I have been requested I write a bit about I2S – it seemed like a good topic about which I have never written. Briefly, the common methods of connectivity, in order of introduction, are first S/PDIF and AES/EBU. These are bi-phase modulated connection protocols, similar except for balanced (AES/EBU) and single ended (S/PDIF). There are also provisions for encoded meta data which differ, the AES containing more data appropriate to the professional recording studio, and the S/PDIF containing data primarily useful in home gear. The balanced nature of AES/EBU lends itself to longer runs. In practice, the subcode data is largely ignored in current implementations. This system was designed exclusively as a digital audio data transmission system.
We then have USB Audio and High speed USB Audio, the former suitable for sample rates up to 96K and the latter for higher rates yet. Today we have UAC2 (audio over USB2) as the commonest standard. It is common knowledge that USB is a "Universal" Serial Bus, commonly used for everything from printers to scanners to audio. Kinda like multipurpose rooms in schools, so multipurpose there is prejudice about using them.
But wait! Way back in 1983, when I built my first D/A converter for high-end audio, there was little need for connectivity, as early digital boxes were single purpose appliances. At that time, I had what I thought was a unique idea to separate the mechanical and D/A converter sections of CD players. It did take a while, as we had to build our own digital signal processing circuits, since no special digital audio chipsets were available. SPDIF did not exist as of yet (and this was decades before USB).
The irreducible minimum of digital audio signals are Bit Clock, Word Clock, and Data; BWD. These signals are multiplexed and demultiplexed in all other connection protocols, even in I2S. Three clocks, 6 wires with grounds: 16, 24, or 32 bits of data, which are clocked into D/A converters with each bit clock until one channel of the sample is complete and the word clock starts it over with the other channel. With data overhead, 3MHz at 48K sampling, doubling to 6 and 12 Mhz at 96 and 192KHz sr. This is the most complicated but by far the best sounding way to hook a digital source to a digital converter. Period.
Within a year or two, SPDIF (actually S/PDIF) Sony/Phillips digital audio interface coax and later TOS (Barf) connections were available on CD players. I added the SPDIF to my prototype and Theta Digital was born, back in 1985. I had a way to hook my D/A converter universally to CD players. Soon, Wadia joined the party, and then there dozens of companies making them within a few years, as cookbook chipsets became available. This enabled anyone who could read to build an so-so D/A converter, which was good to legitimize the idea of the D/A converter, but the chipset designs typically sounded like ass.
With S/PDIF came a quandary. Even though SPDIF enabled any CD player (the source of the era) to be hooked up to any D/A converter, it did not sound as good as BitWordData. The problem with BWD was inherent fragility; digital audio signals were exposed to static and the possibility of shorts which could and would blow the phuc outta some very fundamental parts of the circuitry that in many cases could render the source junk. S/PDIF solved the fragility.
Then it got worse – Phillips, yup Phillips, invented I2S and I2C as a novel way to reduce wire counts in interfaces. Since I2C is for control, I2S for music. I2S was their method of reducing 3 signals down to two. The worst is that I2S sounds like ass compared to BWD (but still arguably better than SPDIF), and it has all of the same fragility, closed loop, and non-standard problems as BWD. Since it was simpler and cheaper, I2S won out over BWD as an internal standard to hook up digital audio appliances. Figures.
So, you may ask, since I2S is better and BWD waaaaaaaaaaaaaay better than SPDIF - Why doesn't Schiit use it? Schiit is a company selling big bang for the buck products. We have no sales dept and only one guy for customer service, as that is expensive overhead which flies in the face of our "bang for the buck" theme. If you frequently connect and disconnect, you will sooner or later have trouble. It also closes up the system and makes it awkward to audition other types of gear. Customer service intensive.
Am I closing the door on I2S as an external connectivity standard? You bet – it is a wet dream compared to BWD.
Take heart! All of our digital gear is upgradable and it is not out of the question that I may offer BWD as an external connection between our gear. The only problem is that I would have to make a digital source component first.
How does USB-Gen 5 talk to the DAC section? I2S or S/PDIF?
