On 2006-03-18, Don Pearce wrote:
On Sat, 18 Mar 2006 17:54:15 GMT, "don" wrote:

dbFS is "decibels full scale". It is an abbreviation for decibel amplitude
levels in digital systems which have a maximum available level (like PCM
encoding). 0 dBFS is assigned to the maximum possible level. There is still
the potential for ambiguity, since some use the RMS value of a full-scale
square wave for 0 dBFS, and some use a sine wave.

No, no ambiguity, dB below full scale does not depend on wave shape,
merely how many digital levels remain unused.

This puzzled me.

The first quote (from don, not Don) is the opening part of the DBFS
entry in Wikipedia - see http://en.wikipedia.org/wiki/DBFS. I think it
is correct at least up to the final sentence about ambiguity. Then it
becomes at least ambiguous itself.

The actual ambiguity seem to be whether, when a waveform is said to
have amplitude x dBFS, you mean the peak amplitude of the waveform or
its RMS amplitude. Thus I think the fundamental ambiguity is not as
stated in the Wikipedia article about whether you use a sine or square
wave as reference.

Like Don (not don) I always assumed with dBFS you implicitly meant the
peak value of the waveform because of the nature of its representation
in a system having a waveform-independent overload level of 0 dBFS.

I had to think about this a bit when doing some FFTs (which usually work
in power/energy terms) on quantized signals. Maybe some people are more
comfortable to think of waveforms in power or energy terms however they
are represented, even when power or energy is probably no longer relevant.

--
John Phillips

On 19 Mar 2006 09:37:00 GMT, John Phillips
wrote:

On 2006-03-18, Don Pearce wrote:
On Sat, 18 Mar 2006 17:54:15 GMT, "don" wrote:

dbFS is "decibels full scale". It is an abbreviation for decibel amplitude
levels in digital systems which have a maximum available level (like PCM
encoding). 0 dBFS is assigned to the maximum possible level. There is still
the potential for ambiguity, since some use the RMS value of a full-scale
square wave for 0 dBFS, and some use a sine wave.

No, no ambiguity, dB below full scale does not depend on wave shape,
merely how many digital levels remain unused.

This puzzled me.

The first quote (from don, not Don) is the opening part of the DBFS
entry in Wikipedia - see http://en.wikipedia.org/wiki/DBFS. I think it
is correct at least up to the final sentence about ambiguity. Then it
becomes at least ambiguous itself.

The actual ambiguity seem to be whether, when a waveform is said to
have amplitude x dBFS, you mean the peak amplitude of the waveform or
its RMS amplitude. Thus I think the fundamental ambiguity is not as
stated in the Wikipedia article about whether you use a sine or square
wave as reference.

Like Don (not don) I always assumed with dBFS you implicitly meant the
peak value of the waveform because of the nature of its representation
in a system having a waveform-independent overload level of 0 dBFS.

I had to think about this a bit when doing some FFTs (which usually work
in power/energy terms) on quantized signals. Maybe some people are more
comfortable to think of waveforms in power or energy terms however they
are represented, even when power or energy is probably no longer relevant.

Think of it this way:

By how many dB would you need to increase the signal level to hit the
limit of the ADC?

That is how many dB below full scale you are, and it ties in perfectly
with my definition. You don't concern yourself with what shape the
wave is - merely how tall it is. So yes, it is the peak-to-peak
amplitude that determines this, not the RMS. The former can be derived
from the latter for known wave shapes, but not for music.

d

Pearce Consulting
http://www.pearce.uk.com

"John Phillips" wrote in message
...
On 2006-03-18, Don Pearce wrote:
On Sat, 18 Mar 2006 17:54:15 GMT, "don" wrote:

dbFS is "decibels full scale". It is an abbreviation for decibel
amplitude
levels in digital systems which have a maximum available level (like PCM
encoding). 0 dBFS is assigned to the maximum possible level. There is
still
the potential for ambiguity, since some use the RMS value of a
full-scale
square wave for 0 dBFS, and some use a sine wave.

No, no ambiguity, dB below full scale does not depend on wave shape,
merely how many digital levels remain unused.

This puzzled me.

The first quote (from don, not Don) is the opening part of the DBFS
entry in Wikipedia - see http://en.wikipedia.org/wiki/DBFS. I think it
is correct at least up to the final sentence about ambiguity. Then it
becomes at least ambiguous itself.

The actual ambiguity seem to be whether, when a waveform is said to
have amplitude x dBFS, you mean the peak amplitude of the waveform or
its RMS amplitude. Thus I think the fundamental ambiguity is not as
stated in the Wikipedia article about whether you use a sine or square
wave as reference.

Like Don (not don) I always assumed with dBFS you implicitly meant the
peak value of the waveform because of the nature of its representation
in a system having a waveform-independent overload level of 0 dBFS.

I had to think about this a bit when doing some FFTs (which usually work
in power/energy terms) on quantized signals. Maybe some people are more
comfortable to think of waveforms in power or energy terms however they
are represented, even when power or energy is probably no longer relevant.

