Two-valve push-pull stereo

[Lucien Nunes]

Anybody fancy building a proof-of-concept demo unit?

[Kat Manton]

I might "build" it in LTSpice...

(Has anyone got any idea about transformer ratios? "n:1" and "p:1" aren't sufficiently informative; was there an expression of how n and p are related in the original text? It saves me thinking too hard...)

[Skywave]

Originally Posted by murphyv310

Hi.
Either I am stupid or this is circuit is way off the mark. T2 has one core, Correct? so how on earth can there be any marked stereo seperation?
Any subtraction of the signal will in this circuit cause cancellation (no voices on stereo) or Addition thus all middle

Well, not necessarily exactly along those lines, but that Post does encompass the spirit of some of my bewilderment about this circuit . . . . which is why I made my Opening Post in the first place - and its request therein.

Based on Posts made to date - for which thank you - I'm going have a detailed think about this later and come back, if I come up with anything that might constitute a worthwhile contribution.

But don't hold your breath

Al.

[TuningIndicator]

Originally Posted by murphyv310

Audiofools!!!!!!

Steve Bench audiophool? No!
Hugely talented engineer, innovator, designer, creator, writer and holder of several patents? Yes!

I think his site gives the best treatise to valve circuit design on the net.

http://greygum.net/sbench/sbench101

[kalee20]

Originally Posted by murphyv310

Either I am stupid or this is circuit is way off the mark. T2 has one core, Correct? so how on earth can there be any marked stereo seperation?

Suppose you apply a signal to just one valve (V1, the 'left' valve). Then, T2 the centre-tapped transformer, will obviously see this amplified signal, and it'll be applied to one terminal of both loudspeakers. But, T1, the upper transformer, will also see this signal. The phasing of the secondaries is such that the T1 output adds to the voltage seen by one loudspeaker, and subtracts from the voltage seen by the other. If the turns ratio is a critical figure, the subtraction results in complete cancellation in one loudspeaker, so only the other one responds.

The converse happens if V2 is driven. So lo and behold, if both valves are driven by separate signals, each speaker gets only the signal destined for it.

The down side is that if the transformers are not quite ideal, the separation will not be complete. In the alternative configuration (a single valve and single transformer per channel), lack of matching of transformers will merely manifest itself as a slight imbalance between the volume of the two channels.

[Synchrodyne]

The said circuit does seem to have been used in a production chassis – schematic on pages 14-42 and 14-43 of “Radio and Television Engineer’s Reference Book”, 4th Edition - and it was probably a Murphy chassis. (The author of chapter 14 was L. Driscoll of Murphy.) I don’t have the referenced Audio magazine article, but Audio Anthologies” Volume 5 has another article, from the November, 1958 edition, that I think provides a clue, namely “For Stereo, The Bi-Ortho Output Circuit”.

It would seem that the circuit at interest is an application of the phantom channel transmission concept, in which effectively (L+R) is carried as the main channel balanced signal, and (L-R) is the unbalanced phantom signal injected and separated via centre-tapped transformers. In this case an amplifier is inserted in each side of the balanced line between input and output. The amplifier output transformers also serve as the phantom channel separation transformers, and follow the same basic pattern.

In this case input transformers for phantom injection are not needed. If single-ended L and R signals, one inverted with respect to the other, are directly applied respectively to each side of a balanced pair line, the notional centre-tap carries the (L-R) signal with respect to earth.

The apparent advantage of this circuit – according to the above-mentioned Audio magazine article, is that the total available instantaneous power for each channel can approach the amplifier capacity, this varying inversely with the correlation between L and R.

On the face of it, the circuit might work more-or-less as expected, but as has already been said, the difficulties of maintaining balance in any system that uses matrixing of L and R would be inimical to success, the more so in a mass-produced receiver where cost-control was paramount. That’s why, for example, time-division rather than frequency division has been preferred for FM Multiplex decoders. (See for example: https://www.vintage-radio.net/forum/...456#post349456.)

The main output transformer, T1, would, I imagine have been fairly standard except for the centre-tapped secondary – I would guess that for the chassis at interest, each half was intended to work into a 3 ohm loudspeaker.

From the schematic in RTVERB4 it does seem to be something of a mixed bag – 4 SW bands as well as MW, but no RF amplifier (cost-cutting perhaps), but a double-wound mains transformer (not cost-cutting). The 6LD12 triple diode triode valve is used somewhat unusually. The independent diode is used as the AGC rectifier. One of the common cathode diodes is used as demodulator, the other is unused. The triode is used as the left-channel phase inverter.

Cheers,

Originally Posted by Kat Manton

I might "build" it in LTSpice...

(Has anyone got any idea about transformer ratios? "n:1" and "p:1" aren't sufficiently informative; was there an expression of how n and p are related in the original text? It saves me thinking too hard...)

It's going to be (ignoring the impedence match, you can have a 2k spice 'speaker!) 1:1, 2:1 or 1:2 in some combination. I wouldn't think about it, it's quicker to give it a go and tweak the values.

[G8HQP Dave]

Think of driving just one channel. Then the output has to be in just one speaker, so the secondary voltages must be equal. I think this means that n and p are equal, if you assume that the non-driven pentode does not load the OPT. The anode signal current goes through half the P-P primary and all the SE primary. We need the same output voltage at half the P-P secondary and the SE secondary, so they cancel at the other speaker.

[Synchrodyne]

The original diagram on page 14-47 of RTVERB4 does show the same turns ratio, namely n-to-1, for both transformers.

