[HBR] The long, SLOW HBR project

[Walt Hutchens]

> 19JN8 RF/band edge marker
> 12DK6 local oscillator
> 19JN8 mixer/unused
> 19JN8 1st IF/bias rectifier
> 19JN8 2nd IF/AGC (plate detector)
> 12AX7 product detector
> 19JN8 1st audio/BFO
> 117N7 2nd audio/B+ rectifier
> 
> A couple of changes are in progress; I'm currently testing a completely
> separate power supply for the LO so that's another tube -- you can see it in
> the pictures and it's the only one that's clearly 'none of the above.'  The
> mixer may change again as well.

Here's the new lineup:

> 19JN8 RF/band edge marker
  6BH6 local oscillator (now with 290V HV supply)
  12AT7 push-push mixer
> 19JN8 1st IF/bias rectifier
> 19JN8 2nd IF/AGC (plate detector)
> 12AX7 product detector
> 19JN8 1st audio/BFO
  12AL5 local oscillator HV (voltage doubler)
> 117N7 2nd audio/B+ rectifier

The big change is the separation of the LO plate/screen power from that of
the rest of the set.

I was surprised to see that the 6/12AL5 is rated as a 'real' rectifier --
330V peak heater to cathode voltage.   Rated output is 9mA per section which
with this full-wave voltage doubler translates to 9mA output.   Not bad, and
with about 3 mA drain it needs only the two 100 mfd caps in the voltage
doubler circuit for filtering.  No trace of either warble (from hum) or
frequency change due to AGC action -- it doesn't take very much of that to
hurt voice quality on strong signals.  AND no problem with the temperature
coefficient of the regulator tube.

I'd call this change a success.

Second to that, the switch back to the push-push triode mixer (which has a
greater dynamic range than any pentode mixer) and the substitution of the
6BH6 oscillator for the 12DK7.   The basing is identical so oscillator tube
swap was easy.   

The sensitivity with this configuration was unsatisfactory on 80M and
terrible on 20M.   Signal for 10% on the S-meter, 11 uV on 80 and on 20M, 50
uV.   (A common definition of S-9 is 50 uV!)

Looking at the various LO signals with a scope shows that 80M has about five
times the mixer drive voltage as 20M.   That accounts for the difference in
sensitivity.   Plus it is logical:  The Q of the oscillator tank circuit is
necessarily a lot lower on 20M than on 80M because the caps are nearly the
same but the coil is a fraction of the size.

There are three ways to go from here:

1.  Back to the 12DK6 LO.

2. Back to the pentode mixer, possibly with gain improvements in the IF
stages.  

(I had regeneration in the RF stage.  While that was eliminated with a
tinplate shield along the lead from the antenna coil to the RF tube, I don't
think this stage is a good candidate for more gain.  There's also the
general principle that gain should be as low as possible until you get past
the selective filter -- in this case, the half-lattice in the plate of the
mixer stage.)

I don't think either the tube change or IF gain increases would be enough.

3. LO circuit changes.   I like the Colpitts-derived circuit that I'm using,
but 'like' doesn't trump making the radio work properly on all bands.
Mohammed is willing to go to the mountain if that is the only way.

In concept a push-pull oscillator circuit could be tried.  With a single
ended RF amplifier and push-pull oscillator drive to the mixer the IF signal
would appear in push-pull on the plates: That's easily accommodated with the
1st IFT and crystal filter that I'm using.

An advantage of this circuit is that signals from the antenna are rejected
at the plates of the mixer because they are in phase.   This not only helps
with strong local BC stations (mainly on 160M) but also reduces the amount
of intermediate frequency noise coming in from the antenna; this could be a
significant effect if you use a multi-band antenna.

The reason for changing the oscillator circuit is that leveling the
oscillator output across all bands requires being able to adjust the
feedback on each.   One can do that in all of these circuits, however the
Colpitts circuit would require adding and adjusting capacitors while the
Hartley and tickler feedback circuits allow changing turns or taps on a coil
which is cheaper.   Additionally, with either of those circuits I could
presumable tweak existing coil designs rather than having to start from
scratch.   
  
An additional benefit of changing to an oscillator with better control of
the feedback level is that it should be easy to push the tube into class 'C'
on all bands, which will tend to level the output.

> What am I doing in this swamp?  Ohyeah ...

Still digging, still trucking in more mud ...

Walt
KJ4KV