Compressor/limiters may look similar on the surface but the difference in the way that any two models perform is most pronounced.

Most studio engineers would agree that a good compressor/limiter comes second only to reverberation as an 'essential' processor, and that no modern studio is really complete without one. Why is it so important? Well, the compressor/limiter serves three main functions. Firstly, it enables the wide dynamic range of most acoustic instruments and voices to 'fit onto' recording media of limited dynamic range. This was its main function in the early days of recording, when tapes and records could only manage a 50-65dB dynamic range. If an orchestral performance covered an 80dB range, there was nothing for it but to turn up the quiet bits and turn down the loud bits, while recording. Originally done by the engineer's hand (and known as 'gain riding'), the job was taken over by the earliest compressors. In this context, the aim of compression is effective dynamic range compression in the least aurally obtrusive fashion. With modern tapes, noise reduction and digital audio, this application is less necessary than before. Compressors used as peak limiters are still valuable in PA use for speaker/amplifier protection and in broadcasting, where transmitters must not be overloaded.

The second application can be broadly described as dynamic range control and modification. Here, compression balances the characteristics of different instruments in a recording to achieve a solid and cohesive sound. Thus, vocals are always compressed to greater or lesser extent, depending on the singer's own dynamic range and microphone technique. Uncompressed vocals tend to bob up and down like a small boat in heavy seas; words get lost and the whole thing becomes unsettling to listen to.

Uncompressed bass guitar can sound lumpy and boomy so a good compressor can help to iron things out here too. This sort of corrective control is where a studio compressor earns its keep every day.

The third operational context is that of evident effects. Heavy compression alters the characteristics of individual instruments, giving guitars extra sustain and sparkle, bass a clear 'solid' sound, and drums tightness and impact. Processing entire mixes increases their subjective loudness by raising the average power as indicated on a VU meter, while controlling peaks. For this reason, AM radio stations invariably compress program material by 10-15dB in order to make it sound as loud as possible. A variety of complex frequency- and program-dependent units have been specifically designed to do this without adversely affecting the overall sound quality.

Compressors have lots of uses other than those mentioned above, but this being a review and not a book, I recommend you read your back issues or wait until the subject crops up again.

Compressor Technology

Being level-dependent gain control devices, compressor circuitry naturally includes gain control and level detection circuitry, both of which affect the 'sound' of the device. For specification fans, gain control VCAs inevitably come in for most attention, since they can be measured like any audio circuit and until the DBX chip arrived, most had indifferent performances. The earliest versions used photocells as gain control elements, which at least had low noise and distortion, but were slow to respond and highly non-linear. Valves can vary their gain according to anode current and valve compressors were, and still are, valued for their 'warm' sound (due to generous level-dependent, even harmonic distortion). Many compressors have been designed around FETs used in their low-level guise as variable resistors, but apart from their dynamic range difficulties, FETs are awkward to linearise and match for stereo.

The fourth and most recent major class of VCA is that based on transistor transconductance. Transconductance amplifiers such as the LM13600 actually work by using the logarithmic input/output characteristic of a transistor pair at a very low level, where it approximates linear behaviour. The approximation fails (ie. distortion appears) above about 25mv p-p input, so these devices have limited dynamic range. However they are inexpensive, easy to use and with careful design can perform quite well. 'Proper' VCAs all use some variety of multiplier, whereby a signal is converted to logarithmic form, added to a constant (the control signal) and then antilogged. Maths fans will recognise this as multiplying in logs, and since transistors can follow a log law over six decades, dynamic range is not such a problem. These devices have a further advantage in that the control law is very close to logarithmic, that is dBs per volt, which suits subjective loudness control nicely. Originally constructed from arrays of painstakingly matched transistors, these VCAs are now available in integrated form. Devices varying in detail but of a similar order of performance are available from DBX, Valley People and of course Aphex.

Control Circuitry

Less clear-cut and more subjective than VCA performance is that of the control circuitry. The purpose of this is to derive a VCA gain control signal based in some way on the input signal and 'in some way' means 'according the designer's judgement', (and hopefully on the users' needs). Without repeating the principles of attack, release, threshold and ratio controls, it is worth noting a few special points. Attack and release times are adjusted only by front panel controls on basic compressors. Some do incorporate program-dependent time constants (the Auto setting) where a slow-attack slow-release time constant is added to a fairly fast primary one. In this way, the compressor constantly applies some gain reduction according to the average program level while still being able to limit sudden peaks quickly. This gives a smoother effect with the minimum of 'pumping' which is suited to complete mixes and radio transmission of records.

One bugbear suffered by almost all compressors is their habit of 'dulling' sounds. Three factors operate here. When used as transient suppressors they must lose gain during the transient, thus lowering the rate of change, hence the HF content of the signal. Secondly, with combined or wide-range instruments, large LF signal components cause compression, hence gain reduction of the HF end whether needed or not. Thirdly, both pitch detection and frequency response in the ear is level-dependent. Sophisticated units such as the Aphex Dominator split the audio range into several bands and treat them separately, and even offer 'harmonic resynthesis' (a form of Aural Excitement) to restore brilliance. Simpler designs modify the side-chain frequency response, provide external access to insert side-chain EQ, or make the sidechain sensitivity and response speed program-dependent. Nevertheless, achieving that 'zingy' acoustic guitar or piano sound usually needs extra-large helpings of top boost. This naturally brings us to the question of noise. Remembering that every dB of added compression corresponds to a dB lost in signal/noise and so many compressor/limiters incorporate a noise gate or low-level expander. Those which simply offer a threshold control for the gate are however not tremendously useful. An effective gate has even more controls than a compressor.

