No 0dbFS has meaning to refer to 0 dBFS (decibels relative to full scale) represents the maximum possible digital signal level in a digital audio system. It's the "clipping point" where the signal is as loud as it can be without distortion. Any signal exceeding 0 dBFS will be clipped, meaning digital information will be lost. Getting lost in translation in post #40 is that the PC volume is analog, so does not correlate.
It depend from the gain structure of your signal path ( more often than not dependent of your application, you won't have same requirements for a 'late night' configuration using 2w amp or a PA system for a 100.000 people crowd) and the design choice you make.
Eg you want to have max digital resolution for your converter ( use it up to 0dbfs not to loose bits of resolutions in digital realm) but want to have provision of voltage gain in case you'll be using an amplifier with low voltage gain then you might target a little bit higher than the 2 volts...
Or you'll use pro 1kw amp for sub and you need high spl at 32m in sub range then you'll need way higher output voltage...
It's all application dependant and without design goal set by yourself it can be anything....
That's why i gave the previous explanation about looking for a dbspl value at listening point... by seting the parameters for good you then know what to design and the choices you'll have to make to reach your goal... otherwise your into the wild.
without knowing which 'commercial gear' you target you won't figure out anything! 😉
Let me take another example from hifi world:
Let's say you are a Nelson Pass aficionados ( things in life can be much much worse than this) and you have an F-5 and an F-4 amps.
One have circa 18db voltage gain to reach it's full 25w rms into 8 R, the other 0db voltage gain ( even little negative gain!) as it's a (power) buffer..
The requirement for your converters voltage output requirements will be fairly differents... a 2volts max voltage will drive the F-5 no issue. But you'll have low ouput with the F-4...
A pro dac ( max output voltage +24dbu) will drive the F-4 ok and if your loudspeakers are efficient enough you might need attenuation even...
No. It's dependant from the amplifier voltage gain. If you have a 'small power' amp ( let's say a chipamp like a tda7293) the voltage gain will be +30db to obtain 100w into 8r.
If you have a large powerfull amp with the same +30db voltage gain but with 1kw into 8r capability obviously the max input voltage for max clipping will be very different...
...
Okay.. So if I put my focusrite at full volume (with the knob) and I turn the PC volume to 100 and play music from tidal/spotify. Then what the h*** does the focusrite output? the full 4.8Vrms(16dBU) or closer to 1.4Vrms with peaks approaching the 6.7Vp of 16dBU?
Or somethiing completely different?
Okay, I understand what you are saying and I understand that depending on the amplfiers gain and output capability, speaker efficiency and listening conditions the ideal input voltage may be very different.. This all makes sense for me.
But what I still do not understand is what actually comes out of my DAC and when..
Therefore its impossible for me to say how much gain I need.. Regardless of my amps gain being +30 or 0dB...
When the PCM5102 says it has 2Vrms output and the focusrite says it has 4.8Vrms output is this then under the same input condition? and what input condition is that?
What peak voltage do they actually output when I play Tidal music at full volume? This is all I need to know now in order to understand it in my head i think.
Then I can easily figure out how much gain I want in my DAC or amplifier input stage... I think xD
You can't know for sure with musical contents... as each tracks are differents!
That's why we calibrate with 'known' signals: 1khz waveform or pink noise at a given level ( digital (eg:-20dbfs) or analog ( 1,23v/+4dbu).
Let's take your pcm and focusrite:
For a known -20dbfs signal send through both, the analog signal they'll output depend from the 'standard' the designer choose to base the analog part of the circuit.
So if your pcm is 0dbfs=2v rms it'll out -20dbfs @ 0,1946V
If your scarlet is 0dbfs= 16dbu it'll out -20dbfs @ 0,48V
Both are the same digital signal value but scaled differently in the analog realm.
What tidal will ouput at max is 0dbfs. But it's not true for all tracks played as it will differ from the kind of music played (and how the 'levelling' algorythm does it things... but forget about this for the moment):
A metal or electronic music track will have low dynamic range. The styles ask for this: so they have very high rms value and low peaks. When you look at them in a software the visualisation looks like constant 'blocks'.
A classical track at the inverse will have a lot of variation about dynamic, with very quiet passages and forte... the average level will be poor ( rms) but there will be high peaks...
If tidal play the metal track it'll lower it's level and no peaks will ever reach 0dbfs. But if it plays the classical track to have same perceived level (rms) than the metal track then there will be 0dbfs peaks...
