California Audio Labs Sigma D/A processor

Originally published October 1, 1992

Though their first CD player featured a vacuum-tube output stage, California Audio Labs is recently known for making good-sounding, moderately priced solid-state CD players, like the Icon Mk.II that Jack English reviewed in July 1992 (Vol.15 No.7). The Sigma, a $695 tubed D/A converter, furthers their reputation in both areas.

The Sigma is a straightforward product: no front-panel switches, buttons, or controls. A single LED on the 8¾"-wide front panel indicates that the unit is locked to a digital source. The rear panel has a pair of RCA output jacks, digital inputs (one coax, one TosLink), an on/off switch, and an AC jack.

How can anyone make and sell a tubed D/A converter for $695 retail? Look inside the Sigma and you'll find out. The unit is a model of simplicity and low parts count, made possible in part by a new chipset from Crystal Semiconductor in Austin, Texas.

The input receiver is the Crystal CS8412, a device that boasts much lower jitter (up to 30 times lower) than the ubiquitous Yamaha YM3623 (in a standard implementation). The CS8412 is fast becoming the standard in high-end converters. The rest of the digital section is composed of the Crystal CS4328, a chip that combines digital filtering, noise shaping, a Delta-Sigma DAC, output driver, and analog output filtering in a single package. One other chip, a common TTL device, finishes off the digital section.

Most digital processors have at least a dozen support chips—called "glue logic"—that make the whole thing work. The Sigma has the fewest ICs (a total of five, including the muting timer) of any processor I've seen.

The left- and right-channel DAC outputs drive a discrete FET circuit that has a 12AX7 vacuum tube integrated into the FET gain stage. The tube, split between left and right channels, is in the overall feedback loop and functions as a voltage follower. The circuit is direct-coupled, with a 741 op-amp used as a DC servo for each channel. The relatively high output impedance of this design requires that the input impedance of whatever component the Sigma drives must be at least 15k ohms. Use with a passive level control is therefore discouraged. Interestingly, CAL chose not to use the CS4328's on-board MOSFET output driver (which can deliver 2V output into 600 ohms) to directly drive the final analog output, instead replacing it with their discrete stage. Note that no current-to-voltage (I/V) converter is necessary with Delta-Sigma DACs.

De-emphasis is passive, switched in with a relay. All capacitors in the analog stage are high-quality Wima polypropylene or polystyrene types. A relay disconnects the output when the unit isn't locked to a digital source, and a mute timing circuit prevents the Sigma from putting out audio until the tube is warmed up.

In the power-supply section, a single large transformer feeds six regulated stages plus the high-voltage (100V) plate supply for the tube. Separate regulation is used on the +12V tube heater supply, digital section supply, and ±15V analog rails.

Build quality, though not lavish, is first-rate. The board is well laid out, parts quality is high, and the chassis is very sturdy. The Sigma was made more economical by virtually eliminating point-to-point wiring; the digital input and analog output jacks are pcb-mounted, as is the transformer. The only point-to-point wiring in the unit is the two wires to the power on/off switch. This technique significantly reduces labor costs.

A few other notes: By the time you read this, CAL will offer an AT&T ST-type optical input on the Sigma. Current owners can have their unit retrofitted at the factory. A price was not available at press time, but expect it to cost between $250 and $350. CAL's new transport, the Delta, will also be offered with ST-type optical interface.

The Sigma had the most distinctive sonic signature of the four processors I review in this issue, a signature one would expect from a tubed product: smooth, laid-back, and relaxed. Specifically, the Sigma leaned toward conveying a sense of ease and musicality rather than revealing the music's every nuance. Its presentation lacked the resolution of detail heard from the other processors, yet the Sigma had a remarkable warmth and smoothness that was uncharacteristic of low-cost digital processors.

The mids in particular were warm and round, in contrast to the Sumo Theorem's slightly leaner, thinner portrayal. I'd even describe the Sigma's midrange as lush, an unusual characteristic for such an inexpensive processor. There was a slightly fat character to the mids that conveyed a sense of body and roundness. Saxophone had a very pleasant bloom and smoothness, conveying the instrument's warmth and richness. Many processors seem to emphasize a sax's upper harmonics, adding glare and making the instrument sound too thin. The Sigma's midrange warmth contributed greatly to the processor's musicality.

Similarly, the bass was round and fat—overly so, in my opinion. Low frequencies were somewhat amorphous and slow rather than fast and tight. This characteristic reduced the music's sense of pace and rhythm, imparting a slight "slowness" and diminished feeling of tension to the music. On "Little Wing," from Stevie Ray Vaughan's The Sky is Crying (Epic EK 47380), for example, the music lacked the cohesion and drive heard from the Theorem. Overall, the Sigma had the least sense of pace and rhythm of the group.

