-20 dbFS Meaning????

Ok we were having a conversation about render quality but we started to drift off so time to start a new thread.  What confuses me still is this VU meter thing and I think it confuses a lot of people.
 
Ok I am reading my book and we all know that -20 dBFS on a DAW corresponds to 0 on a VU meter, why? and how?  I understand that if we hit 0dBFS on our DAW that means all 16/24/32/64 bits are on.
 
Now I was advising the person who started the render thread that we need to hit around -18db (this is the average level) when we use the Sonar RMS meters, I know this is correct because I have done many tests and this is the optimum level for "most" music before we master.
 
Ok so how do we differentiate 0 VU and our Sonar RMS meter's, is 0 VU even important any more in terms of mixing in the box? Should we only worry about 0 VU when we are using outboard equipment?  The last statement does not even make sense to me I use outboard gear and I never calibrate the meters (the tracks and masters all work well on all stereo's and the like), I know that I have to punch the +10 button on my outboard compressor to get an accurate reading of what is coming out of my DAW and into my compressor.
 
Can we get an explanation of this concept who understands it better than I do?  

mike_mccue


"Ok I am reading my book and we all know that -20 dBFS on a DAW corresponds to 0 on a VU meter,"

No we don't...

but you can say so if you want to... that's called "calibration".

Good Luck!

best regards,
mike

 
Hey Mike, Ok I understand that it's caibration, but explain why, how and when and if it really is important in todays digital age please.

Thanks Drew, I thought that this was the case, it's just people make you second guess yourself.

There is some confusion in kicksville's post. Firstly it is important to show that the word peak can have two meanings. In a continuous rms sinewave the peak is the value of the very top of the sinewave. The rms value is -3db down from that. So in the analog world a sinewave's peak and rms values are not the same.

This peak is also not the same as a peak transient eg a bass guitar slap followed by a smooth note. The peak in this case is where the signal may travel well above the average rms level for a short time.  And it is what our DAW peak meter is showing us.

Any signal will have an average or rms value and a peak transient value. (BTW average level and rms level only differ by about 1 db so they can be considered to be the same)

We elect to record our rms levels well down from 0 db FS in order to avoid digital clipping. What goes on at 0db FS is of no relevance. It is just good to avoid it period. But it is good to know that in certain circumstances it may not be clipped. Avoid it instead.

In the digital world a standard has been agreed that the very peaks of a continuous sinewave are the level that we call rms. So if you generate a signal at -20 db FS it is the very tops of the sinewave that are making it up to -20 db FS. The Sonar meter will actually show the true rms value which is 3 db down from that so a correct -20 db FS signal will show up as -23 db on the Sonar meter. Very hard to read down there! In the digital world the rms value is considered to be at the very peaks of a continuous sinewave. (not 3db down from there)

Hence the concept of K system. An rms meter is calibrated to read 0 db VU while the digital level is down at -20 db FS. You need a meter to actually show you something that is hitting 0 db Vu when the (rms) audio is down at -20 db FS. A real VU meter can do this and that is what I have and use. But there are digital plugins like the BlueCat meter that can do this for you as well on tracks and busses. Studio One has all this built in and the main stereo buss meters can be put into K system mode at any time without the need of a plugin. (the transient peaks are still shown)

The VU meter will miss a transient that might jump up 15 db or so (because it needs 300 ms to reach FSD) above the rms level but as we have 20 db of headroom then we don't have to worry about it so much.  K System has two other ref levels that one can work at and they are -14 db and -12db as well. Any meters need to be recalibrated as this digital ref level is altered. Because if you raise the average level in a digital system the meter has to be recalibrated to still show 0 db Vu at the new ref level.

Bob Katz tells us that you need 12 db of headroom minimum to preserve a reasonable transient but 20 db is the better option. So that is why -12db FS is as high as you should go as a ref level. But -14 db FS is common for a lot of work and - 20db for quality work. Don't forget you also have to recalibrate the monitor ref levels in your room as you alter you digital ref levels.

In digital we have great S/N ratio so we dont have to worry about keeping away from a noise floor. We can afford to work at -20 db even at 16 bit, but of course 24 bit is much better again for those reasons.

