Thanks Keith that pdf provides a lot of explanation of all the factors that make Cohen's work unique.
I think one of the sources of confusion was that virtually any preamp using the "op amp 'round the transistors" bias scheme was being coined as being the "Cohen" topology in other fora - even examples which were not fully-balanced. That was certainly the source of my confusion reading those threads and that point I was trying to make was we had seen the bias scheme before. I've never seen a fully-balanced topology in the literature that predates Cohen's and I've seen quite a few since crediting him. Now, we have Cohen himself to set the record straight and I appreciate him doing that.
What had me confused was in Cohen's original paper this section:
I've shortened this for reading, the web site link has the full AES document.
That description is accompanied by this figure:
My first thought was "I think I've seen that somewhere." I looked at the Burdick schematic I had and wondered "did these guys talk?" That's what started me digging and I also found the Harrison (1978) , JR and bcarso the Mothchenbacher-Fitchen (1973) with Samuel Groner finding the Demrow citation. (1968)
Mr. Cohen has now clarified below what he meant by "new" when he wrote the 1984 AES paper:
OK, that explains the confusion. There are indeed many unique attributes that when summed in the entire preamp make it unique. But looking at figure 4 it's easy to draw the conclusion that figure 4, in isolation, was thought to be new.
The post here, which was copied from the original discussion at Prodigy-Pro, doesn't contain explanation of the differences in the various examples we cited. The early citations are instrument amps not preamps. But, each are examples of the biasing scheme some of which, the instrumentation amps in particular, wouldn't make very good preamps. The biasing was the object of comparison at Prodigy though admittedly that's not made clear here in this hastily-copied thread.
The Harrison preamp wasn't a particularly good one either and I hated servicing them. The current sources did make them noisier. Fortunately, because of that however I have more than a few LM394s. On the Benckmark the main thing that had Al going was the 2SB737 and yes it's not DC-coupled nor fully-balanced. I don't recall, but I'm fairly certain he had seen the Harrison. But, they are all similarly biased.
The point is that the biasing scheme was pretty well known (though not necessarily to Cohen) prior to 1984 without a clear point of origin. The trail seems to stop at Demrow in 1968 but who knows where he may have seen it.
I'm glad Mr. Cohen has told us what he thinks was "new" about it after so many years and thank him for sharing the great material and provding clarification and insight. I think most of the confusion was ours. I've done more than few things myself thinking they were "new" when there was prior work. I found a sum/difference network I used in the M/S decoder recently in a 1995 Burr-Brown catalog.
He's lucky to have been able to do a hybrid. Reading some of Derek Bower's papers on the SSM preamps he points out that SSM didn't have thin film resistors - only diffused - and that drove the topology of the SSM2015 and 2016. Once production of SSM moved to Analog Devices he had a laser-trimmed process available and it allowed him to make the SSM2017 with on-chip resistors.
Thanks again Keith for getting Mr. Cohen to contribute here.