Drivers (Part 3)
The third instalment on Drive Unit speaker protection.
Last month, we looked at some of the disparate ways in which a speaker can come to grief. Knowing the mechanics behind dead speakers is simply not enough - we need to see how the limitations relate to using speakers safely in practice.
By far the easiest way of putting this information across is in the shape of a graph. To the best of our knowledge, the only professional monitor manufacturer who freely publishes 'maximum output' curves is Electro-Voice. The maximum power handling curve for any speaker is not the straight line you're lead to expect, so presumably this data is thought to be too embarrassing by the other monitor manufacturers...
The graph in Figure 1, shows maximum acoustic level (or 'welly' as it's known North of Cambridge) versus frequency for a typical monitor speaker. To put this curve into perspective, it's loosely based on the Electro-Voice Sentry 100A monitor, which has a 25 watt tweeter, well above the 3 to 10 watt rating of most other small monitors. Thus the high frequency section of the curve is considerably above average. At the bass end, meanwhile, the 8" driver has a long-throw cone, which is not far removed from the LF performance of some 12" and 15" units.
To see this graph in the context of other bass/mid drivers then, the maximum output curve below 100Hz is below par in contrast to 18" monsters, but above average when compared to a 6½" driver, for instance. Indeed, with smaller drivers, or indeed larger ones with limited mechanical excursion capabilities, the frequency at which excursions come into effect will be well above 55Hz, which is the frequency where the Sentry 100's excursion limit takes effect.
Looking now at the regions on the graph in detail; in Region A (below 55Hz), the maximum bass level is governed by things mechanical. Any excessive signals here will cause the cone to hit the end stops and/or rip long before any thermal overload arises. For the tweeter, similar limitations apply (Region C), but the crossover's high-pass filter normally guards this driver from over-excursion.
With this knowledge, it's easy to see how the crossover can influence a tweeter's maximum (safe) acoustic output, especially if one is trying to bring it in at the lowest possible frequency. Using a steep (-18 or -24 dB/octave) slope enhances power handing by curtailing over-excursion. Conversely, we can infer that if the manufacturer has chosen an inappropriate crossover and/or too low a crossover point, tweeter failures are largely down to overstressing the mechanics rather than burnout.
High-pass filters for the bass driver, are of course, invariably absent in conventional speaker systems with passive crossovers, but they're commonly encountered in active systems in the guise of 'sub-bass filters', 'bass-bin filters' or 'sub-sonic filters'. The upshot of this is that when set up correctly, these filters can largely prevent over-excursion, but at the cost of losing out at the bass end. After all, we only need to curtail the low bass on the special occasions when the drivers are reaching their mechanical limits; low bass and high levels are not harmful in themselves - it's the combination of the two that proves destructive!
In Regions B and D, the limits on acoustic level are thermal. In other words, the only way you can damage the driver is by overheating the voice coil. Given that things take a while to heat up, it follows that for short (but musically significant) bursts of sound 10 to 50 milliseconds in duration, considerably greater output is possible - up to 10 times the mean power, as a general rule (Region F). It's this fact that allows us to use nominally overpowered amplifiers with impunity.
To place this main graph into perspective, we ideally need to view it against the spectral content of the music itself; that's a graph showing the energy/power at every frequency. Every music is different though, so the best we can do is to make two general observations: 1) Around 70% of the power in music arises from fundamentals and is concentrated in the midband, between 100 Hz and 1 kHz. In this region (B), the main limit is thermal, and there's plenty of headroom here (region F).
2) Short term signal levels below 100 Hz and above 1 kHz are frequently as big (or bigger than) the demands made in the midband. Manic percussion, for example, can unleash enormous energies at spot frequencies. Averaged over a period though, the power involved is relatively small (Figure 2a). Thus thermal overload - which is down to excessive power, rather than bursts of energy should be less of a problem here than mechanical over-excursion (especially in the low bass unless you overload your amplifier - or are using limiters. This takes us on to the special problems surrounding tweeters.
