Keeping Digital Clean
Digital audio has provided radio with a means to deliver a higher-quality sound without requiring significant additional cost. It was quickly learned that digital was a different world than the analog domain it slowly replaced. The promise of a signal free from the customary obstacles associated with analog circuitry and transmission was not the holy grail it was expected to be. In the end, we traded concerns over noise floor and distortion for jitter and bit-error rate.
See the sidebar to this story:
Manager's Perspective: Keeping Digital Clean
When several basic rules are followed, analog is easy. It is also forgiving. Analog has a graceful point of failure. As signal degrades, our ears can usually provide clues that something is wrong.
On the other hand, digital signals make their way through the path, gracefully overcoming changes in impedance, capacitance, bit errors and noise. Through the most analog-challenged circumstances, digital continues to sound as good as it always does — that is until the degradation is so severe that the signal disappears completely.
The nature of digital audio is to provide a signal unimpaired by the transmission medium. Analog signals are an inseparable part of their medium. Digital signals ride the transmission medium, almost without regard for its flaws. As noise increases, impedance changes, capacitance varies, the integrity of the digital signal suffers, but in many cases, the error correction allows the signal to be reconstructed at the receive end. As error correction works harder to fill in the missing pieces, it eventually reaches a threshold where it cannot put it back together. In this case, the signal mutes. The point where this occurs is referred to as the digital cliff. For all the robust properties that make digital stand over analog, digital signals still require careful attention to many of the same physical and electronic aspects.Avoid the fall
Some potential digital pitfalls cross physical and electronic concerns. The most basic step in preserving the integrity of digital signals is to use the proper wire. Most facilities distribute digital signals that are defined by the AES3 standard. There are other formats, such as S/PDIF, but these are rarely used widely in a facility.
The AES3 audio standard defines a stereo signal carried over a single pathway. This pathway can be a balanced or unbalanced signal. The balanced signal requires a twisted pair cable with a characteristic impedance of 110Ω, ±20%. The unbalanced standard, referred to as AES3-ID, requires a coaxial cable with a characteristic impedance of 75Ω. Because 75Ω coaxial cable is commonly used for RF applications, and RF requires a high level of consistency, cabling is usually not a major concern.
Because AES3 uses twisted-pair wire, it is easy to try to get by with traditional analog twisted-pair wire. This works for short cable runs, but the impedance mismatch is severe, and problems are likely to occur.
Ideally, AES3 wire should be used. Facility wiring systems also exist that use CAT5 cable. The CAT5 Ethernet standard calls for a characteristic impedance of 100ohms, +/-15ohms. This results in a range of 85ohms to 115ohms. Compare this to the AES3 extremes of 88ohms to 132ohms. Except for the very low end, the CAT5 range falls within the AES3 standard. If CAT5 cable is used for AES3 audio, it is a good idea to use a CAT5 cable with a tighter impedance tolerance. A range of +/-7ohms works well.
Because an AES3 signal has more in common with an RF signal than a DC signal, connectors, terminations and splices are important. The best rule is to keep these to a minimum. When some type of physical connection must be made, keep the wire twisted until the very end to minimize the change. Some connectors and punch blocks are designed to meet the rigid AES3 standards. Likewise, tighter-tolerance CAT5 hardware can also be used reliably.Change for the worse
A common feature of many digital audio devices is sample-rate conversion. This allows various sample rates to be interchanged between devices without the need to manually set each input and output to a specific rate. Regardless of the simplicity, a facility should choose a single sample rate and use it as much as possible. Typically, the rate is chosen by using the most commonly used sample rate of the equipment. Since lower sample rates result in smaller files, lower rates such as 32kHz or 44.1kHz are chosen over 48kHz. 96kHz is being used in many recording studios, but because of radio's bandwidth limitations, 32kHz, 44.1kHz and 48kHz are the most common.
Ideally, a master clock reference is used within a facility, and all the devices are synchronized to it. This eliminates the need for each device to reclock the incoming signal. There is little loss when a signal is reclocked at the same sampling rate, but there are significant concerns when a signal undergoes a downward sample-rate conversion.
