In most cases, there is a dominant peak in the open loop magnitude response. The system will then oscillate at this frequency if the loop is closed with the effective loop gain at or exceeding 0dB. Similarly, in cases where there are equally high dominant peaks, multiple frequency oscillations may occur when the loop is closed with sufficiently high open loop gains.
Going back to Figure 11, to improve the acoustic feedback gain margin, a high pass filter at 200 Hz was activated on the mixer microphone channel to take care of the 166 Hz oscillation. A 4.25 kHz notch with Q=5 and -4 dB gain was also employed to improve the 4.25 kHz margin.
The open loop function was then re-plotted and the result is shown in Figure 13. The 166 Hz peak and the 4.25 kHz peak are now reduced with the introduction of the high pass filter and the notch filter. These filters provided a better sound reinforcement experience as the reinforced voice when using the microphone now had less “ringing”.
If the mixer channel fader continues to push up further, acoustic feedback will still occur, but at other frequencies. These likely acoustic feedback frequencies can again be predicted by studying the magnitude and phase plots.
Also from Figure 13, a good gauge can be established as to how much more acoustic feedback gain margin is obtainable with associated equalization. By using this open loop plot, appropriate filters can be determined and deployed all at one go.
The distance of the magnitude plot from the 0 dB line represents the amount of acoustic feedback gain margin for the respective frequency. This represent the amount of gain that can be applied for that frequency before the acoustic feedback becomes self-sustaining. At frequencies where the magnitude response shows a peak, the acoustic feedback gain margin is correspondingly reduced.
The yellow highlighted area shown in Figure 14 represents the acoustic feedback gain margin. Notice that the acoustic feedback gain margin is highly frequency dependent.
This open loop measurement technique is based on the fundamental principle on why oscillation occurs. It is able provide the whole acoustic feedback gain margin picture at once versus the usual method of “ringing out” one frequency at a time.
This method can be employed to study the acoustic feedback gain margin of the house or monitor loudspeaker system. Multiple microphones or loudspeakers can be switched on and off to understand their contribution to the acoustic feedback gain margin.
Other investigations can be made with this measurement methodology to study the effect of sound reflection back into the microphone by a human head in front of a microphone or by cupping the front of the microphone windscreen with the palm of a hand. These response plots are shown in Figure 15 and Figure 16 respectively.
In many microphone applications, the microphone is expected to be moved around. In this case, the phase response will also be altered accordingly. Imagine that by just moving the microphone one foot distance directly away from the loudspeaker, the phase will cycle through 360 degrees at 1 kHz and even more for higher frequencies. Since phase response can change drastically with microphone position, it is safer to assume that there may be a zero phase situation corresponding to that magnitude peak when predicting if a magnitude peak can result in acoustic feedback.
The objective of course is not to flatten out the open loop magnitude response as this will have drastic consequences on the tonal response from the loudspeaker. Although a carefully chosen filter may improve the acoustic feedback gain margin, the filter will also directly affect the intended on-axis sound performance from the loudspeaker. Hence, such filters are chosen with highest possible Q and just sufficient notch depth to limit the damage to the loudspeaker on-axis response. Trade-off decision has to be made between acoustic feedback gain margin and the on-axis loudspeaker performance.
To improve acoustic feedback gain margin, common sense approach still prevails. The common guidelines include avoiding placement of microphones in the active coverage areas of loudspeakers, making use of microphone directivity to let microphone “hear” less from loudspeakers, close distance miking whenever possible to reduce gain requirement, appropriate acoustic treatment and optimization of loudspeaker system including equalization.
Only when all the common sense approaches are exhausted, appropriate equalization can then help improve the acoustic feedback gain margin but it will come with trade-offs. Commercially available feedback suppressor/eliminator can also be deployed to help reduced unwanted acoustic feedback.
In Part 3, the simulated measurement methodology of the closed loop performance of the system will be discussed and the results under various feedback gain margin scenarios will be examined.