None of this applies to mastered levels either. This is all about working track, buss and master levels only. Mastering brings the rms levels up above this again as high as -6db or -3db etc. But here the transient headroom is being smashed as you do this, hence the reason very loud mastered music sounds bad.

0 db VU is the 0 db marking on a VU meter. 0 dbu and 0 dbv are the actual rms levels of the analog signals that may be floating around in and out of sound cards and external analog gear. 0 dbu is actually 0.775 v rms but it can also be a level that is higher eg +4 dbu or 1.23 v rms. -10 dbv is the domestic level. eg 0.316 v rms or -7.8 dbu to just add to the confusion!

Jeff, thanks for a very informative post!

BEST THREAD EVER!!!

AT LEAST FOR 2011.

It's an actual technical discussion where each post built on each other rather than tearing apart previous statements with posturing.

OMGoodness... I'm feeling flush.

YOU GUYS ROCK!!!!



best regards,
mike

There is a basic reason we can't all standardize our "0" level.

Audio Engineering existed long before someone attempted to apply a self aggrandizing brand name to a generic system of calibration.

Hi all.
Completely agree with Jeff Evans.This is my summary of his post in other thred.

Greetings

http://forum.cakewalk.com...?m=2336328&mpage=2

Jeff Evans


We elect to record our rms levels well down from 0 db FS in order to avoid digital clipping. What goes on at 0db FS is of no relevance. It is just good to avoid it period. But it is good to know that in certain circumstances it may not be clipped. Avoid it instead.  

I'm sorry Jeff, but perhaps you've never had any exposure to plugins that saturate or alter their sound based on level, that you might want to drive harder because they sound better that way. Or are you just like the old BBC engineer back in the 60's telling the Beatles or the Stones or the Who that they shouldn't overload the console?

In the digital world a standard has been agreed that the very peaks of a continuous sinewave are the level that we call rms. So if you generate a signal at -20 db FS it is the very tops of the sinewave that are making it up to -20 db FS. The Sonar meter will actually show the true rms value which is 3 db down from that so a correct -20 db FS signal will show up as -23 db on the Sonar meter. Very hard to read down there! In the digital world the rms value is considered to be at the very peaks of a continuous sinewave. (not 3db down from there)

Unless this is very recent, I don't believe this is an actual "standard", and thus not all programs currently agree on what "RMS" means. Or it's a "standard" that not everyone has adopted. But peak level is always based on the actual sample level of individual samples.

In digital we have great S/N ratio so we dont have to worry about keeping away from a noise floor. We can afford to work at -20 db even at 16 bit, but of course 24 bit is much better again for those reasons.

And under 32/64bit floating point digital inside Sonar, we don't have to worry about headroom either! So, if we know what we're doing, we can afford to work at +20dB too. The only reason to always avoid 0dB, is if you don't understand what's going on (and for people who don't yet understand where you might run into problems, I recommend always staying below 0dB). 

But here the transient headroom is being smashed as you do this, hence the reason very loud mastered music sounds bad.

Jeff, I know you understand this, but you're misstating it here: Headroom is completely irrelevant until you run out of it. Modern overloud music sounds bad because of ridiculously excessive digital lookahead-peak-limiting reducing the peak to average ratio, not the headroom.

Jeff, thanks for catching that. I should have said a square wave measured in the analog domain has a crest factor of 0, not a sine wave.

However, the OP's question has nothing to do with the K System. If Ben wants to know how to reference levels between his DAW and outboard compressors or other analog whatnots, then we're talking about understanding the relationship between dBFS and analog reference scales (dBu, dBV, etc.). While the K System is a useful descriptor for a standardized approach to make mixes consistent from session to session, as Mike pointed out, it has nothing to do with the practice of gain staging during recording or signal transfers from analog to digital.

In the scenario I set up (sending a feed from an analog console to a digital console for broadcast), understanding the concept of nominal levels is the key. In Ben's rig, sending a signal from SONAR into an outboard compressor, the same thing applies. I have to do this kind of signal-matching every day in contexts ranging from feeding broadcast trucks to patching touring companies' consoles into house systems and so on.

