Gain Staging as I understand really isn't so complex... is it?

M

Musicianaire

New member
Okay, after much reading and videoing all over the Web the past 2½ days, I've come to this conclusion about Gain Staging (which I'd never heard of until this year): Basically, just make sure nothing clips, while maintaining good headroom! (bold text added after Farview's reply because I neglected to mention it as a given :) )

Whatever method is necessary to accomplish that goal is personal preference, right? I've seen many ways of doing what pretty much comes down to my stated conclusion. I was starting to get mind boggled when I stopped and thought about it. It's only a matter of doing what I've been doing for live sound for 40 years: Set the Master to unity (-0dB), and adjust the levels of faders and perpiherals accordingly; just don't let anything clip.

Am I right, or have I still missed a point or two?

Even my past recording projects have worked around that principle: don't let stuff clip. I was even a bit wary of clipping with analog sound, though I knew I could get away with it.

Just wanted to get some confirmation - or correction - of my thoughts before going forward. I totally understand how to keep everything from recording too hot, and if that's what it boils down to, then I'm good to go.
 
Last edited:
Kind of.

I look at gain staging as keeping everything humming along at line level between processors. So, you set the mic preamp so it outputs line level, then the signal goes to something else which you will adjust the output of so the signal is coming out of it at line level, and so on.

That's all it is.

Clipping is when you run out of headroom, so being close to clipping means that you are using up most of your headroom, which isn't always good.
 
Musicianaire said:
It's only a matter of doing what I've been doing for live sound for 40 years: Set the Master to unity (-0dB), and adjust the levels of faders and perpiherals accordingly; just don't let anything clip.
Click to expand...

In a nutshell, yup. It's that simple.

Digital audio was designed with line level signals in mind. The same line level used with analog gear. There is no standard for where this will land on your AD converters, and the gain staging you use on your input chain is a critical first step. RMS levels for a calibrated line level signal could land somewhere around -18 dBfs to -22 dBfs. Depends on where your converters are calibrated. The end goal here is, as you said, "Basically, just make sure nothing clips, while maintaining good headroom!" Mixing can be a lot easier if everything is basically close to where it needs to be when you start.

A modest prosumer set of converters will run out of headroom faster than high end stuff. Typical levels for pro recording or broadcast using high end converters with more headroom tend to be around -20 to -22 ish. Conservative, even though the gear used can handle more if necessary.

With analog tape, there is a Goldilocks zone or sweet spot where if the signal is not too hot or cold then you can maximize signal to noise ratio and the focus of the sound. Digital signal to noise is way better, and there is no sweet spot so either you've clipped or you haven't. Digital is a lot more forgiving if you land on the cold side.

The level between processes is another consideration. If you have an outboard compressor or preamp or something that distorts in a good way when you drive it hard, that's fair game if you can pad it down to the right level for the next device in the chain. Provided you want that sound. A good example would be a high end preamp with transformers in it. Modest preamps and converters aren't likely to do you any favours that way.
 
Know the limits (top and bottom, but also left and right) of your gear and use it as you see fit. In analog, I try to get as much gain as I think I'll ever need as close to the source as possible and then try not to turn it down much but especially don't turn it down just to turn it back up unless there's no way around it. Gain staging is making sure the output of my interface can't push my power amp to clip.

But now in analog, things often start to round off a ways before you actually hit the rails, and that is actually important. It can just plain sound good, but before you even hear it, it is just subtly controlling the dynamics of the signal. At multiple stages along the way. You said you can "get away with" some clipping here and there, but I say it's sometimes essential. You will have trouble fitting a kick drum into a typical rock mix if you just let it have all of the dynamic range that it wants. Compressors and even limiters aren't usually fast enough to really catch some of the loudest parts of the transient. If you let it clip a lot of times it's too fast to even hear, but it lets you turn up the fun part that little bit more.

One might ask why you don't instead turn everything else down. Well, whether we're recording to tape or mixing for YouTube release or reinforcing a band in a club, there are real limits (upwards and downwards) to the amount of dynamic range you get. The difference between typical bar chatter and as loud as the PA gets is sometimes not much more than 20db. On YouTube you're supposed to shoot for 14db. If you want people to listen to your stuff in the car...
 
