Print Page
The Importance of Specifications in Design
10/19/2005

As audio systems designers, we rely on manufacturers' specifications not only for a list of features, but for how a piece of equipment can be integrated into a system.

As an example of how good spec sheets can help us and missing specs can hinder us, let's start to design a hypothetical but typical audio system for a TV station where there will be both analog and digital subsystems.

The nominal analog audio level is +4 dBu. The lower case 'u' means that the reference voltage is 0.775 volts.

The nominal digital level is -20 dBfs or "decibels below full scale." Full scale is the digital clipping point--the point where there are no longer any bits left to encode the signal.

(Another common digital operating level is -18 dBfs, used more in Europe than in the United States. Check the specs on what digital operating levels the equipment provides. Often, the user will be able to select the desired level.)

The digital sampling rate for our system is 48 kHz, and we're going to use 75 ohm coax for digital audio wiring.

Let's take a look at some spec sheets.

First, a digital mixing console--according to an actual spec sheet of one particular mixer, the digital (AES/EBU) input and output connectors are XLR. The spec sheet doesn't specifically indicate that these are XLR-3 connectors, nor that these are balanced signals. It may be implying that by indicating the digital I/Os follow AES/EBU standards, but to verify, you would need to look to a diagram of the connector panel, and for any wiring information the manufacturer provided.

Unfortunately, this level of information can often only be found in equipment manuals, which means spending more time to dig it out. Fortunately, more and more companies are putting manuals online, but it's very helpful to have basic connector and signal information right on the spec sheet.

Back to our mixer. Since it uses XLR connectors for digital I/O, we know we'll need baluns to convert to/from 110-ohm balanced XLR to 75-ohm unbalanced BNC for our distribution scheme.

The spec sheet does not spell out the sampling frequency or frequencies that this console operates on. That's not good. It does offer one clue in giving the total delay between input and output at 48 kHz sampling frequency. So we may infer that this is the sampling frequency, but it would have to be verified.

For this console, the delay is 2.5 milliseconds, less than a video frame, so one pass through the console shouldn't be a problem. But audio delay needs to be looked at on a systemwide basis, so this number should be kept in mind as the system develops.

The spec does indicate that the mixer has a reference video in through a BNC connector (a good thing), and what kind of reference it will accept. It also describes the reference word in and out (often referred to as "word clock" in other specs).

We will use the reference video for our system to make sure all devices with digital audio are synchronized.

On the analog side, the spec sheet lists the various inputs and outputs, levels, and the type of connector used for each.

Some examples--the A-side inputs are balanced, use XLR connectors, have an input impedance of 4,700 ohms, and accept input signals from -60 to +10 dBu with +24 dBu maximum. The level range indicates that these inputs can be used for mic and line-level inputs, but I would consider the impedance on the low side for line-level inputs when using a bridging topology.

The B-inputs to this console have the same specs, except that the connectors are 1/4-inch tip-ring-sleeve (TRS) connectors instead of XLR, and that the input impedance is 10 kohms. Since the console uses different connector types for the various I/Os, I would make up wiring diagrams for each of the connectors to ensure that they are all wired correctly.

The program outputs use both 1/4-inch and XLR connectors (simply in parallel or fed through an internal DA? The specs don't say). Also output impedance is spec'd at 150 ohms and level at +4 dBu (10 kohms). I take this to mean that we are to use this console in a bridging mode as opposed to terminating the outputs. This is just what we want, although it may not be written as clearly as it could be.

GREATER CLARITY

I like the way Shure writes its specs for the M367 mixer. The spec sheet indicates the actual internal output impedance, and the impedance for which the output is designed to work.

The Shure specs also indicate which signal paths through the mixer are inverting (polarity reversed from input to output).

Mic-line in to mic-line out is noninverting (same polarity for input and output) but for mic-line in to mix bus, and from mix bus to mic-line out, each is inverting. If the mix bus from one unit is connected to the mix bus to another unit, as per its intended application, then noninverted end-to-end signal polarity is maintained. It's only if the mix bus is used for another purpose and fed to another device that the out-of-polarity signal could cause problems.

It's not uncommon for out-of-polarity signals from input to output to be designed in, whether by intention, as the Shure mixer, or not. I've seen this on some audio distribution amplifiers, where half of the outputs are in polarity, and the other half are out. This is a very useful thing to know, but I rarely see it on spec sheets.

Another place where polarities may be reversed is the audio mixer insertion I/Os. The inserts also are often unbalanced and at a different level than the main program level. This is very typical for low- to mid-priced consoles, although not all.

While we aren't told about polarity for our first example console, we are told that this mixer uses 1/4-inch TRS unbalanced connectors. This implies that either one of the contacts and the shield is used for the send and the other contact and the common shield is used for the return. But spec sheet doesn't give details on which way these connectors are wired. This is something else to note, especially since this console uses 1/4-inch TRS connectors for both balanced and unbalanced signals.

The insert levels are listed at 0 dBu, +20 dBu (10 kohm) with send 150 ohms and return 10 kohm. Is this a little unclear to you too? I'm assuming that +20 dBu is the maximum level before clipping. I would need some clarification about the way the impedance is spec'd, but what is clear is that we have to deal not only with an unbalanced signal, but one that is 4 dB lower than our system nominal operating level.

The most common solution is to install a balanced/unbalanced converter for each insert point. I would look for an active box that had a level control to bring up the send level and drop the return level. An audio mixer with unbalanced and lower-level inserts may seem like a great deal until you factor in the cost of these add-ons.

Converters take up a lot of rackspace and power strip space for all the wall-warts, especially as the number of input channels increases. Frame-mounted converters connected to a central power supply are a big help in this situation. To keep unbalanced lines short, converters should be mounted close to the audio mixer.

Next time, there'll be a bit more to say about our mixer specs. Then we'll add some more gear to our system, and continue to explore the wonderful world of specifications.

Mary C. Gruszka is a systems design engineer, project manager, consultant and writer based in the New York metro area. She can be reached via TV Technology.   Print Page