Powerpoint Slides with embedded audio samples (6.2Mb
Download
Report
Transcript Powerpoint Slides with embedded audio samples (6.2Mb
The Decibel
Why all the Confusion?
Mark Rogers, The Greenbusch Group
AES - The Decibel Revisited
© 2016 Mark Rogers
First, a little Quiz:
What
is the decibel, really?
AES - The Decibel Revisited
© 2016 Mark Rogers
The decibel defined:
The
Dictionary
Webster’s
American Heritage
Wikipedia
Manufacturer’s
IEEE
Web Sites
Dictionary
AES - The Decibel Revisited
© 2016 Mark Rogers
History of the Decibel
Once
upon a time----
Telephone
companies had a need to
measure signal levels
And to compute losses from the new
cables being strung around the nation
AES - The Decibel Revisited
© 2016 Mark Rogers
Origin of the Decibel
Standard
to 1923
Cable Mile, used by Telcos prior
based on 19 gauge copper
resistance of 88 ohms per mile
capacitance of 0.054 microfarads per mile
AES - The Decibel Revisited
© 2016 Mark Rogers
Origin of the Decibel
Meanwhile,
Harvey Fletcher was doing
hearing research
To measure the minimum perceptible change
of human hearing
He called it the SU, which turned out to be
close to the SCM
AES - The Decibel Revisited
© 2016 Mark Rogers
Origin of the Decibel
1923
article by WH Martin of Bell Labs
defined the Transmission Unit (TU)
To replace both SCM and SU
1929
article by Martin renamed the TU to
be the deci-Bel, named after the company
founder, Alexander Graham Bell
AES - The Decibel Revisited
© 2016 Mark Rogers
Uses of the decibel
To
quantify the signal strength of
electricity
sound
light
radio waves
For
many types of systems
AES - The Decibel Revisited
© 2016 Mark Rogers
Basic principles
Based
on human hearing
Logarithmic
Intended to accommodate very large
ranges of levels, for which logarithms are
convenient
And allows computing gain/loss by
addition/subtraction
AES - The Decibel Revisited
© 2016 Mark Rogers
Why are we confused?
The
decibel is often:
Misused and abused (especially by those
darn audio people)
Even many manufacturers do not use it
correctly
Some apparently do not even know how!
AES - The Decibel Revisited
© 2016 Mark Rogers
What is a decibel, really?
A
decibel is 1/10 of a Bel
Why?
To make it nearly equal to the old TU
And because the Bel is too large an increment
to be useful.
But
what is a Bel??
10 decibels
AES - The Decibel Revisited
© 2016 Mark Rogers
It’s very simple!
The
decibel is a POWER RATIO,
expressed logarithmically.
It is defined as:
P1
n dB 10 log
P2
AES - The Decibel Revisited
© 2016 Mark Rogers
Wait a minute!
My
book says its 20 log something… not
10 log…
And I don’t want to measure power… I
have seen dBs based on voltages
AES - The Decibel Revisited
© 2016 Mark Rogers
The ANSI Standard
definition
decibel
is 1/10 of a
Bel
The ratio of two
amounts of power
being ten times the
log of this ratio.
Abbreviated dB
n = 10 log (P1/P2)
AES - The Decibel Revisited
When
the ratios of
currents or voltages
(or other quantities)
are the square roots
of the power ratios,
then the decibel is
also
n = 20 log (E1/E2)
© 2016 Mark Rogers
But, but, but
How
do I get to the voltage form
And what are its limitations?
Let’s
do a little algebra...
AES - The Decibel Revisited
© 2016 Mark Rogers
Derivation of the voltage
form
P1
n dB 10 log
P2
but,
2
E
P
Z
AES - The Decibel Revisited
substituting...
© 2016 Mark Rogers
Derivation of the voltage
form
therefore,
n
E12
Z1
dB 10 log
2
E2
Z2
E12 Z 2
10 log 2 x
E 2 Z1
AES - The Decibel Revisited
© 2016 Mark Rogers
Derivation of the voltage
form
and,
E1
Z2
20 log
10 log
E2
Z1
This last term is the one
most texts omit, many are
unaware of, and we all
forget!
AES - The Decibel Revisited
© 2016 Mark Rogers
Derivation of the voltage
form
E1
Z2
20 log
10 log
E2
Z1
Hmmmmm …. What if Z1 = Z2?
E1
20 log
10 log 1
E2
E1
20 log
0
E2
AES - The Decibel Revisited
Aha!
© 2016 Mark Rogers
When is a voltage ratio OK?
When
the impedance is the same for both
voltages!
AES - The Decibel Revisited
© 2016 Mark Rogers
Remember!
dB
is always a POWER RATIO!
