Is there a better way to set limiters?: Using some methods for setting peaklimiters is like playing Russian roulette with your equipment.
Jul 1, 1997 12:00 PM, Tom Walker
"I just know how to do it." Set ting limiters, it seems, is just one ofthose things you know how to do. It is hard to explain to someone else, andheaven only knows how you learn to do it in the first place. The realquestion is, should it be this way?
The purpose of peak limiters is to let the system operate at maximum outputbut not go over. In other words, we want to protect the loudspeakers fromamplifier clipping and other damage so it is reliable. Yet at the sametime, we want the maximum sound-pressure level possible so we do not have alot of excess equipment and its associated cost. It's a fine line.Depending upon the circumstances, such as renting out equipment to be usedby others, you may want to err on one side or the other.
So how do you set limiters for your systems? How do other people setlimiters? In asking these questions of a great number of people, wereceived some interesting answers a few years ago. Responses that set usthinking.
Method 1: Play the system slightly above normalThis is the method sometimes recommended in the manuals of limiters. Itgoes like this. Play the system at a level slightly above normal. Now,adjust the limiter so that the threshold LED just lights during peaks inthe program material. In many ways, this is a sensible method if yoursystem is well designed and has been reliable. Also, it should be the samesystem every time - you and you alone are the operator. The final conditionis that you are not concerned about toting around too much equipment. Notonly does this method ignore the limits of the equipment, but these are agreat many conditions to meet. In the real world, few systems fulfill allof these conditions in even a small percentage of the performances. If youcan not meet the requirements, then using this method is playing Russianroulette with your equipment.
Method 2: Use amplifier clippingIn brief, this method can be described as no amplifier clipping, ever. Hereyou adjust the limiter based upon the gain of the amplifier and its maximumwattage output. You can find the value of that adjustment by calculatingwhat the maximum amplifier signal input should be. Here is an example ofhow to calculate what signal level would cause amplifier clipping.
Loudspeaker nominal impedance: 4 V
Amplifier maximum power: 400 W into 4 V
Amplifier gain: 26
To find the amplifier maximum voltage, take the square root of theamplifier maximum power multiplied by the nominal impedance:
sqrt(400 Wx4 V) = 40 V
To find the input voltage limit, divide the amplifier maximum voltage by gain:
40 V/26=1.5 V RMS
Therefore, the dBu of the signal into the amplifier is:
20 log (1.5 V/0.775 V) = +5.7 dBu for amp clipping
If you do not want to do the calculations or all the numbers are notavailable, you can believe the clip lights on the amplifier. Simply run upthe signal level until those clipping lights come on and adjust the limiterso that the threshold LED is just lighting about the same time.
There are some good reasons for using this method, chief among them beingkeeping any clipping distortion out of the music. However, with thismethod, you are letting the loudspeakers fend for themselves. Back someyears ago (and it was not that long ago), when a 50 W amplifier was stateof the art, this no-clipping method was much safer for the loudspeaker.Today, when amplifiers are so powerful, this procedure is without a doubtdangerous and sure to shorten the life of even rugged modern loudspeakers.
Here, we should mention why we are concerned about putting too much powerinto the loudspeaker. Or to put it another way, what is power compression(also sometimes called thermal compression), and why is it a problem?
You want a loudspeaker to convert electrical energy into acoustic energy.At low power levels, most loudspeakers do a fairly good job of energyconversion. As power increases however, they do not do as good a job andthe corresponding rate of increase of acoustic output increases moreslowly. At some point, there is a significant difference between theelectrical power into the loudspeaker and the acoustic power out since theefficiency is not as good as at lower power levels. That is, at higherlevels, the loudspeaker is less efficient and into power compression. Thedifference in power in (electrical) vs. power out (sound) must gosomewhere. Most of it is turned into heat, which is bad enough, butunfortunately that heat is in the loudspeaker voice coil. Anyone who hascaught a loudspeaker on fire knows about this problem and never forgets theexperience.
Consequently, the reason power compression is a significant problem is theheating of the loudspeaker voice coil and the wasting of power. Modernloudspeakers are very tough. The loudspeaker engineer and consultants havedone a great job with new glues, formers and methods. In fact, loudspeakerengineers and manufacturers have done a remarkable job, in general, overthe last 10 years. However, we do not want to test their work or wastepower into something (heat) different than what we are trying to do (sound).
Back to the just-short-of-clipping method of setting limiters. Now that weunderstand that this method can be fatal for the loudspeaker and its voicecoil, our natural tendency is to adjust the limiter very conservatively.Conservative means more reliable. But conservative also means not usingthose modern, incredibly powerful amplifiers to their full potential. Ifset conservatively, you have more expense in extra equipment and theincreased expenditure of moving that equipment. It would cost much less ifyou could operate safely and reliably closer to the limit of every piece ofequipment in the system.
Method 3: Use published specifications of the loudspeakerThe previous method centered on the amplifier. This one focuses on theloudspeaker. Here we use the published specifications of the loudspeaker tocalculate the safe level that we can pump into it. This is, in reality,pretty hard to do.
