Gain staging important in tracking, too…

I’ve been intentional about using proper gain structure when mixing lately, and I think that it has made a big difference in my mixes.
For those of you who are not too familiar with gain structure, here are a few things to keep in mind:

1. If you have one track peaking at -3dB and add another track that is also peaking at -3dB, you will be peaking at 0dB (Digital CLIPPING!) at the point where you sum the two together (the buss, master fader, or soundcard output they are both routed to first. Each track will appear to have an ok level, but your mix will still distort.
2. In general, your Track faders should not be higher than your Buss faders, and everything should be lower than your Master fader.
3. In general, try not to have faders above Unity.
4. Be careful to check the Input and Output levels of your plug-ins, too, as they can drastically alter your volume
5. Be careful adjusting levels when you have a dynamics processor or other threshold-driven based effect (a compressor, wah, etc…) You might have dialed in the compressor threshold perfectly before you tweaked your amp simulator, but gain changes in the amp simulator could require adjusting the compressor again. Automation can be used to make this brainless in some cases.

I was recently recording a classical piano and ran into some gain stage issues while tracking. I had 4 different stereo configurations setup for the piano (which means 8 mono channels. In the end I only used 2 stereo pairs, but was experimenting with some other techniques as well.):

1. With digital recording, some have thought it best to get as close to 0db as possible without clipping. This gives the greatest dynamic range and signal to noise ratio possible for later processing. However, this is not always true. Recording tracks in 24-bit where the eventual output is going to be CD quality (16-bit) is the most common example. 24-bit audio has a dynamic range greater than 130 dB. 16-bit audio has only 96dB of dynamic range. Due to this difference and the incredibly bad things that happen when you overload your converters and introduce digital clipping, many experts recommend a different philosophy when setting levels. Peak levels of that reach ¾ of the way up your meter (-12 dB or so) are thought to be sufficient to maintain dynamic range when processing digitally and dithering down to 16-bit. For more information on this, check out Bob Katz’s Mastering Audio.
2. When multi-tracking, keep in mind considerations regarding gain stage.
a. With the Piano, I heard some clipping while setting gain levels. 4 of my channels were coming in through a preamp with its own meters, so I could tell that those channels weren’t clipping. The other four didn’t have meters on the preamp, but all of the meters per channel seemed to indicate levels that were not clipping. Nevertheless, I turned down the preamps until I did not hear any more clipping.
b. The problem ended up being that I was overloading my Master fader. So I ended up with tracks that were individually softer, noisier, and more limited in dynamic range than they would have been at optimum settings.
3. Thus, it is helpful to keep in mind the different places where clipping can occur when tracking, so you can find the source of the clipping and deal with it properly. The following steps can help you find and eliminate clipping quickly:
a. When multi-tracking, turn down the software gain by 3dB for every new track you add: Turning down the levels in software will not change the volume each track is recorded at, but will affect your cumulative volume at summing points. You want to hear when your tracks are actually clipping, not when you’re crashing your busses, master fader, or soundcard output. If you are using input monitoring with a compressor inserted into one or more of your tracks, you will want to use the volume setting instead of gain…)
b. Have your musicians play slightly louder than the loudest they are going to play once you get going. (They’ll usually play louder than this when actually performing, so you’re giving yourself some psychological headroom.)
c. Solo the channel you want to set levels for.
d. If you have input monitoring enabled, open any plug-ins you are monitoring on that track so you can check their meters and make sure you are not clipping the plug-in (this goes the same for any hardware inserted after your preamp in the chain regardless of input monitoring.) CHECK THESE METERS AT EACH STEP BELOW IF YOU HEAR CLIPPING.
e. Turn the pad on your preamp on and turn the preamp way down. Slowly bring up the level until your meter peaks above ¾ of its measurement without overloading. If you still hear clipping, and you have REALLY loud sources, first make sure your mic can handle the SPL levels being produced. Then try using the pad on your microphone if it has one. If you don’t have a pad and are still clipping, turn down the amp, hit the drums more softly, or move the mic further away. With the software levels set, your preamp’s pad on and low levels on the pre, you know you’re clipping at the source (microphone.)
f. If you are able to do both of the above without hearing any clipping, turn down your preamp all the way, then disengage the pad. If you’re getting clipping or overs on the preamps meters with the preamp turned all the way down (or barely turned up) then use the pad again and adjust as necessary. If you can get good levels without clipping when you are not using the pad, then don’t use it.
g. Move on to the next channel and repeat steps c-e.
h. Now listen to all tracks you are recording with no solos. If your input meters are fine but you still hear clipping, check your buss, master fader, and soundcard output meters again and bring down the software gain of your tracks again if necessary.

