dBu to 0 dBFS

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

kambo

Well-known member
Joined
Apr 24, 2009
Messages
1,975
Location
CA
am i thinking wrong to assume that,

if my stereo AD converter's analog input is +28dBu = 0dBFS

and if i mix on an analog console from a multitrack tape machine to hit +28dBu to see 0dBFS on my DAW
my mix would be 8dBu louder then AD convertor with +20dBu = 0dBFS input ?
 
It is not clear what you are asking.. the different 0dBFS maps to a maximum analog input level to equal digital full up.

A +20dBu = 0dBFS a/d convertor would not be happy with a +28dBu feed from an analog console.

JR
 
a +28dBu = 0dBFS a/d converter feed with +28dBu from analog  console
vs
a +20dBu = 0dBFS a/d converter feed with +20dBu from analog console....

would there be any loudness differences on final mixes...

 
kambo said:
a +28dBu = 0dBFS a/d converter feed with +28dBu from analog  console
vs
a +20dBu = 0dBFS a/d converter feed with +20dBu from analog console....

would there be any loudness differences on final mixes...

No.

Both mixes will exhibit clipping, too.
 
I *think* you're asking 2 questions that aren't directly/easily  related to each other.
+1 to Andy, but with one tiny caveat, if the analog signal is somehow different at +20dBu vs +28dBu. The box that is outputting this signal may introduce more distortion/noise in the +28 signal (or maybe less), whether perceived loudness is more/less wrt more/less distortion/noise may be subject to golden ear'd opinion.

"0dBFS" is just an abstracted concept. Consider an analogy (with some flaws): If you have 24 bits--imagine shelves for filing papers stacked from floor to ceiling, where each shelf  represents a certain volume level. 0dBFS would be the ceiling...you can't really stuff papers into the ceiling, doing so would mean that you're stacking papers above the ceiling (clipping).
The same is true for any number of bits regardless of the analog signal you're trying to represent digitally. If it's 8-bits, imagine 8 shelves(*see note below) Likewise, 16-bits just means more shelves from floor to ceiling, but the distance from floor to ceiling is still the same.

So, going back to the analog signal, if you calibrate your analog box output and your converter so that +20 or +28 is the "ceiling-bit" of the ADC, you're trying to stuff papers (signal) where there really isn't a "shelf" to accommodate it (clip).

*8-bits is actually 255 "shelves", or  256 minus 1 for the "ceiling", bc in this analogy bit-0 ("shelf 1") is the lowest level 
Reading over my post, I'm not sure if I've helped or hindered...:-X
 
Andy Peters said:
No.

Both mixes will exhibit clipping, too.
It depends on whether the meters are reading peak or VU (average), 0dBFS is the maximum peak digital saturation (no more bits).  Good peak reading meters might keep you from clipping.

Nominal 0VU (average ) should be way below full scale (20dB down? ).

JR
 
Ethan said:
If it's 8-bits, imagine 8 shelves(*see note below) Likewise, 16-bits just means more shelves from floor to ceiling, but the distance from floor to ceiling is still the same.

I think it might be worth mentioning that the practical implication of increasing bit depth would actually be analogous to the ceiling remaining where it is but lowering the floor. So the distance between floor and ceiling increases. In digital audio the result is increased dynamic range (i.e. from full signal down to the noise floor).

As John mentioned converters typically (as far as I know) are calibrated so that -18dBFS or -20dBFS = nominal operating level.
 
thank you all, but i still dont get the
difference or why/when would you use 28dBu = 0dBFS or 20 dBu = 0dBFS

where does it fit in the room floor and the ceiling... at 24bit/48khz recording u dont really care for your floor, as its almost deeper than u could reach anyway... but ur ceiling is pretty much the same ceiling...
whether u use 28dBu = 0dBFS or 20 dBu = 0dBFS

pretty much every mix engineer i know of are set/use 18/20dBu = 0dBFS
most of them claim music sounds better at 18dBu = 0dBFS 
but no technical reason...
why not use 28dBu = 0dBFS  ::)


 
You could think of the dBu to dBfs relationship as a reference level. Like nW/m for tape or cm/sec for disk recording. With tape there can be multiple reasons for choosing a reference level. The  tape type and user preference are two reasons.