But what if you have to pick one or the other? ;)

I said somewhere recently that gain staging is not about absolute levels but always relative to the limits of the gear on hand. It's mostly about paying attention to and managing the way your signal levels affect the circuits or code that is processing them.
 
Be sure and recheck level if in digital, especially if you’re tweaking EQ,s and compressors. I just did a rookie mistake by not rechecking a compressor output on AC guitar that fed into a bus compressor for all ac guitars. Things started sounding bad. I had to scratch my head and go back to find out why. The output gain on the channel compressor was punched way up, hitting The buss comp way to hard. Check I/O on every piece in the chain.
 
ashcat_lt said:
But what if you have to pick one or the other? ;)

I said somewhere recently that gain staging is not about absolute levels but always relative to the limits of the gear on hand. It's mostly about paying attention to and managing the way your signal levels affect the circuits or code that is processing them.
Click to expand...

Yes sir. That is an apt description of the process. If your capture device is the DAW then you must also understand just what those little ladder lights are telling you. They are all "look ahead" to some extent and here is where your ears become an important part of the equation. You want to 'fill up' your track but not to the inclusion of digital hash and distortion.

And then there's the real tricky part of "gain staging"....Where do you get your gain from? Think about it for a minute....Is it from the input of the preamps? Is it from the output of the preamps? Is it from the input of the track? Is it from any devices you may have inline after the preamps?

Enough times in the barrel and you get to where you can hear the effect of each point in the signal chain on each other point in the signal chain. ANYTHING with an input and an output control has to be considered and each will affect the signal at the capture point. The level may be the same with all sorts of settings, but the overall sound will be different with each pieces' influence on the capture.

And that's what real gain-staging is all about.
 
cavedog101 said:
The level may be the same with all sorts of settings, but the overall sound will be different with each pieces' influence on the capture.
Click to expand...
In analog (and emulations) this is always true. Obviously anything that's deliberately non-linear like compressors and such. But many "pure" digital processes are actually linear and it completely doesn't matter. If I send +18dbfs peak signal through ReaEQ and turn down its output control 18db, it's exactly the same as if the input was 0dbFS and I left it at unity.
 
clipping is desirable in the analog domain recording to tape, you can go over 0dBVU, but we need a buffer zone in digital, I aim for average levels of around -20dBFS sometimes louder, with peaks hitting around -3dBFS to -6dBFS, remember that digital audio has a noise floor just like analog but it's a lot lot quieter, almost invisible, the cleaner your power and preamps e.t.c the less you'll notice it, that's why high headroom analog mic preamps are so desirable.
 
Btyre2013 said:
clipping is desirable in the analog domain recording to tape, you can go over 0dBVU...
Click to expand...
Please don't confuse 0dBVU with 0dbFS.

In analog, the mark that says 0 is usually the nominal or normal expected operating level. It tells us nothing about the actual limits of the circuit. Most times there is 15-20db above that mark before you hit the rails and start actually clipping. Things often get curvy before that point. Sometimes we like how that sounds.

In digital, the 0 mark is the limit. That's exactly as loud as a fixed point digital file can get, and exactly as far as your converter (ASC or DAC) can get. The digital part is perfectly linear right up to that point and then just suddenly clips off. We usually don't like how that sounds. Sometimes the analog electronics get curvy before the digital part clips off, so sometimes it's not so bad.

This is exactly the reason we have that -18dbFS rule of thumb. Our converters are designed to top out at more or less the same level as the analog gear we attach to it. We said earlier thats usually +15 to +20dbVU, and very often +18dbVU. So the limit of our digital system (0dbFS) is set to the limit of our analog system (+18dbVU) and therefore the nominal analog level of 0dBVU hits digital at -18dbFS. But actually that's just a rule of thumb, and if you want to know how your converter works, you have to look at its specs and usually do some math. They never make it easy. I think my line inputs are calibrated to -20dbFS, but I've seen some with as little as 12db headroom above analog nominal.

