The real-time analyzer (RTA) has a number of applications in which it is the best tool for system optimization: zero. There is nothing that the stand-alone RTA can do for us that cannot be done better by other analyzers. Signal path, noise floor, hum, THD, polarity, latency, and frequency response analysis are the mainstays of the verification process. Equalization, delay setting, level setting, speaker focus, and room acoustic property analysis are the mainstays of the calibration process. The RTA is inferior in all these categories to more advanced tools.
This is not to say that the functions of the RTA cannot be put to use. It can serve as an introductory ear–eye training tool. We can listen to music and match what we hear with what we see. Feedback can be induced and identified to train us to be able to hear it before it gets out of control. But even such time-honored functions of the RTA can be replicated by modern fast Fourier transform (FFT) analyzers that duplicate the functionality of the RTA without being burdened by the RTA’s inherent limitations. The RTA, like so many things, has been replaced by a computer.
In the past, the RTA was the only spectrum analyzer in use for most sound reinforcement applications. It was affordable, small, and easy to operate. These have now been replaced with dual-channel FFT analyzers, which are affordable, small, but require training to operate. The reason the FFT analyzer has taken over is engineers have learned that having highly accurate data is worth the trouble it takes to learn how to operate these systems. What is an RTA, and what is it that makes it so inferior?
The RTA is a bank of parallel log spaced band pass filters (octave and one-third octave are typical) at a standard set of center frequencies. The output of each filter is sent to a full wave rectifier circuit that creates a waveform that represents the absolute value of the filtered waveform. The next step is integration, the process by which a DC value is derived from the absolute value. The integration time constant, derived from a resistor–capacitor (RC) circuit, can be set to fast (250 ms) or slow (1 s). The end result is that the 31 one third octave bands each contains an integrated value representing the average value in that frequency range in the current stretch of “real time.” The term “real time” connotes that the displayed values represent a continuous stream of time, with no gaps as might occur with an analyzer, which takes samples of time. Unless the RTA is paused, the data continually streams in time.
The RTA is designed to show a flat response when driven by a source containing equal energy per one-third octave, i.e., pink noise. Since pink noise has random components, the RTA requires averaging to settle down to a flat response. The RTA can make coarse frequency resolution evaluations of the system response. But one-third octave resolution is far too low to be considered for equalization use, as will be discussed later in this section. Low-resolution data is only sufficient for measuring the noise floor of a device, or exploring gross response trends such as traffic noise in urban areas.
The crippling limitation of the RTA is its rendering of its middle name: time. It cannot measure time, so it is useless for setting delays or identifying reflections. It has no phase response and, therefore, we cannot understand the nature of speaker summation, which, as we know, leaves us blind. It cannot discern between multiple arrivals within its integration time, and therefore cannot separate early arrivals from late. Thus it has no means of mimicking the ear’s perception of tonal change vs. discrete echoes. Because it sees all the energy in the room, regardless of timing, the low-frequency addition is greatly exaggerated above the ear’s perception. The longer the reverberation time in the room, the more the RTA low-frequency response hangs on. If the response is equalized to flat in a reverberant space, the system will sound as if the low frequencies are massively deficient. An RTA user only makes this mistake once. After that, they learn not to trust the RTA. Unfortunately, the failings of the RTA caused many engineers to close their minds to audio analyzers as a whole. “Analyzers? We don’t need no stinking analyzers” became a familiar position taken by front of house mixers.
The RTA presents a one-dimensional reading of the complex questions of the sound system response: amplitude over frequency. This has led the users to assume that a one dimensional solution is applicable, the modification of amplitude over frequency, otherwise known as equalization. This is the worst of the RTA’s attributes. The mentality of equalization as the primary or solitary solution leads to an endless repetition of the same mistakes. Since the real nature of the problem is not revealed, the solution is not sought, much less proved. The RTA cannot bring the data needed to frame the questions of how our sound systems are reacting to themselves and the room. The sources of variance are not found, and the principles of minimum variance are not applied. In many applications, the examination of the system is conducted only at a single location (the mix position), and therefore level, spectral and ripple variance over the space are not even considered in the equalization of the system. Hence the same mistakes are made night after night, year after year.
Phase, as we know, is the holder of the puppet strings that control our sound system’s performance. Without knowledge of phase, we are destined to be surprised at every turn. With an RTA, evidence of polarity is wiped out by the rectifier, and evidence of the phase is removed by the integrator. Once lost, these cannot be recovered. This erasure deprives us of our best evidence toward finding the causes of variance.
This is not to say that there is no use for the RTA. There may be applications in related fields such as urban or industrial noise analysis where random, uncorrelated sources are to be measured in low resolution. It is doubtful, however, that even these fields of study would not find advanced analyzers to be superior. In our field of sound system optimization, the RTA was never the right tool even when it was the only readily available tool. The final word is this: the advanced analyzers can mathematically duplicate the RTA computation and display should the need arise. The RTA, however, cannot return the favor.
About the Author
Bob McCarthy specializes in the sound system optimization and design as the president of Alignment & Design, Inc. He is the foremost educator in the field of sound system optimization and has conducted training courses worldwide for over twenty years. Bob’s clients have included esteemed companies such as Cirque Du Soleil and Walt Disney Entertainment, as well as, many of the world’s best sound designers, such as Jonathan Deans, Tony Meola, Andrew Bruce, and Tom Clark, among others.
Excerpt from Bob’s Sound Systems: Design and Optimization