Now I finally understand how the DR = 168 dB comes from. One of the D-1 specs also says
- Self-Noise 20Hz-22kHz bandwidth:
- -146dBu unweighted (calculated)
The self-noise is "calculated" under no load (signal) at idle when integrated over the entire frequency bandwidth (noise level is freq. dependent). Conversion back to the ’ratio’ = 10^(-146/20) = 50 nV (nano-volts=10^-9V). So, by definition, DR = 20 log(Vfs / Vn) = 20 log[10 / (50x10^-9)] = 166 dB. (slightly deviate from 168 dB due to the calculated self-noise inaccuracy) Note that Vfs is the full-scale (max.) voltage from its line out and Vn is the self-noise voltage.
I don’t know how other high-end DACs are compared with D-1 in terms of self-noise measure (again, at idle) simply because other DACs do not publish it. BUT, based on the DR / SNR converted from THD+N measure, D-1 (-114 dB) is actually no better than the top measured DAC such as Benchmark DAC 3 (128dB), Topping D90 III Discrete (131dB) or dCS Bartok (135dB). *** Note I do not imply how D-1 sounds compared to these top measured devices.

