Sonic Frontiers Transport 3 CD transport & Processor 3 D/A processor Measurements
Due to logistical difficulties, the samples of the Transport 3 and Processor 3 that I measured were not the ones auditioned by Shannon Dickson. Unfortunately, my first sample of the Transport 3 wouldn't play discs, so Sonic Frontiers FedExed a replacement. Except where indicated, measurements were made from the Processor 3's balanced analog outputs.
The Processor 3's output was noninverting from its unbalanced outputs, and from its balanced outputs with pin 2 wired as "hot." The maximum output level was a high 5.82V balanced, 2.9V unbalanced. Its output impedance was 290 ohms balanced and 125 ohms unbalanced, both of which should be low enough to drive difficult combinations of cable and preamp load impedance. However, I did notice that, into a 600 ohm load at high levels, the waveform peaks were visibly rounded, suggesting that very low input impedances—such as the Bow Technologies Wazoo integrated amplifier, or some of the Jeff Rowland Design Group preamplifiers—are probably best avoided.
The Processor 3's frequency response into 100k ohms (fig.1, top trace) was perfectly flat over most of the band, with a very slight droop apparent in the top octave. The channel balance was superb, though this did worsen into 600 ohms (fig.1, middle traces). The response with de-emphasis (fig.1, bottom traces) was the same as without, implying zero error. Channel separation (fig.2) was excellent from both sets of outputs, though there was less crosstalk from the balanced XLR jacks. The rise in crosstalk above 1kHz is due to a very small degree of capacitive coupling between channels, but this is superb performance nevertheless.
Fig.1 Sonic Frontiers Processor 3, frequency response at -1dBFS into 100k ohms (top), into 600 ohms (middle), and de-emphasis response (bottom) (right channel dashed, 0.5dB/vertical div.).
Fig.2 Sonic Frontiers Processor 3, crosstalk (from top to bottom): L-R, R-L, unbalanced; L-R, R-L, balanced (10dB/vertical div.).
Fig.3 shows the spectrum of the Processor 3's balanced output while it decoded data representing a dithered 1kHz tone at -90dBFS. The top pair of traces were taken with the Audio Precision's digital generator set to a 16-bit word length. Increasing the word length to 20 bits dropped the level of the noise floor by just over 20dB, unmasking a trace of power-supply noise at 60Hz. At -135dB (left) and -140dB (right), this won't be bothering anyone. Increasing the word length to 24 bits did drop the noise by another 2dB or so—this is one of the highest-resolution D/As we have measured, with ultra-low noise and a true 20-bit dynamic range. (A "24-bit" DAC may take in 24-bit digital words, but its real-world resolution is limited by its analog noise floor.)
Fig.3 Sonic Frontiers Processor 3, spectrum of dithered 1kHz tone at -90.31dBFS, with noise and spuriae (from top to bottom): 16-bit data, 20-bit data, 24-bit data (right channel dashed).
Not surprisingly, the processor's linearity error (fig.4) is basically zero down to well below -100dBFS. And feeding it 24-bit data representing an undithered sinewave at -90.31dBFS gave a waveform that had an excellent sine shape (fig.5). This also indicates excellent low-level resolution and ultra-low noise, both well in advance of the 16-bit CD standard. Performance-wise, the Processor 3 is definitely "24/96 ready."
Fig.4 Sonic Frontiers Processor 3, left-channel departure from linearity (2dB/vertical div.).
Fig.5 Sonic Frontiers Processor 3, waveform of undithered 1kHz sinewave at -90.31dBFS (24-bit data).
The analog circuitry seems very linear, to judge from fig.6, which shows the spectrum of the player's balanced output while it reconstructed a low-frequency tone at full level. The second harmonic is the highest in level at -86dBFS (0.005%), though the fifth, sixth, and seventh harmonics can also be seen. These harmonics remain the same in level from the unbalanced outputs (fig.7), though the third and fourth now make an appearance. All are very low in level, however. Looking at the output spectrum (fig.8) with the "torture" signal of 19kHz and 20kHz tones, each at -6dBFS, a number of intermodulation products can be seen, though the highest in level are still at or below -90dBFS.
Fig.6 Sonic Frontiers Processor 3 balanced output, spectrum, DC-1kHz, 61Hz at 0dBFS (linear frequency scale, 20dB/vertical div.).
Fig.7 Sonic Frontiers Processor 3 unbalanced output, spectrum, DC-1kHz, 61Hz at 0dBFS (linear frequency scale, 20dB/vertical div.).
Fig.8 Sonic Frontiers Processor 3, HF intermodulation spectrum, DC-24kHz, 19+20kHz at 0dBFS (linear frequency scale, 20dB/vertical div.).
I examined the Sonic Frontiers' jitter performance with the units connected both by a conventional AES/EBU datalink—6' of 110 ohm Canare cable—and by the I2S-E datalink, using the Miller Audio Research Jitter Analyzer. The foreground trace in fig.9 shows the narrow-band spectrum of the Processor 3's unbalanced output while it decoded data representing an 11kHz tone with an average level of -10dBFS (-6dBFS peak). This is overlaid by a 229Hz squarewave at the LSB level. The peak-peak jitter was the lowest I have ever measured, at 128.5 picoseconds (ps), while the clock error was a low +36 parts per million. What jitter is present mainly consists of data-related sidebands (indicated by the red markers), though there are also some sidebands present at a low 15.6Hz fundamental frequency (purple markers).
Fig.9 Sonic Frontiers Processor 3 driven by Transport 3 via I2S-E datalink, high-resolution jitter spectrum of unbalanced analog output signal (11kHz at -10dBFS with LSB toggled at 229Hz). Center frequency of trace, 11kHz; frequency range, ±3.5kHz. (Grayed-out trace is the Processor 3 driven by the Transport 3 via 6' of 110 ohm Canare AES/EBU datalink.)
For comparison, the grayed-out spectrum in fig.9 is a spectrum taken under identical conditions, except now the Transport 3 is driving the Processor 3 via the Canare AES/EBU link. Note the rise in data-related jitter—the 229Hz sidebands marked with a red "4" have risen tenfold, from 43.5ps to 456ps—with the total peak-peak jitter now measuring 493.4ps. While the higher harmonics of the 15.6Hz jitter have disappeared, note the very strange symmetrical rise in the noise floor either side of the central peak. This is presumably a characteristic of the UltraAnalog data-receiver circuitry, as it was not changed no matter what datalink or transport I used. Substituting a Meridian 500 transport for the Sonic Frontiers with the Canare AES/EBU link gave 509ps of jitter, while using a TosLink datalink increased the weighted peak-peak jitter to 603ps. In each case the double-winged noise floor remained the same, the differences in data-related jitter leading to the differences in measured jitter levels.
Predicting the subjective effects of this kind of low-frequency, random-noise jitter is a foolhardy task, but it has been conjectured that it will increase the apparent size of individual images within the stereo soundstage, as well as lead to a decrease in pace and rhythm. I do note that SD found the Processor 3's bass "a bit too bloomy and diffused" when it was driven by a non-I2S-E datalink; this impression may well correlate with this jitter signature.
Finally, the Transport 3 managed quite well at tracking the Pierre Verany Test CD, which has intentional gaps in the data spiral. It played through Track 32 without problem, but started to mistrack on Track 33, which has a 1.5mm dropout. —John Atkinson