There is no loudspeaker equivalent of the clipping light on the amplifier(unless you consider that half-second distress scream before a loudspeakerfails to be the equivalent). It is likewise somewhat difficult to verifywhat power the amplifier is really putting out at different input levels.Couple this with the fact that specifications are at static impedances,such as 8 V, and loudspeaker impedances vary with frequency, making thecalculation more difficult. If this is not complicated enough, theenclosure of the loudspeaker acts as a filter network, which means thespecifications (which are often for free air) may not apply to yourparticular case unless you are using a box from the loudspeakermanufacturer.
Then there is the issue of the specifications from the loudspeakermanufacturer. Often after a loudspeaker (or any other product) has been inproduction for a number of years, changes are made in the materials,production methods or production location. Experience shows that the newermodel is usually better than the older one of the same model number. But weall know, too, that specifications are written to put the product in themost favorable light. And those tested conditions may or may not be thesame as what you are experiencing.
The consequence of all this is simply that the method using calculationsbased upon the loudspeaker specifications is a process with a great numberof variables. And the only way to make the calculation is to assume answersto some of those variables. Assuming is another name for guessing and thatmeans we are probably not reliably operating at the limit of the equipment.
Method 4: Actually measure the onset of power compressionAll of the previous limiter-setting methods have the disadvantage of beinga destructive test until you get it correct. They are also not precise andrepeatable in their methodology, and they do not take the venue or acousticenvironment into account. In this modern age, we certainly should be ableto do better than what boils down to using experience for quality guessing.Unquestionably, too much money is tied up in equipment for the success ofthe sound system to be based on guesses, no matter how wise.
The perfect answer would appear to be testing the actual loudspeaker andamplifier system in the venue where it is going to be used, or at leasttesting the loudspeaker in the enclosure it will be operating in as opposedto relying upon data that could be for different conditions.
The ideal test would seem to be one that could be re-run again with thelimiter operating to verify that the peak limiting is active where youthink it is. Think about that last statement for a minute. For yourloudspeakers, the peak limiter is equivalent to a rigger's safety harness.A rigger wouldn't suspend himself four stories above a concrete floor ifthe harness system were put together with guesswork, and a sound-systemdesigner shouldn't trust limiter settings based on guesswork either.Wouldn't it be nice to know your safety equipment will function as intendedbefore you entrust your life or your loudspeakers to it?
To take the guesswork out, AudioControl Industrial has developed an audioanalyzer that actually measures the loudspeaker and the onset of thermalcompression while monitoring the amplifier output. To see how thisanalyzer, the Iasys (pronounced like "I assist"), solves the problems ofother methods just discussed, let's first understand how it uniquely tests.
The Iasys analyzer connections are depicted in Figure 1. They consist of amicrophone that is placed on-axis in the coverage pattern at a typicaldistance and is then connected to the input of the analyzer. The output ofthe analyzer goes to the amplifier with the gains of the amplifier set atmaximum (full power). The amplifier output goes to the loudspeaker undertest with a "Y" connection to the analyzer loudspeaker level input. Prettymuch the type of hookup you would expect.
With these connections, the Iasys analyzer can monitor the sound-pressurelevel from the loudspeaker, watch the amplifier output for clipping andcontrol the level and other parameters of the test signal. To put thisanother way, with these connections, the analyzer can increase the signalto the loudspeaker and literally measure, not guess at, the onset of powercompression. Since the built-in, fast microprocessor is constantlymonitoring the test, the test halts if there is any loudspeaker distress oramplifier clipping prior to power compression. Measuring the loudspeakerallows the Iasys analyzer to make an exact recommendation for the limitersetting. That recommendation is in dBu of signal to the amplifier basedupon actual performance, not assumptions. See Figure 1 for an example ofwhat the test results from the analyzer will be.
This testing is automatic because the analyzer does the setup. Iasysadjusts levels, picks test frequencies and checks that connections are madeproperly. To do this automatic setup, the Iasys measures the backgroundnoise level, sets its microphone pre-amp gain and regulates the test signalaccording to the background level. During this process, the operator can godo something else, including get a cup of coffee. Then the Iasys conducts aprerequisite test of the loudspeaker's ability to reproduce differentfrequencies. After that test, the self-contained analyzer knows whatfrequency is appropriate, and safe, for the limiter recommendation part ofthe test. It then continues its automatic operation, while the operator isstill finding that cup of coffee, and runs the test for the onset of powercompression. The entire procedure takes about five minutes, less if testingmultiple units of the same loudspeaker.
After receiving the recommendation for setting the limiter, it is easy toverify that the limiter unit works as intended. You add the limiter to thesignal chain before the amplifier and adjust the limiter to the settingsrecommended by Iasys. Actually, you probably should adjust the peak limiterto settings slightly less than recommended to take into account the "knee"built into most contemporary limiter models.
This second test, with the limiter active in the signal chain, allows youto test particular loudspeakers' actual performance. It also allows you totest the real-world system performance of the limiter, amplfier andloudspeakers together. You know your system is operating at peakperformance. You also know you aren't using too much or too littleequipment and that the system is operating at a safe and reliable level.