DACs & ADCs Simplified
DAC = Digital to Analogue Converter [or Digital to Analog Converter]
ADC = Analogue to Digital Converter [or Analog to Digital Converter]
Converters are the point at which two domains meet. Whichever direction the conversion takes the analogue and digital realms converge at these strategic parts. Whatever is lost prior to conversion is largely unrecoverable. Any subsequent stages may be constrained by the failings in either domain or the converter itself.
Audio enthusiasts may be enthralled by the prospect of 24 bit audio. In other worlds 10 or 12 bit DACs define the levels of video resolution for TV, graphics and DVD. UK Nicam (stereo) TV is often considered to be equivalent to 14 bit audio resolution.
On the face of it, the number of bits defines the number of discrete steps that exist between minimum and maximum signal. A high resolution requires a high number of bits and theoretically offers the ultimate performance.
Reality is usually a little more perverse than straightforward theory. The world is apparently digital; at least from a marketing perspective. The bold claims of great resolution may be real in the digital domain. All that this means is that the number coming from the ADC or going to the DAC has the claimed number of digits.
Most digital descriptions of a continuous event (sound, picture or even temperature measurement) are based on values of size with defined intervals of time. The time structure is based on a clock, which in the electronic world is often ticking several million times per second. Any variation in this clock would be like a pendulum clock with an erratic swing.
An erratic pendulum swing (or jitter) causes some time intervals to be too long and others to be too short. This applies even if the average time is right. The variation in time is creates an error in the representation of events, despite the resolution available.
Although we appreciate that electronics requires a power source, it is often relegated to a necessary but uninvolved part of the system. The truth makes this oversight rather crucial. Power is distributed around a product and is required by every chip on the circuit board. Any disturbance created by high speed electronic clocks or a conversion process tends to find its way onto the power supply lines.
Many of the devices powered from these common supply lines are relatively resilient to power disturbance and noise. Resilient they may be but immune they are not. Consequently, a small amount of the disturbance is carried from the supply into the signal chain.
Worse still, the artefacts of conversion are directly related to the signal and these contaminations are very hard to remove. It is easy for the supporters of digital technology to point to the shortcomings of analogue circuits. There is a requirement for some analogue on one side of either conversion process. The performance of such circuitry is of the utmost importance if the product is to benefit from high resolution digital internals.
Unfortunately, degradation is not confined to the analogue world. There are errors and issues in the digital domain too. Whether the signals are stored and later recovered or processed via a DSP (Digital Signal Processor) there is considerable potential for calculation errors. Data losses can readily occur in a continuous stream of data (streaming audio or even playing a CD). In many instances digital filters are used, with great benefits to simple signals but of varying success with real world complex dynamic signals.
There are occasions where the whole process of conversion and digital processing falls down. In these cases a return to pure analogue techniques is necessary. This does not mean that digital is bad. We would not have the progress in some areas without the evolution of digital techniques.
What has become apparent is that the application of good quality analogue design is becoming more scarce. The overall performance of a system or product is often defined by the electronic clock circuits, power supplies and precision analogue parts. The drive for higher resolution and speed has made the creation of genuine high performance circuits even more important.
A working testament to this comprehension is encapsulated in the Kinshaw DAC, originally designed in 1993. This audio converter was based around an 18 bit device at a time when 20 bit parts were emerging. Even now, in the face of alleged 24 bit systems the product acquits itself well. The reasons for this are based on the calibre of the components used, double regulated power supplies and optimised analogue design. The digital design uses high quality digital filter, short signal paths and separate digital power supplies.
Kinshaw DAC (1993) aptly described as sound by definition, electronics by design.