Yep, well, I guess that's the difference between subjective vs objective testing. Both have their good points. As you mentioned earlier SQ is pretty sensitive to input level balance & errant phase shifts. Just a little input balance to left or right could really alter the rear output. I used pure test tones and I think real world play back of, say, analog LP's could only go down hill from there.
How balanced was my input for QS/SQ center front to center back separation test? I used a single channel with a Y splitter to the SM input. The only potential imbalance would be the tracking between the sections in the dual gang input pot.
The concept of Axial Tilt has always appealed to me. A project I'm working on has an RIAA preamp built in & I might add an Axial Tilt circuit for that. Anyway, the Tilt circuit is built into your Space & Image Composer, right? Does it output also to be used with your Sony & SM?
In the QSD-2, QRX and other Sansui units with SQ decoding alterations were made to the direction sensing Phase Discriminator chips as well as the Matrix chip. This was a way to conform SQ more to the format of QS and the result was not quite as good as the QS format. Essentially it was a basic SQ decoder with complimentary front/back blending. Where as SQ VariBlend only worked in the rear chs to decrease center front bleed to the rear. Again, subjectively, the Sansui SQ decoding was considered by many superior to Full Logic because of the smoothness of decoding.
Now in regard to the SM SQ decoding, it was a generous after thought by Involve after much pleading from QQ members. Whether that accomplishment is a band aid like Sansui's implementation, only Chucky could say.
Myself, a three band high separation SQ decoder with no audible artifacts? Pretty friggin' cool!!