• Speaker power handling

    One of the many confusing issues in professional audio is that of loudspeaker power handling. On one side, manufacturers use a variety of terms such as peak, RMS, average or program power. On another side, there exist differing methods to determine the power handling of speaker system or component which yield different results. We will try to throw some light into this subject..

    1. Power


    Power is energy per time. It is measured in watts. Power delivered by an amplifier to a load (speaker) is normally determined by dividing the voltage (V) squared by the impedance (Z) :

    Power = -----
    Z
    The resulting type of power will depend on what voltage we use. If peak voltages are used, then the result is peak power. If RMS voltages are used, then average power (often wrongly referred to as "RMS power") is obtained. RMS (root-mean-square), gets us something similar to a mean value from a signal (typically an alternating one, i.e. with negative as well as positive values). More precisely, the RMS (also referred to as 'cuadratic mean') value is equal to the the direct current (DC) that would generate the same average power dissipation into a resistive load.

    2. Power tests


    Para determinar el aguante de potencia (que también podríamos llamar potencia admisible) de un altavoz, se lo ha de someter a una prueba de potencia. Ésta consiste en alimentar el altavoz con señal de prueba, que normalmente consiste algún tipo de señal de ruido con un margen dinámico controlado, durante un tiempo determinado, habitualmente entre 2 y 100 horas.

    La señal de prueba suele ser alguna forma de ruido rosa. El ruido rosa es una señal aleatoria que posee la misma energía en todas la bandas de frecuencia. Por otro lado el ruido rosa no es constante, sino que posee una cierta dinámica. El ruido rosa nos permite de esta forma realizar estudios donde se pone a prueba no sólo el aguante térmico del altavoz, sino también el aguante mecánico.

    El margen dinámico de una señal se expresa con el factor de cresta, que es la relación entre la potencia de los picos y la potencia de la media de la señal. La figura que puede verse a continuación muestra una señal de ruido rosa con un factor de cresta de 6 dB, es decir, que la potencia del pico es 6 dB mayor que la potencia media de la señal. Ello equivale a una relación de 2 a 1 entre el voltaje de pico y el RMS, que corresponde a una relación de 4 a 1 entre la potencia de pico y la potencia media ("rms"), puesto que la potencia se calcula en base al voltaje al cuadrado. Esta dinámica es la especificada habitualmente por la normas internacionales. Antiguamente el factor de cresta de la música comercialmente grabado era elevado (del orden de 20 dBs), pero hoy en día el pop y el rock vienen muy comprimidos con factores de cresta que oscilan por los 10 dBs, pero que pueden incluso acercarse a esos 6 dBs de las señales usadas en las pruebas de potencia de laboratorio.


    There exist several standards that specify power test procedures. The most relevant are :

    2.a. The AES2-1984 standard

    This is a standard for loudspeaker components by the Audio Engineering Society. It is very commonly used, and, although meant for components, it is also often used for individual ways of an active system. It specifies a 6 dB crest factor pink noise signal, with a bandwidth of one decade. For example, a bass loudspeaker could use a 50-500 Hz band, whereas a high frequency unit could use 1000-10000 Hz. The illustration shows the spectrum of both AES signal spectrum examples. The duration of the test is 2 hours, after which the component should not show appreciable damage.


    Comparative spectra for different power test signals as seen by a constant percentage bandwitdh (RTA type) analyzer (Pink noise would show a flat line)

    2.b. The AES2-2012 standard

    The AES2-1984 standard has now been superseded by AES2-2012. Crest factor for the pink noise signal has been increased to 12 dB (4:1) and nominal impedance is now used for the calculation of power. The latter means that the resulting power level for a given loudspeaker component will now typically be around 20% lower than with the 1984 version of the standard (that used minimum impedance for the calculation of power), which may cause confusion. Also, the band-limiting filtering has now been changed to 24 dB per octave (1984 version used 12 dB/octave, as seen on the graph above).

    2.c. The IEC268-5 (1978) standard

    This is a standard by the International Electrotechnical Commission from 1978 and reaffirmed in the eighties. It specifies a 6 dB crest factor pink noise signal over which an IEC programme filtering has been applied. This programme spectrum tries to approximate the frequency content of real music, and shows reduced lows and highs. The illustration compares this spectrum to the AES ones. The terms "Rated noise power" and "power handling capacity" are used.

    Test duration is 100 hours, after which the speaker should not show appreciable damage.

    NOTE: To make matters more confusing, there's another 268-5 standard from 1972 that specifies a different signal spectrum and time, but this is rarely used.

    2.d. The EIA RS-426-A (1980) standard

    This is a standard by the (USA) Electronic Industries Association. The duration of the test is 8 hours, after which the speaker should not show appreciable damage. The signal is also 6 dB crest factor pink noise signal, but with programme filtering that is different to the IEC standard and is also shown on the previous illustration..

