Making Use Of Limiters and Compressors

Certainly in any discussion of this sort, the first consideration is to point up the basic differences between compressors and limiters. This one area has been the cause for many cases of misapplication, to the detriment of the broadcast signal, with accompanying disappointment to both engineering and management. In Fig. 1 Curve A illustrates a linear input/output relationship, or no compression, whereas Curve C shows the shelving-type compression achieved with typical peak limiting devices. While the compression ratio shown is 10:1, which means that for a 10 db increase in _input signal the output increases but 1 db, compression ratios of 30:1 are not unusual for peak limiters. Peak limiters offer a very fast attack time, typically on the order of a few hundred microseconds, some even faster. Abrupt loud program passages and steep wavefront transients of high amplitude will be caught quite effectively by the peak limiter, and held to a level which, while perhaps briefly audible, does prevent overmodulation and splatter. Thus, there is no doubt about the value of the peak limiter at the transmitter site.

However, the use of such a fastattack, severely-shelved compression characteristic, as a method for increasing modulation density, would be highly objectionable to the listener. The dynamic volume range of music would sound squelched to an unnatural degree. A brass or tympani forte would be completely frustrated by the shelving action of the limiter curve. Curve B of Fig. 1 illustrates a compression curve which, from a listening standpoint, is much less objectionable than Curve C. Note that the knee is less abrupt, and the compression ratio of 3:1 is more gentle in action. As a matter of practical fact, 20 db or more of compression of this type can be used without the listener being aware that considerable compression is in use. Moreover, average modulation level would be increased considerably, with an accompanying increase in effective radiated power on AM and in apparent loudness on FM.

The shape of the compression curve is most important in differentiating a compressor from a peak limiter. To impose the shelved curve on your studio program levels would be onerous; to apply the more gentle compression curve as a means for catching the troublesome program peaks which cause overmodulation would be inadequate. Let us continue a bit further in this vein.t further in this vein. There is another aspect of compressors and peak limiters which is quite important, and this is the release time, or the finite time it takes for the device to recover from a considerable degree of compression. If the device recovers too rapidly, the abrupt upand-down variations in program level produce the familiar "pumping effect," quite objectionable to the ear. When coupled with a fast attack time and adjusted for plenty of compression, a slow release time will chop distinct holes in the audio if triggered by severe level changes such as pistol shots, audience reaction, or an excited announcer. The recovery time of either a peak limiter or a compressor should be adjusted to the nature of the program. But remember that despite fancy names, the recovery circuit is basically a capacitor discharge curve, and the charging voltage is a direct function of program level and degree of compression used. While this obviously offers some degree of program-controlled recovery, there is no optimum release time setting for all program types. 

Most of the peak limiters currently available, or in current use, are listed in the accompanying chart. These units are all designed for rack mounting at the transmitter. The two solid-state designs use quite sophisticated circuitry for at least two logical reasons. The first involves elimination of conventional audio rectification to control the gain of a variable-mu tube. There are no variable-mu transistors as yet! The second is based upon other advances in circuit technology offered by semiconductors.All of the units except one provide considerable gain, which must be taken into account. The studio line is delivering +8 dbm, and the transmitter audio circuits require nominally about zero dbm for full modulation. Therefore, much of the gain supplied by the peak limiter must be dumped, or it will adversely affect signal/ noise ratio and overall distortion. Because physical size is not a very significant factor in rack-mounted gear, the choice of tube-type or solid-state design is a matter of weighing circuit features and convenience controls vs. price. While we are all familiar with the reliability of well-made tube equipment, solid-state devices have a definite edge in freedom from annoying maintenance problems, and their stated characteristics do not change as easily as tube equipment.

Only two of the listed units are supplied for FM stereo in one integrated package. Where two independent units must be purchased, they must be connected so that the degree of compression and the time constants are identical. Otherwise, the stereo effect will be degraded.

Of the various compressors listed in the chart, all but two include an amplifier in the package. Thus, in addition to method of mounting and power supply requirements, you must also consider how to cope with the amplifier gain. Two of the peak limiters listed also appear in the compressor section. They are optionally designed to do a dual job, accomplished by adjusting the several controls as directed in the instruction manuals. Several of the units are designed for rack mounting and are self-powered from a 117v source. If they are wired to follow the usual console output, gain must be dumped, except to make up for compression losses. The maximum output capability of a few is borderline if used in place of the regular console program amplifier. 