Cheers,

[murphyv310]

Yes of course we know that it "may" work under perfectly ideal situations, but and it's a huge but, is impractical. OK designers are clever people I am fully aware of addition and subtraction, I worked with Audio for a good number of years but really didn't like it, I was more into TV.
This circuit can not reproduce proper stereo and I would think that even the designer would have to agree.
I would like to give it a try but where would you get transformers wound to a very high spec for a start, I am still not convinced though!

[Skywave]

Originally Posted by Skywave

Based on Posts made to date - for which thank you - I'm going have a detailed think about this later and come back, if I come up with anything that might constitute a worthwhile contribution.

And now having reviewed all the Posts to date on this Thread, I cannot think of anything additional & constructive that I might add: my original Q. has been answered; a lot of other interesting & relevant aspects have come to light.

So I'm now going into 'read only' mode on this Thread; thanks to all who have contributed.

Al. / Skywave.

[Synchrodyne]

Something that has since occurred to me is that is that this circuit seems to depend upon interleaving in much the same way as does the Zenith-GE stereo coding system for FM broadcasting. The push-pull input is M (aka L+R), whereas the parallel (common-mode?) input is S (aka L-R). Thus when M is large, S is small and vice versa, allowing either to exploit the available output power. Allowing that in practice the hard-to-avoid errors probably – or almost certainly - outweighed any benefits, nonetheless the idea would have seemed attractive for equipment where cost constraints outruled the possibility of a pair of push-pull amplifiers, yet the designers wanted a push-pull output, perhaps for continuity of features from a preceding mono model.

As at least circumstantial evidence in that direction, the receiver schematic shown in RTVERB 4, inclusive of the two-valve push-pull output stage using a pair of 6P15 valves, is described as an “export radiogramophone chassis”. The corresponding chapter in RTVERB3 (1960), same author, also has a schematic for an “export radiogramophone” in about the same place, in this case a mono model with the output stage being a pair of 6P25 valves in push-pull. Both appear to be Murphy designs.

Cheers,

[Top Cap]

This is the technique exploited in the Emerson E502, just 3 valves (including rectifier) for a complete stereo system.
A note about the system is on page 12 of the 1959-60 R&TV servicing books. The circuit was developed from the CBS Simplex system One high quality push-pull transformer is used and one cheaper transformer as its bass component is unimportant. The idea is to amplify the major component of the stereo signal in push pull, while the minor component is amplified in push push arrangement.
(Their description - not mine!)
It is claimed that any theoretical limitation is offset by the economies offered.
Apologies for the poor quality of source material - its just getting old like me

[neonlamp]

This circuit works, but seems pointless other than to show that a designer can contemplate a puzzle for fellow engineers to solve!
As one input is in opposite phase to the other, the speaker phases are inverted unless one of the speaker terminals, which would be unusual. As G8HQP Dave said, the lower transformer is L-R, the upper, because both signals pass through, is L+R. So feeding this to the centre tap provides L and R outputs.

However, if the transformers are not perfect (and of course they won't be) some breakthrough signals from one channel could appear on the other because of imbalance or incomplete coupling in the windings.

In addition, it seems to me that the combined transformer (L+R) would have to handle the total standing current of both sides, and therefore have to have a big inductor with an air gap, even if the difference transformer is balanced and could be a smaller ungapped core. Which makes for practical balancing even harder.

Therefore, I concur that this imaginative circuit is impractical for a stereo system and
agree we would all be better off with a conventional output transformer and valve for each side. As normal. Probably a radio engineers view of what an audio circuit should look like!

John

[PJL]

Worse than 2 class A outputs in almost every aspect. Twice the winding resistance, twice the winding capacitance, twice the DC current in the difference transformer, twice the AC power handling requirement in both transformers, and the primary inductance still needs to be the same.

However, for monaural (or central signals) it should deliver a reduction in distortion from the push pull benefits.

There has been mention of transformer accurcay but this seems a minor issue when compared to the driver gain. I guess you would need a balance trimmer to set it up.

We need someone to have a go!

[peetleetech]

Folks, this set would provide the illusion of something if you sat just right. If we assume the phasing on the speakers were kept correct then at the exact middle you would hear two signals. It would tend to sound a bit different as each 'speaker would be out of phase. I think that's the secret.

[Neil F]

Hi, all - I've just come across this thread and haven't had time to go through it in any detail, but I can confirm that by a spooky coincidence a couple of weeks ago I serviced a Ferguson 658 RG which appears to use this circuit. The service information is in R and TVS, and is available up top. It's a big, bold RG from around 1960 and Ferguson incorporated a fair few bells and whistles in it, so it wasn't a cheapy. Sounded fine!

[Skywave]

And that last comment by Niel F seems to sum up the design philosophy behind these unconventional circuits: the targetted profit strictly dictates the design. Re-arrange the costs so that the quality & functionality - as perceived by the user / listening customer - results in maximum profit for every £ so spent on the design & consequent build costs.

In other words, never mind the (perceived) quality - feel the width!

Al. / Skywave.

[peetleetech]

I have looked into this thing some more. The primary would be around 3000 ohms. They say there is nothing worse than a person with a little knowlege. There are plenty of examples. Take a push pull output, remove the inverter and you have 2 amps, okay? A little biasing so that one valve is off whilst the other is on=push pull. Place the center tap to ground and I might like to built it. Can't be hard. I'll find some dual valve and try. report when the doctors have finished with me.

[G8HQP Dave]

No, you need to keep the inverter for one channel.