BTW, Ben: when you say you have to engage the "+10" button on your compressor to make the levels match your meters in SONAR, do you mean a switch between +4dBu and -10dBV? What comp are you using? Here's another good example of why understanding these fiddly bits matters....if you're engaging a -10dBV switch, your output from SONAR is either unbalanced or something else in your gain structure is wrong. A line level signal using a balanced connection should be referencing +4dBu (assuming of course, we're in the US )

I have stated that staying clear of 0db is a good idea. I am referring to the safety of the converters when I say this not so much what happens inside your DAW. You may know what is going on inside your DAW but what you dont know is how your converters are handling things that are very close to 0db FS though so why take the risk.

Even though the K system may not necessarily be required for tracking it can be used anyway. Why not use a standard to track by as well. It is simple, when I am setting levels of incoming signals for tracking it is just a simple adjustment of your input gain to achieve a nice 0 db VU reading. (on the rms part of the signal of course) So why not use it. I bet many here have NO reference or standard for tracking. The reason too is if I was working with Mike (and I hope to one day) and he was using a K -14 standard it just means any tracking I did for Mike would import perfectly and playback at the right level in his system without adjustment.

The K system is excellent for gain staging within all stages of a DAW session. Busses can be easily monitored and kept at the K ref level as the final masterbuss too. Some people say they are getting red lights on everywhere whereas I get none! Of course you can open up all sorts of possibilties of overdriving plugins/stages etc within your DAW and that is perfectly OK to do so for effect if required. But be aware of the final levels your converters are having to deal with though.

A good point picked up here is what analog levels are floating around outside your setup while you are using any internal digital reference. (kicksville is someone who obviously has to deal with this more than most)

The best way to find that out for Ben's interest is to measure them! The only thing about K system referencing is that in all three cases (-12db, -14db and -20 db) the levels running through your system are all different. And the levels leaving/and expected from your interface will all be different as well.

So Ben get out a decent digital true rms volt meter and find out. It was the only way I could do it. Set up a - 14 db FS sinewave inside your DAW and while it plays measure the output signal present on your sound card. If it has balanced outputs as well make sure you measure the voltage across pins 2 and 3. That is the level. Many soundcards allow you to switch (not vary) the output levels as well. Often in software.

I use a Yamaha digital mixer instead of an audio interface. It has a built in oscillator and it can send out any level. When I set the level of the oscillator to -14 db FS I get +4 dbu (1.23V) on the main outs of my mixer. Hence the reason I like the K-14db as a general rule. I know that +4 dbu is leaving the mixer when I use that ref level. So it seems Yamaha have chosen that as its internal ref level.  So at K -20 I am 6db down from that eg -2dbu and at K -12 the desk goes higher at + 6 dbu.

Another way to look at it Ben is to try various ref levels and see what ref level you need to be at in order to get +4 dbu (or -10 dbv depending on your settings)  at your outputs. BTW watch out some devices may output the actual correct level of 0.775 Volts when ref levels are achieved. (some Tascam mixers for example) You don't need to interface at all with any outboard analog gear of you don't want to either. All this only becomes more relevant if you do. But it is good to know what levels your sound card is sending at various ref levels.

I actually meant the reason why a lot of music sounds bad is because of the peak to average ratio and I just hope it can be restored back to normailty. There are some fine examples of modern jazz recordings that have a great peak to average ratio and they just sound fantastic because of it. They are not destroying all genres of music you know with mastering only some.

John T it is relevant in a way because I agree with you in that many younger guys are trying to emulate in the mixing stage what they are hearing actually took place in a separate mastering phase. You will never get any form of mastering at the mix stage to sound as good as mastering at a separate stage. Even with your same tools you will get a better result. I honestly believe that.  Once you let go of trying to make things loud during production you are free. You can track nicely and mix nicely. All you have to do is crank up your monitor volumes. The louder your monitoring levels are, the lower all the levels all become in your productions. The other secret is if you are playing loud reference material and I recommend that too, just make sure the levels that are coming from the mastered reference material are lowered to match your unmastered mix levels perfectly. That is where many go wrong. They don't level match those two things and hence the reason they are chasing volume at the mix stage.

Some mastering engineers are bad also. I have sent them lovely dynamic mixes that were well clear of 0db FS and they somehow still managed to destroy them! But you are mainly right in that many mastering engineers are good and don't do that but only make it better.

