[HBR] The long, SLOW HBR project

[Walt Hutchens]

> Basically the problem [of very low sensitivity on 20M] comes from low
> oscillator injection.   <SNIP>
> 
> The Colpitts-derived circuit I'm using is likely to be the poorest for
> harmonic generation because both grid and place circuits look into substantial
> capacitance. <MORE SNIPPED>
> 
> So, there are two ways to go here [to get more second harmonic generation]:
> One is to redo the oscillator as a tickler feedback circuit.   That has the
> appeal of a tested circuit with already-designed coils and I already have the
> right tubes in place.  There's still the issue of low plate voltage but the
> voltage could be raised if necessary.
> 
> The other approach would be an oscillator specifically designed to generate a
> second harmonic.   You can do that with a push-pull oscillator if you take the
> signal from a common cathode.   Push-pull has other advantages: The use of two
> tubes allows even looser coupling to the tank circuit and since the tube
> capacitances are effectively in series you get another factor of two help in
> that department.

There's yet another variation:  I could adopt the Hartley circuit that was
developed late in the HBR cycle to simplify construction and improve
stability.  

However ...

The tuning cap and already calibrated dial that I am using came from a
G3RKK-design receiver that used a Colpitts-derived oscillator.   (That was
one of the main reasons I chose a Colpitts circuit.)   All three of the
variations above would require adding parts to get a three point fit.  And
the amount of work to get that fit would be greater.

For example an additional problem with the Hartley is that making the
cathode 'hot' is likely to increase hum/warble coming from heater-cathode
leakage.   I used a grounded cathode variation of the Colpitts.   However
with the cathode grounded, the neutral point on the coil becomes part of the
tuning equation and  you may have to move it as part of the adjustment
process.  

Moving the neutral point on the coil for a Colpitts is a change of one or
more capacitors; for the Hartley it's moving the tap on the coil.  If you're
very close, squeezing/spreading turns will do it, otherwise you're talking a
rewind.

These bands are so narrow that this issue wouldn't arise except for the fact
of that calibrated dial.   Any two-point fit would be linear enough to work
if you were doing your own calibration.

The layout of the oscillator area of the set is awkward for the other
circuits.   

Realizing this situation led to reconsidering the use of fundamental-only
oscillator coils.   THEN I realized that I had been thinking incorrectly
about the stability issue.   The full range of the tuning cap is used
regardless of fundamental or harmonic operation, so roughly the same fixed
capacitors are needed.   Doubling the oscillator frequency is (roughly) a
matter of cutting the coil inductance by a factor of four.   Arguments about
parasitic capacitances thus don't hold much water, although lead lengths
certainly become more important.

Transit time effects also become proportionately greater but these are
minimized with high-Q coils and VHF-capable oscillator tubes used in these
sets.   

I wound a new 20M oscillator coil tuning 12.3-12.7 Mcs and in the course of
a couple of hours got it tracking the dial within 1 kcs or so from end to
end -- about as close as is worth doing, considering the dial resolution and
calibration precision of this mechanism.   Sure enough the fixed capacitors
came out in the same ballpark as for 6.15-6.35 Mcs and the coil has exactly
half the turns.   

Bingo!  No more problem with sensitivity.   Strong 20M signals are punching
the meter into the 80-90% range.   (THANK YOU, 20M, for being open last
evening!)  The band edge marker still doesn't move the needle but evidently
that's just a wimpy 4th harmonic and that's  a separate matter.

At a later time I'll make actual measurements on each band and tweak coils
to level the gain if necessary.

Warm up drift is a bit high compared to 80M -- down about 1.6 kcs over the
first hour or so.   That can probably be improved if I redo the oscillator
coil as things in there are a bit sloppy owing to the difficulties of cut
and try inside a 1-1/8" hole and the need to glue in the APC cap, test, and
quickly push the cap out and redo if a further change of the silver mica
located on the trimmer is needed.

Okay, replace 'a bit sloppy' with 'plenty sloppy.'  I will redo the coil in
the next couple of days.   Fortunately that's just a matter of time and PVC
pipe -- the plug and APC cap (even the wire, sometimes!) can be salvaged and
reused.  

There is a bigger problem with plate voltage variation resulting from AGC
action -- maybe 50 cps change of oscillator frequency with the full AGC
range -- and I don't know  how I'll deal with that.  It's possible that the
5651 regulator tube isn't quite in the right part of its current range.

I did note in my recent readings the statement that pentode oscillators
using the screen as plate for the oscillator are roughly voltage independent
if the plate voltage is about three times the screen voltage.   That might
be a way forward.  

10M will be the same challenge but moreso.   The 10M calibration is 28-30
Mcs so the fixed caps will have to be relatively smaller than for 20M -- say
half the size.  That coil is probably the next project, when I finish
digesting 20M and the band edge marker.

A major source of warm up drift in the early 80M coil work was bulging
(buckling!) of the top of the chassis causing rotation of the tuning cap.
(This effect was much too small to be visible, of course.)   That was much
reduced with the addition of a stiffener under the front side of the tuning
cap mount and  better ventilation around the 117N7 and does not seem to be
significant now.   

The bands above 40M won't be of much real interest until I move my
horizontal loop antenna.    I have the sound of galloping horses from that
'broadband on power line' service radiating from the power co. 20 kV lines
that run diagonally through our property and parallel one side of the
existing antenna.   The noise is only strong within 100' or so of the lines
and I think there is enough land on one of the distant corners for an 80M
loop.  

Yes, they do 'notch' the ham bands -- the interference is far stronger near
the band edges. 

If I were a DX guy with a big beam sitting in a few yards of concrete I'd be
purple over this, but my antenna can be moved fairly easily, and when I do
operate, it's generally only ragchew.

Walt
KJ4KV