The Deficiencies of S/PDIF As A
Digital Transmission Method
Article by Chris Daly
Previously published in Electronics World,
SPDIF (Sony Philips Digital Interface Format) and its cousin AES/EBU
(Audio Engineering Society/European Broadcast Union) Digital transfer standards are both complex multiplexing standards that rely on a clock signal
inherent with transfer to then input to a digital receiver which then extracts the multiplex and reassembles the necessary digital words and clock for input to an oversampling filter, hopefully with minimal time increment skew a term also called
The digital words that relate directly to the recovery of audio from the
processor to the oversampling chip in 80% of players are: DATA, BITCLOCK and LRCK (Left
Right Clock) Conventionally in 1 box players these codes are just wordlengths that interface with that players oversampling chip, however as this article will attempt to explain these codes can be exploited individually to provide enhancement over SPDIF and the author considers AES/EBU. The methods proposed(1) suggest for the use of any outboard
converter where SPDIF and AES/EBU is currently used as a subcode multiplex digital transfer that SPDIF and AES/EBU must be
down rated in their expectation to provide digital transfer at audiophile quality.
Rather this article suggests to transfer digital information without jitter interference the wordlengths that directly relate to reconstruction at the oversampling filter known as BITCLOCK
DATA and LRCK or for that matter subsequent and different manufacturers
labeling of word length protocol must be kept inherent however should be interpolated to preserve their correct timing sequence by the use of D type flip flops clocked at at least 384fs namely 16.9344
MHz upon transmission and reception.
As a listener you may have heard examples of jitter , the recorded material typically becomes fatiguing with less than acceptable stereo imaging and to a well trained ear is unacceptable.
Why has the CD industry gone down this path of Digital transfer that causes
jitter. CD was born as a consumer interface(3) and as such, as far as manufacturers were concerned they were providing a perfectly acceptable medium for people to enjoy music with. As an adjunct to standard 1 box players a digital transfer output namely SPDIF was provided so that consumers may enjoy some choice of how their Digital to Analogue reproduction was achieved.
An entire industry was then developed specializing in Digital to Analogue conversion however whilst these
converters provided some improvement over their 1 box counterparts they also effected digital transfer mostly with SPDIF and some AES/EBU suffering in varying degrees from time skew
i.e. jitter. This caused reassessment from many audiophiles whether individual components were really worthy of audiophile status, and from other parts of the industry looking carefully at clock regeneration convinced that better results were indeed achievable
Other firms have looked at clock feedback methods and whilst offering some merit if used with interpolation fail to tackle the real issue of
word lengths being transferred without loss of integrity and trying to address clock control from a medium already
possessing jitter (SPDIF).
(1) All CD players use an integrated circuit that extracts and processes the Eight to Fourteen Modulation signal with one of its many outputs being SPDIF Multiplex subcodes as
marvelous as they are, whilst offering ease of connection for components sadly cannot reassemble or transfer digital information without jitter being integral with that transfer. The
convenience factor has intruded appreciably upon audiophiles needs.
Also as an output from this same IC that handles SPDIF are three signals namely BCK
(Bitclock) DATA and LRCK (Left
Right Clock) which interface normally to the players oversampling filter. The methods outlined below suggest these three codes should now be used with interpolation on transmission and reception to effect digital transfer to audiophile standards.
In place of SPDIF and AES/EBU this article suggests the use of fast flip-flops such as 74AC74 that can be arranged to interpolate DATA BITCLOCK and LRCK with control being achieved with a clock reference at 384fs (16.9344
MHz) or higher upon transmission with suitable RF terminations to
minimize RFI (CD player manufacturers responsibility) and enterprising Digital to Analogue outboard
manufacturers to provide similar flip flop interface in their converters that can accept DATA,
BCK and LRCK clocked ideally by the oversampling filters
separate clock reference i.e. the use of XTi and XTo with CKO buffered output providing Clock inputs to all flip flop interfaces. The Q or inverted outputs (as such selectable) are then input to DATA BCK and LRCK inputs on an oversampling filter.
A Digital receiver is no longer required to convert SPDIF subcode which should save outboard digital to analogue manufacturers considerable expense, and likewise CD manufacturers would no longer provide an SPDIF or AES/EBU connection, unless it was deemed necessary for convenience similar to Toslink now regarded by most audiophiles as sub standard interfacing.
Further integrity could be achieved in this authors opinion with Balanced interface transfer, however the complexity would increase, which may introduce the possibility of jitter. Also the tying of the players XRST connection to all resets of flipflops can offer advantage, however necessitates the use of a fourth connector.
Why is this method of sending individual protocol signals superior to SPDIF? When transferring digital data , D type Flip Flops being directly controlled by a clock reference offer far greater integrity of a word length being transferred. The clock input to the flip flop is arranged to be running much faster than the
word length and as a result offers that time advantage to its Data input . The integrity of the original
word length is maintained with this method, whereas SPDIF and AES/EBU convert these
word lengths into a complex subcode language that then undergoes eventual conversion back to individual
Although data sheets would suggest that the information is not corrupted in this transfer, listening sessions suggest otherwise. Typically bass frequencies are most effected,
coloring the performance by causing bass frequencies to sound artificial and stereo images to be quite wrong. Having listened to one method over the other it is painfully obvious the deficiencies SPDIF
The information BCK LRCK and DATA in limbo as such between the CD player and outboard
converter remain with precision integrity awaiting to be sampled on the next respective
word length clock pulse, and if suitably buffered and terminated to the transmission cables
inherent impedance remains without jitter or appreciable RFI, offering far superior transfer of digital data. effectively in the authors opinion eliminating jitter in the digital interface.
For this article to create any change, you the audiophile should try the
merits of one system over the other, but ideally to effect some change for everyone,
CD manufacturers should offer this method as an audiophile alternative, building into their machines the aforementioned flip flops and provide suitable buffering to a lower impedance that would eliminate the emission of RFI when terminated at the characteristic impedance.
CD Player manufacturers upon discussion with outboard converter
manufacturers should standardize the connector ensuring it meets performance. for impedance
purposes. Likewise manufacturers should provide quality parts that meet audiophile standards.
My thanks to Jamie Hayes for listening sessions.
(1) C. Daly CD Jitter Bug Electronics World June 1997 Pg 517
(2) Listening sessions Pioneer 801 ,Audio Synthesis DSM SPDIF transfer vs
Pioneer 701 Audio Synthesis DSM connected with 3 wire transfer BCK DATA LRCK
J Hayes / C.Daly March 2001, Hobart Tasmania
(3) Ben Duncan HFN/RR Supertuning CD Nov, Dec 87, Jan 88 and Mar 89
(4) Fourre, R., What is Jitter Stereophile October