JBL Technical Note - Vol.2, No.1A 电路原理图.pdf

Technical Notes Vol. 2 , No. 1A 1/3 Octave Equalization and The JBL/UREI 5547A and 5549A Introduction: The use of one-third octave graphic equalizers in the audio world has expanded rapidly in the last several years. Along with the expanded use have come a wide variety of different models from many manufacturers. While they may appear similar on casual inspection, they do in fact differ. If is the purpose of this paper to describe some of the important design and performance considerations which are common to all graphic equalizers and to discuss the performance advantages available from the new JBL/UREI Model 5547A Graphic and Model 5549A Room Equalizers. This discussion is not a mathematical treatment of the subject. It will, how- ever provide a better understanding of the subject to those who neither have nor need an extensive background in filter theory. Filter Shape and Combining Action: Any discussion of the important parameters of a one-third octave equalizer must start with considera- tions of filter combining action, filter shape and minimum/non-minimum phase behavior. When we talk about filters which combine we are actually talking about two different aspects of the design. The first is the method by which all of the filters are connected together. In the mathematical sense combining filters are those which multiply while non- combining filters add. Or, to put it another way, combining filters add decibels - non-combining filters add volts. To illustrate why combining is desirable, let us take as an example a pair of filters which, because of their filter shape, happen to cover a certain band of frequencies in common (see Fig 1). Combining type filter summing action (Filters 1 + 2 Filter 1 Boost Curve Non-combining type filter summing action 2 volt (+8.2 dB) 1.23 volt (+4.4 dB) 1 volt (+2.2 dB) Filter 2 Boost Curve Figure 1. Summing Action of Combining and Non-Combining Type Filters Let us say that one of the filters is adjusted in such a manner that the level of signal at one frequency in this common band is raised to 1 volt rms (approxi- mately +2 dBu.) The control position is marked and the control returned to zero. The second filter is then adjusted to make the signal level 1 volt rms at the same frequency. Now the first filter is returned to the previously marked position. What is the output level of the equalizer at that frequency? If the filter is combining, it will add the two levels in dB and get +2 plus +2 = +4 dBu (approximately 1.23 volts rms). If the equalizer is non-combining, it will add the two voltages together and get 1+1 = 2 volts rms (ap- proximately +8 dBu). Rather a difference! A one- third octave audio equalizer should have filters which add decibels for smooth, predictable combining action of multiple filter sections. The second aspect of a good combining filter is that the filter shapes should be designed to achieve the smoothest, most ripple-free amplitude response over the widest range of control settings possible (See Fig 2). There is always a tradeoff between selectivity (isolation between filter sections) and smooth amplitude response. It is possible to design a filter set with a wide range of Qs and variation of Q with control setting. The Qof a one-third octave filter is theoretically 4.3, but few, if any, commercial units actually measure 4.3 except at one setting of the filter. This is because it has been found that to achieve good, low ripple combining action, the Q of the filter needs to change with the amount of boost or cut so as to smoothly blend with adjacent filter sections (see Fig 3). The result is that the filter shape is broader at small amounts of Boost/Cut, and becomes increasingly more narrow with greater amounts of Boost/Cut. This creates minimum ampli- tude response ripple for a wide range of control settings. The low amount of amplitude response ripple keeps the phase variation low. At least one manufacturer of one-third octave equalizers is making a strong sales pitch for their units based on the fact that because of the way their units are designed there is little, if any, change in the shape of their filter curves at any position of the Boost/Cut controls. This is an interesting design choice, but we do not believe it to be well consid- ered. While the design does offer increased selectiv- ity at the one-third octave center frequencies, it simultaneously introduces a greater amount of phase shift and increased amplitude response ripple Figure 2* Equalization curwes with Good Poor Combining Action Maximum Boost, Highest Q Lower Boost, Lower Q Figure a frequency response correcting device should not introduce additional amplitude and phase response errors into the signal path. There may be some confusion on the part of the designers of that device and also for many users who do not understand that one-third octave devices are still broad-band devices and should not be used to perform the functions of a narrow-band device. Narrow band filters such as the UREI Model 562 are more suited to the control of feedback with minimal disruption of the amplitude and phase response of the system. In addition, feedback room modes do not necessarily occur on the exact center frequencies of the one-third octave equalizer. The ability to adjust adjacent filter sections of a good combining filter so that the apparent center frequency of the equalizer is between the ISO center frequencies allows for smooth combining action for any response adjustment. Good combining action is clearly preferable to increased amplitude response ripple and the resulting phase shift. Minimum and Non-Minimum Phase: The correct type of filter for use in an audio system is referred to as minimum phase. This means that the equalizer produces only the mini- mum amount of phase shift as determined by the amplitude response variation. There is a class of filters which has this characteristic. The definition of that class of filters is a mathematical statement about the structure of the filters which very strictly limits their design. (In mathematical terms, the filters do not contain poles or zeroes in the right half-plane of the LaPlace transform, and the log magnitude and phase are related through the Hilbert transform). It is the goal of all electronic components incorporated in an audio system, including loudspeakers and microphones, that they approach minimum phase response. The earliest one-third octave filters were not designed for listening to audio; rather, they were designed by manufacturers such as General Radio and Bruel the only solutions lie in the province of the acoustical designer. For small amounts of non- minimum phase caused response irregularity, it may be possible to correct the amplitude response, but it will not be possible to simultaneously correct for the phase response errors - in fact, correction of the amplitude response will probably increase the phase error. There are also minimum phase response anomalies in room frequency response. These are caused by acoustic filters which modify the fre- quency characteristic of the sound reaching the listener. Some of these minimum phase filters include the low frequency response rolloff due to the size and mounting arrangement of loudspeaker enclosures, the high frequency response rolloff with distance due to excess attenuation of short wave- lengths by air, and certain wideband response irregularities caused by the size and shape of