinds of disturbance such as plate voltage fluctuations.

Even in this variation do not bypass the coil tap.

3. Third variation: The push-pull version of this. Coil is center
tapped and goes plate to plate. From each coil end to ground, use two
caps in series: The center tap is connected to the OPPOSITE tube grid.
The four caps in series are the main part of your tuning cap; a double
section cap can be connected from the plates to ground. This is a real
winner with a 6J6 but works well with any of the small dual triodes.

And again, no bypass on the coil tap.  The RF voltage should be small
but it should float.

The advantage of push-pull is that you have the gain of two tubes
effectively in series so coupling to the tubes can be significantly
reduced. However this circuit will most often be used when a mixer
(etc.) requires push-pull drive.

4. Fourth variation: Lift the grid-to-ground caps in #3 from ground
and add a third rather larger cap from there to ground. This point may
be used to drive a mixer cathode or other low-Z load without a buffer.

This oscillator must deliver a bit more power when driving a low Z
load so oscillator parameters should be adjusted accordingly.

I have an 80/40 band imaging receiver using a 12AU7 oscillator, 12AT7
mixer in which the caps are 430 mmf plate to opposing grid, 1330 from
that grid to mixer cathode, and 0.01 mf from cathode to ground. Mixer
cathode resistors are 1k. This is tuned over the (roughly) 1400-1900
kcs range by a double section BC tuning cap. 5400 kcs IF (half-lattice
crystal filter), BFO fix tuned at 5402 kcs.

The 1st oscillator coil is bifilar wound (center tapped) on a slug
tuned 1/2" coil form.

Push-pull oscillator drive to the cathodes, push-pull signal input to
the grids, so single ended output from the two mixer plates which are
connected in parallel to a conventional single tuned IFT driving the
crystal filter.

There is no RF stage; there is plenty of gain for these bands with
just the mixer, two IF stages and a plate detector. The antenna coil
is a toroid and is tuned by another dual section cap which both
selects the band and peaks the stage.

AGC from another plate detector, DC coupled (in effect amplified AGC),
is applied to the mixer grids and both IF stages. This works very well
and produces NO oscillator pulling, not even enough to be audible on
SSB signals. HV is unregulated because the oscillator is nearly immune
to plate voltage variations.

This would make a good back end to a multi-band receiver but I've
never finished that project.

Walt
KJ4KV