Musical Fidelity Tri-Vista SACD player Measurements

Sidebar 3: Measurements

The massive Tri-Vista SACD player had maximum output levels of 2.31V for CD playback and 2.34V for SACD playback, both figures just over 1.25dB higher than the CD standard's 2V. The output preserved absolute polarity, and the source impedance was a very low 48 ohms across most of the audioband. It rose to 416 ohms at 20Hz, which mandates using preamps with input impedances above 4k ohms if the sound is not to become a little bass-shy. The error correction for CD playback was good, the Tri-Vista coping with gaps in the data spiral of up to 1mm without audible or measurable glitches.

The player's response for CD playback was flat (fig.1, top pair of traces) and its de-emphasis error was negligible (fig.2, bottom traces). The player's external digital input successfully locked to data with sample rates of up to 96kHz. PCM data encoded at that rate were decoded with a very slight response rise above 30kHz, before dropping like a stone above 45kHz (fig.2, top traces). With DSD data, recorded on the "provisional" Sony test SACD, the response was still flat within the audioband, but rolled off gently above 20kHz, reaching -3dB at 40kHz. Channel separation (not shown) was excellent, with any crosstalk buried beneath the noise floor in the L-R direction and rising above it in the R-L direction only above 1kHz. Even so, at 90dB of separation at 20kHz, this will not be an issue.

Fig.1 Musical Fidelity Tri-Vista, CD frequency response at -12dBFS into 100k ohms with (bottom) and without (top) de-emphasis (right channel dashed, 0.5dB/vertical div.).

Fig.2 Musical Fidelity Tri-Vista, SACD frequency response at -3dBFS (bottom above 40kHz), and 96kHz LPCM response at -12dBFS (top above 40kHz, bottom below 100Hz), into 100k ohms (right channel dashed, 1dB/vertical div.).

Fig.3 shows 1/3-octave spectra of the Tri-Vista's analog output taken while it decoded 16-bit CD data, 24-bit external PCM data, and DSD data representing a dithered 1kHz tone at -90dBFS. At high frequencies, the noise floor drops by 12dB when the PCM word length is increased, suggesting that the player's true resolution is around 18 bits, which is good performance, if not to the standard set by the Weiss Medea DAC we reviewed in March. As is always the case with SACD players, the medium's intrinsic rise in noise at high frequencies reduces resolution in the top octave for DSD playback. Though it is as good as 24-bit PCM in the midrange via the Tri-Vista, DSD's resolution is actually less than that of CD above 12kHz. However, It is fair to note that the subjective effect of this will in all likelihood be negligible.

Fig.3 Musical Fidelity Tri-Vista, 1/3-octave spectrum of dithered 1kHz tone at -90dBFS, with noise and spuriae. From top to bottom at 10kHz: 16-bit LPCM data, DSD data, 24-bit LPCM data (right channel dashed).

This difference in DSD's and 24-bit PCM's ultrasonic noise floors is dramatically shown in fig.4. Also note in figs. 3 and 4 that the Tri-Vista's resolution at low frequencies is very slightly disturbed by the presence of power-supply spuriae. The 120Hz full-wave-rectified component at 120Hz is almost 120dB down from full level, which is excellent. However, 60Hz and 180Hz components, which presumably result from magnetic coupling from the AC transformer, can also be seen. However, it is fair to note that these are still at -107dBFS, which will be, to all intents and purposes, inaudible.

Fig.4 Musical Fidelity Tri-Vista, 1/3-octave spectrum of digital black, with noise and spuriae, DSD data (top) and 24-bit LPCM data (bottom). (Right channel dashed.)

The Tri-Vista player uses separate digital circuitry for CD and SACD playback, but the linearity error (fig.5) is negligible for both media down to below -110dBFS: excellent performance. Correlating with the low noise floor and excellent linearity, the Tri-Vista's reproduction of an undithered 16-bit, 1kHz tone at -90.31dBFS was essentially perfect (fig.6), though the presence of the power-supply components noted earlier moves the waveform slightly away from the 0V line in this graph. Increasing the word length to 24 bits and using the Musical Fidelity's external data input gave an acceptable sinewave shape (fig.7).

Fig.5 Musical Fidelity Tri-Vista, departure from linearity, 16-bit data (top) below -100dBFS, DSD data (bottom). (2dB/vertical div., right channel dashed.)

Fig.6 Musical Fidelity Tri-Vista, waveform of undithered 1kHz sinewave at -90.31dBFS, 16-bit LPCM data.

Fig.7 Musical Fidelity Tri-Vista, waveform of undithered 1kHz sinewave at -90.31dBFS, 24-bit LPCM data.

Harmonic distortion was low, with only the second harmonic sticking its head above the -100dB line (fig.8). Even then, at -90dB (0.003%), this is not going to disturb anyone. Interestingly, harmonics higher than the fourth were apparent until the player had thoroughly warmed up and the feet turned blue. Intermodulation distortion was similarly low (fig.9), with the 1kHz difference component resulting from maximum-level 19kHz and 20kHz tones lying at -103dB, even with a punishing 600 ohm load. (It can't actually be seen in this graph; I investigated the matter using the higher-resolution A/D converter used by the Miller Audio Research analyzer.)

Fig.8 Musical Fidelity Tri-Vista, spectrum of 1kHz sinewave, DC-10kHz, at 0dBFS into 8k ohms (linear frequency scale).

Fig.9 Musical Fidelity Tri-Vista, HF intermodulation spectrum, DC-24kHz, 19+20kHz at 0dBFS into 600 ohms (linear frequency scale).

Using the Miller Jitter Analyzer to examine the Tri-Vista's rejection of word-clock jitter gave a very low figure of 177.4 picoseconds for CD playback, and an even lower 169ps when the player was fed external 16-bit data from my PC via a TosLink connection. This is excellent performance. The spectrum of the Tri-Vista's analog output for 3.5kHz on either side of the peak that represents the analytical 11.025kHz tone is shown in fig.10. Data-related jitter (red numeric markers) is exceedingly low. The main contributors to the measured jitter figure are a pair of sidebands at ±15.6Hz (purple "1"), at ±120Hz (blue "2"), and at the subcode-related frequency of ±150Hz (gray "3"), but these are all are very low in absolute terms. The slightly lower figure for external data, by the way, resulted from the absence of the ±150Hz sidebands.

Fig.10 Musical Fidelity Tri-Vista, high-resolution jitter spectrum of analog output signal, 16-bit LPCM data (11.025kHz at -6dBFS sampled at 44.1kHz with LSB toggled at 229Hz). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz. (Grayed-out trace is similar spectrum for DSD-encoded 11.025kHz tone.)

I don't have an SACD carrying the test signal for the Miller Jitter Analyzer. However, the Sony test SACD does have an 11.025kHz tone, and the grayed-out traces in fig.10 show the spectral analysis of the SACD player's output while it played back that tone. The noise floor is overall about 2dB lower than for CD playback—I have extended the range plotted in the graph by 2dB to make that clear—and while there are no data-related spuriae to be seen, there were sidebands present at ±15.6Hz, ±75Hz, and ±180Hz. However, these were all extremely low in level and give me reason to doubt Musical Fidelity's own measured figure of 133ps for SACD playback jitter using the Miller Analyzer and an appropriate test SACD.

Summing up, the Musical Fidelity Tri-Vista SACD player offers excellent measured performance, with no shortfall that could be laid at the feet of its tubes.—John Atkinson

Musical Fidelity
US distributor: Kevro International
902 McKay Road, Suite 4
Pickering, Ontario L1W 3X8, Canada
(905) 428-2800