Gain structure explained
When it comes to recording and mixing, optimizing gain structure is a fundamental procedure that needs to be understood in order to make both the correct technical and creative decisions when working with audio. Being a mastering engineer I occasionally see stereo files that are delivered that have gone beyond 0dBFS – end stops – digital zero – clip point. (or signals clipped prior to the master output – maybe in a plug in) This means the gain structure is incorrect.
By definition digital audio systems have a finite way of accurately representing your audio signal at their output. Your musical signal is represented by numbers and these are not infinite, the mathematics ends at 0dBFS on your audio workstations 24 bit stereo master output bus. (and any internal, fixed point maths buses) If your output signal is constantly hitting the end of the stereo master bus meter (or your plug in meters) you need to think about your mix gain structure in order to avoid clipping your audio.
The moment your signal exceeds 0dBFS your music is no longer being accurately represented by the internal maths in your digital audio workstation. Currently it is common for people to operate their workstations with signal levels that are unnecessarily hot without correct attention to gain structure. There are a number of reasons for this which we shall talk about.
One of the reasons is that in the days of predominantly 16 bit digital recording we needed to maximize the use of the digital bit depth and keep the signal away from the lower signal ranges to improve the fidelity of the audio and minimize digital noise (and quantization distortion). This is no longer required and 24 bit operation of your digital audio workstation has become standard and it is recommended that you use 24 bit depth for all recording and mixing purposes.
In short : When starting a new track, peak drums no higher than -12dBFS on your channel/master output, then you are very unlikely to clip the audio as you build your track/mix. For the longer technical explanation read on.
Optimizing the gain structure of an audio system has always been very important in audio recording and mixing from the days of large format mixing consoles (such as SSL and NEVE) and large tape machines. Before recordings were made in a studio the audio engineer would take some levels from the band. This would be done when they were playing as loudly as they would be likely to be playing during a recording.
Optimizing gain structure meant ensuring that the audio signal was mid way between the noise floor (hiss and hum) on the tape or a mixing board and its distortion point (overload). This is known as PFL ing a signal (pre fader listen), the audio signal would be peaked at around 0VU on a VU meter. This signal level was determined as being a nominal signal level and was important in optimizing the gain structure.
The electronic components in an analogue system determine the both the noise floor (all electronic circuits generate thermal noise) and the point of distortion. Distortion is caused when any given component has exceeded its linear gain capability and a signal amplified no longer closely represents the input signal. By optimizing gain structure you avoid both the unpleasant hiss and distortion which is unwanted and distracting to the ear. (as a generalization for a clean recording)
With the common use of digital recorders everyone can record at home but it also means that many people have not had any formal sound engineering training and may not comprehend some of the basic recording and mixing principles. This text will now focus on mix down gain structure inside a DAW at 24 bit depth output.
This tutorial is based on starting a new mix of a piece of music in your Digital Audio Workstation and optimizing gain structure in the channels. You should set your workstation session to operate at 24 bit depth. If you normally use a limiter on the stereo master output please remove it. You will not require a limiter on the master output during mixing using this technique.
The first thing we are going to do is start with your kick or snare drum. I want you to move your kick drum/snare drum channel fader on your mixer so that the drum signal peaks at approximately -12dBFS on your track channel meter. Once the signal is peaking at around -12dBFS you can then get a rough mix together based around the kick or snare drum as a reference signal. Bring in your synthesizers or guitars, vocals etc. and build a rough music mix.
At this point do not be tempted to push up the reference drum level. If during the course of mixing you have added compression or other gain altering plug ins then make sure your peak level is back at around -12dBFS for the reference drum.
You will now notice that the stereo master output meter is not peaking as high. This is great, the gain structure has been correctly set and has a number of benefits to your audio quality. It means you will not clip the stereo output and that means no distortion. It also means that when you export or bounce your mix down it will be ready for your own self finalizing or professional audio mastering services.
You may have noticed that the music is somewhat quieter to monitor than when you worked at a higher signal level. This new gain structure also has some benefits to the clarity of your monitoring. In a worst case scenario your sound card could be a budget one which is powered from a 5 volt USB 2.0 line. This normally means that the analogue headroom in the small op amps driving the audio output may be compromised.
This would mean that distortion can creep in before getting close to the maximum signal level an operational amplifier can be designed to reproduce. The low operating voltage creates a deficiency in the circuit powering the small op amps in your sound card. (an op amp is a small amplifying circuit built into monolithic chip which is typically 4mm square).
