I have a question about normalising. I mix from Cockos Reaper through an M-Audio ProFire 2626 interface into an Allen & Heath ZED12 FX mixer, and then back into Reaper. I often find that the end mix level is lower than expected, and I have to push the master fader on the desk up over zero. I do set the gain correctly for each hardware channel, using the PFL button, and I have the outputs of the 2626 up at maximum. My worry is that if I start pushing up the levels of the individual track faders on the mixer, I'll start introducing unwanted hiss from the hardware, so I tend to mix to a maximum of unity gain on the individual channels — but maybe I should start going beyond that? Also, if I do normalise my mix, should I do it before I apply my master-bus processing (a bit of compression and limiting), or should I apply the master-bus processing first and then normalise that processed file to take full advantage of the available digital headroom?
Timo Carlier via email
SOS contributor Mike Senior replies: The first main issue here is avoiding unwanted noise and distortion from your analogue components, and a good basic principle to bear in mind is to maximise your signal-to-noise ratio as early as possible for each piece of equipment, and then to leave any subsequent gain controls at their unity-gain position if possible. So in your situation, make sure the signals you're routing out of Reaper are making full use of the digital output headroom, so that you're sending the maximum level out of the ProFire 2626's sockets. (Check also that the faders in the ProFire's own mixer utility are set to unity gain, so that they don't undermine your efforts in Reaper.) Boosting the gain in your DAW at this point shouldn't incur any significant side-effects or additional background noise as long as you don't clip the output buses in Reaper.
With a good level coming out of the ProFire 2626, you may well find you need very little, if any, boost from the ZED12's channel Gain trim to give decent PFL readings, so from that point until you record the mixdown signal back into the computer, you should only need to turn things down. Clearly, this is what the faders are for, but you may also wish to use the channel Gain trims too, in some cases, in order to keep the channel faders for quieter sources closer to the unity-gain mark, where there's better control resolution. From that point, the trick is to build up your analogue mix so that it naturally fills the console's available output headroom. If you start with your first instruments too quiet, you'll end up with a low output level and therefore more background noise than necessary. If you start things too hot, you'll start clipping the mix bus before all the instruments have been added. It's a bit of a knack, so don't sweat it if you don't nail it exactly first time. If you under/overshoot, the best thing to do is adjust the channel faders en masse to redress the situation. Any channel insert processing will be left unchanged in that way, as well as any post-fader effects levels, so the amount of rebalancing you'll need to do is likely to be fairly small.
My main tip for getting it right first time is to fade up early any channels with strong transient peaks or powerful low-frequency energy: typically bass and drums in modern commercial productions. These elements take up the lion's share of the headroom in many mixes, so they provide a good early indication of the headroom the full mix is likely to demand. In fact, with a little practice, you should be able to discover 'rule of thumb' starting levels for your drums and bass on the ZED12's meters, which will usually lead to a good final mix level.
If you're doing your master-bus processing in the computer, just get the hottest clean output signal from the mixer into the audio interface. If the signal's too hot for the interface, turn down the mixer's master fader; if it's not hot enough, push up the console's master fader (if that doesn't clip the ZED12's output circuitry), or apply additional analogue input gain on the interface. Once the signal is digitised at a good level, it shouldn't need any normalisation before further bus processing; even 24dB of headroom should be perfectly fine at 24-bit resolution. If you're bus processing through analogue gear before digitising, just keep the same principles of gain-structuring at the front of your mind, and you shouldn't come too far unstuck: feed the processor as hot as possible without clipping it, and avoid adjusting the gain unnecessarily until you need to set an appropriate output level for the next piece of gear in the chain.
In addition to technical considerations, the level at which you drive any piece of analogue circuitry also affects the subjective sound in less tangible ways, and this concern may justify creatively modifying some of the above generalised tactics. If you hit the Allen & Heath's input circuitry a little harder than strictly recommended, for instance, you may find you like the resulting saturation harmonics. Perhaps your analogue bus compressor provides slightly different release or 'knee' characteristics (for the same gain reduction) if you process a high-level signal with a high threshold than if you work on a lower-level signal with a lower threshold. Or maybe the mixer's output bus might sound more transparent on acoustic music if fed more conservatively. Finding out that kind of stuff is one of the really fun bits about analogue mixing, so if you've already made the effort to get stuck into the hybrid analogue/digital approach, I imagine you don't need too much convincing to get your hands dirty there!
Finally, it's tempting to think that such signal-level concerns are pretty much redundant in the digital domain, but that's not really true, given how many emulated analogue plug-ins most people now seem to use. If this kind of processor is faithfully modelled, it will usually track the non-linear characteristics of its analogue forbear, so you still need to be aware of what level is hitting it. This is one of the reasons why my Mix Rescue projects are full of lots of instances of GVST's freeware GGain plug-in, a simple +/-12dB VST gain utility that gives me more control over the internal gain-structure of my channel plug-in chains. (There are also various built-in gain plug-ins within Reaper's Jesusonic set if you're not on a PC as I am.)