Peak Levels - Yet Another Thread

  • Thread starter Thread starter chrisharris
  • Start date Start date
chrisharris

chrisharris

King of Bling
I understand tape. I understand recording as hot as possible to tape so that all the little ions get used up with sound instead of hiss. I understand the natural compression capabilities of tape.

But like most hobbyists today, I record digitally. I generally have my levels peak no higher than -12dBFS. I'm generally happy with my recordings, so I'm not going to change the way I do things anytime soon, but I've never been able to understand the "WHY" when it comes to digital recording. I've read pages and pages of math discussions. None of it makes any sense to me.

I don't think it's a hardware issue. I mean, whether you turn up your preamp before you hit record or later, whatever noise is generated by your preamp is going to be increased.

Anyway, it has something to do with math. I suck at math, but now that I'm trying to teach my son how to do this, I'd love to be able to give him a fairly simple explanation for this, if one exists.

So there's your challenge. Dumb it down for me, please?
 
Make sure it doesn't clip and press record. You're welcome.
 
Some folks feel that when you go for really hot digital levels (-6 dBFS to -1 dBFS), even if you don't clip, it's not good, and that you are better off keeping them in that -18 dBFS to -8 dBFS range, with only the most extreme peaks hitting around -6 dBFS.
The general logic for them thinking that comes from what might happen on you "front end" (mic, pre) in order for you to get such high levels...that in most cases you are going to be pushing real hard on your front end to spit out very hot levels, but by keeping them in the -18 dBFS to -10 dBFS range, you are most likely also keeping your front end in its good working range instead of pushing it real hard.
Also...when you end you with a lot of tracks all recorded at -1 dBFS...you then have to work to mix them so they don't overload your stereo bus when they are summed.

I get their point...however, I've not found any issue with signals at -1 dBFS VS -18 dBFS....mostly because all of my preamps are capable of very high gain and output...and I generally set the sound I want at mic & preamp, and then I look at the digital signal at the converter. As long is I'm not hitting the top....I'm going with whatever I have, because again, I'm setting the level based on the *sound* I want from the mic/pre...not about what my meters are reading at the converter (as long as they are not clipping.
AFA mixdowns...I don't sum digitally (I mix OTB)...so basically the same signal that went into the DAW comes back out, track by track, and I mix through a console. That said, when I do need to adjust levels for ITB mixing...it's easy enough to first set the level balances of the whole mix...and then just lower the stereo bus master faders so that it's not clipping.

One thing...by following the -18 dBFS to -10 dBFS "rule"...there's a lot less chance for a newbie to screw up, and they will be able to mix ITB a lot easier without fear of clipping the stereo bus...but other than that, digitally speaking, when it comes to levels, it's all math. You can raise or lower it digitally without any affect to the audio for a given track as long as you don't clip. You can raise or lower it a hundred times...it's all just math...it's not *audio signal*.
 
I disagree with the premise of recording to analog tape as hot as you can. Your levels are supposed to be bouncing around 0db on the VU meter. 0dbVU is line level. Line level is where every piece of equipment you have is designed to be run.

Line level is an average level, not a peak level. On analog tape, 0dbVU (line level) is 12-15db below the peak level that the tape can handle. The difference between the nominal level and the peak level is the headroom. The headroom is there to accommodate the peaks.

With digital, 0db is the ceiling and line level is around -18dbfs on your DAW's meters. The digital meters are reading peak level instead of average level and 0db is no longer the target.

So, gain staging with digital is exactly the same as with analog, but the meters are reading different things.

In other words, if you sent a signal to both a tape machine and a DAW, it would not read the same level. Depending on the instrument, 0dbVU on the tape machine could read anywhere from -18dbfs to -3dbfs in the DAW.
 
