Hi-Fi digital audio network

This page is about the evolution of Hi-Fi systems from purely analogue set-ups to completely digital systems using only digital form for storage. The point where the digital information is converted into an analogue signal is pushed as far as possible using modern technologies:


Analogue sources

For a long time audio sources were only available in analogue form on a material support, as vinyl long play, radio waves, or all kinds of tapes. The audio information stored as an 'analogon' to the audible pressure signal on the material was retrieved in a specialised device by converting it back to an 'analogon' in electrical form, amplifying it and making it audible again through the speakers. Sources are in general a turntable, a tape deck or a a radio tuner.


Digital sources with individual DAC

With the advent of digital sources, the approach did not change that much for several reasons. First, users added new devices to their audio systems that were now able to use digitally represented audio signals and convert it into an analogue signal using its own Digital-Analog-Converter (DAC). This approach allows to gradually upgrade the system, by adding A Compact Disk (CD) player, a Digital Audio Tape (DAT) player, or a Digital Audio Broadcast (DAB) tuner. Most importantley, however, the paradigm did not have to change, i.e. digital systems were merely used to emulate and imitate analogue systems; their full possibilities were not exploited.


Digital sources and centralised DAC

As a simple step towards fully digital systems, the DAC can be centralised into a pre-amplifier which is fed by several devices capable of reading digital information. The resulting system is still hybrid as the central stereo unit is half digital, half analogue, and it does not eliminate separate reading devices such as CD players.


Decentralised DAC, amplifier and speakers

The boundary between the digital and the analogue world can be moved even further away from the center, towards the speakers. The fully digital music system resides as center piece on a system bus such as Firewire. The peripheral intelligent speakers are connected to this system bus and integrate a power amplifier per speaker drive. The follow setup is close to a classical approach using two speakers, though integrated and active.

The central unit is in essence an embedded computing system, using only solid state memory devices (no moving parts), possibly running the GNU/Linux operating system with Real-Time enhancement, and possibly equipped with a touch screen. It can read from any digital source, radio, internet, CDs, DVDs, USB memory sticks, etc, and can stream to and from a Network Attached Storage (NAS) device. The latter can be remote such that its operation does not interfere with the audio pleasure. A Human Machine Interface (HMI) can be used to control the digital music server in addition to a small LCD on the device.

The audio server reads the digital music information from a selected source, in general, two stereo digital signals. Being aware of all active devices on the system bus, a network of digital filters generates the appropriate signals for each speaker drive, thus implementing a digital crossover network. It is these signals that are then sent using the system bus to the speakers that are nodes in the network. Firewire, i.e. IEE1394, is proposed here as network medium, even though ethernet with time synchronization according to IEEE 1588 is also a possibility.

Firewire is inherently appropriate for distributed time-critical systems, even used for industrial systems such as in SoftServo's FireWire Servo/Drives, ServoWire Soft Motion, or Parker SSD's 890 system drive. It is possible to share a global time in order to synchronise DAC to a few nanoseconds at a high data bandwith. Using isochronous messages, data packets are sent at 8kHz to the speaker, which takes care of the conversion of all channels associated for the node, preamplify with digital volume control, and pass the signal on to the power amplifier, one per drive.

The crossover network could even go beyond the requirements for splitting the frequency range. The digital audio server can generate a signal close to white noise and send it separately to each drive, one after another. By placing a microphone near the place where the music will be enjoyed, the transfer function for each cahnnel can then be identified. Having done this for all channels, shortcomings in the frequency range can be compensated by the appropriate filter in the filter network.


Decentralised DAC, modular amplifier and speakers

The concept presented in Decentralised DAC, amplifier and speakers can be taken further by modularising it. Every drive can have its own amplifier, pre-amplifier, DAC and network adapter. In addition, it is thinkable to share a common DC bus. It is thinkable to use the FireWire cabling also as DC bus for modest power. In the following, the speakers are presented as pair, but do not have to be collocated as such.


Decentralised Sigma-Delta pulse driving half bridges

Simplifying the concept Decentralised DAC, modular amplifier and speakers further would factor out the DAC at all, by using a pure class D amplifier as introduced in Sigma Delta modulation for power electronics. The idea is to use the digital signal and pass it through a Sigma-Delta modulator (SDM) implemented in and FPGA or an ASIC. The pulse stream could then -through appropriate circuitry, of course- drive a half bridge consisting of two MOSFETS (or even IGBTs).


Back to Miscellaneous technical ideas
Back to homepage
Copyleft 2002, Peter Wurmsdobler.