The grayed-out spectrum shows the performance of the Meridian 508.24, which, at 144.2ps, is similar to the Elgar's. But note the rise in noise either side of the Meridian's central peak, as well as the relatively high-level, low-frequency sidebands. The Meridian, however, has slightly lower levels of data-related jitter, which is to be expected given the fact that the Elgar has to receive its input via an inherently jittery AES/EBU link.

Inserting the 972 in the chain but keeping the sample rate to 44.1kHz dropped the Elgar's jitter even lower, to just 132ps peak-peak, which is superb performance. However, increasing the sample rate to 96kHz increased the jitter level to 150ps (fig.11), with the dual AES link a little worse than a single, double-speed link. This surprised me—all things being equal, you'd expect the single link to be more prone to jitter, because of the doubled bandwidth requirement. More important, some discrete components appeared just below the frequency of the central peak, though these are all at a very low level (below -120dBFS).

Fig.11 dCS Elgar/dCS 972, high-resolution jitter spectrum of analog output signal, Meridian 500/Canare, 44.1kHz upsampled to 96kHz, single AES/EBU link (11kHz at -6dBFS with LSB toggled at 229Hz). Center frequency of trace, 11kHz; frequency range, ±3.5kHz.

Increasing the 972's output sample rate to 192kHz with a dual AES/EBU link gave a very similar jitter result, shown as the grayed-out trace in fig.12. A (The foreground trace in this graph is the spectrum of the Elgar's analog output taken with the 972 in the chain but leaving the sample rate unchanged at 44.1kHz.) Again, the spurious components with teh updampled data can be seen.

Fig.12 dCS Elgar/dCS 972, high-resolution jitter spectrum of analog output signal, Meridian 500/Canare, 44.1kHz sample frequency (11kHz at -6dBFS with LSB toggled at 229Hz). Center frequency of trace, 11kHz; frequency range, ±3.5kHz. (Grayed-out trace is with the 972 upsampling to 192kHz with dual AES/EBU links.)

None of these measurements indicates a clear reason why Jonathan, I, and other Stereophile writers should find upsampled, word-length-increased 24/96 digital data to sound so much better than the CD Standard's 16/44.1. I found it made CDs sound better even when driving the inexpensive little Musical Fidelity X-24K DAC at 96kHz. So color me perplexed by the 972.

But at least it does no harm. And the functionality it adds to a digital recording-and-mastering setup makes it a bargain for the pro-audio engineer concerned with maximizing digital signal quality.—John Atkinson