When a signal is downconverted, for example, from 44.1kHz to 32kHz, the audio energy between 32 and 44.1 must be removed or aliasing will occur. A filter is used to remove these components. This necessary filtering can cause additional problems in that signal overshoots can result from the ringing of the filter. The complete discussion of this topic situation can easily become quite involved, but in general, it is best to avoid any downconversions when possible.
Another potential source of unwanted digital noise can be introduced by switching digital signals. When signals are mixed, the clock references are tied together. Some digital audio switchers perform a hard swich from source to source. If this switch is made and the two sources are not synchronized, there will likely be a click or pop during the transition. Ideally, a switch such as this is only done with synchronized signals and during a point where the signal crosses the zero-voltage reference. In most cases, the delay of a single frame or two is not a problem.Audio coding and data reduction
Non-data-reduced or uncompressed audio provides the cleanest method of capturing digital audio. The AES3 standard defines the parameters for this coding method. In most cases, simply converting an audio signal to digital is not the only concern. A complete digital audio system covers not only the audio signal, but also its storage and transmission. Because of these issues, various methods of reducing the required storage space or transmission bandwidth have been developed.
The audio on a CD, sampled at 4.1kHz with a 16-bit resolution, requires about 10MB of storage space and 1.4Mb/s of transmission bandwidth for every stereo minute. For a very high quality 96kHz/24-bit signal, about 33MB of storage space and 4.5Mb/s of transmission bandwidth is required per stereo minute. With files sizes such as this, disk space and transmission bandwidth requirement would be out of control.
To reduce the storage or bandwidth requirements, data-reduction algorithms can be applied to the digital signal. Some of the algorithms more commonly used include G.722, APT-x, MPEG-1 Layer II (MP2), MPEG-1 Layer III (MP3), MPEG-2 AAC, ATRAC and PAC. These algorithms function by trading file size for some reduction in audio quality. The key is that the trade-off is not noticeable. The debate between the appropriate use of APT-x, MP2, MP3 or AAC can become a heated one. Choosing the algorithm for a given situation should be done by considering the source material and the individual settings of the encoder. The final choice will be based on how the resulting audio sounds to you.
When coding algorithms were first introduced, there was considerable discussion about the effects of passing a signal through several codecs. Some demonstrations showed that the resulting audio after just a few cascaded conversions had little resemblance to the original source. Fortunately, most algorithms have improved, but the same care should still be taken to minimize the number of conversions being made.Proving performance
While digital audio has easily provided a cleaner signal from start to finish, one obstacle has been in assessing its quality. Clipping onto a signal and listening to it is not possible with a telephone butt set. The new digital equivalent is needed. Analog audio could be evaluated with butt sets, oscilloscopes and distortion analyzers. Digital audio has its equivalents, but instead of confirming signal to noise and harmonic distortion, we now measure for bit-error rate and jitter.
Tools to evaluate the physical medium are still required. Because most of the media used for digital audio came from the computer industry, cable checkers and hard drive diagnostic tools have become regular tools.
Fortunately, by planning the system carefully and using the proper equipment, maintaining high quality digital audio can be easy.
Today in Radio History
The history of radio broadcasting extends beyond the work of a few famous inventors.
EAS Information More on EAS
The feed provides feeds for all US states and territories.
Need a calendar for your computer desktop? Use one of ours.
Information from manufacturers and associations about industry news, products, technology and business announcements.
After 57 years in the same building, CKUA was ready for a move. But it hasn't forgotten its history.
Browse Back Issues[an error occurred while processing this directive]
Also in the June Issue
- The Radio magazine Pick Hits of the 2013 NAB Show
- The Settlement/Amendment Process for FM Translators
- Side by Side: Headphones
- Field Report: iZotope Insight
- Field Report: Rode iXY
- Better Coaxial Cable Runs
- 20 Years of Radio magazine: July 1994