Gain staging ok but main bus clipping??

  • Thread starter Thread starter Jack of Alderaan
  • Start date Start date
J

Jack of Alderaan

Member
I'm trying proper gain staging to mixed result...i'll explain : i was careful that each track in my DAW would not reach past -6dbfs for true peaks (and around -20/-18dbs RMS). I did this before and after each plugin added to a track (comp, EQ, reverb, tape emulator etc...). Having managed that i thought levels would be ok. But checking the master bus, the mix is clipping a few dbfs past zero. But the gain staging i have done was supposed to prevent that too or am i mistaken?

So what i did is to add gain unity at the end of each track's chain of plugins and remove 8db (8db equally in the dozen tracks). Now back to metering the main bus and i have this value: around -20dbfs RMS in average, with maximum true peaks of -4dbfs.

Question: is this normal and you are supposed to lower the gain on each track, even after gain staing, as to not overload the master bus? Or does this mean i did not apply enough compression on each track and the transients are too high? Did i mess it all up somewhere along the mixing?
 
Hi,
It sounds to me like what you're describing is perfectly normal.
You can be as careful as you want on a per track basis but the master is the sum of the parts so, eventually, the more tracks you add the closer you're going to get to clipping it. (Assuming some degree of frequency overlap in your tracks)

I suppose over time people get used to their own workflows and have an idea of how much room to leave on their individual tracks but it's not uncommon to have to go back and pull down each of your tracks/bus faders a little bit to make more room.
 
One area that has helped me a lot is EQ'ing out on each track areas that are not needed for the mix (focus freqs). Example is, on guitar, I EQ out those frequencies that sound like poo or has little sound at all. Same thing on the bass. But every track is going to add to the sum, the trick is making sure it adds what it is supposed to. EQ is your friend, not just for what you want, but for what you don't want that includes noises that you can't hear, but are there.
 
Hi,
It sounds to me like what you're describing is perfectly normal.
You can be as careful as you want on a per track basis but the master is the sum of the parts so, eventually, the more tracks you add the closer you're going to get to clipping it. (Assuming some degree of frequency overlap in your tracks)

I suppose over time people get used to their own workflows and have an idea of how much room to leave on their individual tracks but it's not uncommon to have to go back and pull down each of your tracks/bus faders a little bit to make more room.
Well it sounds normal to my ears at least. I understand that the frequencies will be cumulative but i didn't expect it to clip so easily with 6db of headroom on each track. I knew i would have to i remove a few dB here and there, but 8/10db on each track sounds like a lot, so maybe i messes up somewhere? That's why i Wat to hear from other people 's experience, thanks!

One area that has helped me a lot is EQ'ing out on each track areas that are not needed for the mix (focus freqs). Example is, on guitar, I EQ out those frequencies that sound like poo or has little sound at all. Same thing on the bass. But every track is going to add to the sum, the trick is making sure it adds what it is supposed to. EQ is your friend, not just for what you want, but for what you don't want that includes noises that you can't hear, but are there.
The thing is i have already cleaned the traks and removed unused/Bad frequencies. Maybe not in a perfect manner but having to remove 8db of gain on each track sounds like a lot for supposedly gain-staged, cleaned up tracks. Or maybe it's normal? As you can see, i'm not expert at all haha..
 
Well, that is part of the art/science of it all. If you have been sculpting the freqs, then look at your compression on each track. Maybe something is popping it head out making you peak over. Also, give yourself some head room on the master, and put a limiter on it but set it just to tame the peaks (I am sure you have done this, but ...)

I also think you will just have to pull everything down and then push it back up with a compressor for final. Compression is OK as long as you don't use it like a hammer and everything is a nail. Just enough to achieve the desired results. Some compromises might be required.
 
I'm no "mixer" but my understanding was, the reason you track at -18/-20dB for each track was to avoid just such an outcome?
Each pair of tracks at -18 will sum ~3dB higher, so -16dBFS and four summed tracks, 6dB higher...You can see this will eventually crash into 0dBFS! THEN, there is something called "inter-sample peaks" which can burn your bum but I confess I don't really understand it.

I repeat, I don't mix but I don't really see the problem with ending up at -8 or -12dBFS? Just bump up the master fader? Normalize?

This was of course just as much a problem with a multitrack analogue mixer.

Dave.
 
How many tracks do you have?

If you have only two tracks that peak at -6dbfs at the exact same time, you will be at 0dbfs. If you are doing something like stacking kick drum sounds, and they are all at -6dbfs, you will clip instantly.

You don't necessarily need to add a gain plugging at the end of your chain, you can just lower the faders.
 
Well, that is part of the art/science of it all. If you have been sculpting the freqs, then look at your compression on each track. Maybe something is popping it head out making you peak over. Also, give yourself some head room on the master, and put a limiter on it but set it just to tame the peaks (I am sure you have done this, but ...)