Worth mentioning too that the analog VU meters are showing you an average very much like (close enough to) a short-term RMS or LUFS level. It's normal for actual peak levels (the ones that actually matter when we're worried about the limits) to be much higher, and on some sources if you have then right on 0dBVU will still end up distorting or clipping. This confuses a lot of newbs. The digital meters they're looking at are almost always peak levels, but they heard -18dbFS, so they shoot for that and end up quieter than they really need to be. That means nothing in digital really, but it means you're running the analog end of things much lower than it's nominal level and much closer to its noise floor. That's only a problem if it's a problem, but in a lot of cases it can be noticeable. It's complicated quite a bit by the fact that for most of us at home with all in one interfaces, we don't get VU meters. The only meters we have are those in our DAW, which are usually peak based.
 
ashcat_lt said:
Please don't confuse 0dBVU with 0dbFS.

In analog, the mark that says 0 is usually the nominal or normal expected operating level. It tells us nothing about the actual limits of the circuit. Most times there is 15-20db above that mark before you hit the rails and start actually clipping. Things often get curvy before that point. Sometimes we like how that sounds.

In digital, the 0 mark is the limit. That's exactly as loud as a fixed point digital file can get, and exactly as far as your converter (ASC or DAC) can get. The digital part is perfectly linear right up to that point and then just suddenly clips off. We usually don't like how that sounds. Sometimes the analog electronics get curvy before the digital part clips off, so sometimes it's not so bad.

This is exactly the reason we have that -18dbFS rule of thumb. Our converters are designed to top out at more or less the same level as the analog gear we attach to it. We said earlier thats usually +15 to +20dbVU, and very often +18dbVU. So the limit of our digital system (0dbFS) is set to the limit of our analog system (+18dbVU) and therefore the nominal analog level of 0dBVU hits digital at -18dbFS. But actually that's just a rule of thumb, and if you want to know how your converter works, you have to look at its specs and usually do some math. They never make it easy. I think my line inputs are calibrated to -20dbFS, but I've seen some with as little as 12db headroom above analog nominal.

Worth mentioning too that the analog VU meters are showing you an average very much like (close enough to) a short-term RMS or LUFS level. It's normal for actual peak levels (the ones that actually matter when we're worried about the limits) to be much higher, and on some sources if you have then right on 0dBVU will still end up distorting or clipping. This confuses a lot of newbs. The digital meters they're looking at are almost always peak levels, but they heard -18dbFS, so they shoot for that and end up quieter than they really need to be. That means nothing in digital really, but it means you're running the analog end of things much lower than it's nominal level and much closer to its noise floor. That's only a problem if it's a problem, but in a lot of cases it can be noticeable. It's complicated quite a bit by the fact that for most of us at home with all in one interfaces, we don't get VU meters. The only meters we have are those in our DAW, which are usually peak based.
Click to expand...

I knew all that already but thanks for lecturing me about it, I'm not confusing anything I never stated that 0dBVU is the same as 0dBFS, not once.
 
Btyre2013 said:
I knew all that already but thanks for lecturing me about it, I'm not confusing anything I never stated that 0dBVU is the same as 0dBFS, not once.
Click to expand...
Yeah, I was pretty sure you knew what you meant, but felt the actual wording of your might post confuse some of our less experienced readers, so thought I'd just cut it off at the pass.
 
ashcat_lt said:
Yeah, I was pretty sure you knew what you meant, but felt the actual wording of your might post confuse some of our less experienced readers, so thought I'd just cut it off at the pass.
Click to expand...

fair enough, I did not mean to come off as stand offish either so sorry if I did, I think gain staging is quite complex in many ways.
 
Ideally, if people could work just for a short time in the analog-only domain...they gain staging thing would be realized and understood much quicker and easier, and where equipment connectivity has a more clear, signal-chain path. Moving to digital after analog, proper gain staging is carried over.

Digital blurs things quite a bit, allowing for weird signal paths that wouldn't make sense in the analog world...and as much as there is a finite, 0 dBFS ceiling, it's not always obvious within a DAW environment with internal math that audibly masks things, and only the meters are telling you the truth.
 
You must log in or register to reply here.
Back
Top