It doesn't matter if you have calibrated your main listening point at a known level target: the algorythm in Tidal will make it constant.
If like me you don't see the point to pay again for something you already have ( music on physical media) then analyzing your tracks/album to know their rms value and having a calibrated level knob will do the same as the algorythm in tidal... you just operate them manually and take a bit of time for the analysis... 😉
That's why we calibrate with 'known' signals: 1khz waveform or pink noise at a given level ( digital (eg:-20dbfs) or analog ( 1,23v/+4dbu).
Let's take your pcm and focusrite:
For a known -20dbfs signal send through both, the analog signal they'll output depend from the 'standard' the designer choose to base the analog part of the circuit.
So if your pcm is 0dbfs=2v rms it'll out -20dbfs @ 0,1946V
If your scarlet is 0dbfs= 16dbu it'll out -20dbfs @ 0,48V
Both are the same digital signal value but scaled differently in the analog realm.
What tidal will ouput at max is 0dbfs. But it's not true for all tracks played as it will differ from the kind of music played (and how the 'levelling' algorythm does it things... but forget about this for the moment):
A metal or electronic music track will have low dynamic range. The styles ask for this: so they have very high rms value and low peaks. When you look at them in a software the visualisation looks like constant 'blocks'.
A classical track at the inverse will have a lot of variation about dynamic, with very quiet passages and forte... the average level will be poor ( rms) but there will be high peaks...
If tidal play the metal track it'll lower it's level and no peaks will ever reach 0dbfs. But if it plays the classical track to have same perceived level (rms) than the metal track then there will be 0dbfs peaks...
It doesn't matter if you have calibrated your main listening point at a known level target: the algorythm in Tidal will make it constant.
If like me you don't see the point to pay again for something you already have ( music on physical media) then analyzing your tracks/album to know their rms value and having a calibrated level knob will do the same as the algorythm in tidal... you just operate them manually and take a bit of time for the analysis... 😉
Last edited:
Okay @krivium , I think I understand.
So basically the loudest possible peak from Tidal could potentially be 0dBFS, or at least very close to?
So the worst possible peak from:
So basically the loudest possible peak from Tidal could potentially be 0dBFS, or at least very close to?
So the worst possible peak from:
- The PCM would be close to 2.8Vp?
- The Focusrite could be close to 6.7Vp?
Yes.
There is confusion because: we (human) judge level on rms level ( sort of, it's a shortcut) but the musical contents have peaks ( which we do not want to modify too much as they often contain important informations) and as we can't predict signals we have to assume things and from there use signal and standard to calibrate the circuit despite they don't really mirorr the musical signal ( few things are pink noize or sinus ( but it can exist in musical context though).
And to make things even more abstract to grab the absolute limits in analog and digital are at the opposite ends of each scales ( odbfs is a max volume for digital, in analog the absolute limit is noise in the very low levels...).
I totaly get it makes headache! 🙂
There is confusion because: we (human) judge level on rms level ( sort of, it's a shortcut) but the musical contents have peaks ( which we do not want to modify too much as they often contain important informations) and as we can't predict signals we have to assume things and from there use signal and standard to calibrate the circuit despite they don't really mirorr the musical signal ( few things are pink noize or sinus ( but it can exist in musical context though).
And to make things even more abstract to grab the absolute limits in analog and digital are at the opposite ends of each scales ( odbfs is a max volume for digital, in analog the absolute limit is noise in the very low levels...).
I totaly get it makes headache! 🙂
Last edited:
So basically if I want to connect the pcm to an amplifier, that amplifier has to accept more than 2.8Vp to guarantee no clipping appear at full volume from PC?
And for the focusrite it would have to accept more than 6.7Vp.
Did I get that right?
Yes ( your math and approach seems valid to me) and no ( real life is the ultimate test):
One of the 'best' practice is to have all stage in an audio path to all clip at the same time.
That said having headroom within the amp guarantee you the amp won't clip>generate square wave> push up the harmonic content> fry tweeter.
But it ask to have either an operator aware of the limits ( don't bring level up to the max when driving a 250w max driver with a 5kw amp...) or some kind of attenuator in the path.
And it won't protect against the dac clipping either... it's all about choice and compromise you are willing to make.
Being in the engineer role is not easy task. 😉
One of the 'best' practice is to have all stage in an audio path to all clip at the same time.