I also think you will just have to pull everything down and then push it back up with a compressor for final. Compression is OK as long as you don't use it like a hammer and everything is a nail. Just enough to achieve the desired results. Some compromises might be required.
Well, the compression seems ok, nothing that peaks too much. Actually i didn't use a limiter, i'd rather leave that for the mastering stage. But i've put a bit of compression on the main bus, slightly. Well it sounds ok to my ear, if i had to judge without looking at all the waves and meters and stuff i'd say i'm fine and i like the mix. But then i could miss some mistake i did and i wouldn't know...

I'm no "mixer" but my understanding was, the reason you track at -18/-20dB for each track was to avoid just such an outcome?
Each pair of tracks at -18 will sum ~3dB higher, so -16dBFS and four summed tracks, 6dB higher...You can see this will eventually crash into 0dBFS! THEN, there is something called "inter-sample peaks" which can burn your bum but I confess I don't really understand it.

I repeat, I don't mix but I don't really see the problem with ending up at -8 or -12dBFS? Just bump up the master fader? Normalize?

This was of course just as much a problem with a multitrack analogue mixer.

Dave.
-18/-20 because most plugins that emulates vintage hardware are working at that level which translates to 0db VU. Sweet spot. It also leaves a lot of headroom so you're sure peaks are not clipping. I'm not surprises that i had to lower the dB on each track, but -8dB sounds alarming to me!! Maybe it's not, but i thought i wouldnt need to remove so much.

Well the problem is that if i don't decrease 8db on each track then the main bus is overloading/clipping past 0db. I didn't expect to have tons of headroom, but not to the point of clipping while each track has a lot of headroom.

I've read stuff about inter-sample....i still don't get it haha.
 
How many tracks do you have?

If you have only two tracks that peak at -6dbfs at the exact same time, you will be at 0dbfs. If you are doing something like stacking kick drum sounds, and they are all at -6dbfs, you will clip instantly.

You don't necessarily need to add a gain plugging at the end of your chain, you can just lower the faders.
13 tracks. Yes i could use the faders but then if i lower them too much it's less precise to handle IMO. Ok i get it, all these db cumulates....does it mean it's normal and i'm making up problems that aren't there? The mix sounds fine tonmy ears, it's just the metering that makes me doubt about it and think "did i screw something up?"
 
Well it sounds normal to my ears at least. I understand that the frequencies will be cumulative but i didn't expect it to clip so easily with 6db of headroom on each track. I knew i would have to i remove a few dB here and there, but 8/10db on each track sounds like a lot, so maybe i messes up somewhere? That's why i Wat to hear from other people 's experience, thanks!

To me that isn’t necessarily a lot - but across the board is a bit of a bludgeon to the mix - if it sounds right to you than it is right -

The thing is i have already cleaned the traks and removed unused/Bad frequencies. Maybe not in a perfect manner but having to remove 8db of gain on each track sounds like a lot for supposedly gain-staged, cleaned up tracks. Or maybe it's normal? As you can see, i'm not expert at all haha..
How did you decide what were ‘Bad Frequencies’? Did you trying Mastering Compression on the Main Bus? I’ve used Pulsar MU to lower my output 1db -2db - when things are poping over - it is pretty clear which tracks are doing that in my mixes - Usually it’s the drums - Specifically the Kick and Snare - once in while it is the Bass and sometimes it’s the guitars - I rarely have overs though so what do I know.
 
You can also just turn down the master if you want to keep the other faders around unity.

Since every daw is floating point, nothing actually clips until the converters. You can simply turn down the master to keep it from clipping.
 
Do a little experiment... load a single track, and copy it to 4 different tracks. Mute the extras, and set your level to be -20. Now unmute each additional track and watch the main level jump each time you add another track.

This same thing happens with your mix, just on a more random basis.
 
Do a little experiment... load a single track, and copy it to 4 different tracks. Mute the extras, and set your level to be -20. Now unmute each additional track and watch the main level jump each time you add another track.

This same thing happens with your mix, just on a more random basis.
Yes, if you add two identical tracks the resultant is an increase of 6dB (as it would be for an analogue mixers) Different tracks will still add but AFAIK to about a 3dB over?

Dave.
 
Different tracks will still add but AFAIK to about a 3dB over?
Well 3dB is a rough estimate/eyeball/rule of thumb approximation. It’s what we’d expect if the source were uncorrelated noise. But I think it’s more about the average levels. Peak levels can actually accidentally line up just right so you get the full 6db. True peak is just made up guesses about what the signal might do coming out of a DAC so how that might work out when mixing things is kind of anybody’s guess, though it generally follows the real sample peaks to some extent.

I personally think the OP’s initial premise regarding “gain staging” is misguided to begin with, but it’s common misinformation propagated all over the internet at that point, and I don’t really want to pick that fight atm. I will say that if this really is how you set the final mix levels, it can’t sound good except on accident. Different tracks are supposed to have different levels!
 