I refuse to master material that has already been smashed. I just say take it away it is already mastered! There is very little repair work you can really do to that. Some maybe but not a lot overall. The Hip Hop guys are guilty of it a lot. But I am in touch with them also at mixing stages and they are now getting the message and printing nice mixes without any mastering being applied. (unless I get to mix as I am doing now too and I love that because I have total control then)  I think its OK to hear your mixes under compression sometimes as it can change mix decisions but take it off before printing or patch the mastering processors between your mixer and your monitors and use it there instead.

Great discussion and great that for a change we are keeping it all pleasant, ok I am almost lost on this one and I am suppose to be educated on this stuff.  Here is what I am reading from all your posts in particular Jeff (hi Jeff) and Drewfx.  From what I am reading we use these measurements to calibrate the input stage of our outboard gear so as to have the maximum amount of headroom.  Correct or not?
 
There seems to be two schools of consensus here, on the one hand Jeff advocates the tried and true method but I still do not understand why we need to calibrate anything these days esp if a) we use our ears and b) we know our desk or our compressor has a certain amount of headroom.  If I am correct and all we are doing is measuring how much headroom our analog device has, as long as we don't clip our converters everything should be hunky dory.  Once again am I correct?
 
On the other hand Drewfx seems to be saying what I am saying, that is we don't need to calibrate anything.  This digital thing has got some of the older analog heads confused and people are still carrying around the myths from the bad old days when digital first appeared.  Digital has moved on from those days and Drewfx is one hundred % correct on this matter, I know because I have done so much research into this topic. 
 
This is where we have the biggest problem in the audio industry at the moment is the analog heads (and this is not a gripe or an attack just an observation) or the older engineers are still mixing as if they are in analog land even though we are in digital land.  Digital is this totally different beast and we are only now realizing how different it is but on saying that it is also funnily enough the same as analog.  Huh? you are saying what do you mean?
 
Well DrewFx is correct in saying that to get the best out of our UAD Fairchild or our UAD 1176 you have to use the device as if it were "real" this means slamming it home, into the red.  Which is totally the opposite of what the older analog heads are saying and whilst in your box you may not have "infinite" headroom it is much more than the +20db of most analog devices.
 
So what are we measuring? (I know I pissed people off in one of my classes, because I could not get my head around this concept)  I am pretty sure we are talking about the +20db headroom our analog devises are suppose to have but in the digital realm this has no relevance because we can't clip the converters. 
 
This leaves us with the RMS meters included with Sonar, with some of you saying that they are about 2db out in terms of accuracy.  I have been advising people what we need to do is when we are recording and depending on the type of music to have peaks around -6db to -3db and the average should be around -18db.  This is around about what you want with the master as well.  When you are mixing the game is to get all the tracks to have a similar average across the board.  Am I correct again?
 
Finally I use an A2 TL Audio compressor (nice bit of kit, don't listen to the reviews, it is just a little hard to use and you have to run it in before it works) on this device it has a +10 button in regards to the meter (it has nothing to do with unbalanced or balanced) I read and I will find this information at some point, that to get the VU meter to accurately measure what is coming out of your DAW you have to press this button.  It seems to be a truth to me (someone correct me if I am wrong) this is because when I use my UAD LA2 plug I turn the meter calibration button to plus 10 and all of sudden it is accurately keeping pace with Sonars RMS meter.
 
Maybe what we have here is this analog world bleeding into the digital world and nobody having any real idea of what is fact and fiction and whilst the calibration thing in the past was a must and still is if you are doing recordings the way Jeff is but if we are doing it the "modern" way then the most important measurement is the RMS value within Sonar.  I say this because, I know my levels are great and I use the -18db as a reference point.  All my mixes are now starting to transfer from the studio to the boombox.  I don't have the loudest masters on the planet in fact I usually end up with masters that average between -12db and -9db, I do this on purpose, yes I could easily push the levels higher but I don't join in the loudness wars.  I have also got that dynamic range meter and it says I have an excellent amount of headroom.
 
Ok I hope I did not go too far off topic and everyone understands my ramblings, I think what we have here is this confusion of digital and analog, whilst digital is starting to behave more like analog, it still has some nuances that we are yet to get our heads around.
 