With a digital interface if you are a mastering engineer and you are dealing with slammed mixes all day then 8-12dB of headroom may be enough. If you are recording live instruments all day then you might prefer to work with 24dB of headroom. If you are dealing with slammed mixes all day and have the interface set up for 24dB of headroom then you will be creaming all the analog stages with the slammed mixes.
 
kambo said:
thank you all, but i still dont get the
difference or why/when would you use 28dBu = 0dBFS or 20 dBu = 0dBFS
The difference lies in the ability of analog equipment to deal with such a high level as +28dBu. Most of vintage equipment cannot deal with it. In fact, not much modern equipment can either.
The recommended standard stateside is 0dBfs<=>+24dBu, which can be dealt with only by scaling down the input voltage by 6 dB and scaling it up by 6dB at the output. That's because most of modern equipment runs on +/-15 or 18V rails, resulting in a clipping level of +20 to +22 dBu.
The EBU recommandation is for 0dBfs<=>+18dBu, which means the analog signals can be dealt with by most existing (including vintage) equipment.


pretty much every mix engineer i know of are set/use 18/20dBu = 0dBFS
most of them claim music sounds better at 18dBu = 0dBFS 
but no technical reason...
Yes there is; even with 24dBu, most of the gear has a hard time dealing with transients.


why not use 28dBu = 0dBFS  ::)
For the reasons I just exposed. Already there are many issues with 24, so 28 would be even worse.
 
mattiasNYC said:
I think it might be worth mentioning that the practical implication of increasing bit depth would actually be analogous to the ceiling remaining where it is but lowering the floor. So the distance between floor and ceiling increases. In digital audio the result is increased dynamic range (i.e. from full signal down to the noise floor).
I understand the point you are making ;), but I've never liked the common audio analogy of "lowering the noise floor" with more bits, because even though the end result could be you have better SNR, that's not inherently the case just because we have more bits.

In a 1-bit ADC, we can only convert our analog signal to 2 possibilities ("shelves"/steps/etc):
    1 == signal_on_full_blast
    0 == signal_off
In a 2-bit ADC, we have 4 possibilities:
    11 == signal_on_full_blast
    10 == signal_loud
    01 == signal_soft
    00 == signal_off
In a 3-bit ADC, we have 8 possibilities: 
    111 == signal_on_full_blast
    110 == signal_very_loud
    101 == signal_loud
    100 == signal_medium
    011 == signal_soft
    010 == signal_very_soft
    001== signal_very_very_soft
    000 == signal_off

Having more bits allows us to capture more in-between steps.
These days, with music being crushed to 'signal_on_full_blast' :( from beginning to end, it wouldn't matter if we used a 128-bit ADC when all we're only using 1-bit resolution, signal_off and signal_on_full_blast.

But getting back to Kambo, where you decide to set your corresponding analog reference level is up to you, and whomever else you need to work with. In my studio days, we calibrated ADC inputs to -18dbFS as 0VU, that way even if the analog outputs danced above 0VU, you still have the "headroom" of 18dB before clipping.
 
The discussion about bits is a bit off topic (bad pun), when dealing with modern conversion the noise floor is not defined by word length theoretical constraints, and I won't feed that veer further.

I think Abbey may be on a richer trail. A question to inspect is does Kambo's analog console cleanly pass +28dBu..? My old console designs were only rated for +24dBu to +26dBu or so (sorry don't remember)...

I can imagine some transformer outputs hitting +28dBu cleanly (? as clean as transformers are) but most conventional solid state drivers are not going there. The well respected THAT corp 1646 output driver spec's 27.5dBu @ 0.1THD+N  so at the edge of clipping, it delivers much less than 0.01 THD+N at more normal levels (+/-18V ps rails.) Note: the 1646 has +6dB of internal voltage gain because any single ended analog path inside the console will be limited to more like +20dBu max. (even my old consoles had +6dB voltage gain in the output stage to deliver the hot rated peak output.

This may be TMI but on reflection +28dBu=0dBFS sounds a little optimistic to me, while that pretty much insures the analog path will clip before the digital path saturates. Perhaps we could have a discussion wether it is preferable to have a digital path that cannot be overloaded. In the early days of digital, overload could be very nasty, modern convertors are more tolerant. Of course nominal mix level should not be near digital 0dBFS.  :eek:

JR

PS: For the record I do not advocate clipping in either domain.
 
"analog path will clip before the digital path saturates."

that might be a good thing for me!

clipping is part of the game for me.... i do sound design work!
i use cubase master fader as my clipper, and its better than anything out there for my work....

my plan was, do the work in cubase clean, and monitor through new mixer i am building
connected to high dBU/dBFS AD to the second computer,
and monitor/record that!


attached is sample output for a custom work!

 
kambo said:
"analog path will clip before the digital path saturates."

that might be a good thing for me!
Probably not. Analog clipping takes all sorts of different shapes, depending on the topology. Also clipping due to current limit is not the same as clipping due to voltage limit, or hitting the slew-rate limit. Generally speaking, most of audio gear is not designed for euphonic clipping, which is rather serendipitious.


