Sumo Theorem D/A processor Follow-Up

Follow-Up, from April 1993, Vol.16 No.4

Since my review of Sumo's $799 Theorem D/A converter in October 1992, the company has incorporated a jitter-reduction circuit in this popular processor. My assessment at the time was very favorable: I called the Theorem the "best-sounding converter under $1000." When I heard that Sumo had improved the Theorem's sonic performance, I was eager to audition it again. In addition to finding out how good the newer version was, it would be possible to hear the effects of lower word-clock jitter—if indeed the revision lowered jitter as claimed.

Thanks to the Meitner LIM Detector described in January's "The Jitter Game" (p.114), we can now quantify how much jitter exists in digital processors and CD players. It is thus possible—for the first time—to independently verify or refute manufacturers' claims of low jitter and the efficacy of "jitter-reduction circuits."

I auditioned the updated Theorem (in the identical reference playback system—see my original review) to assess the musical performance of current production. I also measured its jitter performance and compared it to the original sample.

I don't know if the newer version sounded that much better than the first or if I had just forgotten how good the Theorem was, but its performance was superb, and far above what one would expect at this price level. The Theorem's best quality was a beautiful transparency, clarity, and excellent sense of space. The Sumo processor also had another attribute important to musical reproduction that is uncharacteristic of low-priced digital: the ability to present instrumental images as individual entities. In these areas, the new Theorem was competitive with units costing five times as much. Moreover, the Theorem's treble was very pure, clean, and free from hash.

The Theorem's main weakness was its lightweight presentation, the bass lacking enough weight, power, or authority. This characteristic extended to the lower mids, making some instrumental textures slightly lean and lacking meat. Another liability was the Theorem's somewhat etched treble. Although highly detailed, the presentation tended to be on the analytical side, a factor that could make the Theorem fatiguing.

For a more complete description of the revised Theorem's musical characteristics, see my review of the $599 Cobalt 307 DAC (to which the Theorem is compared) elsewhere in this issue.

On to the jitter issue. The second sample indeed had much lower word-clock jitter than the first, which had a fairly high level of 1.25 nanoseconds at most input levels and a whopping 25.9ns with certain input signals. The latter figure was apparently a result of Phase Locked Loop (PLL) instability in the receiver circuit.

The word-clock jitter of the new Theorem measured a much lower 124ps (picoseconds) with no input signal, rising to 371ps when driven by a full-scale, 1kHz sinewave, and peaking at 563ps (worst case) when driven by a –70dBFS, 1kHz sinewave. These figures are significantly lower than on the first unit measured, and verify Sumo's claims of lower jitter. Not only was the RMS jitter level lower, but the jitter spectrum was much cleaner. The number and amplitude of periodic jitter components (jitter with a specific frequency, seen as vertical lines in the trace) are vastly reduced, shown in the comparison between figs.1 (first sample) and 2 (second sample). The presence or absence of these periodic components is likely of more sonic importance than the overall jitter level. Although not extraordinarily low, the Theorem's jitter was within reasonable levels, particularly for a moderately priced processor.


Fig.1 Sumo Theorem (revised), jitter spectrum, 1kHz sinewave at 0dBFS.


Fig.2 Sumo Theorem (revised), jitter spectrum, 1kHz sinewave at –70dBFS.

Incidentally, the lower jitter was realized with a simple modification to the PLL associated with the input receiver. Owners of the first production can have their units upgraded by Sumo at no charge, provided the owner pays the freight in both directions.

Sumo's revision of the Theorem strengthens its standing as the best D/A converter under $1000. It is not only a good-sounding product, but its $799 price establishes it as the bargain of affordable digital processors. Its lean tonal balance and tendency toward an analytical treble may not suit all listeners, however, making an audition mandatory before committing to a purchase.—Robert Harley

Sumo Products Group
Agoura Hills, CA 91301
Company no longer in existence (2019)

Ortofan's picture

... "mediocre" - not even damning with faint praise this time.
$800 would have been much better spent on a Sony CDP-X339ES or, for something different, a Pioneer Elite PD-65, or even a JVC XL-Z1050TN.