--
John Phillips

The wave-shape doesn't matter when talking about digital signals. 0dBFS is
reached when any part of the waveform sets "all the bits to 1"
This can be the crest of a sine-wave, the tip of a sawtooth or the flat top
of a square-wave. If you have a meter that indicates dBFS, with a true-peak
characteristic, you will get the same indication whatever the waveform.
However, if you have a conventional rms reading analogue meter, driven from
a D-A converter, then the waveform will affect the indication, just as it
will for analogue waveforms that *all have the same peak value* The
commonly-used EBU standard of +18dBu=0dBFS is only valid for sine waves.

As an aside, in radio, digital metering is still done on conventional BBC
style PPMs, which under-read by anything between 1-4dB depending on the
programme content.(some will say even up to 7dB) I and others have tried
persuading radio stations to use a true-peak meter, even if it is calibrated
with the familiar BBC 1-7 scale. The universal reaction was that the signal
was too quiet, and everyone prefered to go back to a meter they were
familiar with, even if it didn't tell the truth, and rely on the 10dB
headroom between the +8dBu UK peak operating level and the +18dBu maximum to
accomodate any unseen peaks. US practice is even less precise as they still
use VU meters and rely on the 20dB headroom between 0VU (+4dBu) and their
+24dBu=0dBFS.

S.

In article ,
Serge Auckland wrote:
As an aside, in radio, digital metering is still done on conventional
BBC style PPMs, which under-read by anything between 1-4dB depending on
the programme content.(some will say even up to 7dB) I and others have
tried persuading radio stations to use a true-peak meter, even if it is
calibrated with the familiar BBC 1-7 scale. The universal reaction was
that the signal was too quiet, and everyone prefered to go back to a
meter they were familiar with, even if it didn't tell the truth, and
rely on the 10dB headroom between the +8dBu UK peak operating level and
the +18dBu maximum to accomodate any unseen peaks. US practice is even
less precise as they still use VU meters and rely on the 20dB headroom
between 0VU (+4dBu) and their +24dBu=0dBFS.

The great beauty of the analogue PPM is that it gives a good indication of
perceived loudness as well as the electrical value. It's the Holy Grail to
find something which does this better - but it hasn't happened yet.

--


Dave Plowman London SW
To e-mail, change noise into sound.

"Dave Plowman (News)" wrote in message
...
In article ,
Serge Auckland wrote:
As an aside, in radio, digital metering is still done on conventional
BBC style PPMs, which under-read by anything between 1-4dB depending on
the programme content.(some will say even up to 7dB) I and others have
tried persuading radio stations to use a true-peak meter, even if it is
calibrated with the familiar BBC 1-7 scale. The universal reaction was
that the signal was too quiet, and everyone prefered to go back to a
meter they were familiar with, even if it didn't tell the truth, and
rely on the 10dB headroom between the +8dBu UK peak operating level and
the +18dBu maximum to accomodate any unseen peaks. US practice is even
less precise as they still use VU meters and rely on the 20dB headroom
between 0VU (+4dBu) and their +24dBu=0dBFS.

The great beauty of the analogue PPM is that it gives a good indication of
perceived loudness as well as the electrical value. It's the Holy Grail to
find something which does this better - but it hasn't happened yet.

--


Dave Plowman London SW
To e-mail, change noise into sound.

It's relatively trivial to make a PPM with an LED analogue scale, arranged
in an arc if that's what's more familiar. The PPM's software can be set for
BBC dynamics, both rise and fall, or true-peak rise and conventional fall,
(or any other dynamics that you may care to think of). When we supplied
digital desks to various radio stations, we started with the PPMs indicating
true-peak rise, but within a week or two, the user always reset them to
mimic conventional mechanical pointer rise and fall. It seems that nobody's
actually interested in what the real levels are, just what it looks like -
as you say, they have a mental map of perceived loudness, and that's more
important than the actual level - after all, isn't 10dB headroom enough to
catch any nasties?

S.

In article , John Phillips
wrote:
On 2006-03-18, Don Pearce wrote:

No, no ambiguity, dB below full scale does not depend on wave shape,
merely how many digital levels remain unused.

This puzzled me.

The first quote (from don, not Don) is the opening part of the DBFS
entry in Wikipedia - see http://en.wikipedia.org/wiki/DBFS. I think it
is correct at least up to the final sentence about ambiguity. Then it
becomes at least ambiguous itself.

The actual ambiguity seem to be whether, when a waveform is said to have
amplitude x dBFS, you mean the peak amplitude of the waveform or its RMS
amplitude. Thus I think the fundamental ambiguity is not as stated in
the Wikipedia article about whether you use a sine or square wave as
reference.

Like Don (not don) I always assumed with dBFS you implicitly meant the
peak value of the waveform because of the nature of its representation
in a system having a waveform-independent overload level of 0 dBFS.

Alas, this is another one of the areas where it is easy for statements to
be ambiguous. Partly due to the confusions between instantaneous peak
levels versus rms, partly due to unspoken assumptions at times that you are
dealing with a sinewave.

