Sound On Stage
The PA Console
"The desk is designed to route, and control the level and tonality of a large number of signals, both singly, and in groups, without spurious interaction or the addition of distortion, hiss or other noises."
The most fundamental — and the original — purpose of the mixing console in a PA system is to sum all the instruments, enabling a common amplification system (the outfront PA) to cover the audience. Conjecturally, if naively, this could be a box with half-a-dozen faders. However, in the process, it becomes necessary to accommodate the unnatural sounds and overload problems that are brought on by close-miking, so our conjectural 'six fader' mixer is found to need some extra controls, namely equalisation facilities and a means of gain control/attenuation to enable us to steer between the evils of noise an overload-crunched sound.
Next, the drummer assures us indignantly that our console must produce two, differently summed outputs, so his percussive meanderings can be made to ping-pong across the stage; in other words, 'stereo'; and to make things easy for the mixing person, requires a master fader so he can adjust the overall level without the risk of upsetting the balance of individual instruments.
So far, our conjectural mixer has acquired some 36 knobs, 6 switches and two extra faders. Yet this mixer is styled in the most fundamental format imaginable, for PA at least.
Having attained a fair cost and complexity, it's tempting to kill several more birds. Firstly, with just a few more knobs and sockets, the mixer offers the opportunity to subject acoustic and electric instruments alike to Space Stations and other FX gizmos. Secondly, it's relatively easy to 'drop in' additional mixing facilities so that each musician can hear the exact blend of instruments necessary to play competently and with attunement over his (stage) monitor. And finally, to make life even more comfortable for the sound engineer, some LED and/or analogue metering to assure him that, despite the guttural horn sound, the signal hasn't crunched (viz, peak/overload metering) and that the sound level really is much louder than his weary ears suggest (Average or 'VU' metering), plus a headphone amplifier and a means of switching around a maze of signal access points within the mixer, so that signals can be traced and inspected despite the bedlam raging all around. And Sub-grouping too would be useful...
Religiously detailed overviews of mixing console function can rapidly lead to vertigo, so we'll confine our examination to a more lucid, albeit slightly less accurate, step-by-step tracing of a signal's path, looking at potential pitfalls and underlining the key requirements as we go along.
Today, most mixers utilise IC op-amps, and operate from standard +15V supply rails. This determines the maximum signal level the desk can handle internally. In theory, +15V rails allow a signal of some 8V RMS (+20dBU), but in practice, rail voltages may be slightly less, and audible distortion may set in several dB below this figure, especially in the case of op-amps driving low impedance loads. Thus, +15dBU (4½V RMS) is a conservative maximum. This figure also represents the fundamental limit on the signal magnitude applied to the input.
In desks of high calibre, however, the maximum is modified by the presence of an input transformer, which, in exchange for conferring the status of balanced line operation, and curbing hiss on the occasions when a high degree of amplification is called for, provides an extra 15dB or so of voltage gain. As a rule of thumb, any desk with a transformer coupled input will overload with signals peaking in excess of 0dBU (0.775V RMS). The transformer may also overload at these levels, but the effect is less dramatic and the modified sound tantalising, being akin to that produced by a valve instrument amplifier, and yet only a hair's breadth away from being swallowed by the brute distortion of an overdriven op-amp.
How do these abstract figures interface with reality? Much depends on the microphones you use, the instruments, the miking technique and the frenzy within each musician. But the short-term levels derived from close-miked drums and horns is rarely exaggerated. With a capacitor microphone, the level on a Tom, snare or Kick-drum can easily reach +10dBU — that's 2½ volts without any help from the mixer! Using low impedance dynamic microphones, peak levels from percussion are typically 10dB lower, but then a well blown trumpet is 10dB louder...
At sound pressure levels in excess of some 120dB, the head amplifiers and transformers inside cheap capacitor microphones overload, and many dynamic microphones begin to compress the dynamics at this point, viz: A 6dB increase in SPL ISN'T reflected as a 6dB increase at the microphone's output. The theoretical perils of deriving truly over-the-top voltages from close miked instruments are therefore restrained when certain low cost mics are in use. But despite this qualification, it's good practice to regard low impedance dynamic and capacitor microphones as tools potentially capable of producing up to 2 volts and 5 volts respectively on stage.
Returning to the console, it's clear that with a transformer coupled input, overloading is potentially inevitable. And even without a transformer, we're sailing uncomfortably close to the sudden manifestation of unpleasant crunch up noises. Rule No. 1 for sensible PA consoles then, is that a switchable attenuator must be provided before the front end. If we assume a +10dBU input signal to be the most we're likely to encounter then adding 15dB (the transformer's voltage gain) and subtracting the +15dBU maximum input level leaves us with a 10dB disparity. Rounding this up to 20dB gives us a bare minimum figure for the attenuator — or 'pad' as it's colloquially known. Some up-market mixers have a choice of pads, typically selectable in 10dB steps, but a single -20dB attenuation step will cover the majority of requirements, at least for microphones. The -20dB pad also has the beauty of being a 'Reduce signal to 10% button', hence readily comprehended by the inebriated or those braggart desk-drivers who invariably lack an intuitive rapport with the decibel!