    2.e. The EIA RS-426-B (1998) standard

    426-B means quite a deviation from 426-A. The result of this test is not a "power handling" specification anymore but an "optimum amplifier power", which is the maximum input power at which the product under test is rated for acceptability under all three limit categories: a power compression test with a fast variable rate 40-10 kHz sweep sine wave that gets played continuously in a loop, a distortion test and an 8-hour "accelerated life test" with 6 dB crest factor pink noise at half the rated optimum amplifier power and with the spectrum shown on the graph above. None of the tests should result in appreciable damage or change to the unit. The measurement procedure for this standard is rather complex, tedious and subjective at times; at the moment it has not been widely accepted by the sound reinforcement industry and I have serious doubts it will be, with the possible exception of the accelerated life test, which is not significantly different to 426-A except for the wider spectrum.

    3. Types of speaker power specifications


    3.a. Average power. Often wrongly referred to as "RMS" power, since it is derived from RMS voltage readings. Power (which is only positive, since it goes from the amplifier to the speaker, not the other way round) and is already RMS-like, so it would not make sense to apply RMS to it. "Average power" is therefore the right term for a power level that uses RMS voltage for its calculation.

    3.b. Programme power. It is an archaic term that derives from old swept sinewave power tests. Nowadays, it does hold no real meaning. For most manufacturers, it is simply twice the average power, although other manufacturers may use ratios other than 2:1. It may be used as a guideline to the selection of amplifier power. For instance, a speaker with 300W average power and 600W programme power (2x300W) might use an amplifier with 600W output. This is for carefully controlled conditions; for more usual applications with some degree of abuse that amplifier would be too large.

    3.c. Peak power. Corresponds to the calculation of power based upon peak voltages. For a 6 dB crest factor signal, peak power is four times the average power.

    From all the above, it can be concluded that, for power test signals with 6 dB crest factor, the ratios of the three types of power would be as follows::
    Power
    Ratio
    Example
    Average 1 300W
    Programme 2 600W
    Peak 4
    (¡not always!)
    1200W

    3.d. Continuous. Simply specifies that the power signal is applied all the time, since there are some standards that specify intermittent signals.

    4. Causes of speaker failure


    The causes for speaker failure can be either thermal or mechanical.

    The causes for thermal failure are :

    • too much (average) input power
    • excessive power outside the speaker bandpass (radio frequency, subsonic frequencies, deep bass ...). Energy not to converted to sound ends up as heat
    • amplifier clip, the most common cause of thermal failure
    • direct current (DC) at the amplifier output, although this is uncommon in today's amplifiers, which feature DC protection
    • excessive equalization on the ends of the bandpass, mostly high frequencies, since these frequencies exhibit low transducer efficiency and generate lots of heat. Extreme gain settings on ubiquitous basic LF and HF shelving EQ (or the "U" shape in a graphic equalizer) will make thermal failure more likely when the speaker is being driven hard


    To prevent thermal failure, avoid amplifier clip, use LF and HF shelving EQ in moderation, and ensure that the speaker is only receiving frequencies within its bandpass, using high-pass and low-pass filters to limit the frequency content being fed to the speaker.

    The causes for mechanical failure are always linked to excessive diaphragm (cone) movement. The speaker shows greater excursion (backward and forward movement) the lower the frequency. Hence a signal low enough in frequency and large enough in level may cause the voice coil to exit the gap, resulting in the coil rubbing, and possible ending up shorting or opening. The worst case scenario happens when the coil former hits the bottom pole piece ("bottoms out") and gets deformed. To prevent mechanical failure, avoid using signals below a speaker's bandpass, and use an amplifier of the correct power output.

    5. Selecting amplifier power


    In general, the amplifier power needs to be larger than the speaker's rated power. This is because an amplifier only delivers its rated output power with sinewave signal, and delivers much less with a real signal with dynamics.

    As general guideline, it is recommended to use an amplifier delivering 50% more power than the speaker's average ("RMS") power. For example, for a speaker with 450W average power, an amplifier with an output of 700W may be used. If a small amplifier is used, sufficient level will not be reached, nor the perception that it is attained, so the signal will tend to be clipped to compensate, thus endangering the integrity of the speaker.

    In general, the amplifier power needs to be larger than the speaker's rated power. This is because an amplifier only delivers its rated output power with sinewave signal, and delivers much less with a real signal with dynamics. As general guideline, it is recommended to use an amplifier delivering 50% more power than the speaker's average ("RMS") power. For example, for a speaker with 450W average power, an amplifier with an output of 700W may be used. If a small amplifier is used, sufficient level will not be reached, nor the perception that it is attained, so the signal will tend to be clipped to compensate, thus endangering the integrity of the speaker.

    Annex. The amplifier's volume control is not a power control


    A common misconception is that an amplifier's volumen controls allow for adjustment of its output power, which would enable, in theory, the use of an amplifier with too much output power for a given loudspeaker by adjusting the amplifier's volumen controls to, say, 50% (twelve o'clock).

    And that is far from the truth. The amplifier is basically a "multiplier". A signal goes in and it comes out with a voltaje that is X times larger. A power amplifier's volume control is an input attenuator; if we bring it down, we just need to send a larger signal to the power amp from the mixer or processor.