Hello,
Thought I might chime in since there seems to be some interest in this right now. I typically do not interfere in equipment threads because it is the users forum. Anyway .....
The units for Bruce are significantly different from the units that I demoed at RMAF. Bruce's requirements for level and eq calibration required me to figure out a way to achieve his needs and my requirement for audio quality !
So, to specifically answer a couple of issues, the input chip is indeed the ssm2019, I did significant listening to the head amplifier from the RMAF unit which was built around a discrete low noise pair of transistors on the front end and this preamplifier and found that this chip offered sound quality comparable to discrete if properly implemented, additionally, i was able to drive 2 eq sections directly without a follower and coupling caps etc so as in life, there are pros and cons, its just a matter of making a informed choice.
As far as feedback is concerned, even a tube with a cathode resistor has a certain amount of local feedback. The proof should be in what sounds good. I'm not talented enough to build a unit with a lot of feedback so I tend to stay away from it ;-)
Lastly, there have been some posts regarding input transformers, In my experience, I have found that either approach when properly engineered is capable of superlative results, it is however, a little harder to interface transformers to a large variance of head types since the power transfer is optimal around one set of values.
On a personal note, thanks to all on these forums for the kind words over the years. keep listening !!