Cary Audio Design CAD-5500 "CD Processor" preamplifier
By Dick Olsher   •   November, 1990 A "CD processor," is how I distinctly heard Cary Audio's Dennis Had describe it. The venue was Stereophile's High End Hi-Fi Show in New York last April. Nothing really unusual in today's digital marketplace, I thought to myself, though a bit out of character for a company dedicated to vacuum-tube technology. But wait a minute. Dennis had described it as an analog CD processor. Analog!? Well, yes, the unit processes the analog signal from a CD player.

Why would anybody want to operate on the analog signal? I do not for a moment doubt that digital processors are essential high-end tools. Most of us have been amply educated as of late about digital data corruption and the ills of nonlinear DACs; the concept of better digital circuits making an audible sonic difference is now widely accepted. Even a quick listen to the recent processors from Theta Digital (the Pro and Pro Basic) was sufficient to profoundly change my opinion about the sonic potential of CDs. Retrieval of low-level detail, spatial resolution, and soundstage transparency were dramatically better than what I'd heard before from any CD player—even those super-priced Japanese players. The concept of analog processing left me cold, however. What could possibly be the matter with the analog signal? A great analog section internal to a processor such as the Theta, fed to a Class A line-level preamp or a passive preamp, would surely take care of business.

Not quite, according to Dennis, who claims that what's typically wrong with the analog signal from a CD player is RF contamination and the distortion products generated by it. He posits that RF-induced distortion in the preamp and/or amp is responsible for the bright, harsh, edgy sound of many CD players.

RF contamination
It is important to realize that the inside of a CD player is inundated with RF leakage energy. The culprit is not the DAC operating at 44.1kHz, 88.2kHz, 176.4kHz, or 348.8kHz, but rather the master clock for the digital bitstream, which normally operates in the range of 2.8–3.05MHz. This RF energy is real and potentially troublesome. The manufacturer is required to certify that his product complies with FCC regulations limiting the radiated RF from the product. The following is a quote from the Owner's Manual of a well-known manufacturer of digital processors: "This equipment has been tested and found to comply with the limits for a Class B digital device, pursuant to Part 15 of the FCC Rules. These limits are designed to provide reasonable protection against harmful interference in a residential installation. This equipment generates, uses, and can radiate radio frequency energy and, if not installed and used in accordance with the instructions, may cause harmful interference to radio communications. However, there is no guarantee that interference will not occur in a particular installation."

While the problem of radiated RF is well known to the point of having been regulated by the bureaucrats, little attention has been paid by designers to the RF energy that leaks out of the chassis along the interconnect cable. This RF garbage can literally piggyback the analog signal along the interconnect cable into the preamp or amp where it might overload the front end or induce intermodulation distortion products in the audio band. This is the problem that the CAD-5500 is said to address. The solution it offers has been dubbed "Reverse Phase Canceling" by Cary Audio, or RPC for short.

Reverse Phase Canceling
The input signal from either the CAD-5500's CD or Aux inputs is routed through two separate pathways. One pathway is full-spectrum through a purely resistive voltage-divider network. The other path is through a high-pass filter section which uses ceramic RF trimmer capacitors to pass ultrasonic and RF energy above 20kHz. This high-frequency signal, which contains the RF distortion, is fed to the grids of a 12AX7 dual-triode tube. One of the triode sections is used to drive the LEDs on the front panel, thus indicating the presence of ultrasonic and RF spuriae. The anode signal from the other triode section, which is 180$d out of phase with its input, is level-matched with the original signal then combined with it. The end result should be a steep rolloff of all frequencies above 20kHz, including the RF component.

Note that the action of the RPC circuit is similar to that of a high-order low-pass filter, the signal level said to be already down a robust 8dB at 23kHz. Some of you might be wondering by now, as I was, if a standard active LP filter could not be used to achieve similar rejection of RF garbage. Dennis Had maintains, and I'm inclined to agree, that his circuit is free from the objectionable sonic aberrations of traditional brick-wall filters. Figs.A and B, provided by Cary Audio, show "before and after" spectrum analyses of the noise in a typical CD player's output. Note that the marker in these pictures is at 20kHz. With the processor in the chain, there's a steep rejection of RF noise energy.

Fig.A Typical CD player, noise spectrum, 20Hz–200kHz (linear horizontal scale, 20kHz/div, 10dB/div. vertical scale).

Fig.B Cary CAD-5500, noise spectrum from fig.A after being processed by CAD-5500, 20Hz–200kHz (linear horizontal scale, 20kHz/div, 10dB/div. vertical scale).

Following the RPC section, the now–bandwidth-limited signal is routed through a volume control to a 7025 vacuum tube (a beefed-up 12AX7) which provides dual-stage line-level amplification with a nominal overall gain of 12dB. There's a version of the processor available with a phono input module (the 5500S for $1995) which provides an additional 46dB of gain for an overall gain of 58dB—more than adequate for any MM cartridge. For a mere $255 more, the processor may be had with a beautiful Penny & Giles pot, which is how my unit was equipped.

The output impedance is very high at over 11k ohms, which will make the CAD-5500 far less than ideal for driving long cables. Expect some treble rolloff with long interconnect runs, and perhaps even with quite short runs of highly capacitive cables. With both Cardas and Lindsay-Geyer interconnects, 20–25' long (footnote 1), the treble through the processor was consistently soft and lacking in air. That the extreme treble was closed-in did not bother me much—especially in view of the magical clarity and harmonic integrity with which the 5500 imbued the lower treble and upper mids.