To the engineer, one essential aspect of a compressor is its ease of use. Compression requirements can vary widely. You may want to slightly compress (using perhaps a 1.2:1 ratio) over the whole range of inputs for piano recording. For PA use, you may want no compression below almost maximum output, followed by hard limiting to protect the system. To optimise operating conditions, some compressors therefore offer input gain, output gain and threshold controls while others have only variable input gain. The more knobs, the more versatile (but generally more difficult to operate) the system becomes.

The final operational aspect of compressor/limiters is the effectiveness of their gain reduction indication. This can vary from a single glowing LED (anyone remember 'magic eye' indicators?) through meters to bargraph displays. Comprehensive displays are more costly and, up to a point, more helpful.

The CL52

This Ashly compressor is housed in a steel box, the dimensions and colour scheme of which needs no mention though it does have trimmed front corners to prevent clumsy users scratching things. Pulling it to bits revealed quite solid construction. Six boards (two VCA, two meters, one PSU, one input/output) are interconnected with ribbon cables and powered from a generously sized copper-shielded transformer. The controls are very solid, using brass-shafted potentiometers and Allen-keyed knobs. The case is filled with foam sheets which is very sensible; one doesn't really want dozens of clanking cases resonating in one's expensive control room. It's a pity more manufacturers don't adopt this feature. The VCA chips used are DBX type 2150.

Returning to the front panel we find an ample but not excessive number of controls. Each channel has input gain variable from -30 to +30dB and an output level control calibrated from to +18 dB, which indicates output voltage (ie. dBu) at full limiting. The internal side-chain threshold is fixed relative to the VCA operating level, but since side-chain insert sockets are fitted to the back, you can alter the side-chain frequency response and gain, therefore threshold, as you wish. Attack and Release controls both offer a sensible range, covering 0.2-20mS and 0.1-23 respectively. The compression ratio control is calibrated from 2:1 to ∞:1. This I found a bit disappointing since ratios around 1.5:1 are very useful. After all, DBX noise reduction only uses 2:1 compression (admittedly over the whole dynamic range) and that isn't subtle. Judgement was however reserved for a listening test.

Apart from the knobs, four very solid grey push-buttons did the honours as channel 1 and 2 In/Out, Mains On and Stereo Link. No less than 48 LEDs adorn the front panel (that explains the large mains transformer): one for each switch and eleven for each meter. Four meters? Yes, both gain reduction (2 to 20 dB) plus a 'threshold' indicator and output level (-18 to +9 dB) plus a +20 dB 'clip' indicator are shown simultaneously for each channel. Due to a sensible colour scheme, very useful they are too.

In Action

Connection via ¼" unbalanced jacks is easy enough, though you will have to lop off the American 3-pin plug and contend with American wiring colours. Black is live and that's easily remembered as it is the colour you go if you touch it.

Immediately evident is that the In/Out switches are not bypass switches. In fact they merely disconnect the sidechain, leaving input and output controls and level meters operative. In the 'out' mode, no audible degradation could be detected using a direct condenser mic source, which is to be expected from the high technical specification.

In operation on all sorts of musical material the unit acquitted itself very competently. Both Attack and Release controls were predictable and smooth, allowing very fine tuning between the impact and ring of a bass drum, for instance. On complete mixes a significant increase in loudness was obtainable without much tonal alteration so long as fairly long time constants were employed. One special feature of the unit was valuable here; it's virtually impossible to overload. At high input levels, the compression ratio increases and the attack time decreases, automatically. Relatively slow attack times can therefore be used to obtain 'punchy' percussive sounds, safe in the knowledge that excessive, overload-inducing transients will be suppressed. Obviously this sort of automation removes some versatility in the sense that it is difficult to get OTT superpunchy effects. On balance though, I'd say it was a good thing and certainly makes the compressor easy to use. The minimum compression ratio of 2:1 was, as feared, too high a minimum for my taste, but you can of course circumvent this using two desk channels; simply mix some of the compressor return with the straight signal.

A word of praise for the LED displays; they were very bright and clear, and much better than the dim, apologetic 3-LED specials of most Oriental equipment. In practice, a +20dB output is hardly ever indicated due to the overload suppression mentioned earlier. However, since the Out mode disconnects the side-chain, the circuit can then overload like any normal stage.

Conclusion

The word to describe the overall sound of this unit is 'neutral'. It does lose a bit of brightness but not too much and this is to be expected in a compressor operating on this principle. It lacks the character, that is to say distortion, of some FET and valve compressors but it is essentially viceless which has got to be a big point in its favour.

This then is a standard compressor format with a few very welcome refinements and it's sensibly built. All you need is one of these, an Aphex Compellor and a grotty old guitar sustain pedal and you should be able to cope with anything.

The Ashly CL52 costs £495 including VAT.

Further details may be obtained from: Sound Technology, (Contact Details).