Long post - apologies!

I first posted this to rec.antiques.radio+phono as that was
the focus of my question (and I didn't want to cross-post.)
However, I received a very interesting answer from John
Byrns that I thought was worth sharing with the denizens of
RAT.

I have copied the original below, plus John's reply (also
Bill's). Hope it all makes sense!

Cheers,

Roger

-------- Original Message --------
Subject: Old PP outputs stages
Date: Sun, 31 Aug 2003 15:32:14 -0700
From: Engineer
Reply-To:
Organization: Lorna & Roger at home
Newsgroups: rec.antiques.radio+phono

I've been perusing the schematics in Ed Kraushar's excellent
"Vintage Canadian Radio Schematics" CD (OK, get a life, I
know...!) and am a bit bemused by one of the phase splitter
designs employed in a couple of the sets that have push-pull
audio output stages.

The no-brainer solution is, of course, the coupling
transformer: primary in the audio amplifier plate circuit,
secondary centre tap to ground (or output tube -ve bias) and
the anti-phase secondary ends to the two output tube grids.
Apart from quibbles over transformer frequency response,
hardly a problem with AM radio, it's perfect, i.e. compact,
economical (in quantity) and provides some free gain to
boot. BTW, I would not use interstage coupling transformers
in "modern" tube Hi-Fi designs even though, IIRC, Williamson
and others did it for years, and there was at least one
solid state amplifier in the late 1960-70's that use it (The
Ravensbourne, I think it was, plus another in the same line,
the Ravensbrook - I had one!)

However, quite a few of the [old] radio receiver designs
don't use coupling transformers. They use one of the
following methods:

1. A triode phase splitter tube with equal cathode and plate
load resistors - this is the classic design still used in
some high end audio, so clearly OK for AM radio.

2. A separate phase inverter tube: the "in-phase" signal
from the directly driven output tube grid is taken through
an attenuator to the grid of a triode tube so that the
"attenuation x gain" is equal to unity (we hope), then the
output of this triode drives the second output tube grid. I
guess it works when the design is done right, the gain is
unity and the tube is new. But it's all open-loop so PP
balanced drive is dependent on component values and tube
gains remaining the same. This only gets a "B-" in my
design book, and that's generous.

3. This is the odd one that prompted this posting. There's
no interstage transformer, phase splitter or inverter tube,
as such. Here, the audio amplifier output goes to only one
output grid - via a coupling capacitor, of course. The two
output tubes share an undecoupled cathode resistor
(typically 330 ohms in the case of a pair of 6V6's.) So far
so good - we presume the second O/P tube is cathode driven,
providing the grid is grounded to signal by a large
capacitor. But it's not! In at least two designs (Astra
1946-47 Model DR103-47 and Brunswick 1931-32 Model B-H) the
second output tube grid goes to ground via 470K but without
any decoupling capacitor.

My questions on method (3), above, for the vintage radio
experts a

A. How does the second output tube get it's grid-cathode
(inverted) drive if the second grid floats?
B. If it somehow still gets driven, is it balanced?
C. Why not parallel the 470K grid resistor with, say, a 0.1
uF cap. to ensure the cathode space charge cannot move the
grid at signal frequency and unbalance the drive?
D. Even if the second grid was properly decoupled, would
just a common 330 ohm resistor provide a balanced output
stage? Or would the second output tube always be
chronically under driven?

Cheers,

Roger

PS. Anyone [who] wants to see "state of the art" phase
splitter design, see the Mullard "5-10" (also "5-20")
schematics. Here's one source, the "5-10":
http://www.bonavolta.ch/hobby/en/audio/el84_7.htm
Note C3 to ground. But we're well away from vintage radio
now! [BTW, I built two of these a long time ago, wish I
still had then!]

Bill Sheppard replied:

In article
,
(Bill Sheppard) wrote:

"John Byrns gave an excellent synopsis on the common-cathode, grounded
grid scheme 3 or 4 years ago. Turns out the thing will always be
lopsided because the grounded-grid tube will never 'see' as high a
signal level between its G1 and cathode as the driven tube sees. This is
with the grid directly grounded. Maybe John can elaborate further.
John...?"

John Byrn then replied as follows:

"I had forgotten about posting on that circuit. Without
looking it up, I
think my main complaint was about the construction articles
that were
popular for a while, and featured this circuit. They
basically just used
an unbypassed common cathode resistor, whose value was
essentially the
same as normally used with a push pull pair of the same
tubes. This
doesn't really work at all well because the resistor is way
too low in
value, and the second tube receives very little drive. The
circuit should
work very well for class A operation if the cathode resistor
is replaced
with a current source though. Since a current source would
be more
complex than the phase inverter we were trying to eliminate,
I did some
simulations and found that the circuit worked fairly well if
the cathode
resistor were increased in value, to maybe 3 or 4 times the
normal value,
I forget exactly what value seemed to be a good compromise.
The grid
resistors of the two output tubes must then be connected to
a tap on the
cathode resistor to provide the correct bias, basically the
upper part of
the cathode resistor would be equal to the normal cathode
resistor.

"Someone in this group recently posted about a radio, I
forget what it was,
that had what I thought was the cleverest phase inverter for
a radio that
I have ever seen. I forget what tubes the actual radio
used, but
basically the detector and audio amplifier tube was
something like a
12SQ7, with the normal plate load resistor split between the
cathode and
plate circuits the same way as in a split load phase
inverter, with the
two output stage grids connected to the plate and cathode as
with the
split load inverter. A normal split load phase inverter
would provide no
voltage gain, so an additional tube would normally be
required to provide
the needed gain. In this circuit the detector circuit and
volume control
are floated up from ground, and referenced to the cathode of
the 12SQ7.
This effectively drives the 12SQ7 between grid and cathode,
providing
normal gain, which is cut in half to each output grid by the
effect of the
split load.

"The most obvious problem with this scheme is that the
detector can no
longer be used to provide the AGC voltage for the radio
because the diode
detector circuit is flapping around with the amplified audio
at some
positive potential. This problem is easily overcome by
using a tube like
the 12SF7, which includes a diode, as the IF amplifier. The
diode can
then serve as a shunt AGC rectifier feed from the primary of
the IF
transformer, as is often done with a second diode anyway.

"This circuit is so clever I wonder why it wasn't commonly
used? Does it
have some subtle problem that is not obvious? Actually now
that I think
about it, I have also seen it used in some old British
construction
articles too.

"Regards,

"John Byrns"

Me again...

So, there it is - I thought quite fascinating. I'd really
like to hear what the RAT-ers have to say about that very
poor PP "design" in several commercial radios that prompted
my first post, also John's neat idea.

Cheers,

Roger
--
Roger Jones, P.Eng.
Thornhill, Ontario,
Canada.

"Friends don't let friends vote Liberal"