If you own a mixer without this facility (certain manufacturers still have strangely fettered minds in response to the heretical concept of 2 Volt microphone signals), don't despair: pads are nothing more than a simple pattern of resistors, and with a little guise and some rubber sleeves, it's possible to hide them safely inside an XLR connector. Alternatively a small diecast box with an in/out switch, chassis mounted sockets and a boldly marked attenuation figure is a more flexible add-on aid, and also more obvious in the average jungle of cables; it's certainly no fun unscrewing dozens of XLR plugs to discover which cable had 'That 20dB pad John made up last week'. Ready made 'inline' pads can also be bought, but they tend to be outrageously expensive. Instead, Table 1 delineates resistor values for most PA requirements; note that 'balanced inputs' refers also to the 'transformerless' variety.
Having dispensed with the immediate hassles of high level mic signals, another, more sensitive and adjustable form of controlling signal levels is called for. Looking at Figure 1, the input gain control, or 'gain pot' appears either after the input amplifier as VR1, which is strictly a passive attenuator (i.e. in the conventional 'volume control' configuration) or is built into the amplifier's feedback loop (VR2) to control the gain of this stage. Although in academic circles the latter active gain pot is more elegant, it suffers from a serious disability in that the gain can never be reduced to below unity. So if a comfortable +5dBU signal appears at the input to IC 1, the minimum level transmitted to succeeding stages will be +5dBU.
This may appear quite satisfactory if these succeeding stages have unity gain, but, of course, any attempt to boost a portion of this signal by, say, 10dB with the EQ controls will carry the risk of overload. Worse still, one famous UK manufacturer produced a mixer with this active gain control arrangement, and finding that the op-amp merrily oscillated when 'shut-down' to unity gain, provided a gain pot which couldn't be wound down to produce less than +10dB of gain. Whilst this ruse cured the oscillation, it did make life rather awkward for the unlucky purchasers — particularly as the wretched mixer had no pad either.
Using the passive attenuator arrangement, the signal level can be shut down to zero or minus infinity — at least in theory! A more realistic figure is -70 to -100dB, depending on the residual resistance in the pot when the wiper is pressed hard against the end stops. But this order of level reduction isn't as clever as it sounds, as here we're comparing it against the full gain of IC1. So if the op-amp's gain is set at +30dB, a -70dB attenuation on VR1 is only equal to -40dB below the original input signal. Still, it's 40dB better than VR2's unity gain restriction. Nevertheless, VR1 isn't a practical arrangement, because the input stage amplifies 'flat out' all the time and easily overloads. Practical input stages therefore make judicious use of both active and passive gain control techniques.
After the input control, a sensibly designed mixer channel will have a 'unity gain-structure' — which means, in essence that succeeding stages and controls won't be able to significantly boost the signal level. (On posh consoles, there will usually be 2 or 3dB of 'make-up' gain in further stages, but this is only to compensate for losses in faders, etc.). The only exception to this concept is the EQ stage. Whilst it's perfectly possible to utilise a 'cut bass and treble' arrangement in lieu of fostering the midrange frequencies with extra amplification, this isn't very convenient, and one has to live with the EQ's ability to put mixers into the overdrive mode when tweaked by heavy hands. In other words, when initially setting up the input gain, any likely boosting of the EQ settings will have to be foreseen, and taken into account.
The channel fader works purely in the shut down mode (like VR1 previously discussed), so its setting has no bearing on overloading within the channel. Noting that the input gain setting is fiddly, if not critical, the channel fader offers a means of making routine or ad-lib level adjustments during a set without worries.
At the same time, the fader should be in a mechanically convenient position (typically 2/3rds up its travel), rather than hard against either end-stop when the mix is roughly set up. In this way, adjustments in either direction will be feasible without unnecessary restriction. Needless to say, the need to set the channel fader at a physically convenient position does govern the input gain pot's position to some extent.
Once the fader and EQ settings have been roughly established, the gain pot is then advanced to provide a level roughly commensurate with the level of the instrument in the mix. If, winding it up a little more results in overload, then other adjustments will be required. As soundcheck and rehearsal levels rarely approach the intensity of their audience inspired counterpart, it's a good rule of thumb to set the gain control so peaks occur at least 15dB below the overload threshold; the channel and master fader(s) are then progressively tweaked with the aim of placing all knobs in usuable positions, i.e. well away from end stops. Of course, this procedure isn't something that can be described cogently and meaningfully in words; rather, it's an intuitive skill that can only be picked up by intelligent practice. To sum up, the general aim is to steer between hiss and hum (too little gain at the front end) and crunch-up noises (too much gain at the front end), leaving enough room either side for useful improvisation en-route, whilst at the same time balancing these requirements against the relative levels (i.e. mix) of another 5, 11, 15 or 23 channels, and knowing immediately which knob to strike if something distorts, or howlround threatens.
In the next part, I'll discuss the usage and abusage of LEDs as well as some thoughts on sound levels.