[HBR] Another Receiver Project -- HBR-4, Part 20

[waltah at earthlink.net]

'A realization, continued.'

A push-pull RF stage is a good idea but at first it seemed to require 
a major front end redesign, mechanical as well as electrical.   But 
on reflection, maybe not.

Push-pull is a good idea because the natural curvature of tube 
characteristics causes generation of second harmonics at large 
signal levels.   

If you look at the characteristic curves in a tube handbook, the 
spacing of the grid voltage curves isn't uniform.   When you draw a 
straight 'load line' relating plate voltage to grid voltage for a 
particular load, the output voltage steps along that line marked by 
equal changes in grid voltage will not be equal.   In other words, 
distortion will occur and the worst of it will be the second harmonic -
- the big contributor to the third-order intermodulation distortion 
which sets the practical limit on the dynamic range of a 
communications receiver.  

Some tubes are better than others, but such distortion is a fact of 
life in amplifiers and when a wide range of signals must be handled, 
there's no way around it.   

In a push-pull amplifier, however, the two halves swing in opposite 
directions.  Adding the two signals with reversed polarity causes 
the second (and all even-order harmonics) to cancel out.   And in a 
receiver front-end, the result should be much reduced third-order 
IMD.   

The G2DAF Mk II receiver is the only ham set I'm aware of to use a 
push-pull RF stage and it was claimed to hit 100 db dynamic range 
-- the only ham design I'm aware of to make that claim.

Remember the self-inverting amplifiers used in some of the early 
oscilloscopes and some medium-fi and intercom applications?   
The output tubes are connected in push-pull with an unbypassed 
cathode resistor; one grid is grounded and the other one is driven 
by a single-end signal.   Why couldn't the same be done with a 
receiver RF stage?   The signals in the two tubes aren't perfectly 
equal and opposite, but they should be close enough for 
considerable cancellation of distortion.   And no push-pull input 
circuit required.

That leaves the output.   That transformer has to be push-pull, 
because that's where the two distortion signals get cancelled.   But 
why would the push-pull part have to be the tuned part?   What 
about letting the tuned circuit be single ended and just adding a 
center tapped auxilliary winding on the same coil, connected to the 
two plates?   Push-pull output can be taken from the two plates, 
single ended is available from the tuned winding.  

The gain of such an amplifier is that of the single tube with an 
unbypassed cathode resistor of the same impedance but it is 
divided between the two plates.   Should be roughly the same as 
what I have now, since half of the tube is used as a cathode 
follower and thus provides no gain at all.

I had a spare slug-tuned coil and it was the work of a couple of 
hours to tear it apart and add a bifilar center-tapped winding for the 
tube plates.   The next step is rewiring the RF stage to use that 
transformer;  should get that done today.

Depending on the 'luck of the draw' I may have serious instability in 
this stage but reversing the plate connections should deal with it, 
since the poor shielding is between the two tuned circuits.   Even 
in the present non-phase inverting form it is somewhat unstable on 
40 meters where the Q of the coils is at the highest.   I'll load the 
tuned circuit for that band with a resistor as the gain is way higher 
than other bands, anyhow.

In another minor inspiration, I changed the cathode resistor of the 
6JH8 mixer stage from the 330 ohms I had guessed at, to 1200 
ohms as indicated in the recommended balanced mixer circuit for 
the 7360.   That improved the sound of the receiver and in the past, 
better sound has always meant better numbers, once I figured out 
how to measure 'em.   Evidently the 6JH8 and 7360 are more alike 
than different.

Obviously I need to plan to go step-by-step to optimize that 
resistor.   

Signal to noise (sensitivity) is now perfectly satisfactory, I think.  
On any band, disconnecting the antenna greatly reduces the noise 
level.   

This is the best-sounding receiver I've used -- a real pleasure when 
listening in and around the DX window on 80 where most folks take 
clean, clear signals very seriously.   Lots of 'professionalism' (for 
want of a better word) there -- except for that piece of garbage that 
drifted from pile-up to pile-up, tuning up.  Heard last night, 4N600A 
and a few G's.   And that's on my 160M full wave loop at 15' height, 
fed with a twisted pair and tilted (by the slope of the land) toward 
the west.   Gee ... maybe I'll put up an 80M dipole at a little greater 
height.

Looking at the G2DAF transmitter discussion in the RSGB 
Handbook 5th edition, I found a reference to a Mk III version of his 
receiver.   Has anyone ever seen any more than this?   *That* 
would be an interesting design -- unless it's solid state.

Walt
KJ4KV