DACs

Bryston BDA-1 Digital to Analog Converter and Halide Design USB to SPDIF Bridge

ARTICLE INDEX

On the Bench

Measurements were made with my new M-Audio Profire 610 Firewire audio interface with Spectra Plus FFT analysis software. The coaxial SPDIF output of the Profire 610 was used to drive the BDA-1 for some tests, and the Halide Bridge for others. The frequency response of the Profire 610 was measured and corrected for using a loopback test (output directly connected to input). All the plots are scaled to 0 dB Full Scale (dBFS) This means 0 dB on the plots is the maximum digital output level, with all 16 or 24 bits set to 1. I have no jitter analysis equipment, so I was unable to measure the jitter spectrum of the Halide Bridge directly.

Total harmonic distortion plus noise (THD+N) was measured using a 0 dBFS 1 kHz tone. Measurements were made at 16 bits 44.1 kHz, 24 bit 96 kHz and 24 bit 192 kHz, both with and without the SPDIF bridge driving the BDA-1.

At 16 bit, 44.1 kHz driven by the Profire 610, the THD+N is a very low 0.0037%, which is near the noise floor of the measurement system and fully consistent with the 0.002% THD+N quoted by Bryston. The push pull analog output stage of the BDA-1 yields odd order harmonics all lower than 90 dB down, but visible out to past the 11th harmonic.

The same test done with the SPDIF bridge shows a slightly lower THD+N measurement (0.0002% less) and a very slightly cleaner distortion spectrum.

At 96 kHz the spectra are similar, but cleaner. Driven by the Profire the THD+N is 0.0037% (top) and driven by the Halide Bridge the THD+N is 0.0028% (bottom).

At 192 kHz, the measurement was made only with the Profire as the digital source since the Halide Bridge tops out at 96 kHz. THD+N is 0.0034%.

THD+N vs frequency was measured with a 24 bit 192 kHz signal from 20 Hz to 96 kHz. The result was extremely low across the board. The bump at 40 kHz is an artifact of the measurement setup. The sharp rise near 96 kHz is due to the anti-aliasing filter roll-off.

Intermodulation distortion was measured at 24 bit 192 kHz with a pair of tones at 60 Hz and 7 kHz. IMD was a spectacularly low 0.002% (top). Another test was made with a 19 kHz and a 20 kHz tone. The 1 kHz intermodulation product was about 100 dB down from the test tones (bottom).

Frequency response was measured at 16 bit 44.1 kHz (top) and 24 bit 192 kHz (bottom). Deviation from flat is very small, and is most likely due to imperfect correction of the measurement setup's frequency response. The anti-aliasing filter at the higher sample rate is much more gentle than the lower sample rate, starting to roll off at 48 kHz.

Time domain response was measured at 24 bit, 192 kHz with a 10 kHz square wave. Very little ringing and round off of the square wave is evident.