JBL Technical Note - Vol.2, No.3 电路原理图.pdf

Technical Notes Vol. 2, No. 3 Applications for the JBL/UREI 7922 Digital Audio Delay Introduction: Digital Delay devices have been around since the mid 1970s. Initially, their high cost and limited capa- bilities restricted the use of such devices to the highest quality professional sound reinforcement systems, where the signal fed to remote fill speakers was delayed in order that people at the rear of the audi- ence would not hear an echo from the main, front speaker array, and would, instead, perceive sound to be localized at the front of the Auditorium. More than a decade later, this remote speaker application remains the primary one for digital delays but, as the available tools and sophistication of the industry have advanced , other delay applications have developed. Certain audio delay devices have been designed solely or primarily to create special effects; such effects signal processors are outside the scope of this paper. The more advanced applications with which we are concerned include: time correction for non-equal arrival time of the acoustic wavefront from dissimilar drivers on either side of a frequency dividing network within a given loudspeaker system, and correction for non-equal arrival time of the wavefronts in the area of overlapping coverage from similar drivers which are mounted on dissimilar horns in a single array. The latter application, in particular, requires very fine delay time resolution as well as the effective elimination of propagation delays from the devices input to its output. This technical note explains the use of the JBL/ UREI 7922 Digital Audio Delay, and its salient fea- tures, as they pertain to the aforementioned applica- tions. Why Time Delay Is Used on Remote Fill Speakers: Lets examine why a typical sound reinforcement system for a theatre, auditorium or church will end up with one or more remote fill speaker, and why the fill speaker(s) benefits from time delay. We begin with a single primary loudspeaker system up front, the main speaker array. Unfortunately, even though this array is mounted very high, near the ceiling, it is almost impossible to achieve the desired intelligibility at seats in the rear of the area without (a) causing deafness in the first 10 rows, and/or (b) causing very loud, but completely unintelligible noise in the rear of the area due to the highly reverberant nature of the sound field back there. Clearly, the best solution is to place another loudspeaker (or several) somewhere toward the rear of the audience (see Fig 1). Unfortunately, when the rear speaker is simply installed as an extension of the main speaker, its sound arrives at those rear seats well in advance of the front speakers sound. Also, since the fill speaker is much louder (due to proximity), the audience now hears a very confusing echo arriving from the main speaker after the sound arrives from the fill speaker. Whats worse, they perceive the sound source to come not from the front of the venue, but from the rear fill speaker. The solution is to delay sound to the rear fill speaker. The psychoacoustic result of delaying sound to the rear speaker, if done properly, is to fuse the sound into a single image which is localized up front. The rear fill speaker contributes the greater percentage of sound level to the rear audiences ears, yet they Sound At Rear 1$ Mostly Reverberant Energy A. With a single front cluster, as the sound level is increased to cover the rear seats, it becomes uncomfortabley loud up front, while reverberant energy prevents a significant improvement in Rear Area intelligibility. B. With a rear fill speaker, the sound level up front can be maintained at a reasonable level, and sound at the rear can be increased to a level where intelligibility is improved without excess reverberation. 2 Figure 1. Why Rear Fill Speakers Are Used In Large Sound Systems Sound Wave Kev Low Level Medium Level High Level Very High Level Ear Shredding Sound Up Front is Reasonable in Level Predominant Sound in Rear is From Fill Speaker perceive the sound to come from the front. This precedence effect, described by Haas 1, is best real- ized when the sound from the rear fill speaker is delayed so that it arrives from 10 to 20 milliseconds after the sound from the main speaker system. More- over, the sound from that rear fill speaker can be as much as twice as loud (10 dB) as the front speaker, yet the delay will create masking which prevents the listener from perceiving the rear speaker as a separate sound source. 