Once you get used to using these techniques with your system, you’ll start to anticipate where your preamp’s levels should be set, and where your problems are. You will be able to make sure any clipping you hear will actually be recorded, and then can systematically eliminate clips before you start tracking for keeps. You will be able to capture the cleanest, loudest, most dynamic sounds possible with your equipment.


Anyone want to add from their experiences?

Peace,

Charlie,

Thanks for the help with the MBOX2 problems I was having. Your suggestions helped me set it correctly and I was able to use it without worrying about the problems. Now I have my 002 Rack back, and it seems to be working correctly, so it looks like I'm good on this.

Thanks for your help.

Peace,

Bob,

Thanks for your comments. However, I wonder -- even though there is no HARM in allowing levels to peak way above 0dbFS within Sonar's mix engine, could it still be a good idea to always work within the 0dbFS limitation as a habit? I record live and use a Yamaha HDD recorder. As you mentioned, my biggest concern (except for proper mic placement) and the factor into which I put the most effort is making sure I never clip. As a matter of principle, I always follow the same rules when recording & mixing in Sonar -- for consistency.

As far as I know, there is no reason to CHOOSE to allow levels over 0dbFS in Sonar. Is there some hidden benefit (to allowing +0dbFS levels in Sonar) that I'm missing?

I'm not suggesting that you implied that there IS a benefit. I'm just wondering if there is something I'm missing in the flexibility of a 32 or 64 bit double-precision mix engine -- other than the huge safety buffer that basically eliminates clipping in the digital mix realm. Otherwise, I'll probably continue to follow traditional mixing technique, never allowing anything in the signal flow to break 0dbFS (...to the best of my ability, using Sonar's fabulous metering).

Yeah, I was not suggesting that everyone run their projects at +200dB . In fact, you will find that some "floating point" plugins do indeed arbitrarily clip at levels that are only "a little" over 0dBfs. (notice in my experiment, I left out all plug-ins). But I've seen many people spot a +2.6dB "over" at some point in the song and go "oh no! I must reduce all my track gains now!", when really, it can be done at the bus level with absolutely no loss. This is the "liberation" I speak of compared to my old analog console.

Bob,

Thanks for weighing in. It is nice to have someone with your background and expertise around to clarify some of these issues. It sends my mind more at ease to understand the way the gain structure works in Sonar more thoroughly. I'm glad you brought up the plugins in retrospect, though, as these can often clip in a rather ugly way.

When plug-ins are involved, I will probably still do my best to follow proper gain structure according to the old analog rules since I use plug-ins a lot, and find that following the rules of proper analog gain staging helps me reduce the likelihood of clipping them. However, I appreciate very much understanding the new digital exceptions to the old rules as you pointed out, and will certainly take advantage of the freedom these exceptions provide where applicable.

Thanks for providing some great information on this topic. This kind of contribution is exactly what I was hoping for when I posted!

Peace,

Thanks for the clarification Bob. Yes...I do find that being able to lower a bus level instead of 17 separate faders is MUCH easier/faster (even despite the Quick Group tools in Sonar). I route my big projects into multiple 'section' busses (i.e. strings, percussion, brass, etc.) and then route the section busses (in parallel with any Aux/FX busses) to a Master which is routed to a Main Out. So for the most part, I can deal with levels pretty early in the signal chain. But it's still nice to be able to adjust one section fader (if things are getting hot) instead of 5 different string section faders...

John:

Here's the encyclopedia version....