Fixed bias Class AB Push-Pull to Single Ended, Power reduction.

[varaso]

Great, I'll open the amp up as soon as I have the time and mess with it !! Thanks a lot !

[martin manning]

Yes, the Pr:Sec winding ratio is halved, and for reasons I explained earlier, the load to the ‘on’ tube is halved ( because only 1 tube is delivering AC- through half of the primary winding). So it’s the same as running a tube ‘1-knotch’ away from the optimal load. Output power will be reduced. The load line will be steepened, so you are less likely to risk overdissipating the screen grid. However, the plate will be running into a lower load (which means it won’t be as efficient).

I disagree, as mentioned above:

The off tube will be conducting, just sitting there idling and not responding to the signal coming from the phase inverter. That's the same as being in cut-off as far as the other tube is concerned, so that one will run on its normal Class B load line at 1/4 Ra-a. It will clip early, though, because it is not center biased like a single-ended amp would be. That may be ok for what you are looking for.

[varaso]

Well I tried it just like martin manning said, the overal headroom was minimal but the clipping characteristics were not great.. it was clipping in a weird non musical way..
could that be crossover distortion? With the volume higher, the distortion was heavier and sounded kind of decent with a lot of sag, it was hard rock sound though, I didn't really dig it..
I guess it should be cathode biased to sound better.. ?

any suggestions?

[martin manning]

Have you thought about an on-board attenuator? I guess you have used a L-pad, but I have a Marshall 2x 6V6 4001 amp that has the L-pad network shown below in it, and it seems to work pretty well. The headphone out with the speaker connected there produces about 1W. You could play with the values if you want more output, or perhaps add a cap if you want to boost the high end.

[tubeswell]

Yes, the Pr:Sec winding ratio is halved, and for reasons I explained earlier, the load to the ‘on’ tube is halved ( because only 1 tube is delivering AC- through half of the primary winding). So it’s the same as running a tube ‘1-knotch’ away from the optimal load. Output power will be reduced. The load line will be steepened, so you are less likely to risk overdissipating the screen grid. However, the plate will be running into a lower load (which means it won’t be as efficient).

I disagree, as mentioned above:

Okay - this is my line of thought.

Say P-PZout = 8k, that is for 2 tubes, one on each side of the Primary. The Pr:Sec impedance ratio for an 8R speaker is 1000:1. That's a Pr:Sec winding ratio of 32:1. Each tube on its own effectively 'sees' 1/2 of the 8k P-P reflected load through its respective 1/2 of the Primary (between the plate and the centre-tap) when both tubes are conducting AC*, because the impedance is shared, and each 1/2 of the Primary terminates at the CT. If one tube is conducting only DC, then there is no change in current on that 1/2 of the Primary, and it will not transform. So the Pr:Sec winding ratio for the 'on' half become 16:1. (16:1)^2 = 256:1 impedance ratio, which gives 2k load with an 8R speaker. This is half of the 4k load normally 'seen' on each 1/2 of the Primary when both tubes conducting. (this is what I was meaning. )

*Albeit that the conduction of each Primary 1/2 is 180 degrees out of phase with the other ('positive' or 'negative' current is still current, and change in positive or negative current is still change in current). The direction of the turns on each 1/2 of the Primary is reversed w.r.t. the other 1/2, so the opposing signal phase in each half transforms into a single magnetising current phase in the core. In this regard, both tubes are effectively in parallel, and so the impedance seen by both must be 1/2 of what it normally would be for one tube under equivalent voltage and current conditions into a SE OT with the same overall impedance ratio, because there is 2 x the current in P-P.

[Tony Bones]

[Deleted because the post made no valuable contribution to the thread.]