Some of the compressor units are designed for plug-in console or rack-shelf mounting, and require an external power source, such as the console supply. Obviously, these units can be readily substituted for the console program amplifier with little problem. If they do not already match the mounting system of the console amplifiers, it is not difficult to accommodate them within the confines of the console shell. Their gain and input level requirements have been made compatible for direct substitution with regular program amplifiers, and they do an adequate job of compression at reasonably low cost. Unfortunately, these units do not lend themselves to stereo use, as they are not normally supplied with a means for ganging compression and time constant characteristics. If identical but isolated units are placed in each stereo program channel, the one which is driven hardest by a program peak will compress more than the other (and will take longer to recover). This tends to degrade the stereo effect at the listening end. 

It is difficult to state the case for the new LDR (light-dependent resistor) compressors without sounding prejudiced in their favor. They have many practical advantages over more conventional compressor amplifiers, and only one significant disadvantage. An LDR is a type of cadmium sulfide or cadmium selenide photocell that greatly varies in resistance depending upon the amount of light which reaches it. This simple component therefore lends itself admirably to applications for controlling gain in an amplifier or in a system. It is easily adapted for remote control. The LDR cell, if properly manufactured, contributes no noise to the circuit in which it is inserted, assuming that its net output level after compression is not so low as to be below the system input noise. Distortion is a function of the nonlinear resistance characteristics of the cells. Typical measurements show 0.55- THD or less at +4 dbm output after 20 db of compression. The distortion is below 0.1%THD when the LDR is inserted at typical internal system points, such as those shown in the block diagram of Fig. 2. Insertion loss is no more than 3 db when looking into a 600-ohm load, and less than 1 db when looking into the non-loading input of some amplifiers.

The single fault with LDR's lies primarily in the light source. An incandescent light has a very definite thermal characteristic which delays activation of the cell. This means that for compression (gain reduction) use it must be classed as a slow-attack device when driven by a conventional lamp. If the lamp is powered by AC, ripple voltages will be superimposed on the audio circuit which the LDR is controlling. Of course, the answer to thermal delay caused by the lamp is to use an electroluminescent source, and this has been done successfully in the Teletronix Model LA-2A. This rack-mounted unit utilizes an LDR ahead of a conventional amplifier, and although it is available adapted for stereo, its gain must be taken into account as mentioned earlier. Although called a leveling amplifier, this unit can also be used as a peak compressor because of its fast attack time. Two other units, the Melcor Model C-20 and the Fairchild Model 663, offer unusual versatility if you consider adapting them to existing systems or designing them into new systems. Attack time of about 15 milliseconds is a function of their incandescent light sources. Because these units have no gain to contend with and virtually no insertion loss, and also because they are packaged for control panel mounting, they are unusually easy to incorporate into an existing system.

in the block diagram of Fig. 2, a typical portion of an FM stereo mixing console is depicted. The two LDR's (labelled Comp. 1 and Comp. 2) are inserted between the master gain controls and the program amplifiers. System levels are indicated, including 15 db of gain reduction due to compression. The common light amplifier is bridged off the output of the program amplifiers through a resistance network to insure proper stereo isolation; at the same time it allows sufficient driving level to accomplish compression. As the two LDR's are included in one compressor module (as many as 4 are possible), each channel will be compressed the same amount, regardless of which has the higher program level, and the common light amplifier provides the same time constant and controls. Because the threshold adjustment is a panel control, we can take advantage of a feature not normally available in many other units. Without disturbing the system levels existing at the point of insertion, the threshold can be adjusted so as to start compressing below this level. If we assume that we are only interested in compressing levels above normal, there will be little need to adjust for compression loss.

The LDR compressors, as with many others in the list, act upon average program content rather than peak program energy. They may be set for some degree of compression at all times. During periods of prolonged levels which fall below the compression threshold, expansion back to normal system gain will take place, depending on the setting of the release time adjustment, a panel control. However, background noise will never be any higher than normal system noise, as there is no extra gain supplied. Adjusting the units to deliver a specified curve is simply a matter of strapping terminals or changing a resistor. The stepfunction curve of D in Fig. 1 is possible for those applications where compression is wanted only over a 6 db range, with return thereafter to a linear gain characteristic.

By raising the gain of the announce mic channel and adjusting the compression threshold appropriately, the program can be made to fade down under the announcer ( sometimes called "ducking"). It will automatically fade back up again after he stops talking. The fade-up time is set by means of the release time control.