1. Helmut Haas, The Influence of a Single Echo on the Audibility of Speech, Journal of the Audio Engineering Society, March, 1972. Reprinted from doctoral dissertation submitted to the University of Gottingen, Germany, under the title, Uber den Einfluss des Einfachechos auf die Horsamkeit von Sprache in December of 1949. How To Align A Remote (Rear Fill) Speaker: Basically, the process is very simple. Set up a sound system with a separate amplifier driving the rear fill speaker, and install a digital audio delay ahead of that rear amplifier. Then determine how much later the sound from the front main speaker arrives at some ideal (or at least average) listening position in comparison to sound arriving from the rear fill speaker, and dial in that amount of delay plus an additional 10 to 20 milliseconds. The trick is determining the correct delay setting. (See Figure 2.) One can calculate the distance from speakers to a given seat using simple geometric manipulation, though in the real world the speakers may not be installed precisely where they were specified to be installed on paper. Then, again, you could run a tape measure, but that is not necessarily practical, either. Besides, there is a very insidious factor that is seldom Main (Front) Speaker Array Remote (Rear) Fill Speaker Power Amp Equation for estimating Rear Delay Time X in Milliseconds (Distances D given in feet): X (ms) = (Di-D2) +(10 to 20 ms) 1.13 0 ms Audio Analyzer Capable of Time Measurements Pulse Generator Mic Preamp This may be a single piece of test equipment, as made by Ivie (IE-17) or Crown (Techron). Figure 2. Use of A Digital Audio Delay For Fusion of The Image in A Sound System With A Rear Fill Speaker Power Amp Measurement Microphone at Ideal Listening Position D 2 Graphic Equalizer Graphic Equalizer X ms Digital Delay Mixer 3 discussed in the literature; the actual path which sound travels from speaker to listener is not usually a straight line. Due to temperature gradients indoors (as well as wind gradients out of doors), the sound may take some undefined curved path. Therefore, for optimum tuning, it is preferable to measure the differ- ence, in arrival times, using calculated times only as an initial guide to delay setup. How can the difference in arrival time be meas- ured? First, connect the sound system normally (with the Digital Audio Delay Bypassed or out of circuit). Verify correct polarity through the system. Adjust all, crossovers, equalizers, and amplifiers for the best possible sound. Then youre ready to make measure- ments and time adjustments. Commercially built test equipment made by several companies is available, including the Techron TEF (Time/Energy/Frequency) analyzer from Crown, the Ivie IE-17 Sound Analyzer, and various dual-channel FFT (Fast Fourier Transform) analyzers. 2 With such equipment, it is possible to inject an impulse (a click) into the amplifier that drives the main front speaker(s) and read out the time interval between the moment the impulse is generated, and the moment it arrives at a measurement microphone at our ideal listening position (actually, youll want to evaluate several positions before you learn which is typical). Then, with the time delay unit bypassed or physically removed from the circuit, inject an impulse into the amp that drives the rear fill speaker(s), and measure the time interval between impulse generation and arrival at the test mic. Simple subtraction gives you a value which you can use as the basis for initial setup of the Digital Audio Delays output; add 10 to 20 milliseconds to this value, however. By the way, the measured time difference between the two uncorrected speakers should be something like .885 milliseconds per foot distance in the sound paths from the two speakers. 3 Once you have dialed in the estimated time delay, it will be necessary to do some listening tests to determine the optimum delay. Initially, try using a repetitive clicking sound. You can use the output of a phase checker (though youre not using the measure- ment capability of that device). You can also have someone tap a drum stick on a wood block at a stage microphone. Just get a sharp leading edge waveform that you can listen to, and then experiment with slightly different delay settings until you hear one precise, loud click instead of a smeared double click or a too-fat clock. When the correct delay is achieved, the sound 2. TEF is a trademark of Techron (a division of Crown Interna- tional), and Techron is a trademark of Crown, International. 3. The speed of sound in dry air, at sea level, at 59F, 29.92 inches of mercury atmospheric pressure, is 1128 feet per second, or 1.13 feet per millisecond or 0.885 milliseconds per foot. However, sound travels more slowly in less dense air. This means that above-standard temperature, humidity or altitude, as well as below-standard atmospheric pressure, will slow down the sound and throw off time delay values based on simple calculations. should appear to come from the main, front speaker, and there may be an increase in apparent loudness. Finally, play a variety of music through the system (unless it is strictly for speech reinforcement, in which case youll want to talk the system). Select music with a lot of transient attack since a continuously bowed violin or a long organ note will give you no real means to evaluate the delay setting. The optimum