Just remember it took 40 years for The Beatles, Pink Floyd and Queen to get analog tape and equipment to the point where they could make the masterpieces they did.  Before 1966 recordings were to of a pretty substandard (I am really talking about stuff from Bing and the like) from 1966 they knew how to make great analog recordings.  It is same with digital, the first digital recordings came out in 76, it's now 2011, that is 40 odd years, so we must be close and understanding how to make great digital recordings.  This topic and subject is one of the 1st and most important things we should learn and understand because this gives us the correct levels when we go to mix.
 
Peace, and feel free to correct me if I am wrong!   
 
 

You don't want to maximize headroom; you just want to make absolutely sure you always have enough. 24bit converters mean you don't have to worry much about quantization errors if your signal is too soft, so you can leave plenty of headroom at the converter.

And inside Sonar, or any 32/64 bit floating point DAW, you don't have to worry about the headroom of the DAW itself.

I'm not sure that Jeff and I are advocating entirely different things. I just want to make absolutely clear that there are places you can go over 0dB with no issues - there's a difference between saying "you can" and "you should".

But if Jeff or someone says "you shouldn't go over 0dB" as part of a comprehensive mixing strategy (which I think is where Jeff is coming from), then I want to make sure that we differentiate between that and "you can't". Otherwise people just get confused when you tell them something like, "Nothing bad will happen if you go over 0dB inside Sonar, but you shouldn't do it anyway". What?

But I do recommend trying to push your plugins so you learn what happens with them. If they add saturation/distortion it's almost always because the programmers did so intentionally, which implies it at least might be desirable distortion. I want to know what distorts and what doesn't, and whether I like it or not.

And it's amusing how you'll find people who will swear how certain plugins (including some of the UAD ones) add analog-style distortion/saturation, and then go on and on about how great it sounds - when in fact they don't distort at all, no matter how hard you push them. Which you'll find out, if you actually test them at high levels (or analyze what they're actually doing to your signal).

The RMS thing is just a matter of whether 0dB RMS is calibrated to a sine wave or a square wave. To get a sine wave with the same area under the curve as a square wave, you need to increase the amplitude (or peak level) of the sine wave +3dB, so that's where the 3dB difference comes from. 

Sonar, and most DAW's (but not necessarily all), use a square wave to define 0dB RMS. So if you feed it a sine wave, it's RMS level will be -3dB compared to the calibrated RMS level (and peak level). But if you test with a square wave, everything will match.

Hi Ben You are not coming across as thick. It can all be a bit confusing for sure. If you PM me with an email address I can shoot you off some test level wave files at various levels. (through You Send It) It is pretty straight forward actually. You simply import say the -14 db FS file onto a track. It will be stereo. Just make sure there is no FX or anything on tracks and busses. And all faders need to be at unity gain. If it is a sine wave it will show -17 db on the Sonar rms meters both on tracks and the masterbuss. (I might make it a square wave which should show -14 db fs then)

So that is our initial test level and it is at -14db referenced to 0db FS. Now if you can get a hold of an rms voltmeter it would be good to measure the output voltage that appears on either output 1 or 2 from your MOTU. And the voltage across the balanced output eg the hot pin and the return pin. That will tell you then what the output level is. Say it puts out exactly +4dbu (1.23v) then we know that -14db is also your ref level as is mine with the Yamaha digital mixer.

But suppose it is not +4dbu. Then it is a matter of looking at that voltage to see wether it is above or below +4dbu and by how much. You might get say +10dbu out which means that your internal ref level is actually -20 db FS and not -14 db FS. So if you drag the -20 db wave in which I will supply you with, then the output should come back to +4 dbu. So that means you should be thinking of -20db FS as your average rms level reference. (I am not talking about transient peaks now, they can be anywhere above your rms levels) I doubt the output would be lower than +4 dbu with a -14 db FS test signal. That means you will need to run at a higher ref level in order to get +4 dbu out which I doubt very much.

Suppose though with a -14db fs test wave you are getting +8dbu out. That means we are 4 db higher. So it means MOTU are thinking that -18 db FS is the correct test level for +4 dbu out.

I will include full range pink noise (at -20 db FS. You can also download this from Bob Katz's website FYI. I have created two extra files from this at -14 db fs and -12 db fs). Then you can also calibrate your monitors. The pink noise is better for setting SPL levels in your room.