On same tape machines, it's possible to rewind or fast-forward the tape with the electronics in the playback mode. If you do this without first muting the monitors, the result may be a "curl of smoke where the tweeter used to be". Another day, a particularly nasty bout of CB breakthrough can have the same, sad result. It might also 'take out' your amplifier. For the tape machine, the answer is that in the RW/FF mode, the entire spectral content on tape is shifted up several octaves. So the tweeter dies because it's overwhelmed by the high power content that properly belongs to the midband region. For CB - or any other powerful Radio garbage, the cause is subtly different; it's a downwards translation from RF to relatively low frequency intermodulation products, but the result, a vapourised voice-coil, is much the same.
In domestic speakers, tweeters are grossly underrated as a rule. Providing the spectral distribution is normal, with an average power of a few watts, including occasional high energy bursts, all is well, but should this happy relationship be altered, there is little leeway, and burnout becomes inevitable. Looking at clipped signals from a new angle, overload can also be seen as a special case of altered spectral distribution, with midband power again being translated upwards into regions where it does not belong.
Ironically, the limiters intended to protect your monitors can also precipitate burnout. When the system is overdriven, limiting causes the difference between the peak (short term) and mean levels (long term), (typically 10 to 15 dB without any compression) to close in to 6 dB or less, and so the power dissipation soars dramatically. Figure 2 displays this. Just as with clipped waveforms, the area under the graph shows us how power, dissipated in the tweeter, rises.
Ideally, we need to know the true continuous sine-wave or pink noise capability of the speaker in the midband. Sadly, only a few manufacturers give this data, for instance, ATC and Electro-Voice. Despite this, it's good enough to know that the order-of-magnitude is around 20 to 70 watts for most small monitors. Using amplifiers with up to ten times the speakers' long-term power handling (ie. 200 to 700 watts) is then perfectly safe with three provisos: a) The amplifier must never be driven into overload for any length of time (even marginally). Use an accurate LED peak meter (See November 83 HSR, for a DIY project) or a true clipping indicator to guide you. Calibrate the LED metering so it enters the red region 1 or 2 dB below the amplifier's quoted input sensitivity. If the speaker's impedance dips below 5 ohms, a 3 dB safety margin is a good idea. Much depends on by how much the clipping threshold drops when the amp is driven into a low impedance. This information is often stated in our amplifier reviews.
b) When the music has significant low bass content, the speaker cones may reach their excursion limit well before clipping. This is always audible; every drive unit has a characteristic 'You are hurting me' noise. Don't ignore it. Either turn down the level, or take off some of the bottom end, using the EQ on your mixing desk.
c) If the speaker's power rating has been wrongly assessed, (or guessed) your amp may be too powerful. If this is the case, all will be well, providing you listen out for the point where the quality of the sound changes. Sometimes it is gradual, with other speakers the change is more dramatic, but in all instances, you should be listening for sound colouration, distortion and compression. Basically, if it hurts your ears, it is probably hurting the speaker. Beyond this point, you risk thermal overload, assuming the LF excursion limit hasn't been reached beforehand.
If you do not have accurate peak metering (VU meters are hopeless), use your ears. Clipping is another name for 100% distortion, so you don't need to be overly perceptive to know when it's happening. If your system is underpowered,it will be especially tempting to drive it close to overload, because distorted sound is subjectively so much louder. Buying a suitably powerful amplifier is cheaper in the long run (after you've zapped a few drivers), but remember that using more efficient speakers is an equally valid approach. For example, starting with 91 dB @1 watt @1m speakers and a 100 watt amplifier, you can attain a (useful) 6dB increase in SPL capability either by (a) moving up to a 400 watt amp, or (b) going for speakers with a 97 dB @1W @1m sensitivity.
Feature by Ben Duncan
Previous article in this issue:
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!