Proper gain structure is important because it affects signal-to-noise performance and available headroom within a sound system. Every sound system has some inherent noise, whether it be self-generated by the internal electronics or induced into the signal path by external sources. Therefore, unnecessarily low gain settings can result in signal levels which are significantly closer to the noise floor, potentially causing a sound system to appear noisy. Conversely, excessive gain settings may cause the audio signal to overdrive the electronics, resulting in severe distortion due to clipping of the audio waveform. Besides being audibly undesirable, a distorted waveform can also cause damage to some system components, such as loudspeakers.
In addition to its influence on signal-to-noise and available headroom, gain structure can affect other aspects of sound system behavior. In particular, some audio components rely on signal strength as part of their normal operation. These components may not perform as expected if they receive signals that are lower, or higher, or even just different from what is anticipated. Examples of such components are: Auto Mixers, Duckers, Levelers, Comp/Limiters, Ambient Noise Compensators (ANC), and Acoustic Echo Cancellers (AEC).
Auto Mixer, Ducker, Leveler, and Comp/Limiter functions are triggered by input signals that exceed a specified threshold. With Levelers and Comp/Limiters, signal levels below threshold are not considered unusual (they simply are not affected by the component). However, Auto Mixers will not pass signals that are below threshold, and Duckers will not automatically attenuate program signal if the sensing input signal is below threshold. Furthermore, signals containing a large amount of background noise can falsely trigger these components, if the level is set too high and/or the threshold is set too low. It should also be noted that any real-time control of signal levels should not occur before these types of components. For example, control of individual Auto Mixer channels should not take place ahead of the Auto Mixer. Instead, the Auto Mixer Input Level controls (which are post-threshold) may be used for this purpose.
Ambient Noise Compensation (ANC) relies on a continuous and accurate model of the program signal level, to differentiate it from changes in the ambient noise level. So, real-time control of levels should not occur after this type of component. Acoustic Echo Cancellation (AEC) relies on a continuous and accurate model of the signal to be cancelled from the microphones. So, any real-time control of signal levels at the loudspeaker output should be duplicated for the AEC Reference. If these signals are different, a 2-channel ganged Level control may be used.
What is proper gain structure? Generally speaking, proper gain structure is establishing and maintaining good signal strength throughout the entire sound system. In most cases, this means that the relative volume of loudspeakers should ultimately be determined by adjustment of the power amplifiers (after prior system gain settings have been established). Other system outputs (such as recording feeds) may require lower levels, which should be established by selecting an appropriate reference level at the output itself. Other than real-time level control (as described previously), signal attenuation within the system should be avoided.
To establish proper gain structure, the primary element is input gain. Each system input provides adjustable Gain In (trim), with an associated Peak indicator. For best performance, increase gain on a given input until the Peak indicator just begins to flash on normal signal content. The Peak indicator first comes on with 6dB of headroom remaining (before clipping occurs). To provide additional headroom (allowing for louder input signals), it is recommended that gain then be reduced by 12dB (two 6dB decrements). To monitor system levels, Peak Meters should be connected at strategic points in the signal path, including at the inputs. With gain settings as described above, input meters should indicate peak levels between 6dB and 12dB on normal signal content (12dB ~ 18dB of headroom). This will provide a nominal level of approximately 0dBu, with good signal-to-noise performance and a safety margin of available headroom.
Throughout the rest of the system, various components will include level adjustment capabilities (faders). By default, these faders are typically at 0dB (unity gain). This is a very good setting for most applications, and does not necessarily need to be changed. However, any of these faders can be used for real-time level control (as described above). Of course, faders can be used to mix multiple signals at differing levels (i.e. microphones at 0dB and music at -10dB). Faders can also be used to compensate for gain reduction (Leveling, Comp/Limiting, etc.) which may have occurred earlier in the signal path. Note: Floating Point DSPs allow this type of gain staging without the danger of clipping distortion or the loss of data bits. As long as signal levels do not exceed maximum at the inputs or outputs (A/D & D/A converters), extremely high and low level signals can be tolerated without negative impact on quality. However, remember that some system components do not function well without proper signal strength (as described previously).