Thursday, November 1, 2001
Avionics System Design: Voodoo Audio: Part I
Recent flights and discussions with flight and technical personnel have convinced me that cockpit audio remains a poorly understood area–one in which a good design requires some innovative bat wing dust and pilots need deep understanding to stay out of trouble. So in this column and in my column for the December issue, I will relate a history of how we reached this state of confusion.
There are applicable standards for cockpit audio–especially TSO-C50 in all its revisionist glory and DO-214 for signal guidelines. But general aviation (GA) manufacturers seem to treat this information as largely suggestive, rather than as actual design rules to follow. Air transport operators, military avionics manufacturers and airframers have a far better grasp of the guidelines’ importance–especially in the larger, system sense.
Many GA manufacturers and operators feel certain that they have found the ideal path toward audio happiness. However, they frequently ignore the impact their choice of audio has on other avionics systems in the cockpit.
How did it get this way?
Early standards were simple and clear. A microphone was a carbon element, with a 150-ohm impedance (borrowed from the telephone industry). It required external DC excitation of at least 8 to 12 volts DC to produce audio of about 150 millivolts (mV). Headsets and radio outputs had a 600-ohm impedance and required at least 50 milliwatts (mW) of power to produce acceptable levels. Both microphone and headset signals were normally referenced to the airframe ground.
If a speaker existed, it was understood to have an impedance of 4 ohms and be driven at 5 watts. Typically, it also was referenced to the airframe ground.
Everyone worked in this initial design environment. It was a world in which the weight and cost penalty of extra wire (for grounds) was thought to be significant. Lights, instruments, pumps and inverters used the ever-present airframe as their return path–and so did audio.
The result of these early concepts was poor but acceptable audio in an environment in which ambient noise was extremely high. The audio was actually very noisy; ground loop interference was a permanent fixture. But it was easy to understand, conceptually, and simple to interconnect in the airframe.
After the Korean War, however, this mediocre level of performance failed to impress military operators. They moved away from these standard practices in an attempt to control acoustically coupled cockpit noise, as well as reduce airframe interference. Military operators adopted "noise canceling" microphones like the M87 and M101, which had a dynamic microphone (mic) element built like a studio microphone. The M87 had an active diaphragm with balanced openings on both sides of the mic housing. The concept has the voice, mixed with ambient noise, entering one side of the diaphragm and noise-only entering the other side. The noise cancels mechanically in the diaphragm, and the resulting signal is voice only.
The early noise-canceling dynamic mics were only partially effective. But they produced far less self-generated noise than carbon mics, which sounded like angry rattlesnakes under vibration. Equally important, this new type of mic significantly reduced distortion.
The noise-canceling mic introduced issues that were not so attractive, however. Dynamic self-excited microphones offered low impedance (about 5 to 8 ohms), which should have benefited noise rejection. Unfortunately, they operated at such a low signal level (often under 1 mV) that any induced noise suddenly became significant. In addition, the wire loop in the mic sometimes would function as a small antenna and thus became susceptible to interference when in strong magnetic and radio frequency (RF) fields.
This type of microphone’s big plus was its isolation from the airframe ground–since no DC excitation was required. This greatly improved the ground loop control of a sensitive signal. Military designers discovered that a low-level microphone must be accompanied by a low-impedance headset to keep drive voltages low. Otherwise, terrible cross-talk and coupling often would result in the headset wiring or the airframe interconnect, despite shielding. Thus was established the 8- to 20-ohm standard headset/dynamic microphone combination commonly used by the military.
Still, not everyone favored the low-impedance microphone. Military users also adopted amplified dynamic microphones and 600-ohm headsets as an alternative. (The amplified dynamic mic has low impedance but also a small outboard amplifier to make it work like a "carbon equivalent" microphone.) Helmets, like those made by Gentex, often included the amplified dynamic configuration. The goal was to gain better sounding audio, low headset bus loading, and no interconnect problems associated with the raw dynamic mic and its low-level signal.
This diversity of approaches meant that military audio systems often had to support different mic and headset standards within a single station (or audio box for each pilot), like the Andrea C6533 or A301. And each user typically had his own station control within the aircraft.
The military quickly realized that multistation grounded audio was futile, especially in a large airframe, and floating balanced lines (often with a single point ground) became the norm for interconnect. This technique allowed clean audio to be delivered over wide distances, with minimal unwanted coupling. The required headset levels were locally amplified at each user’s station, so radio bus loading was minimal and interaction between users was low. The floating balanced lines required shielded wiring and significant system knowledge to be implemented correctly. But in a well-engineered system, the results could be impressive.
In the civilian world, a simplistic "audio panel" usually had all users interfaced at the same signal level. Headsets and mic jacks often had no ground wires but were connected to any available airframe metal for a signal return. Shielded wire (expensive and heavy) rarely was used except for long runs to remote radios. You may gasp in horror, but thousands of general aviation planes were equipped this way for decades–and they sound like it. This interconnect resulted in serious ground-loop noise in every path and could make radio transmissions unintelligible due to serious microphone audio interference.
By the 1960s, a distinct split occurred between civil and military systems. Some bad habits became firmly entrenched in the civil world and have remained so since. The worst of these were civilian radios with single-ended, or grounded, audio for the mic, headset, speaker–everything. Even when designers included an internal transformer to develop a sufficient voltage swing for good 600-ohm drive, they still grounded the output, though grounding was unnecessary. This doomed pilots to noisy audio. In some radios, power ground, mic common, headset common and speaker common were all tied to the same pin. Not a landmark moment in design strategy.
Early audio panels commonly "mixed" audio from different radios by simply shorting the radio headset outputs together (all supposedly 600 ohms) and then running the combined bus to the headset. As a radio was switched "on" to the bus, the existing level would drop (sometimes precipitously). This meant audio bus loading was marginal-to-bad in virtually every condition. Keep in mind, radios typically have a lower source impedance than the impedance they are designed to drive. They do so to provide loading headroom, or sufficient drive capability for operation. They therefore become very unattractive loads for other radios on the bus.
A fully TSO-C50/DO-214-compliant 600-ohm output must be able to drive a 150-ohm load and thus must have a source impedance of 150 ohms or less. "Advanced" audio panel systems use a series resistor from each radio headset line to reduce this source loading effect but at the inevitable expense of signal level. Unattractive as all this is from a design perspective, radios still work reasonably well for a pilot if enough signal level is available.
Tune in next month, when we continue this saga of cockpit audio design and tell about how one company, David Clark, changed the headset market forever.
Walter Shawlee 2 may be reached by e-mail at firstname.lastname@example.org.