To me that isn’t necessarily a lot - but across the board is a bit of a bludgeon to the mix - if it sounds right to you than it is right -


How did you decide what were ‘Bad Frequencies’? Did you trying Mastering Compression on the Main Bus? I’ve used Pulsar MU to lower my output 1db -2db - when things are poping over - it is pretty clear which tracks are doing that in my mixes - Usually it’s the drums - Specifically the Kick and Snare - once in while it is the Bass and sometimes it’s the guitars - I rarely have overs though so what do I know.
Indeed i like the result, but maybe there's a better mix to make, at least in the technical sense? Regarding "Bad frequencies" i meant unused freq like some low-end and even high end. I removed a few overtones here and there that i was not fond of but i did that lightly. In general when it comes to frequencies i try to forget the numbers and meters and just use my rats cannibales : do i like it more? Less? No significant changes? I use the DAW's EQ for subractive EQ and Pultec ému for additive EQ.

I appplied a bit a comp on the main bus but very lightly, i'd rather leave the heavy lifting of mastering to a professional. Yep i'm gonna check the drums, see if i can tame the transients more maybe?

You can also just turn down the master if you want to keep the other faders around unity.

Since every daw is floating point, nothing actually clips until the converters. You can simply turn down the master to keep it from clipping.
That's what i basically did but i uses a gain plugin at the end of each track's signal chain. The faders are at unity, or around it. Now the main bus is not clipping anymore. It sounds good to my ear. But what i wanted to know IS is it alright to remove 8db on each track so it does not clip, is this common practice? Or does that mean i didn't do something correctly?

Do a little experiment... load a single track, and copy it to 4 different tracks. Mute the extras, and set your level to be -20. Now unmute each additional track and watch the main level jump each time you add another track.

This same thing happens with your mix, just on a more random basis.
I see, it's more cumulative than i thought. Then what i did, the 8db (more or less) i had to remove on each track is not shocking? That's common practice?

Well 3dB is a rough estimate/eyeball/rule of thumb approximation. It’s what we’d expect if the source were uncorrelated noise. But I think it’s more about the average levels. Peak levels can actually accidentally line up just right so you get the full 6db. True peak is just made up guesses about what the signal might do coming out of a DAC so how that might work out when mixing things is kind of anybody’s guess, though it generally follows the real sample peaks to some extent.

I personally think the OP’s initial premise regarding “gain staging” is misguided to begin with, but it’s common misinformation propagated all over the internet at that point, and I don’t really want to pick that fight atm. I will say that if this really is how you set the final mix levels, it can’t sound good except on accident. Different tracks are supposed to have different levels!
I'm actually curious about your take on gain staging? To me it serves two purposes : avoid clipping in between plugins and emulating analog levels to suit my plugins algorithms (which emulate vintage hardware). Can you elaborate? Also, i may have oversimplified it in my post but the 8db removal on each track was just the basic start so it does not clip, then i build upon it using the faders to properly mix, all tracks do not end up at the same level in the end!
 
Ive always worked on the old principle that if you turn it up in one place, usually it comes down elsewhere. Are you hearing distortion? That needs fixing, but if you cant hear it, as was said, floating point means at that point in the chain, it might not be a problem at all. I got a bit lost but you seem to have a gain plug-in on each track? What for?
 
Ive always worked on the old principle that if you turn it up in one place, usually it comes down elsewhere. Are you hearing distortion? That needs fixing, but if you cant hear it, as was said, floating point means at that point in the chain, it might not be a problem at all. I got a bit lost but you seem to have a gain plug-in on each track? What for?
Actually i only use a gain plugin at the end of the signal chain for each track. So the faders stay at unity and then i can build upon that. What i meant is, between each plugin within a track, i use a metering plugin to check the levels and tweak the plugin output accordingly so i maintain the same level all along. Then once each track is created that way i use the faders to mix.

I didn't hear any distorsion, but it was past zero dB, so i'd rather not take any chances and avoid bad practices if i can. The purposes of my post is not to "rescue" this track per se, but to learn things about mixes and proper audio level.
 
Is that not the problem? I do the opposite in cubase. I use the level adjust in the channel strip, essential the dynamics section where the slope is 1:1 and I use the makeup gain to get the channel working with sensible fader levels. This works transparently as far as i can tell?
 
I think it's a matter of taste, both do the same thing in the end i think?
 
Don't know if it helps but there is a snap of the mixer for the demo song in Samplitude ProX6. You can see that most of the tracks are just at or below -20dBFS with just the odd one or two hitting neg ten. The master output is crowding 0dBFS but the sound from my MOTU M4 is perfectly clean.

Of course, a screen shot is just a moment it time but the tracks do just sit mainly at -20ish.

Dave.
 

Attachments

  • Sam 6 demo.webp
    Sam 6 demo.webp
    156.8 KB · Views: 47
Back
Top