To make things even more confusing wrt terminology I am currently doing
measurements and statistics of how the 'short term' peak level varies with
time with some audio waveforms. Thus I'm using peak levels, and then having
to say what the 'peak' peak level is, and how often a given 'peak' level
occurs... There are times when normal English can become hard to use to
deal with such things. :-)

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html

In article , Don Pearce
wrote:


Think of it this way:

By how many dB would you need to increase the signal level to hit the
limit of the ADC?

That is how many dB below full scale you are, and it ties in perfectly
with my definition. You don't concern yourself with what shape the wave
is - merely how tall it is. So yes, it is the peak-to-peak amplitude
that determines this, not the RMS. The former can be derived from the
latter for known wave shapes, but not for music.

Also for 'random noise' ... Although all being well, this isn't a
worry in terms of FS clipping. If it is, statisics may be the least
of your concerns. :-)

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html

In article , Serge Auckland
wrote:


When we supplied digital desks to various radio stations, we started
with the PPMs indicating true-peak rise, but within a week or two, the
user always reset them to mimic conventional mechanical pointer rise and
fall. It seems that nobody's actually interested in what the real levels
are, just what it looks like - as you say, they have a mental map of
perceived loudness, and that's more important than the actual level -
after all, isn't 10dB headroom enough to catch any nasties?

FWIW My impression is that R3 at least are generally well clear of
clipping. For example, from DAB I've not yet seen a single sample that got
to the clipping level, or even within a dB or two of it! However unless
they are clipping earlier in the chain, I guess it must happen
occasionally, simply due to the statistics of the real world, and the Laws
of Murphy. ;-

So I guess the answer to your question is similar to that for, "Will I
survive one pull of the trigger when playing Russian Roulette?"... i.e.
"Probably!" Alas, there is a distinction between trying this once, and
repeating it on a regular basis... 8-]

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html

On 2006-03-19, Don Pearce wrote:
On 19 Mar 2006 09:37:00 GMT, John Phillips
wrote:

On 2006-03-18, Don Pearce wrote:
On Sat, 18 Mar 2006 17:54:15 GMT, "don" wrote:

dbFS is "decibels full scale". It is an abbreviation for decibel amplitude
levels in digital systems which have a maximum available level (like PCM
encoding). 0 dBFS is assigned to the maximum possible level. There is still
the potential for ambiguity, since some use the RMS value of a full-scale
square wave for 0 dBFS, and some use a sine wave.

No, no ambiguity, dB below full scale does not depend on wave shape,
merely how many digital levels remain unused.

This puzzled me.

The first quote (from don, not Don) is the opening part of the DBFS
entry in Wikipedia - see http://en.wikipedia.org/wiki/DBFS. I think it
is correct at least up to the final sentence about ambiguity. Then it
becomes at least ambiguous itself.

The actual ambiguity seem to be whether, when a waveform is said to
have amplitude x dBFS, you mean the peak amplitude of the waveform or
its RMS amplitude. Thus I think the fundamental ambiguity is not as
stated in the Wikipedia article about whether you use a sine or square
wave as reference.

Like Don (not don) I always assumed with dBFS you implicitly meant the
peak value of the waveform because of the nature of its representation
in a system having a waveform-independent overload level of 0 dBFS.

I had to think about this a bit when doing some FFTs (which usually work
in power/energy terms) on quantized signals. Maybe some people are more
comfortable to think of waveforms in power or energy terms however they
are represented, even when power or energy is probably no longer relevant.

Think of it this way:

By how many dB would you need to increase the signal level to hit the
limit of the ADC?

That is how many dB below full scale you are, and it ties in perfectly
with my definition. You don't concern yourself with what shape the
wave is - merely how tall it is. So yes, it is the peak-to-peak
amplitude that determines this, not the RMS. The former can be derived
from the latter for known wave shapes, but not for music.

Exactly. I think it's the Wikipedia definition of dBFS that's puzzling.
I was wondering about re-writing the first bit to something like:


'''dBFS''' is short for "[[decibel]]s [[full scale]]". It is an
abbreviation for decibel amplitude levels in digital systems which have
a maximum available level (for example [[PCM]] encoding). By convention
0 dBFS is assigned to the maximum available level.

There is a potential for ambiguity when assigning a level on the dBFS
scale to a waveform rather than to a specific amplitude, since some
derive the characteristic level of the waveform from its peak value
while others use its [[RMS]] value. Consider a sine wave and a square
wave both of whose peak amplitudes reach the maximum avaiable level.

* Both have a peak amplitude of 0 dBFS.

* The RMS amplitude of the sine wave is approximately -3 dBFS while
the RMS amplitude of the square wave is 0 dBFS.

It is conventional to use a waveform's peak value when assigning it a
level on the dBFS scale. This is probably the more useful because -x
dBFS then means that only x dB increase can be applied to the waveform's
amplitude before [[clipping]] takes place. This is independent of the
waveform in question.

Note that there is no direct connection between a level on the dBFS
scale and any analogue signal level. If a connection is required then
a calibration level must be specified and the equipment must be set up
to achieve this. For example +18 dBu RMS sine wave = 0 dBFS peak is a
common European broadcasting calibration for analogue/digital signal
interchange. The calibration may be different in Japan and the USA.


--
John Phillips