the room and any other acoustical spaces which are coupled to it. It is possible to correct for both the amplitude and phase response anomalies in rooms which are minimum phase by using a minimum phase filter of inverse amplitude characteristics. Therefore we build filters to correct what we can correct, and leave that which we cannot correct to the acousticians. Boost/Cut vs. Cut Only: In one form or another, equalization of sound systems has been around since the thirties. How- ever, it was not until Dr. Paul Boners work in the early sixties that sound system equalization came of age. Later in that decade Altec introduced the hardware that began to make equalization a com- mon practice among sound contractors. These early devices were passive loss networks capable of cut- only action. As active devices became available they initially imitated the cut-only action of the passive units. But soon manufacturers produced units which could boost as well as cut. The question then arose: Which do I use - boost or cut? The answer is that, for room equalization, cut is best, but that with a knowledge of the limitations in its use, some small amount of boost may be accept- able. There are several reasons for the choice of cut-only EQ. First, of course is the realization that when performing the equalization there is a good deal of difficulty in determining which frequency response anomalies are amenable to correction. The natural tendency is to lill in the holes of the frequency response with boost equalization. Unfortu- nately, if the cause of the dip is not amenable to correction as described earlier, then no amount of boost EQ will help. In addition, the effect of boost EQ is more easily heard by the ear and sounds less natural than an equalization curve arrived at by cut EQ. One individual has described the effect of boost equalization as similar to looking out over an empty field dotted with telephone poles. Cut EQ is then similar to the same open field except that the poles have been replaced by telephone pole size/shape holes. The poles are clearly visible, but the holes are not! The analogy is sound. The ear is much more sensitive to the effects of boost EQ than to the effects of cut EQ. To some extent boost equalization may be used to smooth the amplitude response. The amount that 4 may be used is determined by several factors including the type of system, the type of program material, the frequency at which the boost EQ would be used and the sensitivity of the listeners to the effects of the equalization. Generally, more critical program material, better listening environments, and more sensitive listeners are less able to tolerate boost equalization. In addition, the effects of boost equalization are more easily heard (to their detri- ment), when the frequencies being boosted are in the middle of the audio frequency band. The effect of excess boost will be heard as artificial and the filters may ring. The JBL/UREI Model 5549A Room Equal- izer is recommended for the correction of room response anomalies. With 15 dB of cut available at each of 30 bands of one-third octave equalization and separate end cut filters it provides the range of control necessary to deal with the wide range of situations found in both fixed and portable sound systems of all types. Creative Equalization: The one-third octave graphic equalizer is obviously one of the many powerful tools in the repertoire of the creative audio mixer and its use is fairly well known. It offers the ability to shape the audio spectrum in an almost unlimited way while simultaneously presenting a front panel display of control settings that makes it very easy to under- stand at a glance what has been done to the EQ (and what yet needs to be done). The preceeding comments about filter shapes and combining action are of special importance to the achievement of a sound that is not only balanced in tone, but remains musical. At the same time it must be said that when it comes to creative equalization, just as with most things subjective, what works is right. For creative tasks we recommend the JBL/UREI Model 5547A Graphic Equalizer. With a boost/cut range of 12 dB, 30 bands, and separate end cut filters, if incor- porates both the high performance standards and the high degree of control flexibility demanded of a studio quality 8 audio product. Inductors: Wire-Wound or Synthesized? The most common method of designing one- third octave filters has been through the use of series L-C filters which offer better stability and lower sensitivity to component tolerances than most other filter types. In the design of a high quality L-C equalizer, therefore, one of the most important tasks is the design and construction of a series of high quality inductors, at least one for each filter section. Inductors have traditionally been made of magnet wire wound on a core of magnetic material. In recent years the wire-wound inductor has seen considerable competition from an electronic circuit called the synthetic inductor. UREI has manufac- tured one-third octave graphic equalizers using wire- wound inductors since 1972, and one octave band graphic equalizers using synthetic inductors since 1975. With this experience in the use of both tech- nologies, we feel that we have a good understanding of the benefits and the limitations of both. With good design and manufacture, wire- wound inductors may be made to a very high level of quality. As used here the term quality is concerned with the following performance factors: 1. Controlled, consistent Q of the inductor (not . to be confused with the overall circuit Q.) 2. Consistent, precise inductance value. 3. Freedom from distortion caused by core saturation or other non-linear behavior. 4. Long-term reliability. Parts which meet these criteria require good design, materials, manufacture and testing. The resulting wire-wound components do the job, but have several drawbacks: size and weight, particu- larly with respect to the low frequency coils; potential susceptibility to hum field pickup; and high cost. The hum field pickup problem can largely be re- duced by appropriate shielding, but at additional size, weight and cost penalties. The cost of the coils and of the associated precision capacitors needed to produce a high quality one-third octave band equal- izer may approach 75% of the materials cost of the product. This has, and will continue to keep the price of such units out of reach of many users. If the product cost is to be reduced significantly the cost of the L-C networks must be reduced. In recent years several manufacturers have developed one-third octave equalizers which use an electronic circuit to simulate the action of the induc- tor. As compared with using real inductors made of wire, the active inductor (also sometimes called a gyrator), at first glance, seems to offer significant advantages for use in a one-third octave equalizer. It is compact, light weight, immune to the effects of hum fields, the Q and inductance are controllable, 6 and it offers significant cost savings over the tradi- tional wire-wound inductor. Additionally the syn- thetic inductor uses components which are already designed for convenient printed circuit board mount- ing with no additional hardware. The advantages of synthetic inductors must be balanced against their disadvantages. The first is their susceptibility to signal overload. In a series L- C circuit at resonance the voltage across the tot