delay time will almost certainly be from 5 to 30 milliseconds greater than the actual wavefront arrival time differ- ence between the speakers, but practical experience shows that the range of 10 to 20 ms extra is most common. When youve got it right be sure to check a few other seats, and then make a note of the settings for ease of recovery in the event the equipment is later misused or removed for service. NOTE: Some people have suggested using an oscilloscope to visually identify the spikes associated with arrival of the impulse sound from the main and rear fill speakers. However, this is a very uncertain method. Its difficult to get a sharp enough leading edge to see what youre doing. Based on our conver- sations with several acoustical consultants, we think the approach presented here is more realistic and practically achievable. Details of the JBL/UREI 7922 in Rear Fill Applications: The 7922 has two modes for delay time adjust- ment, high resolution and normal. In the normal mode, the delay time is adjustable in 1 millisecond increments, which correspond to a distance of about 1.13 feet. This should be sufficient for most rear fill applications, and the normal display mode makes it easier to slew through the available 327 ms range to the desired delay time. However, very fine increments of 10 microseconds are available in the high resolution mode, and you may wish to experiment with this mode for perfectionist sound system setup. The maximum delay time of 327 milliseconds corresponds to a distance (standard atmosphere) of 364 feet. However, if one allows some 20 millisec- onds for additional delay to the rear fill speaker (for best Haas precedence effect), then the maximum distance differential between acoustic paths is about 346 feet. Bear in mind that this distance is not the distance between the two sets of speakers; rather it represents the difference in acoustic paths between the speakers and the listener. There are few installa- tions where the rear speakers even come close to this maximum distance, so the 7922 should be useable almost everywhere. The JBL/UREI 7922 has two independent delay outputs, and therefore it can be used to delay the sound to two different sets of rear fill speakers. For 4 example, one rear fill may be in the middle of a venue, and the other under the balcony. In these cases, the procedure for setup is about the same; each rear fill channel is adjusted independently, against the front channel. Then the entire system should be turned on and checked again as the fill speaker delay(s) may require touch up of the delay time in the event any transient smearing or echo develops due to overlap of delay zones. Time Correction of Drivers on Either Side of A Crossover: Any loudspeaker system which utilizes two or more types of drivers operating in different frequency bands will have a crossover region wherein different drivers reproduce the same signal. If the acoustic centers of these drivers are not precisely aligned, then the wavefronts they produce will arrive at some distant reference point at different times. Instead of construc- tively reinforcing one another, there will be varying degrees of destructive cancellation at different points on and off axis and at different frequencies. The resulting comb filter or phasing effect significantly degrades the audio signal quality. What if the drivers are brought into proper alignment so that the wavefronts from the high and low frequency drivers (in the crossover region) simultane- ously arrive at a reference point in front of the speak- ers? Objectively, you can look at an oscilloscope, TEF analyzer, or the like, and see the dual-peaked display of an impulse merge into one peak. Subjectively, the sound becomes much more distinct. Imaging is significantly improved, and there may well be an apparent increase in sound level of several dB. Do you need to do this? Well, anyone who has aligned a speaker system in this manner will tell you they dont want to go back to listening to a non-time corrected system. The average listener can hear the difference when time correction is switched in and out, and can readily tell you which is the better sound. Given that you want such a correction, how do you achieve it? The apparent solution is to align the acoustic centers of the drivers. This is easier said than done. Generally, the drivers are physically locked in place by constraints of the mounting. If the drivers can be moved, physical manipulation may be awkward, especially when youre dealing with an array which is hung from a ceiling or other relatively inacces- sible location. Even if you have an ideal workbench situation, where you can physically offset the drivers with great accuracy, and you actually know where the voice coils are located, the effective acoustic centers still may be incalculable. Another factor. not all the time error is caused by the difference in acoustic centers between the drivers; there is phase shift assoc