Digital recording and playback techniques were introduced to consumer electronics in 1981. Since then an ongoing debate has questioned the capacity of digital systems to reproduce the nuances and 'taste' of original recordings, capturing the higher-frequency signals and atmospherics which audiophile customers argue are essential to a full appreciation of the music and films they enjoy.
Of course there are many challenges for engineers developing digital audio systems. At the most basic level, engineers have faced the challenge of achieving quality inputs into the process: at the beginning of digital recording, sampling rates were significantly lower than they are today, and even now CD recordings only achieve a rate of 44.1KHz, compared with the much higher potential sampling rate achieved by 'pure' digital formats such as Direct Stream Digital (DSD).
'Digitally remastered' re-issues, although aiding the popularisation of CDs throughout the 80s and 90s, did little to address reproduction issues faced by digital systems. These issues resulted in phenomena such as 'pre-ringing': the light echo of sound heard before the first note is struck, caused by the design of the digital filter in the playback system.
Jitter is another phenomenon that became known to discerning listeners and professionals from the early days of digital reproduction. This occurs when a jitter in the clock used to time audio samples occurs randomly, or as a result of an instable voltage in the power supply. The listener hears these faults as background noise, in the case of a random sample jitter, or as distortion if the jitter is correlated with the audio signal. Figure 1 shows how a smooth analog signal may be sampled either too late or too early in the 'clocking' process, and the steps that can be taken to correct this fault:
Fig 1: Converting a digitised signal back to analog – higher resolution (more quantisation levels) and/or faster sampling reduce the quantisation error. Clock jitter introduces additional errors.
The increasing popularity of MP3 and MP4 formats has compounded the issues of low sample rates in the original source, and 'jitter' caused by faulty clocking with the data compression techniques used in MP3 and MP4 formats. In other words, further data from the original signal was lost through the MP3/MP4 compression techniques and their requirement to grab 'blocks' of data and discard anything considered inessential to the signal in the interests of achieving a tighter file format.
In addition to the challenges caused by jitter, low sampling rates and compression technologies, earlier digital reproduction systems suffered from poor signal to noise ratio reduction in the sampling process and a high percentage THD, or Total Harmonic Distortion, in the reproduction process.
The introduction of the Direct Stream Digital (DSD) recording format in the early 2000s was designed to answer the problem of lower sampling rates from original sources. DSD delivered a super-high rate of sampling from the original signal, but still did not address what audiophiles identify as their key problems in digital reproduction, including the aforementioned pre-ringing and jitter, as well as post-ringing, phase latency and transition band roll-off. Thus the Super Audio Compact Disc (SACD) format delivers outstanding original source quality, but does not resolve the issues faced in the reproduction of that quality by a digital system.
Because of this, the SACD format never really took off, and today we find that only some 4,500-5,000 recordings have been issued in this format, compared with i-Tunes sales of MP3 format recordings reaching over 5 billion during 2008. At one point, there was a risk that owners of the highest-quality systems would begin a slow drift back to analog sources, revisiting their vinyl collections in a search for supreme richness and depth in their audio experience.
Mixed-signal units: the key to quality
The answer for the audio systems industry lay in improving the quality of the electronics used to reproduce analog sounds in a digital environment, including analog to digital converters (ADCs) and digital to analog converters (DACs). However, this is not a straightforward process and several factors have to be considered:
1. Audio filtering in the DAC process
One of the major problems first experienced by users of CDs in the early 1980s, through to digital systems like MP3 today is the digital filtering inherent to DACs. When reconstructing the signal, the DAC introduces an element of error and damages the original signal.
Arguably the greatest and most disruptive of these is pre-ringing, hearing an echo of sound events before they occur, a fundamental issue with FIR filters found in DACs. This occurrence before the event is so unnatural that listeners are highly sensitive to it – and this has been the defining problem of digital playback. Conventional DAC filter packages have previously focused on controlling issues associated with frequency responses, neglecting time domain issues.
Pre-ringing can be addressed through the use of minimum-phase filters alone, however if used in isolation these can lead to increased post-ringing (unnatural sustain of a sound event) and distortion caused by delay to a group of signals. While humans are much less sensitive to post-ringing as sustain and echo are naturally occurring events, clearly the objective is still to minimise any distortion.
By providing a wide range of filters and by using them together this unwanted distortion can be overcome. Linear filters, non-half band filters and minimum-phase filters can all be used. And because manufacturers and end users can vary the impact of these filters it is possible to achieve optimum reproduction quality, tailoring the mix of filters employed to suit the nature of the source signals and the style of audio being listened to.
Fig 2: A range of filters can be used to reduce the effect of pre ringing and other digital distortion
2. Dynamic Element Matching
At Wolfson the wide range of filters are combined with a sophisticated Dynamic Element Matching (DEM) process that provides further processing to the multi-bit signal to reduce in-band noise and distortion to the lowest possible levels.
The multi-bit signal is decomposed into a number of individually Delta-Sigma modulated signals, with these signals then being brought back together following reductions in in-band noise and distortion to produce the output signal. A multi-channel approach can be adopted by the DEM process to ensure maximum clarity and fidelity in each separate part of the signal following the Delta-Sigma modulation process, which in turn produces the highest possible quality of output signal for reproduction.
By adopting a multi-channel approach to DEM it is possible to ensure that there will be significantly fewer errors in reproduction at low frequencies. Moreover, linearity of the signal is improved because each element of the original DAC signal is present at an equivalent frequency at both the input and output stages of the process for maximum fidelity.
3. The Sigma-Delta Architecture
The importance of a high-quality Sigma-Delta architecture to the overall performance of a DAC unit cannot be overemphasised.
Fig 3: Audio DAC system
The Sigma-Delta architecture is responsible for receiving incoming digital signals and monitoring the outgoing pulse, creating an error signal based on the difference between the binary signal coming in and the pulse train going out. The Sigma unit then works to add up the results of the error signals supplied by Delta, supplying this to the low-pass filter, which then makes minute adjustments to the analog signal to compensate for the differences between the binary signal and the pulse train, in effect ensuring enhanced fidelity and greater purity of sound as an end result.
The first DAC systems to include Sigma-Delta modulators were primarily single-bit solutions, however it soon became apparent that this provided very coarse quantisation for an audio signal, with a great deal of unwanted noise generated that had to be filtered out after the DAC process. Multi-bit systems followed, with 16-24 bit signals being fed into an interpolation filter, then a multi-level Sigma-Delta modulator that improved out-of-band noise and signal images.
For Wolfson the key is to use the most sophisticated multi-level Delta-Sigma architecture available, deployed in conjunction with sophisticated DEM techniques and industry-leading signal filtering technologies to deliver a superior sound experience.
ADCs & DACs – The core of the component chain
Although DSD can now deliver the highest quality source signal, every part in the component chain is key to delivering the highest possible quality to audiophiles. From power cables and input devices to speaker cabinets and control units, every element needs to be of the highest possible quality at a competitive price.
And when it comes to audiophiles looking to make a significant investment in an audio system, the most important thing is ultra-high fidelity to the original listening experience ” which means that every element of your reproduction component chain has to be the best there is. At the heart of this chain are the ADCs and DACs.
While modern techniques and design have greatly improved the performance of ADCs and DACs, there will always be pressure to continue improving sound quality. This is especially true at the high end of the audio market where Wolfson continues to be committed to meeting the challenge of delivering high quality audio in a digital age.
Author profile: John Crawford is High Performance Audio manager at Wolfson Microelectronics