Thanks Jeff.

Well, it's not too hard to do - group all tracks and drop the Gain on all of them at once.
 
Of course, any tracks that have compression on them will have to have those compressors' thresholds adjusted accordingly. Any tracks that have an distortion/exciter/amp sim might similarly require tweaking. And you still might have an issue with a plugin distorting because of its own internal gain, as was the case in the anecdote that began this thread.

Working with Nebula I have HAD t track everything at -18db, went to the trouble to calibrate my Dorrough meter to a specific output bus in my Lynx that never gets touched and is always on the master 2 bus.
 
It is amazing how much some of the plug-ins tweak the input/output level...Nebula will vomit in your audio if you hit it to hard, it is painful so I have had to go in and learn how to track, then making sure the plug-ins are spot on as well...reaper makes some plug-ins that can help you set the gain staging and know exactly how much a plug-in is messing your levels theres a two part "pre" and "post" gain plug that will calibrate your signal, once you've learned what the plug-in is doing your gold.
 
There is a growing industry push for the -18db as standard protocol...it gives you plenty of headroom and everyone should find a consistent gain-staging plan...
 
One of my headaches is I set the zoom on my track meters at the extent of -18 but I have to do it for each track, maybe there is a way to set it as default but i haven't found it yet...not that I was looking...it's just habit now to set my zoom level so I can see what is happening in my clips.
 
Davids video was decent, it heps see it to make sense...https://www.youtube.com/watch?v=EaHQJ0T9iHg

I have read this thread and the SOS article. I am dismayed. Too many references with too many associations.
 
Fur shur, lots of good advice for all of us to follow. And I welcome all attempts at helping developing safe-recording practices.
 
But too many of the comments comparing the many different aspects of analogue to digital to playback to processing to summing to exporting to conversion to actual sound are just plain confusing.
 
A few of my loud verbal responses:
 
Recording to 24 bit wav files is not the same as playback of these files.
 
Zero on the sonar meters is NOT 0dbfs.
 
Zero on the meter of any track/bus/main that is output to the input of an audio converter CAN be an indication of 0dbfs.
 
There is no headroom above 0dbfs: FS means full scale.
 
A good reason to track low, and closely monitor plugins, is because some transients are in the nano second range. Meters, and our eyes, do not respond this quickly. But plugins can.
 
(Reference Tape-op http://www.tapeop.com/interviews/btg/102/bill-cheney/)
 
I am not suggesting that seeing-no-red is a bad thing. I am suggesting that item by item comparison of all aspects of digital recording and playback with analogue recording and playback is causing confusion. 
 
Paying attention to not pumping 24 bit plugins beyond their happy zone is reasonable.
 
But to say we should not use the possible 1000 db of headroom provided by floating point calculations seems silly to me.
 
If 24 bit wav files are limiting our use of floating point math should we instead use 32 bit files?
 
Maybe Bob Katz recommendation to always use dither when bouncing to clips and exporting would fix some of what is being discerned as loss of fidelity.
 
Confusion reigns supreme. 
 
 

jm24
But to say we should not use the possible 1000 db of headroom provided by floating point calculations seems silly to me.
 
If 24 bit wav files are limiting our use of floating point math should we instead use 32 bit files?
 
Confusion reigns supreme. 

The point of the original post is that the "1000 dB of headroom provided by floating point" isn't always there in practice. A surprising number of otherwise quality plugins just can't handle > 0dB signals without distorting.
 
I used to believe that 32-bit floating point made it unnecessary to worry about positive decibel values, but discovered that I had been incorrect in that assumption. That was the simple point of this thread, which then took some ancillary side-trips regarding best-practices and metering.
 
 
 
 
 
 
 
 

Recording at a lower level digitally is a very good idea.  Many are still pushing things way too loud and wonder why they get into trouble.  At 24 bit recording even in the real world, 138 dB of dynamic range is still possible.  So even if you are at -20 (or -23) you still have 118 dB of dynamic range under your belt!
 
If you use a decent VU meter plugin (such as the Klanghelm) all this is much easier too.  You can choose a reference that the meter will show 0dB VU.  I tend to work with the K system and work with either K-14 or K-20.  You can use VU meters on tracks, buses and the main stereo buss too.  This way you don't need to consult the loudness meter until the mastering stage.  If you VU meter everywhere especially before and after plugins and plugin chains you will never bring on a clip light anywhere, ever.  (and you will never clip internally with any plugin either)  There is no need to be up near 0 dB FS ever.
 
When you do this all your mixes will all be at the same level prior to mastering.
 
Mastering up to -16.5 or -15 is also a good idea because that seems to be the level that the iTunes sound check feature works at.  iTunes won’t change a thing eg add or subtract gain at this mastered level leaving your track perfectly intact.
 
I have had some mastering clients though that would not accept this (-15) either and they would say it is too soft.  You can go as high as -10 to satisfy that and the mix will still sound punchy and transient but louder.  The trick is to also use a DR meter (dynamic range) and keep an eye on the dynamic range before and after mastering and it will stop you from slamming tracks too hard and making them sound bad.
 
I work with three metering systems now.  VU meters during the production and the DR and Loudness meters in mastering.  The loudness meter is very interesting too.
 
The other aspect of VU metering which people just don't get or want to is the movement of the needle or ballistics.  When the meter swings wildly either on a track, buss or main mix it means there is something wrong.  You can track it down and fix it and it is usually only one or two things causing it.  When you set up a compressor correctly the meter moves in a certain way and when it is not set well it moves in a different way.  This information is just not available with any current peak metering or even rms metering system (bar graph displays are not where it at) used in most DAW's.  DR and loudness meters wont help you either in this area too.
 
When everything is right in the mix the meter just dances beautifully right up to 0 dB VU and rarely overshoots much and rarely jumps over wildly. That is when the mix actually sounds the best. You can even mix with a VU meter bringing in things one at a time and allowing for the needle to start lower and eventually when everything is in, it is just hitting 0 dB VU. I have expensive real VU's and they still have the edge but the Klanghelm VU is very good though and does the job.