Sound bytes

Many professionals are aware of digital audio, or aware that it exists. But they are less aware of what this means and how this affects their choice of things such as wire and cable products.

To begin with, digital audio is more digital that it is audio. That is, it is a bit stream, a series of ones and zeros. These are generated by taking the original analogue audio and cutting it into pieces ("sampling") and representing each slice as numbers, digits.

The Table above shows the most common sampling rates. Also shown is the "occupied bandwidth".

The sampling rate is not what runs down a cable. The actual signal is also determined by the size of each digital "word". You might hear some manufacturers tell you that their box accepts or produces 24-bit words.

AES specifications go up to 32-bit words. The standard also allows the cable to carry one or two channels per cable pair (or per coax cable, if that is used). The choice of one or two channels is made in the transmitting and receiving equipment. The cable doesn't care.

In paired cables, the AES specs have cleverly added something called ‘bi-phase' signaling. This means that, in a connector, where you have two pins to carry that pair, it doesn't matter which wire in the pair goes into which pin, either way will work as long as you have one wire in each pin.

So, if we have up to 32-bit words with up to two channels and a bi-phase signal, the actual bandwidth will be the sampling rate x 32 x 2 x 2 (or just multiply the sampling rate by 128. This is precisely what we have done in the above Table.

But, some of you might say "My connector has three pins, not two!" And most connectors, such as XLR connectors, are three-pin connectors. Pin 1 is ground, connected to the shield in the cable. Pins 2 and 3 carry the signal. The above Table reveals a number of interesting facts. First, this audio signal, which used to extend from 20 Hz. To 20 kHz (20,000 Hz.) now extends into the Megahertz.

48 kHz sampling, as you can see, has a bandwidth in excess of six MHz. What that means is very simple: anything that carries, or processes this signal, will be very different from the old analogue version. This signal is now of a much higher frequency, and things that didn't apply before are now important. Table 2 shows the first of these, wavelength.

A quarter of a wavelength is where a signal reaches its maximum size, so we're most concerned with a quarter-wavelength. You can see for even the highest analogue frequency (20kHz) the quarter-wavelength exceeds two miles. This means we would have to consider an audio cable as a "transmission line" with a specific impedance only at that distance. Since we never go that far, it doesn't matter what the impedance of that pair is, it can be anything.

But, when we convert to digital, you can see the quarter-wavelength is now a very approachable value. At the highest sampling rate (192 kHz), our quarter-wavelength is only ten feet. That means two things:

First, since we can easily have a cable that is the length mentioned for any of the digital sampling rates, these cables will become a "transmission line" and we will need to pick a specific impedance to match between the cable and the boxes at each end. What impedance should we choose then?

The lowest loss for a twisted pair "transmission line" occurs at 150 ohms. The performance (and distance) of a cable is compromised to get the size down to something reasonable. By the time you're down to 110 ohms, an old standard from the telephone company, your cable begins to look quite similar to the old analogue audio cables and will easily fit into an XLR.

The second thing that our short quarter-wavelength means is that the connectors really don't make a difference. An XLR or any similar connector is only a few inches, at most. You could have a bunch in a row and never get close to ten feet (much less forty feet!) So the impedance of the connector means nothing, and you can use whatever you wish.

But now we can understand two "modes of failure" in digital audio. One is where a patch cable or other interconnect is used on a digital device, and when the cable is plugged in, the digital bit stream stops. Why? One reason might be that this new cable is an old analogue cable, not made to 110 ohms. (Most analogue cables are actually 30 to 70 ohm impedance, far from the correct value for digital.) This means there is now an "impedance mismatch" on the line and a huge percentage of that bit stream will be reflected back to the source device.

It could be so much, that too little reaches the receiving device. Or it could be so big a reflection that it affects the sending ‘chip'. Either way, putting old analog cable in the line can be problematic, at best.

And, while you might put real digital audio cable into your installation, you can ruin a perfect install by forgetting to replace your patch cords with digital patch cable. Do you need to change the patch panel? Is it electrically ten feet long? (or forty feet long?) No, it's only a few inches. So you don't need to change your patch panels, just the cables...and anything else that is ten feet or longer!

Steve Lampen is multimedia technology manager at Belden and an SBE Certified Radio Broadcast Engineer.

MORE FROM ARABIANBUSINESS.COM