We got a couple of letters regarding last month's "Theory and Practice" column. So we are all on the same page, the purpose of that article was to impart a basic understanding of what digital audio is comprised of for the guys who are not interested in all of the formulas and math.
That being said, there was some confusion when we referred to "steps in volume" while talking about bit depth. A more technically-correct way of putting it would have been "dynamic range." For the mathematically-inclined, the resolution of a converter is defined by 2 raised to the power of n, where n is the converter's bit depth. A 16-bit converter has a resolution 65,536 "steps" between no signal and maximum signal level. A 24-bit converter has 16,777,217 steps between no signal and maximum signal level. Now back to the real world and the reason we got into the whole digital audio discussion in the first place…-ed.
Last month we discussed what digital audio is made up of. This month we will go over some ways to preserve the digital audio signal once inside the digital domain, as well as some ways to get a better sound of what you may already have in inventory.
In the first part of this article, we talked about sample rates. In a nutshell, a sample rate refers to how often an analog signal gets recorded or reproduced (or converted), usually expressed in how many times per second. A sample rate of 48kHz means the signal is being sampled 48,000 times every second. A device that is recording/reproducing these samples, such as a console or any other digital equipment, needs to know what exactly a second is and at what interval these samples are to be "recorded"/played back. This is what a word clock does.
Word Clocks
Word clocks come in two different flavors, internal and external. Every digital device, from a consumer CD player to a touring-quality digital mixing console, has an internal word clock – it's needed, because without it, you wouldn't hear anything, because the device wouldn't know what to do with these samples. External word clocks, meanwhile, usually come in rack mountable form and support standard digital I/O formats such as S/PDIF, AES/EBU, optical fiber ("LightPipe") and coaxial (BNC).
The precision of a word clock is crucial because it will determine how accurately the original signal will be recorded or reproduced. This can be seen in Fig.1 the top half of which illustrates the nice, evenly-spaced samples on an accurate word clock. This is what digital audio in a perfect world would be. On the bottom of Fig. 1, the samples are not so evenly-spaced. When someone says, "those converters don't sound good," this is likely what they are referring too. Not to discount other aspects of a converter including the quality of the analog path, an inaccurate word clock can and will lead to a false representation of the original signal.
This is what's known as digital "jitter" and sonically, it can be described as "losing the sparkle" or just an overall "muddier" sound. For the analog guys, it just sounds more "digital." To combat this evil occurrence, engineers will sometimes employ the use of an external word clock. (Yes, we know this is a subject of often intense debate and that an external clock does not guarantee better sound but it does help us address synch issues between multiple pieces of digital gear.)
Typically, a digital device such as a digital console or processor will have some sort of interconnect port for an external word clock signal. This is how a digital device receives an external word clock signal. Once an external word clock is hooked up, you're done, right? Nope! On a digital device, there is usually some sort of menu option that allows you to select what sampling rate and bit depth to operate at, as well as what to sync to (internal, external, S/PDIF, AES/EBU, etc.). Once the gear in question is "slaved" to a master word clock, it will not use the internal word clock, and you lose the ability to control sample rate. When it "locks on" to the incoming sync signal, it will record/reproduce samples at whatever rate the incoming sync signal tells it to.
Fewer Conversions
Some digital gear has a pretty bad rap, but it's not necessarily the gear's fault for sounding so "digital." A big reason for this – based on what I've seen out in the field – is that too many engineers/designers don't take advantage of digital connectivity. Why not take full advantage of what digital technology is designed to do? When you compare spending hundreds of thousands of dollars on digital consoles, crossovers and amplifiers to spending a couple thousand (or even less) on a digital snake, it is surprising more engineers and designers don't add it to the budget.
I will admit, I am just as guilty as the next guy for sometimes plugging my computer's analog out straight into a digital console, even though my computer has a S/PDIF output, and the consoles I use usually have the same input. It becomes a convenience and time issue, which, I'm sure, is why most engineers do the same thing.
Signal Degradation
So what's the big deal with all of this conversion stuff? Well, as stated in last month's "Theory and Practice" column, the conversion process is where most of the signal degradation occurs. Remember, when a signal goes out of one digital device and into another, there are two conversions. So when you sit back and take a look at your system with all the new latest and greatest digital technology, you might be doing more signal degrading than you think.
Pictured in Fig. 2 would be a typical system scenario for a sound company that has new digital technology, but does not use digital interconnects. This is wrong in my opinion, but due to convenience, time constraints and even an under-educated crew this is often the bitter truth of how a system is designed and setup.
Fig. 2 starts with a signal out of an iPod, into a digital console, out of a digital console, into a digital system controller, out of a digital system controller, into a digital amp and out of a digital amp to the speakers. The problem is that every time the signal goes from device to device, there is an analog-to-digital and/or digital-to-analog conversion. This is a part of the digital world that will haunt the ears of engineers forever.
From Seven to Two
Now that we know the problem, what is the solution? If you got 'em, use 'em! Some digital consoles, processors and even amplifiers come standard with AES/EBU and/or S/PDIF interconnects. Others have optional cards that are available to add these interconnects. If you look back at Fig. 2 and how it is currently set up, there are a total of seven conversions. When you add all of these conversions in combination with poor gain structure and low-quality converters between devices (jitter), this is a recipe for disaster.
Instead of using all of the analog interconnects, use the digital ones that are provided. Play walk-in music on your computer and use S/PDIF to go into the console. Then use AES/EBU from the console to processor and processor to amplifiers. So now the seven conversions would be down to just two. Now add an external word clock, sync all of your devices to it, and welcome yourself to digital audio paradise.