The line-level stage is designed to clip softly above 2V RMS, thus gracefully compressing the extreme voltage peaks generated by some CD players and processors. If an input voltage of 1–2V is enough to drive any audio power amp into clipping, what is the utility of generating 5V or more, as some players do? Dennis Had believes that this feature is partly responsible for the overall sonic excellence of the unit. Because tubed preamps in general tend to clip softly, this bit of insight may help explain why CD players in many instances sound best with a tubed line-level stage. This brings up a related thought. The bandwidth of a tubed power amp is much more limited than that of a modern solid-state design, primarily because of limitations of the output transformer. Thus, it is likely that a tubed amp will be more immune to RF distortion and as a consequence would be judged more musical than a solid-state competitor when fed by digital program sources (footnote 2) If this is indeed the case, I would expect that the CAD-5500 would make a much more significant difference in a system which includes a solid-state amp and in which the CAD replaces a solid-state line-level stage.

The power supply is housed externally to the main chassis, so what you'll find inside is a single large glass-epoxy board stuffed with quality components such as precision metal-film resistors and Sidereal-brand capacitors. All of the RCA jacks are gold-plated Tiffanys with Teflon insulation. The Penny & Giles conductive-plastic pot was a pleasure to use, and, based on past experience, because I did not directly compare this sample with a standard CAD-5500, I would recommend this option. Overall, the unit appeared to be well-built, with looks and feel above the high-end norm at its price point.

Sonic impressions
There's no point beating around the bush. The Cary completely redefined my perceptions of the compact disc's present and future aspirations as a high-end medium. During its salad days in the mid-1980s, I perceived the CD primarily as a convenience medium. It was destined to be an enormous commercial success for the same reasons the cassette tape made it so big. Small size, low maintenance, no side changes, an absence of infrasonic garbage, no ticks or pops, were all powerful marketing attributes which, together with a massive propaganda campaign, left a popular impression that here was a technologically superior medium that obsoleted the vinyl disc. On the face of the technical evidence, in the form of conventional specifications, vinyl appeared routed. Déjà vu indeed, recalling the early transistor days when the acceptance of solid-state as "technical gospel" led the way to a wave of atrocious transistorized amplification in the '60s.

But, of course, many of us balked at the early CD players. The shrillness, brightness, and edginess of these machines drove me back to the comforts of vinyl. Early digital converts who had just given up on their direct-drive turntables and $25 cartridges maintained that Heaven must indeed be bright and edgy; the truth was being revealed at last. Their fallacious reasoning went as follows: The CD represents an accurate link to the digital data on the master tape. Therefore, if the disc sounds bright it is merely revealing the shortcomings of the recording. The end result was a general condemnation of the human element—rather than the technology. Digital proponents argued that recording engineers and producers just could not get it right. And because a few good CDs surfaced from time to time, opponents could not dismiss CD as potentially incapable of good sound. The frustrating thing was that these few good CDs were usually of minor musical value; something like "Albanian Yogurt Makers Chant Favorite Marxist Lullabies."

Several player generations later, many of the CDs that were previously "unlistenable" were being rendered palatable; it suddenly became obvious to many audiophiles that the player technology was far from mature. Players were being obsoleted at an alarming rate by better filtering schemes, or other technical wrinkles. Finally, with the arrival of the Theta Digital DS Pro Basic, and a little later the Theta DS Pro, CD sound reached a zenith in my system. The Pro Basic was nothing short of astonishing. Never before could I see so far into the soundstage, or localize image outlines with such 3-D focus, or retrieve low-level detail and hall information with such ease. Still, the music's textures were a bit artificial and electronic-sounding when compared to live music, and the highs were not so well integrated with the rest of the range. The treble seemed to have an existence of its own within the soundstage, extended beyond the normal spatial confines of an instrument's harmonic envelope. This tendency of the treble to billow or mushroom outside reasonable image outlines was spectacular because the effect was bigger than life. Very few audiophiles would object to a bit of soundstage Technicolor, but after prolonged listening I grew tired of the Basic's synthetic flavoring.

The Theta Pro improved on the Basic's performance so far as having a more relaxed and liquid delivery. Textures did not appear as mechanical as before. But the treble integration did not improve, and I was still not completely happy with the quality of the treble. There was some low-level hash and a tinge of brightness through the upper octaves that, while tolerable, detracted from the naturalness of the presentation.

This was the environment into which the Cary made its entrance. I was frankly dubious at first about the impact it could possibly have on what was, after all, an already excellent sound. In it went between the Theta and a pair of bridged Classé DR-8 amps, which were driving a pair of Apogee Stage ribbon speakers.

Holy Cow! What a giant step forward! Several impressions rapidly hit home. Instrumental outlines snapped into a cohesive and believable space within the soundstage across the entire tonal range. The treble became integrated with the rest of the harmonic envelope so that instruments appeared to breathe naturally and without exaggeration. It became easier to zoom into the soundstage and examine a particular spatial aspect—as if the soundstage were in fact a hologram. So strong was this impression that I was almost certain that had I stood up or moved laterally, that I could discern new spatial perspectives. The increase in treble purity was dramatic. The hash and brightness I mentioned earlier vanished. I could not believe that CD treble could sound this clean and natural. The extreme treble was slightly soft and closed-in, but this was more than offset by the purging of the lower treble's glare and electronic haze. Midrange textures were also improved, being more liquid, or more tubey if you will. Not mushy or soft, but closer to the balance of hard and soft afforded by live music.

---

Footnote 1: Featuring measured shunt capacitances of approximately 2300pF and 10,000pF, respectively, including connectors.—John Atkinson

Footnote 2: Food for thought considering that during a recent visit to Santa Fe, Sam Tellig noted that the true magic of the Meridian 203 Bitstream processor, discussed by him in his last two "Audio Anarchist" columns, only really became apparent when he used VTL 225W monoblocks to drive his preferred Spendor S-100 speakers.—John Atkinson