ESS more tricky and measured DR of 9038 with AVG-detector instead of RMS-detector(all another measure RMS de facto) that gave 3..2.5db of "improvement". Being placed in a near to ideal environment the 9038q2m shows 126..126.5db(A-weighted RMS or -123db unweighted) dynamic range/SNR, and 3-4% of samples may have result 6db worse. Dual 9038q2m configuration improves that number for 3db i.e. 126..126.5db unweighted or 129db (A).
As you can see, 9068 is exactly the same 9038 + MQA(lossy audio format similar to AAC, mp3, APTX). I didn't see the 9068 datasheet but I'm sure they did measure DR standard way(bravo Martin!) and got 126db(A), but SNR I guess was MUTE=On
Every time when DR and SNR mismatched it is fishy.