By sending a lower signal level into these devices (which tend to generate very low noise) you are likely to hear less distortion caused by the op amps and low operation voltage in the sound cards output. As a result of improved gain structure you will be hearing more clarity in your mix and you will probably make better decisions. Your mixes should have slightly better balance and fidelity as a result.
As you are working at a lower signal level you will find you have to turn up the volume on your power amplifier, active monitors or monitor controller device to compensate. Remember you can use dynamics processing, equalization and effects processing exactly as you used to in your old mixing process. Because you have built in plenty of headroom from the mix down outset you are in a perfect position for exporting your 24 bit audio for professional mastering services. If in doubt please consult your mastering engineer. Gain structure is a very important part of audio mixing.
In addition to these technical points it is worth mentioning that many digital emulations of analogue equipment may have models of analogue gain structure built into their design. This means when operating at the gain structure described you will be closer to sending the correct level through the emulations relative to their analogue counterparts. This should result in a closer emulation for those plug ins that are designed with a nominal signal level in mind. (it is often closer to -18dBFS) but this outlined approach will still be a step in the right direction.
Mastering engineer Barry Gardner in the studio:
Further technical detail
You may be left wanting some further technical details and wondering why you can work this way, the fact is professional audio engineers have always optimized the gain structure of their mixing console and worked at similar signal levels to what has been explained above. Here is some further detail.
When using a large format analogue mixing desk the record and mix level within the mixers channels was often aimed at roughly 0VU. Gain structure was vital to ensure the best sound quality. A VU (volume unit) meter is a standard meter with a needle, it averaged the signal voltage. Historically these meters were largely used in the recording industry and also in broadcast industry equipment in the USA.These meters are vital tools in setting correct gain structure.
0Vu = approx -12dBFS (on a digital meter) 0Vu is approximately equivalent to the electrical value of 1.23 volts RMS referenced to +4dBu. (+4 dBu is the professional recording reference level) By this reasoning you can see that it means you do not need to mix your music into the end of the digital meters on your channels as the electrical level around the same that one would expect to find within a high end professional mixing console.
In most digital audio workstations the meter scaling does not help with making metering decisions as a peak at -12dBFS looks very low relative to the top of the meter. It influences the user and makes it look as if your signal is far too low relative to the full length of the meter. In actuality this is not the case.
You may have heard that it was a good idea to keep the signal “hot” in order to keep it away from the noise floor. But remember we are working (or should be) at 24 bit depth and 24 bit depth audio has a theoretical dynamic range of 144dB. (i.e. the noise is 144dB below 0dBFS)
16 bit systems have a theoretical dynamic range of 96dB.
144dB (24 bit) minus 12dB is 132dB. 132dB – 96dB (16 bit dyn range) = 36dB theoretical signal to noise ratio improvement in the digital realm.
At 24 bit and a peak level of -12dBFS your music will still have a theoretical signal to noise ratio 36dB better than if you operated at 16 bit. There is no problem peaking signals at this lower suggested level when using 24 bit depth.
The above gain structure guidelines are especially suitable for new producers and engineers who are starting out, the above will practically result in repeatable, low distortion recordings and mixes. Once you are comfortable with this way of operating you can deviate as you see fit. This is because you will understand any compromises involved, but this remains a very good and foolproof way of making low distortion recordings and mixes at 24 bit. Make sure your mixes end up sounding as good as possible by understanding why gain structure is important.
Gain structure for mixing desks and mixing consoles
I notice a few searches for this so I will add specifics regarding gain structure on an analogue mixing console. When mixing or recording the process is essentially the same. Use the PFL (Pre-Fade Listen) button on your console channel and adjust the channel gain trim so that your channel signals peak to around 0Vu on your output metering (or PFL metering).
Often the stereo master output Vu switches to PFL metering when the mixer channel PFL button is pressed. Do this for all channels it will optimize the gain structure through the desk. It keeps the signals out of the noise and yet leaves ample headroom to avoid distortion.
If you are finding a disparity between the analogue mixer optimal gain structure and the recording inputs on your DAW you may have to switch your audio interface between the 2 possible reference levels, namely +4dBu and -10dBV.Or if the option exists, on your mixing console. If your mixer outputs are referenced to +4dBu (1.23 Volts RMS) signals as standard and your sound card is set to -10dBV (0.316 V RMS) you will very likely over drive the inputs of the sound card.
For information value +4dBu reference level is approximately 12dB “hotter” than -10dBV.
As an extra tip when recording – peak channel signals at -3 Vu when taking levels as the band will often play more vigorously when you do a real recording “take”.
Analogue mastering (mix review inclusive) : 1T £30 | 2T £60.00 | 4T £112.00 | Album £230
“Gain Structure” by Barry Gardner Copyright 2012-23.