In our previous installment (Part Three, Issue 153) we determined that a truly adequate amount of clean power is advantageous in handling peak volume levels in our music, because this headroom prevents unwanted clipping of the signal and possible damage to our speakers and equipment, while providing better fidelity in our listening position. After all, the way the speaker performs at the listening position in terms of its frequency response at a given volume, along with the influence of the room on the speakers’ performance, is what we end up hearing. Taking the size of our listening room into account, along with considering our speakers’ sensitivity and the amount of power needed to adequately drive them gets us a long way toward choosing a good speaker and one that delivers a tonal balance that we’ll be happy with.
We also considered just how much amplifier power may be necessary, with some illustrative examples using a 6,000-cubic-foot room and a pair of speakers nominally rated at 85 dB. The goal was to better understand how well speakers will pressurize a room with sound and have a better idea of what a particular speaker designer’s take on how their speakers might perform and what kind of design and budgetary restraints they might have been limited to.
At the end of the article, we raised the question that’s on many a speaker buyer’s mind: “How loud will my speakers play?” The perhaps-elusive answer to this question can assist us in better understanding our speaker’s capability in this regard.
Audio system installation specialists, sound consultants, studio professionals and end-user customers such as you and I need accurate information so that we can be sure of what we are purchasing. However, in the commercial space, the data for a speaker’s peak linear sound pressure level may be available, but less often is this information even contemplated by domestic-market customers.
There are many different tests and standards out there for evaluating the output of loudspeakers, but not all are adhered to by all manufacturers. THX has their standards, and then there is the CEA 2034-A-2015 Spinorama (ANSI) speaker test standard (originated from research published by Dr. Floyd E. Toole in the 1980s and refined by Harman International), the Skywalker Sound ISO 2969X cinema standards, the SMPTE 202M standard, and others.
What we often find is that the data that is produced may either be non-informative to the majority of customers, and so it is never requested (or certainly not by the majority of end users who may simply blindly trust a brand), or non-conclusive. Speaker “A” may have a set of measured response parameters that are the same as speaker “B” from another manufacturer, yet sound completely different because of its off-axis response, directivity index, or the total amount of power output it produces.
One example may be that a speaker with a higher directivity index – the ratio of the amount of sound it produces directly forward, compared the amount of sound the speaker delivers off-axis – may be more advantageous for use in a bigger room rather than a small room.
But where can you find directivity index information in the first place, and what does it mean? In upcoming articles we will discuss the critical value of Spinorama testing. Spinorama consists of a series of on- and off-axis speaker measurements which identify the amount of direct sound coming from the speakers to the listener, and also how much of the off-axis sound is being more widely dispersed. Importantly, Spinorama measurements can also predict whether listeners will prefer a given speaker in listening tests.
Being made aware of the directivity index of a speaker can help you decide if it may be more suitable for you and your personal situation and tastes.
There is now a relatively new test signal that has been developed by pro audio company Meyer Sound (see Copper’s profile in Issues 99, 100 and 101). Named M-Noise, its goal is to standardize the measurement of loudspeakers by assessing their maximum power output and the true dynamic potential of musical source material under accurate real-world conditions, while overcoming the discrepancies that can result when using different test signals.
Modeling the dynamics of music is something that a standard pink noise test signal is less capable of doing. Pink noise (a type of wideband noise often used in audio measurements) models the potential bandwidth of audio source material, but does not contain the type of instantaneous peaks at different frequencies that may be present in music and that a speaker may be subject to. Real music is not like pink noise. It is far more dynamic in its range – and power demands.
In particular, pink noise doesn’t have a high crest factor, an indication of how extreme the peaks are in a waveform. The crest factor of music increases at high frequencies. The M-Noise test signal is designed to more accurately represent crest factor, and therefore, music, for a more accurate emulation of how music affects a speaker.
Historically, speaker power-handling capability has been measured at the output of the speaker itself as captured by a microphone set at a specific distance from the driver, usually 2.83 volts at 1 meter for an 8-ohm speaker. What M-Noise achieves is an effective comparison between the input signal and the output at the speaker using a Fast Fourier Transform (FFT) algorithm. A complex waveform (like a musical signal) can, via Fourier analysis, be broken down into the simpler waveforms that make up the complex one. M-Noise uses this technique to break down and convert the audio signal into its frequency components, which may then be analyzed for differences in phase and coherence to detect distortion. This data (the audio output from the speaker) is then combined with the music-emulating M-Noise input signal to draw conclusions about the speaker’s true peak sound pressure level, and how the speaker will perform when used in the field. In performing the measurements, the volume is turned up in specific increments until the speaker begins to exhibit distortion and a deviation in frequency response is noted. Then, a comparison is made against the test’s low-level signal input and the original “undamaged” frequency response.
So, what is obtained is data that is indicative of deviations from the original frequency response when the speaker is played at a high volume, instead of the typical measurement of loudspeaker sensitivity that is obtained from measuring the output of a standard amount of voltage (2.83 V) that is put through the speaker using a microphone at a standard distance (1 meter). The data obtained from M-Noise tests can be quantified and used to predict a loudspeaker’s ability to handle high-power bursts, long-term exposure to various frequencies at high volumes, and optimal and safe operating ranges for a loudspeaker.
What this means is that a given speaker which may, according to its published specs, handle very high power without any physical damage but will distort horribly and produce a distasteful sound under such conditions, can be identified as inappropriate for use at that power rating. Conversely, a speaker may have its optimal practical operating range more accurately mapped, which would enable it to be used according to the real-world demands that will be placed upon it.
Because of the accuracy in identifying the true performance qualities of a speaker’s genuine capabilities, I see great potential in using M-Noise testing to evaluate loudspeakers for consumer audio use. Just imagine how practical it would be to know the true volume capability of your future speaker purchases, with the potential of knowing how long it will be able to play at the highest levels it could withstand before you enter into problem territory. Over in the UK we have to get an MOT test for our vehicle once a year to make sure it is still roadworthy. Similarly, with M-Noise speaker tests, we could be more informed about how “roadworthy” or our speakers are and have a better idea of their true performance.
What the amazing findings have revealed is that once you have a three dB error reading which deviates from the reference signal anywhere on the bandwidth, you are as loud as you should go while maintaining linearity and are entering the realms of unwanted speaker distortion. Going beyond the maximum level by just one or two dB louder again incurs a 90% chance of failure due to thermal overload at the crossover or/and complete amp failure.
Detailed instructions for using M-Noise are available on the Meyer Sound website here. For a brief video explanation of M-Noise, please view the YouTube link below.
If you are interested in testing your own speakers with M-Noise and the REW Room EQ Wizard free downloadable room acoustics software, you can download M-Noise test signals for free and try the test procedure for your own speaker set-up following the link below. For details, please see Adrian Wu’s article in Copper Issue 127.
Postscript: the Audio Engineering Society (AES) has a standard for measuring loudspeakers in professional applications, known as AES2-2012, AES standards for acoustics – methods of measuring and specifying the performance of loudspeakers for professional applications – drive units. While this is outside the scope of most readers, it may be of interest for those readers who would like to dig deep into the subject of loudspeaker measurement.
Header image: 360-degree view of an anechoic chamber at the Acoustics Research Centre, University of Salford, UK. Courtesy of Wikimedia Commons/Daniel Wong-McSweeney.