That said having headroom within the amp guarantee you the amp won't clip>generate square wave> push up the harmonic content> fry tweeter.
But it ask to have either an operator aware of the limits ( don't bring level up to the max when driving a 250w max driver with a 5kw amp...) or some kind of attenuator in the path.
And it won't protect against the dac clipping either... it's all about choice and compromise you are willing to make.
Being in the engineer role is not easy task. 😉
Finally!! I actually think I understand it xD
Thank all you guys for explaining it in this detail!
Exactly my thought. I want to be able to utilize all my amplifier power for the subs for mid and tweeter I may limit the amps even further to ensure that I will stay in low distortions.
For your last two points. My products are only for personal use, so the operator knows.. until he forgets 😛
But knowing what I know now I will add a limit in the dsp (I have full control of the program on it) to ensure that the DACs will never see 0dBFS.
Again thanks everyone!
I know.. I'm a mechanical engineer by trade..
But still trying to get the hang of all the details of audio electronics, purely from interest/hobby use 🙂
Deep Purple Made in Japan " Can I have everything louder than everything else, right ? " 😎
Well, i use dsp and they offer limiters too but never used them.
My own requirements are a bit different though as i use my system for mastering duty (and have analog attenuators in signal path) so i need to be able to 'hear' when converters are overdriven ( it can happen even with limiters with a ceiling of -0.1dbfs).
That said it happened maybe two time in 15 years... so i don't find them mandatory. Even less if your gain structure is well thought.
I would expect from the Tidal algorythm to implement such '-0.1db ceilling' too. So all in all i would first try without and if you find there is some kind of 'harshness' taking place from time to time on wide dynamic range material to engage them. But it shouldn't happen imo.
About operators... well my studio is my living room, so family have access to my mains monitors.
Well theorically: once i explained to my girlfriend how to power on the thing and risks ( for gear but her ears too) she didn't want to use it. 🙂
The issue is with my kids... which are not really afraid ( how fortunate the one which doesn't get he could end up deaf...). So i implemented settings in my attenuators ( they don't know they exist!) to be sure they don't break anything when i'm not at home.
And eventually to be on safe side, i've implemented a pair of smaller monitors way easier to power up and use. In the end it's the one they prefer to use. 🙂
It should output 0 dBFS at that point. Whatever 0 dBFS is for the Focusrite. And "should". 😉 I would assume that the Focusrite provides the specified nominal output voltage at 0 dBFS, but that's an assumption.
The best way to find out is to play a sine wave recorded at 0 dBFS with the volume controls maxed and measure the output. If you get odd results, i.e., results that don't meet the published spec., I'd take a look at the waveform on an oscilloscope to make sure it isn't clipped.
You can generate a tone here: https://www.wavtones.com/functiongenerator.php
I'd go for 400 Hz, 0 dBFS, 5 seconds (the longest they support in the free version). Play the tone on repeat and measure the output of the Focusrite.
That's a good practice from an engineering perspective. Sadly, it results in a system that's perceived as weak because the end user will have to turn the volume knob past 9 o'clock. It can also make it so that the system can't play loud enough on quietly recorded albums/tracks.
If I was to design a preamp I would design it such that it could provide some amount of gain even though 0 dB with 20 dB in the power amp is ear-splitting loud in my setup.
The perception of weakness (or not very powerful) that comes from not wanting to turn the volume knob is really unfortunate because it results in systems that are much noisier than they need to be.
Tom
@tomchr,
I get your point and i should have been more accurate, this comment was made with prerequisite you have calibrated the system as i stated previously: 83dbsplC for -20dbfs signal ( pink noise) for ONE loudspeaker at listening point.
Believe me it's already LOUD.
E.G. It's the way serious movie theater are calibrated, as most have never been into a serious studio. This 83dbspl C level can be adapted somewhat as in typical home environnement ( room size related) it can sound too loud ( level of early reflections often mess up this target).
The way your level knob will be located depends of the record played ( for a constant perceived loudness between different kind of music), with higher dynamic range close to 0, typical pop from 80's close or at -6db, more recent release between -8db and lower.
This is how most of us works, so close to level the engineer used for mixing/mastering (= the initial intention, intended level of reproduction).
Of course there is reason we work at this level ( louder and we destroy our ears with long session, quieter and we tend to overuse compression to counteract the miss in perceived level) and for entertainment nothing is set in stone so people do as they like ( and this is fine to me).
