[HBR] Yet Another HBR Project -- Chapter 3

[hbr@mailman.qth.net]

The oscillator kept drifting up more than I thought it should, also it 
was just a little unstable, wandering a few cps one way, a few, the 
other without apparent reason.   Applying the 'hot soldering iron tip 
under various parts' technique, I found that the 15 mmf fixed cap 
across the tank was not NP0 as labled, but N ... some fairly large 
number.  Swapping it for 4 3.9 NP0's in parallel fixed both problems.   
This combination has a slight positive coefficent -- but it is *slight*.   
The other cap must have been breaking down.  There's 50V p-p 
there, but evidently it was too much for that nominal 100V cap.  I like 
to use 500V or 1KV caps in these jobs but this time I didn't have 
one.  

Now it goes down about 100 cps in 5 minutes, then back up 120 in 
an hour and a half -- not sure if it's stable yet or not, but it's very 
close.  And there are a couple of other parts I haven't checked out 
yet.  

Jim said:
> "Inputs in pushpull" really means "180 degrees apart". So you have Mixer
> Side 1 and Mixer Side 2 operating identically except everything in Mixer
> Side 2 is 180 degrees shifted (out-of-phase) from Mixer Side 1 and
> viceversa.
> 
> So their outputs are ideally exactly 180 degrees out of phase 
> and...cancellation. 

Naw ... think *multiplier* -- a mixer is a *multiplier*.   In the described 
case, + times + = +, and - times - = +  so the outputs add.   (I'm 
neglecting the phase reversal of vacuum tubes ...)

You are focussing on the outputs resulting from the amplification of 
the input signals.   Those do indeed cancel -- that's the beauty of the 
doubly balanced mixer.   But the desired mixing products add.  

One of the few useful things in recent Amateur's handbooks is the 
detailed discussion of theory in reasonably plain language.   There's 
a chapter on mixers and some other stuff -- chapter 15 in my 1995 
book; the first couple of pages of that might clear things up.

Yet another way to see it:  Imagine a perfect mixer, i.e., no distortion 
occurs on either signal path.   (A beam tube is close, within its limits 
-- a very linear pentode amplifier with a second input that swings the 
electron beam from one plate to the other.)   Put the same exact 
signal in on both inputs:  what comes out?   If you draw two sine 
waves in phase and sketch the result from multiplying them together, 
you can see -- a sine wave at 2x the frequency, superimposed on a 
DC component.   (f + f = 2f and f - f = 0 in the frequency domain.)  
Now considering combining the output of two single ended mixers, 
the output at the plate receiving the two positive signals is *identical 
to* but 180 degrees out of phase compared to the other plate sure 
enough -- but at twice the frequency the 180 input degrees becomes 
360 degrees, i.e., in phase.   So the outputs can indeed be paralleled.

I admit to being unable to quickly find an example of the type circuit 
I'm considering in the literature but I think its just that hams didn't 
build a lot of complex vacuum tube circuitry.   

One practical example is the 'push-push' mixer I use all the time -- 
plates and cathodes in parallel, each input single ended on one grid.  
It's not really push-push at all, but push-pull for both signals on the 
same two grids because each grid input causes a cathode voltage 
change that is equivalent to a reverse phase input at the opposite 
grid.  Works like a champ, and if you used a pair of two grid tubes 
(with plates similarly in parallel) and applied one push-pull input on 
each pair of grids, it would be even better -- doubly balanced, which 
is the 'there is nothing better' of mixers.

The mixer tube I want could have been provided in a single 9-pin 
envelope. One cathode connection, two G1's, a G2 for both sections, 
2 G3's and 1 plate connection for both; total seven pins allowing two 
for the filament.  Unfortunately the tube manufacturers didn't see fit to 
do so. Of course since I'm the only one with an application maybe 
they didn't see this as a really big market.

A pair of 6BN6's should be ideal -- I can't see how slight 
microphonism would be a problem in this application.  Hahahaha ...

Quoting myself:

>> But I don't know about a Hahnel circuit with push-pull output ...
>> that would likely require more than one envelope.

The late news is that with some doodling I've come up with an  
elegant circuit giving push-pull output.  Namely a push-pull dual 
triode free running oscillator with a crystal connected as a *parallel* 
plate to grid (Pierce) oscillator.   

The circuit:  Coil plate to plate, center tapped for HV.   Each plate to 
opposite grid via small cap, bigger cap grid-to-ground.   (A 1:5 ratio of 
plate to grid voltages seems to work well.)  Cathode grounded.   
That's your basic push-pull oscillator, though I don't know the name 
of the circuit.   (It's a kalitron with added grid to ground caps.)   Now 
put a choke (or large resistor) in the plate HV connection and 
connect one end of the crystal to the tap on the coil.   Lift the 
grounded ends of the two grid-to-ground caps, join them, add another 
cap from there to ground and connect the other end of the crystal to 
the junction of the three caps. 

Should operate as a crystal oscillator with the tube sections in 
parallel on a frequency at which the crystal appears inductive and 
push pull on a frequency determined by the tuning of the tank, but 
with a strong locking effect since the two grids are pulsed in phase 
by the crystal.   You might think the synch pulses would cancel out, 
but one of the grids is below cutoff, so only the conducting tube sees 
the pulse.   

Interestingly this circuit should lock at 1/2 Fc, Fc, 1-1/2 Fc, and so 
on -- but I can't think of a use for that feature in this application.   

Strictly speaking it could either be a Hahnel or just a locked 
oscillator, depending on how it's adjusted.   The Hahnel is a *keyed* 
oscillator -- oscillation completely stops at the output frequency for a 
brief period determined by the crystal frequency.   But a locked 
oscillator can run continuously, with the crystal signal simply 
advancing the phase very slightly on each cycle -- i.e., the free 
running oscillator must be adjusted to a frequency slightly below the 
desired multiple.   Both types generate spurious outputs in the form 
of a spectrum of harmonics of the crystal frequency but it seems to 
me that the locked oscillator would have the better purity of the two.   
However, it is subject to jumping to another multiple if misadjusted, 
for example, as the tank circuit drifts.   The Hahnel, OTOH, would 
decrease output at the desired frequency and increase it at the 
incorrect multiple.  In a particular application, I suppose one bad 
behavior might be less bad than the other.   Anyhow, you don't get 
away from building stable circuitry by using either concept.

I don't see anything wrong with it, but I had two (former) marriages 
that I saw nothing wrong with beforehand ... I b'lieve I'll breadboard 
the oscillator rather than commit to it indefinitely at this point.   
Hopefully tomorrow.   I see a place on the VFO test chassis with 
holes for a 9-pin socket and a crystal socket ...

> One question on the pushpull 1U4 VFO: Will it have enough output to
> be useful?

5V p-p across a 270 mmf cap.  Since that's the grid-to-ground cap of 
the push-pull oscillator, there's 10V p-p grid-to-grid.   Way more than 
enough for a 6BN6 mixer ('bout 0.5V p-p for that), plenty for any 
ordinary mixer circuit (a couple of volts is usually ample) and well on 
the way to enough to use a beam tube mixer, if I wanted to.   Which I 
don't.   And that's with 60 volts on the plates -- I could raise or lower 
that, as needed to adjust the output.   

I really like push-pull oscillators.   More parts are needed and the 
tuning cap is 'more special' but the gain of the two tubes adds and 
the symmetry largely kills off even harmonics, making pure output 
easier to achieve.   Same advantages, actually, as push-pull 
anything else in a receiver.   

Jim (and occasional others), the continued critique is much 
appreciated!

Walt 
KJ4KV