The Sigma was good in the top octaves, but slightly less smooth and sweet than the lushness heard in the mids. The treble was nicely portrayed, cymbals appearing slightly behind the loudspeakers rather than forced on the listener. There was, however, a trace of grain in the upper mids and lower treble that added a slight coarseness to instrumental textures, massed strings in particular. Although the Sigma had less grain than the PS Audio Digital Link II and Forté DAC 50, it lacked the Sumo Theorem's textural smoothness. Listen to the violins in the Dutoit/Orchestre Symphonique de Montréal's recording of Holst's The Planets (London 417 553-2), for example. While the Sigma presented a richer, fuller string sound, it added a slight roughness to the texture not heard from the Theorem. I must stress that this characteristic was by no means hashy or metallic; instead, there was just a fine layer of grain overlaying midrange textures. This criticism is minor; I don't consider it a significant liability.

In soundstaging, I rate the Sigma second (behind the Theorem) of the four processors in this survey. There was a good sense of space and depth, with a better-than-average ability to differentiate individual instruments from the whole. The music was thrown slightly behind rather than before the loudspeakers. Moreover, the Sigma had noticeably better resolution of spatial information than the Digital Link II, and particularly the Forté DAC 50.

One CD that is particularly revealing of a processor's soundstaging ability is Mike Garson's excellent The Oxnard Sessions, Volume One (Reference Recordings RR-37CD). Through the best converters, there is a palpable sense of disparate instruments hanging in three-dimensional space, surrounded by the recorded acoustic. Through the Sigma, there was a nice sense of depth and distance, but the Theorem had a greater impression of transparency and air.

One aspect of the Sigma's presentation that may not suit all listeners was its somewhat low powers of resolution. The Sigma was the antithesis of analytical, "ruthlessly revealing," or etched. Instead, detail was presented more subtly. This could reduce the sense of life in some music, making the Sigma less engaging. For example, the percussion work on the CD Mokave (AudioQuest AQ-CD1006) lost some of its edge, vitality, and rhythmic intensity through the Sigma. There was just less information in the signal, a characteristic that could make the Sigma somewhat bland.

Putting it all in perspective
After getting to know the CAL processor, I auditioned them in relation to the $1495 Bitwise Musik System Zero (favorably reviewed last month) and $399 Audio Alchemy Digital Decoding Engine.

Starting with the DDE, it was no contest; the Sigma was superior to the DDE, particularly in the treble. Although the DDE had good clarity and resolution—greater resolution than the Digital Link II—its hashy treble was a significant liability.

Although the CAL Sigma was the most colored processor of the four processors I review this month—it departed the most from neutrality—I found it musically engaging and unfatiguing. It was easy to listen to for long periods, and lacked the unpleasant stridency and glare so often heard from inexpensive digital playback. "Warm and friendly" best describes the Sigma. Its flaws—soft bass and lack of detail—tended to be sins of omission rather than of commission. Recommended more for classical music—particularly choral and chamber music—than rock and jazz.

To sum up, the Sigma provides a presentation that avoids the common pitfalls of low-priced digital: hardness, glare, and rapid fatigue. Its primary flaws—reduced resolution of detail and a slow, fat bass—were a greater liability with rock and jazz than with classical, particularly choral works and chamber music. Despite these tradeoffs, I found the Sigma very musical and enjoyable. Given the Sigma's distinctive sonic signature, prospective purchasers are advised to audition the unit for themselves. I suspect that some will greatly value the unit's remarkable ease and musicality.


California Audio Labs
Company no longer in existence (2020)

Glotz's picture

Today $200 destroys it... err, we've come a long way in 30 years.

Most of entry level audio was chalky and lacking resolve. Can't fault CAL for mating it with tubes.

Robin Landseadel's picture

The Topping E30, at $130, destroys it.

Glotz's picture

And the Schiit Modius destroys the Topping E30.. for $200.

PeterPani's picture

When playing CD on my 1984-model Sony CDP-101, comparing it with several audiophile DAC's I purchased since - I am not sure what sounds better. Mostly everything on digital sounds the same. Every song is a repetition of digital characteristics fatiguing my ear canals until I switch back to analog.

volvic's picture

I think the weakness with digital or 16 bit/44.1 isn't the disc itself but the machines. I sound like a broken record, but I was hesitant in 2007 to move on to computer audio, but when I heard how much better it was and not by a small margin, but by a substantial margin, there was no going back. The drawbacks are the poor sorting, metadata and the occasional firewalls preventing certain software from working, but the sound is truly revelatory. Surprised your CDP 101 is working, mine long stopped functioning. Still have two CDP 111s that struggle. Most likely the servo is the culprit.

Ortofan's picture

... Sony CDP-X339ES, a complete CD player with superior D/A performance.