[HBR] Another Receiver Project -- HBR-4, Part 21

[waltah at earthlink.net]

'Shhhh ... it doesn't know it isn't push-pull.'

Okay, so actually, the new RF stage *is* push-pull -- sort of.

I used a 6ES8 with the self-inverting input circuit (drive one grid, 
ground the other, leave cathode unbypassed) and fed the two 
plates into a center tapped (bifillar) winding on the existing slug 
tuned output coil.   It would have been tidier to take the output from 
the plate connections straight to the two 6JH8 deflection plates, 
but because of the external coil added in series with the tuned coil 
on 40M and below, the gain on the low bands was too low.   The 
tube works push-pull, but the output is taken single-ended to just 
one of the mixer deflection plates.

That should have all the advantages of a push-pull RF stage 
(cancellation of 2nd harmonics) and I can't think of any real 
disadvantage.   The idea seems worth trying in other receiver 
designs; you don't get much gain from this sort of RF stage but 
combined with a change to a higher gain mixer it might even be a 
possible 'mod' in a few well-designed and built older sets.

Next up:  The premixer, which (a) did not seem to have enough 
output, and (b) delivered seemingly unnecessary spurious signals 
because it was not balanced to reject the VFO signal.   Yesterday 
I redesigned it to use a 6922 (=6DJ8 but with 300 Ma filament), 
driving the grids from the VFO and the cathodes from a winding on 
the crystal oscillator plate coil.   Both inputs in push-pull and the 
tube plates in parallel so both input signals should cancel out.  

The first problem was that the receiver was dead -- all I got was a 
heck of a lot of noise.  Right ... the 6922 is a VHF tube and I had 
omitted parasitic suppressors.   Adding 100 ohm resistors right at 
each grid connection got things working.   

It does have higher output than the former configuration though I 
have not made measurements yet.   However, it also has even 
more spurious signals.   Most of the new spurs drift and have lots 
of hum; they are all tunable, meaning that the VFO is somehow 
involved.   There are two possibilities:  

1.  They are the result of a VHF parasitic beating with a harmonic 
of one of the oscillators -- most likely the crystal oscillator.   

2.  They are beats between very high harmonics of the VFO and a 
harmonic of the crystal oscillator.  

There's a balance adjustment for the premixer and it greatly 
reduces some of the new spurs but has no effect on others.   It 
could be that this indicates that both types of spur exist or it may 
mean two different paths are being followed for the spur to get into 
the IF.   For example there's obviously the direct path right into the 
grid of the mixer along with the intended premixer output signal.  
(There's only one tuned circuit, so spur attenuation is not that 
great.)   That one would be affected by the balance adjustment.   
But there could also be a radiation path (not enough 
shielding/spacing), a path via the filament circuit, and probably 
others I don't think of that would be unaffected by balance.   

The premixer cathode is 'hot' for RF (both VFO and crystal 
oscillator signals would show up there) so there will be spurious 
signals formed by the heater-to-cathode 'diode.'   

(Because the filament is hot, it forms the 'cathode' of a really 
crummy diode for which the cathode sleeve is the plate.   When 
the cathode sleeve is positive as in the usual cathode bias circuit 
with the filament grounded, a tiny current will flow in that diode.   
That current will be time-varying due to the AC on the filament and 
will be switched by any signal voltage appearing on the cathode 
sleeve.   A perfect (truly lousy) modulated diode mixer ...   The 
cure is to bias the cathode and associated grid returns a few volts 
negative so that the H-K diode never conducts.)

First thing to do is back bias the premixer cathode.  And it 
certainly needs to be separated from the RF stage (which has that 
unbypassed cathode resistor for self-balancing) by a filament 
choke.  

Then more steps as necessary to get rid of the spurs.  There's no 
obvious reason that with a balanced premixer and a high-Q tuned 
circuit in the premixer plate, there should be more than one or two 
faint ones.   

And then, measure performance.  

Walt 
KJ4KV