IK Obi said:
I shoot for -18 and never peak over -12.
Click to expand...
Ahh, come on. And right after Farview's... ;)
-18' what? Peak? '18's pretty low. And peaks are easily 6 over a -18 average, so..
:D :)
 
Last edited:
Ok I'll go.
-24 scale in Sonar, RMS + Peak.
The track hangs out at the bottom most of the scale, and the peaks land where they may.
If they ding' me we go back and listen. Quite often there's no audible damage on a small stray. (Then we have a short chat about 'mic technique or what ever. Sus it out. Adjust or not :D

chrisharris said:
I don't think it's a hardware issue. I mean, whether you turn up your preamp before you hit record or later, whatever noise is generated by your preamp is going to be increased.
Click to expand...
I gladly defer on this as I'm not super tech orientated ok? :p As I understand it at the mic pre level there is the optimum gain vs noise range for a given pre. Then (hopefully) that coincides with a reasonable match at your A/D to get to your nominal target record level. Some A/D's have adjustable sensitivities for their front end, and sometimes you may want to push the pre out of the 'normal' range, and maybe have to pad it back down.
Past that.. The dynamic range of the front end (typically), where ever you may have landed it, is narrower than the 24bit scale so it's placement in the 24 bit scale is (a) rather very flexible, and (b) once in the box' floating point or whatever they call it, that noise is the least of the worries.
 
IK Obi said:
I shoot for -18 and never peak over -12.
mixsit said:
Ahh, come on.
:D :)
Click to expand...
Click to expand...

:D

I don't sweat the numbers at the A/D converter a whole lot...but I do glance just to see if I'm too far in the red zone (starts around -6 dBFS).
I'm usually in the -12 to -8 dBFS area, with peaks tickling the red zone. Again, I pay more attention to what is going on at my mic and pre, and it just ends up in that range most times.
 
Okay, I'll bite and try to give a simple technical explanation.

The first thing to know is that measurements in dB aren't absolute. It's a way of comparing two things which is why you need a letter or letters after dB to establish the starting point. Some of the common ones are dB(u), dB(V), dB(SPL) and, in the case of digital recording, dB(FS).

The FS stands for "Full Scale" and is the point where you run out of available bits and digital clipping starts. Digital clipping sounds nasty and is to be avoided at all costs.

Unfortunately, measuring in dB(FS), while useful on your computer, is different enough from usual working methods to be confusing. When working in analogue, most meters have a 0dB level which is established at a specific "average" level well below clipping. Typically, this means that the 0dB point on an analogue meter is about 18dB below the clipping point. This means that 0dB on a conventional, non digital system is equivalent to -18dB(FS). If you think about how, on an old analogue recorder, you'd usually have the meters bouncing between 0 and +4dB, you can see why people say they record in this range on a computer DAW.

There are two reasons for this. First, you have to leave room so that no brief peaks ever hit the 0dB(FS) clipping point--and keeping your average peaks around -12dB(FS) gives you enough leeway to be safe almost all the time.

There's a second reason though. When you mix tracks together, the overall level gets louder. If you mix two tracks with peaks at -18dB(FS) and two peaks line up, the resulting peak will be at -15dB. Add another one and the level goes up by another 3dB. Don't worry about why (it's to do with logarithms) but any time you add two identical things together, the resulting sum is 3dB higher.

Clearly, when you're mixing, you need headroom to put multiple tracks together without clipping and this is another good reason to record in the -18 to -12 sort of range. Yes, you can always turn things down at the mix stage but life is easier if you can just sit down and mix rather than adjusting muliple tracks downwards first.

So...I hope I've succeeded in being technical enough to give an explanation--but simple enough to understand.

Bob
 
Bobbsy said:
Clearly, when you're mixing, you need headroom to put multiple tracks together without clipping and this is another good reason to record in the -18 to -12 sort of range. Yes, you can always turn things down at the mix stage but life is easier if you can just sit down and mix rather than adjusting muliple tracks downwards first.
Click to expand...

Yes...this is the most obvious reason, and one a lot of newbs learn about quick enough...however, there's always been this underlying "better audio at lower levels" suggestion by some folks who insist that one should keep levels down mainly for that.

I spent time reading a Gearslutz thread with a lot of pages (a WHOLE LOT - there's 127 total, I only got part way through them) about the differences in OTB and ITB mixes and what causes them, and one of the side discussions gets into tracking levels that go into a DAW VS the analog gear world. While some folks agree that -18 dBFS is a "target" to set you sights on, others like to shoot for -12 to -8 dBFS.
Even Paul Frindle, who is a big time gear designer (I believe he designed some of the SSL consoles) and who has such a deep understanding about what is really going on internally with analog and digital gear, kinda suggests that if you go too low at the A/D converter (-20-ish dBFS)...and later mix all ITB, some plugs may not work as well with those lower input levels....so he mentiones the -12 to -8 dBFS range.