If the impedance is the same for both
voltages, then a voltage ratio is merely a
convenient way to calculate the power
ratio to obtain the decibel value.
AES - The Decibel Revisited
© 2016 Mark Rogers
The decibel
Being
a ratio, it has no units, is not a unit
of measure
dB is dimensionless
But
it can be used as a unit of measure
How?
By making P2 an absolute reference level of
power
Written
as dBr, where r is the reference
AES - The Decibel Revisited
© 2016 Mark Rogers
Variations on the decibel
Plain
Old dB
dBr
(generic ref to standard level)
Audio
dBm
dBu
dBV
dBW
dBk
dB FS
AES - The Decibel Revisited
Acoustics
dB-SPL (Lp)
dB-PWL (Lw)
dB A, dB C
Radio
dBmv
dBµv
dBi
dBz
© 2016 Mark Rogers
Handy things to remember:
0
dBu = 0.775 volts
0 dBm = 1 milliwatt
dBu = dBm if and only if z = 600 ohms
0 dBm = 0.775 volts if and only if z = 600
ohms
0
dBV = 1.0 volts
0 dBV = 1.107 dBu
AES - The Decibel Revisited
© 2016 Mark Rogers
Ponder This
The
difference between +4 inputs and -10
inputs is not 14 dB.
It is approximately 12.8 dB
“+4” usually means 4 dBu, which is
referred to 0.775 volt
“-10” usually means -10 dBV, which is
referred to 1.0 volt
AES - The Decibel Revisited
© 2016 Mark Rogers
Sound Levels in dB
0
dB-SPL = 20 microPascals
0 dB-PWL = 1 picowatt (10-12 watt)
Beware, if you enjoy old textbooks, that some
used a different reference
that older reference was 10-13 watt, which is a 10
dB difference
AES - The Decibel Revisited
© 2016 Mark Rogers
True or False?
A
signal added to itself (mixed on two
inputs at equal levels) will double the
power (increase the mixed level by
around 3 dB)
3 dB is twice as much power
10 dB is twice (or half) as loud
10 dB is ten times as much power
AES - The Decibel Revisited
© 2016 Mark Rogers
True or False?
A
signal added to itself (mixed on two
inputs at equal levels) will double the
power (increase the mixed level by 3 dB)
Usually false.
If the signals are identical (voltage and
phase) the voltage will double, giving a 6 dB
increase.
AES - The Decibel Revisited
© 2016 Mark Rogers
True or False?
3
dB is twice as much power
False
Not exactly
Double power is 3.01029995664… dB
3.0 dB is 1.99526231497… times as much
power
AES - The Decibel Revisited
© 2016 Mark Rogers
True or False?
10
dB is twice (or half) as loud
Well... maybe... sometimes
For a standard average human, whatever that
may be...
Because it is subjective, it will vary from
person to person.
And,
it’s FALSE at low frequencies,
remember Fletcher-Munson!
AES - The Decibel Revisited
© 2016 Mark Rogers
True or False?
10
dB is ten times as much power
True, exactly.
AES - The Decibel Revisited
© 2016 Mark Rogers
Correct or Incorrect Usage?
The
signal to noise ratio is 50 dB
This mixer clips at 21 dB
It was loud! It must have been 140 dB!
I really like my Fluke because it measures
dBm directly
The power amp’s voltage gain is 26 dB
The sound pressure level was -4 dB A
AES - The Decibel Revisited
© 2016 Mark Rogers
Correct or Incorrect Usage?
The
signal to noise ratio is 50 dB
This use of dB is generally correct
(although it would be nice to know the
weighting factor and bandwidth)
AES - The Decibel Revisited
© 2016 Mark Rogers
Correct or Incorrect Usage?
The
dB.
spec sheet says this mixer clips at 21
Incorrect!
The reference is missing…..
Is is dBu, dBV, dBm, or dBcooper?
AES - The Decibel Revisited
© 2016 Mark Rogers
Correct or Incorrect Usage?
It
was loud! It must have been 140 dB!
Incorrect!
This says the sound was 1012 times more
powerful than.
Than what?
The reference and weighting are missing!
AES - The Decibel Revisited
© 2016 Mark Rogers
Correct or Incorrect Usage?
I
really like my Fluke because it measures
dBm directly.
Incorrect! It is a volt meter, not a watt
meter.
Therefore, it is measuring dBu, not dBm.
Regardless of how it is labeled on the meter
If the circuit impedance is exactly 600 ohms, they
will be equal, but…
AES - The Decibel Revisited
© 2016 Mark Rogers
Correct or Incorrect Usage?
The
dB.