But if you want to experience what the artist ( and technician obeying their wish) wanted you to hear, you know at which level to listen.
I get your point and i should have been more accurate, this comment was made with prerequisite you have calibrated the system as i stated previously: 83dbsplC for -20dbfs signal ( pink noise) for ONE loudspeaker at listening point.
Believe me it's already LOUD.
E.G. It's the way serious movie theater are calibrated, as most have never been into a serious studio. This 83dbspl C level can be adapted somewhat as in typical home environnement ( room size related) it can sound too loud ( level of early reflections often mess up this target).
The way your level knob will be located depends of the record played ( for a constant perceived loudness between different kind of music), with higher dynamic range close to 0, typical pop from 80's close or at -6db, more recent release between -8db and lower.
This is how most of us works, so close to level the engineer used for mixing/mastering (= the initial intention, intended level of reproduction).
Of course there is reason we work at this level ( louder and we destroy our ears with long session, quieter and we tend to overuse compression to counteract the miss in perceived level) and for entertainment nothing is set in stone so people do as they like ( and this is fine to me).
But if you want to experience what the artist ( and technician obeying their wish) wanted you to hear, you know at which level to listen.
Yeah. I think I will have to play around with some measurment to get a feel for what is what and how loud that is xD
But just wanted to make sure i understood the basics, and thanks to you guys I think I'm at least close 😀
@krivium To make something clear.
I "dont care" about absolut hifi. Of course I strive for flat response and good power response when designing speakers. But I may later colorize it via EQ to my preference. My ONLY goal is to make my system sound like I like it. Not like anyone else, artist or technician think it should be. That said, I am fairly close to hifi response form my speakers but with a slight emphasis on umph 😉
I also do not own a single physical media for music (I do for movies) I listen to so much different music and I often enjoy listening to new music more than something I already know. This makes physical CD's and so on impractical.. I do use TIDAL at maximum resolution though.
Also bold with small kids xD
I just made my stereo child secure in other ways, but certainly always a risk with kids and other people. Hence I wanna make sure nothing brakes at full volume. I just do it via my digital crossover. Once the filters are in place, I lower the global gain to a point where volume knob at max is just a tad louder than I wanna listen to. Then no hjarm done if someone decides to crank it 😛
But again, thanks for all the help you guys!!
Hi DannerD3H,
As i stated it's perfectly fine for me for anyone to do as they wish! All what matter is to please you.
I just gave this info as it's not widely known and put some light on choice we make in proworld ( and why).
Have fun!
Ps:
'This make CD's and so on impractical'
Like you i like to discover things and from whenever it was produced, recently, past century... doesn't matter style and such, i love music! 🙂
For older things ( which have been released on cd) i use discogs and second hand shop to buy and resell. I manage to have small amount of money invested and keep my archive growing.
For music that really are of importance to me: vinyl.
I use 'streaming' too but only to discover things, after that i buy things in physical media: more money for artists ( hopefully) and i own the recordings, so can listen to them even if a blackout over internet or main occurs ( i've got a ganerator just in case).
I know it feels like i'm paranoid but as part of the industry, let's say i don't want to support some practice i find not moral ( like i said why paid for something which have already brought huge amount of money to labels and other 'third party' involved without really support the artists?). Of course ymmv! 😉
As i stated it's perfectly fine for me for anyone to do as they wish! All what matter is to please you.
I just gave this info as it's not widely known and put some light on choice we make in proworld ( and why).
Have fun!
Ps:
'This make CD's and so on impractical'
Like you i like to discover things and from whenever it was produced, recently, past century... doesn't matter style and such, i love music! 🙂
For older things ( which have been released on cd) i use discogs and second hand shop to buy and resell. I manage to have small amount of money invested and keep my archive growing.
For music that really are of importance to me: vinyl.
I use 'streaming' too but only to discover things, after that i buy things in physical media: more money for artists ( hopefully) and i own the recordings, so can listen to them even if a blackout over internet or main occurs ( i've got a ganerator just in case).
I know it feels like i'm paranoid but as part of the industry, let's say i don't want to support some practice i find not moral ( like i said why paid for something which have already brought huge amount of money to labels and other 'third party' involved without really support the artists?). Of course ymmv! 😉
Last edited:
And I greatly appreciate that! Always good to understand why things are the way thet are 🙂