I still think if you gain stage your analog front end well, and get the sound you want there, and then simply take the output from that into your A/D....the levels at the converter will be right where they need to be, but since so many of the all-in-one interfaces that the home/hobby crowd likes to use, pack everything into one box, those folks rarely work with a true "front end" before going to the converter. It's all happening in one box...and with the constant misguided perspectives about the need for loudness, most just record way too hot all the time.

For me, the gain staging is most always focused on the analog...because I often track to tape using all analog front ends...then dump to DAW mainly for edits...but then mix back out of the DAW to analog using mostly analog outboard processing through an analog console. So my analog output levels going to the DAW or ususually what is coming back from the DAW when I mix OTB...with only minor adjustments inside the DAW.
I've usually raised levels in the DAW more than I've had to lower them...but again, without ITB summing...I can do that without too much fear.

Oh...I found the link to that Gearslutz 127-page thread if anyone wants to slog through it.... :D

The Reason Most ITB mixes don’t Sound as good as Analog mixes (restored)
 
Last edited:
The simple answer is that there's no one ideal level for everything. You have to think about the type of material you have, particularly with regard to the crest factor (the peak to average ratio) of the material. I'd never advocate putting the peaks at -18dB(FS) but that's not a bad place for the "average" level, letting the peaks go somewhat higher. (How much higher? Well, it depends!)

If you're old enough to remember analogue metering, I'd probably let my peaks be some place between +4 and +8 on a PPM...this would equate to between -14 and -10dB(FS) which seems to work for me as a good compromise.

Bob
 
Bobbsy said:
There's a second reason though. When you mix tracks together, the overall level gets louder. If you mix two tracks with peaks at -18dB(FS) and two peaks line up, the resulting peak will be at -15dB. Add another one and the level goes up by another 3dB. Don't worry about why (it's to do with logarithms) but any time you add two identical things together, the resulting sum is 3dB higher.
Click to expand...

Are you sure? I thought doubling track count yielded approximately 3dB of gain, and summing identical tracks caused a 6dB increase.
 
Yes, you're quite right. Two IDENTICAL tracks summed will indeed yield 6dB of gain. The OP can try this for himself if his DAW lets him generate tones: two identical -10dB tones mixed will yield -4dB. Luckily, most music doesn't work this way!

In some editing I did in my attempts to keep it simple, I let "identical" creep in where I didn't intend to.

The principle is that each doubling of track count will result in an approximate 3dB of extra gain.

Sigh...I should never try to get too simple!
 
Farview said:
In other words, if you sent a signal to both a tape machine and a DAW, it would not read the same level. Depending on the instrument, 0dbVU on the tape machine could read anywhere from -18dbfs to -3dbfs in the DAW.
Click to expand...

Good to know......cool.
 
The need to build headroom into your digital recording has nothing to do with digital, it's all about the analog side if it. Digital handles levels all the way up to clipping very well. It's just the analog path leading to the ADC and the path coming from the DAC that tend to fall apart when they are run too hot.

Also, the peak level of a piano is going to be different than the peak level of a violin, even when they are the same volume and average level. You have to take into account the average to peak level (the crest factor) of what you are recording when you set the levels in digital.

It's kind of the opposite of setting the levels of drums in analog. if you set the level of the snare drum so that the VU meter read 0dbVU, it would be all distorted because the transient was WAY above 0dbVU. Now with digital, the meter is reading the transient and not the average, so you have to relate to it differently.
 
Farview said:
The need to build headroom into your digital recording has nothing to do with digital, it's all about the analog side if it.
Click to expand...

Ditto.

That's why I'm always paying attention to my analog front end...not really the converter levels. You have to work real hard on the analog side to bury the converter's meters.

Farview said:
You have to take into account the average to peak level (the crest factor) of what you are recording when you set the levels in digital.
Click to expand...

Right about the average to peak....though I still say it's the analog levels you need to set correctly...NOT digital levels. If you match the right reference (-10/+4) on both your analog front and your A/D converter...then set the levels at your analog front for good gain and good sound...the digital level really takes care of itself.
My converters don't even have the ability for input level adjustment...the input level at the converter is set by the analog front end output.
 
You must log in or register to reply here.
Back
Top