1970s Design Indulgence

Can you expand on the 12MHz please? I'm not sure what you mean, or it could be me having a lapse of memory.

The anode or plate bending technique looks interesting but I couldn't get any combination to improve on what the amp simulation does. In case I wasn't understanding the "one-third" rule of thumb I tried it both ways round. It pulls gain down and then lets it peak before decaying, but unfortunately it doesn't meet unity before starting back towards the peak. The decay beyond the peak is then quite steep taking it beyond 180 degrees.

I think that seeing I've just about exhausted every known stabilisation technique (and myself), I ought to take a look at the power supply. Maybe I should snub away bridge rectifier switch-off parasitics? I think I've seen the occasional sporadic burst on the scope, but so far I didn't consider such events to produce the results I've been getting. Maybe I'm wrong (and happy to admit it if it results in a cure)? I can only safely place a 0.1uF across the transformer secondaries in the existing prototype. Let's hope it's sufficient.

Although it's academic because when I tried it, it was perfectly stable until given signal (from the AP - a quite small signal), it blew the fuses, I can model it in simulation, and did. I think I'd be up all night trying to find a balance point if indeed there is one. When you think about it, the idea is for a radio frequency detector which also filters out the carrier. I have a sneaking suspicion it is intended to oscillate.

I will agree it (the amp in its many different guises) is doing something at HF, which no extent of simulation modelling can reveal, or I think it would have by now (it's around 3 months).

It could be argued that the configuration is wrong, and if it is then why does the Proprius based on the same configuration behave impeccably left constantly on for months if not more than a year?

Within the first 30 hours of (this version) being on, after the slightest of modifications, the sound is incredibly good, but at 60 hours it fails to appeal. Perhaps the Proprius design is not scaleable? But I cannot accept that.

The only other difference is the power supply, and I would have thought a linear supply superior, but perhaps the filtering of a SMPS prevents ?whatever is happening?

Not daft at all! We were considering obtaining two 36V 4A switchers (to put in series for 72V) earlier today, but thinking about it, we'd better have four as each amp is "individually" supplied.

I forgot to add that I also tried the Proprius power supply inductor (20uH) and series resistor (0R33) just prior to the final reservoir/smoothing cap, but that made no difference.

Oh, and also limiting input high frequencies with a passive filter that's -3dB at 50kHz... no difference.

Last mod around 30 hours ago was a 0.1uF mains rated filter cap across both bridge rectifier "inputs" (each that is). It is subjected to 50V rms nominal, but rated for 230V rms continuous as an "input filter". We shall see (hear).