My problem with the current batch of servo subs is that from what I can tell, they are late. Great at producing a continuous sine wave, but if the correction is on the next cycle, it is chasing it's tail. I had the idea of an optical sensor that could measure the cone position in effectively real time, but am not sure modern DSP processing could provide the feedback quick enough. My guess would be within 5 degrees.

5 degrees at 100 Hz is 1.4 ms, which is 60 PCM samples at 44.1 kHz. It is not necessary to have that much latency in a DSP processor. Nor is it necessary to use a DSP at all for a servo system.

I would hope not. But to do things with analog winds up with integrators and other things that causes phase delays. My only experience with digital servo was in computer capstan drives. We were only looking to accelerate and stabilize. Easy. I have to believe,with no real knowledge, that inductive feedback, pizo transducers, or what-not will force the overall timing out one cycle at least. That is why I was thinking optical (tapered window/light source ) for absolute position sense digitized before any analog amp. Signals processing has advantages like code that says "don't do that". I asked around some of the digital crossover "pro" companies about latency. 1ms or more was not uncommon. In the days of Ghz processors, I don't get it. Could it be that the mechanics are just not good enough to bother? It has been a lot of years since Rice and Kellog, and we still have overgrown dixi cups for diaphragms.
Between the two basic servo methods, totally unstable and control it, or make it close and force it, it seems like audio is of the second camp. Not sure why. Just things that keep me up at night. I would love to try my sensor design, analog or digital, but I just don't know how to design such a system. My previous attempts were not even very good oscillators. I'm just an old lab rat, not a real engineer.