Incorrect! It is missing the notation that it is
not a power ratio as the impedances differ...
power amp has a voltage gain of 26
Even with such notation, per IEEE/ANSI: this
usage “should preferably be avoided”)
A voltage gain may be properly expressed in
decilogs.
AES - The Decibel Revisited
© 2016 Mark Rogers
Correct or Incorrect Usage?
The
sound pressure level was -4 dB A
Correct!
Yes, Virginia, you can have sound levels
below 0 dB SPL
AES - The Decibel Revisited
© 2016 Mark Rogers
Easily compute dBs in your
head
Take
advantage of the power of
logarithms:
As power ratios multiply (divide), dBs add
(or subtract)
negation is inversion of the ratio
AES - The Decibel Revisited
© 2016 Mark Rogers
Handy things to remember
double
power is 3 dB, half power is -3 dB
double voltage is 6 dB, half voltage is -6
dB
quadruple power is 6 dB, 1/4 the power is
-6 dB
10 dB is ten times the power
20 dB is 100 times the power (10 times
the voltage)
AES - The Decibel Revisited
© 2016 Mark Rogers
Consider that:
The
ratio of 1 watt to 10 watts is the
same as from 100 watts to 1000 watts
Exactly 10 dB
Approximately twice as loud
Or so they say, stick around for the demo!
AES - The Decibel Revisited
© 2016 Mark Rogers
Intermission
Any
questions?
After
break
We will ponder the Fletcher-Munson curves
and listen to some sound!
AES - The Decibel Revisited
© 2016 Mark Rogers
Fletcher-Munson Curves
Equal-
loudness
contours
SPL vs.
Frequency
Using
human
subjects
AES - The Decibel Revisited
© 2016 Mark Rogers
Demonstration
How
much is 1 dB?
Barely perceptible change, they say
26% increase in power, 12% increase in
voltage
AES - The Decibel Revisited
© 2016 Mark Rogers
Demonstration
How
much is 3 dB?
Twice the power, 41% increase in voltage
(1.414 = square root of 2)
AES - The Decibel Revisited
© 2016 Mark Rogers
Demonstration
How
much is 10 dB?
Twice as loud, they say
10 times as much power, 3.16 times as
much voltage (square root of 10)
AES - The Decibel Revisited
© 2016 Mark Rogers
Demonstration
How
much is 20 dB?
Four times as loud?
100
times as much power, 10 times as
much voltage
AES - The Decibel Revisited
© 2016 Mark Rogers
Demonstration
How
loud is 0 dB-SPL?
The threshold of hearing, they say
AES - The Decibel Revisited
© 2016 Mark Rogers
Response of The Ear
3 dB
Just Perceptible
10 dB
Doubling/Halving
Pink Noise
6 dB
Noticeable
20 dB
Much More/Less
AES - The Decibel Revisited
© 2016 Mark Rogers
What about weighting?
A-weighting
is intended to approximate
the human hearing at 40 phons
Use for levels below 55 dB SPL or so
B-weighting
is intended to approximate
the human hearing at 70 phons
Use for levels below 55-85 dB SPL
C-weighting
is intended to approximate
the human hearing at high levels (flattest)
AES - The Decibel Revisited
© 2016 Mark Rogers
Listen to This
Samples
from Bob Smith
Both samples have the same level in dBA,
and similar spectral shapes
Why do they sound so different?
Factory
AES - The Decibel Revisited
Restaurant
© 2016 Mark Rogers
Now see the spectra
AES - The Decibel Revisited
© 2016 Mark Rogers
Summary
The
dB is always a POWER RATIO
n dB = 10 log (P1/P2)
You can also use 20 log (E1/E2) if the
impedances are equal
By
setting P2 to an absolute level, dB can
be used to create a unit of measure
An absolute voltage can also be used, if
impedances are equal.
AES - The Decibel Revisited
© 2016 Mark Rogers
For further study
IEEE
Standard Dictionary of Electrical and
Electronics Terms (An ANSI Standard)
Sound System Engineering, Davis,
Patronis, Brown; Chapter 7 (4th edition)
Handbook for Sound Engineers
The New Audio Cyclopedia, 1.17, 36.9, 37.7
Acoustics,
Leo Beranek, Chapter 1
Effects of Noise on Man, Karl Kryter
AES - The Decibel Revisited
© 2016 Mark Rogers
Thanks for Attending!
Any
more questions?
There are no stupid questions. There are only
stupid answers. If you ask a question, the onus
will be on me to not be stupid.
– Don Davis, SynAudCon
AES - The Decibel Revisited
© 2016 Mark Rogers
Contact me
Mark Rogers
Director of AV Dept
The Greenbusch Group
206.378.0569
[email protected]
AES - The Decibel Revisited
© 2016 Mark Rogers