[HBR] HBR - BFO progress

[Walt Hutchens]

Chris said:

> It turned out that my scramble-wound coil was not able to get
> to 8k microHenries, more like 6k.  That meant adding more
> capacitance to get things at the frequency I want, and the 50 pF
> only gives me a swing of a bit more than 1 kHz both sides of 85.
> I may make another run at that if you all think I need more swing than
> that.

If you plan to receive only CW (and AM) and beat notes not much over 1
kc are high enough that might be enough. If you're going to receive
SSB as well then you'll want to park the center of the IF bandpass on
roughly the center of the sideband which would likely be a little
higher than that -- 1.5 kcs? -- and put the BFO where the carrier
should be. By the time you allow adjusting a couple hundred cps one
way or the other to get a more natural sounding voice AND a bit of
drift in the BFO so the frequency range isn't exactly according to
plan it's probably good to have a total swing of 4 kcs or even 5 kcs.

You'll also (probably) have to tweak the center frequency via your
fixed shunt cap to get lined up properly with those nominal 85 kcs
IFTs. The command set ones seem to me to often be off 1 kc or so one
way or the other, and they don't have very much tuning range with
those itty-bitty trimmers.

Because of this it's best to get three xfmrs from one command set AND
use the BFO assembly from that set.  Mismatches can be worked out, but
it's a tedious job.

My hat's off to anyone who has wound a 800-turn oscillator coil!

> The only other objectionable thing was that the negative peaks on my
> waveform are a bit truncated. It looks like I am hitting a voltage
> supply rail in that direction, or something. I may fool around with
> that a bit. Does the BFO need a nice waveform?

ALL ordinary oscillator circuits operate class 'C', that is, the
current in the tube (and energy input to the tank circuit) is in
pulses. The higher the Q of the tank circuit the better the voltage
waveform across it will look, although it will look better across the
whole tank than from cathode to ground, anyway.

(High quality signal generators and the like are sometimes class 'A'
but they require some sort of AGC circuit to keep them oscillating
without going class 'C'.)

Waveform matters because anything but a pure sine wave contains
harmonics, and these can often go up into the lower ham bands. The
first issue is the 1st IF which should be planted close to halfway
between harmonics of the MEASURED 2nd IF, unless you are using IFTs
with enough tuning range to trim things post-construction. The command
set IFTs do not have that much range.

For example if the 2nd IF is exactly 85 kcs then 20x = 1700 kcs; you
would want to go either ~40 kcs higher or ~40 kcs lower with the 1st
IF. If you're using a crystal controlled 2nd oscillator with command
set IFTs some breadboarding may be needed to figure this out.

However commercial 85 or 100 kcs IFTs have a wider tuning range; with
them you can probably just trim the 2nd IF (and BFO)
post-construction. However you STILL don't want BFO energy getting
into your IFs because it can cause overload and birdies with signals
that ought to be out of passband, also it may mess up the AGC action.

These issues are one disadvantage of double-conversion designs.

IF your are following a W6TC design closely much of the following will
already be done for you.

Since the BFO waveform cannot really be great, treat it like any other
oscillator you don't want interference from. Put it as far from the
input to the 1st IF/2nd mixer area as possible. Usually that means on
the far side of the product detector from the IF strip. Often it will
work out well in or near a front corner of the chassis, making you
good to go with a short lead to the BFO tuning cap. Keep all the BFO
parts close together.

Keep leads short and use shielded wire for the really (RF) hot ones
like the hot end of the coil to the tuning cap unless you can get
them down to the 1" range.

Think about filament wiring too: It's a good way for stray signals to
get around a chassis. Sometimes a 12V twin triode with 12V filament
supply and the 'cold' filament half serving as the BFO, a bypass on
the filament tap, and the 'hot' filament half of the tube as 1st audio
works out well -- the audio tube filament serves as a filter resistor
for the BFO filament.

Decouple the B+ carefully. Do the math for decoupling at 85kcs: .01
mfd caps WILL NOT do the job. The command sets used 0.05's and that's
really the minimum. The BFO switch must switch an (RF) cold lead -- if
switching the B+ then decouple at the BFO location.

Try to confine ALL the 85 kcs energy to that little area. Think about
how you'd add a below-chassis shield if needed -- only sometimes
necessary.

Use a single ground point for the whole BFO stage and especially do
not let tank circuit currents flow in the chassis, for example, by
grounding the coil at one place and the fixed cap somewhere else. If
the tuning cap is small and on the panel close to the coil and main
fixed cap, that (small) current is probably okay if the 1st IF and
mixer are on the far side -- for example, toward the rear of the
chassis.

This may seem over obsessive but I've redesigned from scratch as often
because the BFO was getting into Bad Places as for all other failings
together.

Again, if you follow W6TC closely, you will not have to worry about
these things. But the minute you start rearranging, they become
important.

> I'm really tickled that I was able to make this work at all. I'm
> going to watch at our hamfest later this month for a better 50pF
> trimmer cap, and I have my tube list. I need to make a more complete
> parts lists, that's a fact. There are probably some potentiometers
> that I could pick up.

It's always a kick when an oscillator actually does. :-)

Finally, those who do want to strike out in another direction (from
the W6TC designs) will find that the Heath HR-10 ham band receiver has
just about everything except the coils, coil sockets, and front
panel/cabinet. (The HR-10 front panel isn't tall enough to allow for
the coils above the chassis.) Most of the chassis work is already done
-- you have to add the coil socket holes, and 3-4 more for tubes.
(Mine came out with 10 tubes, mostly double section.) A flat shield
plate must be added above the chassis between the RF and mixer coils.
There's plenty of room on this chassis. I added a 100 kcs calibrator
and there's still ample room.

There's a built in power supply but a separate speaker is required.

The string drive dial is about the second best thing to an Eddystone
898; I found it entirely satisfactory. Your coils can be tracked to
the original dial calibration which is silk-screened on the back of a
sheet of Lucite; of course you use the original tuning cap.

This set uses a half-lattice (two) crystal filter at 1682 kcs and you
have two IF stages (and BFO) at that frequency -- no second conversion
is needed.

Going this route you ditch a whole lot of parts-finding but add the
thrill and frustration of doing your own design. I think there's a
blow-by-blow from from when I did this, in the list archives.

Walt Hutchens