clipping is part of the game for me.... i do sound design work!
i use cubase master fader as my clipper, and its better than anything out there for my work....
A dedicated clipper, particularly a digital one, requires a lot of work. Basic clipping in the digital domain is one of the easiest operations (takes just one flop), and it sounds awful. In order to make it sound nice, it takes a lot of tuning and adaptive EQ'ing. In the end, it's a powerful and CPU intensive algorithm.
 
abbey road d enfer said:
Probably not. Analog clipping takes all sorts of different shapes, depending on the topology. Also clipping due to current limit is not the same as clipping due to voltage limit, or hitting the slew-rate limit. Generally speaking, most of audio gear is not designed for euphonic clipping, which is rather serendipitious.
This is actually a topic of discussion surrounding digital console design.  At least one digital platform I know of (Midas) anticipates the operator clipping the analog mic preamp before the A/D conversion and incorporates their flavor of secret sauce "soft clipping".  Some customers have incorporated this into their signature sound (they like it).
A dedicated clipper, particularly a digital one, requires a lot of work. Basic clipping in the digital domain is one of the easiest operations (takes just one flop), and it sounds awful. In order to make it sound nice, it takes a lot of tuning and adaptive EQ'ing. In the end, it's a powerful and CPU intensive algorithm.
Depends on what the function or definition of that "clipper" is. Clipping is easy, clipping while not sounding like crap***, less easy.  Analog clip limiters that don't suck are a decades old mature technology. At least one manufacturer (dbx) tried (with questionable success) to incorporate a soft limiter into the A/D conversion, IIRC they squeezed more than 6dB of dynamic range into that top bit (probably sounded better on paper)... I haven't heard about it lately so it probably died a quiet death.

I am not a fan of the midas approach either, but the customer is always the customer.

JR

*** not fair to characterize all clipping as making one sound.. a little clipping on very narrow peaks can enhance the presentation. Clipping the LF envelope, will generate much more audible artifacts (think guitar fuzz tone).
 
abbey road d enfer said:
Probably not. Analog clipping takes all sorts of different shapes, depending on the topology. Also clipping due to current limit is not the same as clipping due to voltage limit, or hitting the slew-rate limit. Generally speaking, most of audio gear is not designed for euphonic clipping, which is rather serendipitious.

A dedicated clipper, particularly a digital one, requires a lot of work. Basic clipping in the digital domain is one of the easiest operations (takes just one flop), and it sounds awful. In order to make it sound nice, it takes a lot of tuning and adaptive EQ'ing. In the end, it's a powerful and CPU intensive algorithm.

High quality A/D converters IME handle clipping well. Lots of mastering engineers clip the A/D for level. IME whether clipping will sound good or not is situational. If it's percussive material with a massive crest factor clipping often sounds better than a limiter. If the material has a high average level then limiting often sounds better. I'd disagree that digital clipping sounds awful. It can but used sparingly it can be very helpful.
 
Gold said:
High quality A/D converters IME handle clipping well. Lots of mastering engineers clip the A/D for level. IME whether clipping will sound good or not is situational. If it's percussive material with a massive crest factor clipping often sounds better than a limiter. If the material has a high average level then limiting often sounds better. I'd disagree that digital clipping sounds awful. It can but used sparingly it can be very helpful.
  I was writing in the context of clipping used as a sound design tool, where it's meant to be heard and rather continuous; in this context, basic digital truncation does sound awful.
 
yea, i heard some outboards plugins etc...
they do sound good on very narrow specific material... not really usefull...
ie: sounds good on gritty bass, on certain notes, (C2 and D2 only )  not over the whole scale of bass...
when they promote it, they only use those very specific sounds and notes....
so, they usually die out...
 
Gold said:
High quality A/D converters IME handle clipping well.
JRsez said:
In the early days of digital, overload could be very nasty, modern convertors are more tolerant.
Lots of mastering engineers clip the A/D for level. IME whether clipping will sound good or not is situational. If it's percussive material with a massive crest factor clipping often sounds better than a limiter. If the material has a high average level then limiting often sounds better. I'd disagree that digital clipping sounds awful. It can but used sparingly it can be very helpful.
JRsez said:
*** not fair to characterize all clipping as making one sound.. a little clipping on very narrow peaks can enhance the presentation. Clipping the LF envelope, will generate much more audible artifacts (think guitar fuzz tone).

Did you ever hear a digital path rollover one tick past FS? Modern convertors protect modern users from that, but back in the day a full level step would surely get your attention and ruin the mix.

JR

PS: Clipping an analog path will do the same thing, but I don't have to like it. Good design means all audio paths recover from clipping quickly with no spurious artifacts.
 
Back
Top