Brian's Brain

This post is a supplement to my article 'CAT5 tracks: Audio goes the distance, reliably and on time' in the July 7, 2005 issue of EDN.

AudioRail's Garth Wiebe provides a historical background on digital audio interconnections: “As a first step up from analog audio, digital audio is most commonly transmitted using the following point to point methods:

• S/PDIF (Sony/Philips Digital InterFace)/IEC 60958/AES3: Two channels of 16- to 24-bit audio are serialized into an NRZI data stream of subframes, frames, and blocks that you can easily lock a PLL onto to extract a clock from. S/PDIF, IEC 60950, and AES3 are, for all practical purposes, the same thing in protocol.
• Alesis ADAT ‘lightpipe’: eight channels of 16- to 24-bit audio are serialized into time-division-multiplexed serial NRZI data frames of 256 bits that you can easily lock a PLL onto to extract a clock from.
• TASCAM TDIF: This is an eight-channel digital audio connection.

“The S/PDIF / IEC 60958 / AES3 protocol is implemented in ‘75V coaxial digital audio,’ ‘AES/EBU,’ and ‘Optical S/PDIF.’ ‘75V coaxial digital audio’ is a simple, unbalanced, single-ended transmission line of 75V, usually using RG59 coaxial or equivalent impedance cable, minimally end- and sometimes also source-terminated resistively at 75V, with a signal swing of about 0.5 to 1V.It usually uses an RCA connector, and occasionally a BNC.

“’AES/EBU’ is a simple balanced, differential transmission line of 110V, again minimally end-terminated at 110V, with a signal swing of 2 to 7V.It specifies a standard audio XLR connector. Both ‘75V coaxial digital audio’ and ‘AES/EBU’ use isolation transformers in the interface circuits to allow for long cable runs without ground loops.

“’Optical S/PDIF’ uses the ‘Toslink’ plastic optical fiber, which is just ‘red-LED’ driver/detector technology, driving a cheap, thick plastic lightpipe fiber. It is generally good for on the order of about 30 feet, although that is a very rough number. The thick plastic fiber is highly “multimode” and so the light travels many different paths and gets very “fuzzy” at the transitions (my choice of words) and attenuates in optical signal strength very quickly as cable length increases.

“Alesis used the same Toslink plastic optical fiber transmitter and receiver components, with a more efficient means of packing the data into them. They originally used this to link their S-VHS style digital 8-track ADAT (Alesis Digital Audio Tape) recorders, such as the DA-88, but they made the interface available to the industry and it has caught on as a very popular interface format, because of its elegance and data efficiency. Eight channels of digital audio are packed twice as efficiently at twice the ‘wire’ data transition rate as optical S/PDIF.This comes at a penalty of being good for very roughly half the distance of optical S/PDIF. (We recommend limiting to 15 feet, although that is also a very rough number.)

“The ADAT sample rate is 48 kHz maximum. To get up to 96 kHz, Alesis calls out a method that they dubbed ‘SMUX’ to combine adjacent TDM time slots, so that you get four channels instead of eight out of each light pipe. To get up to 192 kHz, SMUX combines four adjacent time slots to get only two channels out of each lightpipe.

“TASCAM TDIF connects using a cable with 25-pin D-sub connectors (like a computer parallel port or old style comm. port cable, but with tighter electrical specifications). This had its origins in TASCAM’s 8-track DTRS (Digital Tape Recording System) recorders, which used Hi-8 videocassettes and competed with the Alesis machines that used S-VHS videocassettes.”

Continued with 'Audio Over CAT5: Other approaches, Part II'....