[HBR] Another Receiver Project -- HBR-4, Part 8

[waltah at earthlink.net]

The 6JH8 premixer has somewhat higher gain than the 2 x 6J6 
circuit did -- say 5 db more.   However it has *much* larger signal 
handling capability on the push-pull (deflection plate) input.  That 
input is from the Butler crystal oscillator.   

Unfortunately the strengths of a Butler oscillator are stability, 
waveform, and ease of design; it's not a high-output circuit.  Higher 
output circuits for overtone crystals are generally vfo circuits with a 
crystal inserted in series at a low impedance point; they can be 
tricky to get to operate reliably.

I'll stay with the Butler circuit for the time being. 

I got some increase by replacing the 12AT7 with a 12AU7; the 
greater grid voltage swing of the latter allows a higher plate voltage 
swing, hence higher output.   Another doubling of output voltage 
was obtained by increasing the number of turns on the secondary 
of the transformer coupling the oscillator to the premixer.   However 
I'm testing with the 18 Mcs (80M) crystal; there's the risk of self-
resonance at 10M where the frequency is 42.5 Mcs.   Or there may 
be too much stray capacity to allow tuning the plate coil on 10; 
making the coil smaller would hurt 80M output.

In any case the higher turns ratio represents a pretty good match 
to the premixer.  Unlike a grid the deflection plates actually draw 
some power -- about 0.1 mA at +8 volts.   You can bias the 
deflection plates negative a bit and reduce this considerably, while 
increasing the deflection sensitivity at the same time.  In this 
respect the 6JH8 is quite unlike the 7360 and 6ME8 which are 
supposed to be operated at +75 volts on the deflectors and will 
always draw significant deflector current.

There is a spare bandswitch section for the Butler oscillator so it 
would be possible to switch the coil on the high bands.   

Changing from direct drive from the premixer plate to mixer 
deflection plates to a step-up autotransformer circuit picked up 
another doubling of output.   Same possiblity of self-resonance or 
inability to tune the 10M band, also not checked yet.   

Increasing the VFO plate voltage from 38V to about 60 volts raised 
the premixer drive from 0.7V p-p to ~2V -- about the limit for 
reasonable linearity -- with no noticible increase in drift.  

The Butler crystal oscillator plate voltage was already maxed out.   
If I used the 70 volt negative bias on the cathode that might help, 
but the current drain would vary from band to band which would 
require regulating the bias line.   It may be necessary to switch 
oscillator circuits.

With those changes the gain of the mixer is adequate and S9 
occurs at 35-75 uV on 80 meters, the only band I'm testing right 
now.   However, the sensitivity is an unsatisfactory 3 uV for 6 db 
S+N/N -- should be 1 uV or less.  

The poor sensitivity does appear to be the mixer.  There are several 
possible reasons.  The tube is operating at about half the rated 
voltages, for one thing.  Also the drive probably still isn't quite up to 
optimum -- I cannot precisely measure it because of loading by the 
scope.   

The 6JH8 is in this application an extremely linear high gain 
pentode mixer.   Pentode mixers aren't known for being quiet.  The 
noise, however, doesn't come from the grid circuit, which I think 
means that a further increase of gain (=drive) should improve 
sensitivity.

On the positive side the thing is an excellent strong-signal 
performer.   On 80, I can hear anything anyone else can hear, and 
better than most.   No matter how lousy the conditions, if I can 
hear it I can copy it.   

The VFO is outstandingly stable ... drift is down for about 200 cps 
in the first few minutes, then small enough that I haven't measured 
it.  It's in the same ballpark with the crystal oscillator.  

There's still plenty to do.   In addition to the sensitivity issue the 
filter passband is excessively 'lumpy,' the calibrator isn't wired yet, 
only 80 meters is working (the crystals came yesterday), the front 
panel is unfinished, the S-meter zero drifts continuously, there's a 
strong spurious response that looks like a harmonic is involved 
near the low end of 80, the Butler oscillator tuning drifts 
excessively ... I'm sure there's more.  

Lessee ... 4x 5.4 Mcs = 21.6 Mcs; 21.6 -18 = 3.6 Mcs which 
would tune with the VFO at 5.4 Mcs.   That's it, I bet.   There's the 
price of high VFO drive -- it's probably pushing the 6JH8 premixer 
into a bit of grid current.   When you're dealing with signals of 
several volts you don't need a large percentage of distortion to get 
several microvolts of spur.   

The lumpy passband is probably due to mismatching the crystal 
filters.  I was extremely careful about feedback paths and the 
lumps don't change with IF tuning.   Some experimenting with the 
crystal filter loads will probably help.   First thing to do is to check 
my working Tempo One and see what that filter looks like there.   I 
know it's not great.   4 pole, for one thing.   

An interesting possibility for a future design is use of a beam tube 
premixer as a combination VFO-mixer, eliminating the separate 
oscillator tube.   A single-ended oscillator would be substantially 
isolated from both the deflection plate input and the output because 
both of those are push-pull; the circuit is shown in various sources. 

A disadvantage would be that these tubes don't have high gain 
when considered as amplifiers.  Transconductance as a pentode is 
about 4400 and using the screen as a plate for a Hartley or Colpitts 
circuit it would be considerably lower; the 6GK5 is about 15,000.   
This would definitely be a compromise VFO design, although the 
savings of a tube might make it worth trying for compact 
equipment.   

Walt
KJ4KV