Outboard DACs

I discuss the operation of LAiV's R-2R DAC in our boxout [see boxout], in addition to the time domain behaviour of the OS (oversampled) and NOS (non-oversampled) processing modes. It's worth noting here that LAiV's upsampling does not extend to 4x44.1kHz (176.4kHz) or 4x48kHz (192kHz) sample rates where the µDAC effectively runs in 'native' or NOS mode regardless of the OS/NOS setting. The same is true for all DSD inputs. Otherwise, the µDAC behaves largely as expected, offering a full 4.05V balanced output (3.96V with 44.1kHz streams) and a very generous 115.5dB A-wtd S/N ratio from a 41ohm source impedance. Despite the claimed 0.05% tolerance of the R-2R DAC's element matching, and wide S/N ratio, low-level linearity is merely 'good' at ±0.5dB over a 100dB range but not exceptional [red trace, Graph 1]. The drift to +1.5dB at –110dBFs is bettered by most modern ESS/AKM DAC-based solutions, for example.

By contrast, LAiV's use of an Accusilicon AS318-based master clock has proven benefits in the low 22psec jitter (48kHz/24-bit data) and very low phase noise, as evidenced by the lack of spectral broadening in the J-Test driving signal [see Graph 2]. Distortion, on the other hand, is both 'digital', reflecting quantisation errors in the R-2R DAC, and 'analogue' from the Class A output stage. In practice 2nd-4th harmonics dominate the very uniform 0.004% distortion at peak output (re. 20Hz-20kHz) while extended, higher-order harmonics take hold at lower digital signal levels. THD falls to a minimum of 0.003-0.0065% over the top 15-20dB of the µDAC's dynamic range but differences between OS and NOS modes are only apparent at very high frequencies [blue vs. orange infills, Graph 1]. PM