Output Stage Oscillating #2

[abbey road d enfer]

The problem as I see it, in general terms, is that the positive feedback loop is critically dependent on the parameters of the transformer.

Actually, the dominant parameter is the DCR of the primaries. The second one is the stray capacitance. What really makes things more difficult in terms of optimization is that the neg-Z drive is nested in an overall NFB loop. Ideally, the neg-Z drive should be stabilized without teh overall NFB, but the topology of the circuit doesn't allow that easily.

  In addition, the transformer actually being used probably does not need the positive feedback for proper operation so it may as well be dispensed with.

It really depends on the expected performance.  This circuit can reduce THD and linearize the frequency response of any xfmr by as much as 10-30 times. The best example is the AP oscillator, where the already excellent xfmr, custom-designed by Jensen, sees it's 20 Hz THD going from 0.1% to less than 0.01 and its attenuation at 20 Hz from 1dB to less than 0.1.
Now, I don't know why PYE made this choice, since a compressor has generally a significantly higher distortion and I believe the OT was not a source of concern. Probably it was an existing building block (although the "Compression Unit" and "Compression amplifier" use a very different output stage), or just someone in the lab fancying with the idea of creating the best output stage in the world...?

 

[CJ]

Do you mean the transformer the OP is actually using or the transformer of the original design?

Cheers

Ian

OP.

that transformer is very close to the Peerless S-217-D, maybe try the same zobel used i the eqp-1a and see where the smoke comes out,