I guess this is also what you see reflected with the 5102A performance in the Denon 4800 compared to the 1690 in the Onkyo/Pioneer brothers/sisters. However a few observations;
It seems there are other SINAD bottle necks as well.
- how come the 3800 can't bring the same performance with the same input levels as the 4800 and
- why aren't the AKM's in the 3700/4700v1 performing around 107 dB as they theoretically could
Anyway, this remains a theoretical discussion. Still the top-tier AVR's from Pioneer/Integra/Onkyo use expensive ESS DAC's as does Yamaha. And of course Denon/Marantz as well. Interesting to know how such difference it makes in real life. In mine not as I explained in other posts. But the marketing thoughts behind it trigger me...
My two cents to your questions:
1) The 3800's SINAD vs output was measured in a different way as Amir explained, that could contribute to part of the difference.
2) "Typical" is just typical, we don't know the exact definition but obvious it means the "typical" THD+N of -93 dB for the PCM5102A could vary a bit, + or -, I guess depending on the conditions when measured.
3) SINAD, is the inverse of THD+N, noise is included so it is highly possible that the 4800 has less noise, contributing to higher SINAD.
4) Other reasons I don't know about, or understand, would need some dac experts to chime in and educate us curious people.
Also keep in mind that Amir has measured two each of the AV8805 (one with AKM, the other with ESS DAC), AV7705 (same AKM DAC) and one each of the SR7015 and SR6014 (same AKM DAC), even for the same models measured (7705), the measured pre out SINADs were slightly different two. And the SR6014 actually measured better than the SR7015, though only slightly.
That, to certain extent, is the nature of such bench measurements, they may be precise, but there are other variables that make repeatability not guaranteed, even if measured the exact same way on the same day.