Push-Push OTL?

[StoneT]

A couple of days ago I was out walking in the snow and thinking that my class 'A' push-pull amp was about as inefficient a design as you could get. But then I remembered OTL amps! My train of thought chuntered on something like this:
OTLs generally use paralleled output tubes so that each tube seems to see a larger load (amongst other reasons).
Couldn't you get the load to seem larger by driving both ends in phase rather than the opposite phase of a push pull?
If you had two cathode followers (one on each end of the load) and drive them with in-phase signal but attenuate the signal to one (the slave), the other (the master) would effectively see a large load.

I've modified another spice schematic I was working on to test it (was going to be a CF parafeed) It kind of seems to work but needs tweaking obviously. (Better tube selection, better way of providing unequal drive etc..)
The tubes shown are all ones that were being used in the parafeed project purely due to laziness. As I post this I realize that two SRPPs might be more appropriate than CFs but the principle is the same.
Does anyone know of previous designs like this? I'd like to work it up but if someone can tell me now that "Oh yes, that's the old humperknuckle design from the 60's... fundamentally flawed because of X, Y, and F..." it would save me the bother.

[Alastair E]

Err...

Call me daft, but if the signal is IN phase and of the same amplitude, the net output from the load would equal zero...

[StoneT]

That's vey true so you aren't daft. But the signal to the 'slave' tube is attenuated and therefore not the same amplitude. that's the function of the 72K/400K potential divider between the 5998s.
That is the main problem I'm having though. This divider increases source impedance of the drive signal to the top right tube causing HF roll off and a spike on the leading edge of a square wave test signal. That's why I put the .1nF past the 72K. this isn't ideal though.
Another way would be a 7K/40K divider with a bigger cap, but that reduces the load seen by the preceeding stage. Not good for a Mu follower.
The best way might be to have a divider on the output of the volume control and use the attenuated signal to feed a separate drive stage but then things are getting a bit big, which is what I was hoping to avoid to begin with.

[toprepairman]

Interesting idea but it's not really going to work at all efficiently.
back to the drawing board and some PL519's.
The correct approach is to use high impedance speakers, say a quad of 15 ohm types is series. So you now have a 60 ohm speaker, with about 3 amps peak available from PL519's, which is 540 Watts peak, 270 RMS, with a peak output voltage of 180 volts. Of course this is seroiusly exceeding permissable anode dissipation but it points the way forward.
Regards
Henry

[Steerpike]

Or get some of those Philips drivers with the 1000 Ohm voice coil, built especially for this type of amp.

[Sch3mat1c]

Eugh.

[richwalters]

Originally Posted by Steerpike

Or get some of those Philips drivers with the 1000 Ohm voice coil, built especially for this type of amp.

Back in the early 1960's, I remember a design by Philips using EL86's for use with an 800 ohm LS, so it isn't new. With modern day adhesives it should be far more reliable.

[toprepairman]

So we're agreed on the need for HiZ speakers. Do Philips still make the 1000R speakers, and were they any good?
For the sake of practicality and good quality we are back to multiple drivers.
At a london show only a few years back our honoured Tim de P. was demonstrating a speaker he had built which sounded completely fantastic, but has not been seen since or copied. It was a hexagonal column about 70 to 90 cm diameter, and 1.5 meters tall. I'm going from memory on the numbers so can't be absolute.
It had a string of tweeters, maybe eight of them running up the 'front' face and on each adjoining face was a row of mid bass units (6 or 8 inch), possibly six on each side, could have been more.
Anyway the point here is that these drivers could well have been wired up in series and driven by a valve OTL.

[toprepairman]

Just a quick extra note, those speakers were octagonal, not hex, typo on my part.

[DF96]

If you drive one end of the load with signal X, and the other end with signal X/4 (say), then you get the load effectively receiving 3X/4 - this is a partial bootstrap so it will raise the apparent impedance but at the cost of reducing output power. The load will also be wiggled up and down by a common-mode signal of 5X/8. I can't see the point.