In this article we SPICE our approximation of Decware Torii PP Amplifier and other vaersion of a PP design idea with SE OPTs discussed on DIY forums. The design is unusual in a sense that it utilizes 2 single-ended OPTs running in series/parallel. In particular, we study 6n1p driver and EL34 output tubes in class AB pentode mode with very high level of NFB.
2.   Schematics and simulation results
Here is our rendering of Torii's topology, which is based on 6n1p gain tube direct coupled to a 6n1p accordion splitter, running into a pair of el34 power tubes connected in pentode mode to an individual SE OPT, respectively. The outs of the transformers are in parallel and that is used as the source of a deep feedback loop. Note that we do not pretend this is exact rendering but likely it captures the spirit of the design:
Zero level of Negative feedback (R14 is very large)
6db level of negative feedback (R14 is ~ 70k)
20db level of negative feedback (R14 is ~ 22k)
23db level of negative feedback (R14 is ~ 8k)
26db level of negative feedback (R14 is ~ 3k)
Low distortion obtained with very deep NFB comes with a price in terms of stability (as usual). Here is the frequency sweep for the 26db setting, suggesting that motorboating and HF ringing/oscillation are a strong possibility if correction is not applied.
frequency sweep for the 26db NFB setting
3.   Some observations
The parameters of the transformers are likely to determine the shape of the frequency and step responce to a very lareg degree. I changed values of the inductance, resistance, leakage inductance and stray capacitance of the OPTs and saw variations in frequency and amplitued of the peaks on the frequency plot. SE OPT come with failry low Q as the gap reduces primary inductance but not much of leakage inductance and capacitance. Interestingly, I saw that a pair of transformers with different ringing properties frequently give better results due to mutual cancellation.
Motorboating is directly affected by primary inductance (and also should be affected by the value of coupling caps; I did not try changing the caps values). With L>100H it goes away, but that's too much of impedance for a small-core SE OPT, 10..30H is more realistic.
The amp would give better zero-feedback distortion figures and consequently woudl require less NFB to obtain some target THD if the output tubes were triode-connected, see eico-hf87.htm