Reply
 
Thread Tools Display Modes
  #401   Report Post  
Glenn Booth
 
Posts: n/a
Default Distorsion percentage, power or voltage?

Hi,

In message , gecwhite
writes

The twenty comes in because the "voltage dB" is essentially derived from
the "power dB," but power is a non-linear function (square law) of
voltage.


Yes, I got that; I just needed to know which came first :-)

By _Bell System Blue Book definition_:

X(dB) = 10*log(P/P_ref)


*That's* what I was looking for; many thanks. Is the Bell System Blue
book available? Call it sad academic interest.

[Much good stuff snipped to the clipboard for later]

I hope this clears up some of your questions.


All of them. Thanks very much.

--
Regards,
Glenn Booth
  #402   Report Post  
chung
 
Posts: n/a
Default Distorsion percentage, power or voltage?

Bob-Stanton wrote:

chung wrote in message news:e5d5b$4012a5ca$c247604

.............. Even in that case, a lumped model
is good enough, considering all the approximations one uses in that type
of anslysis.


FET output devices can oscillate at a very high frequency. One would
need a speaker cable model that was accurate at very high frequencies

The lumped element speakercable models have a limited bandwidth. I
doubt that you could write a lumped element model, for a 100 ft of
cable, that would be accurate above 200 KHz.


So consider the approximations. How accurately can you model the power
amp? How about the connection between the output transistors to the
terminals at the back of the amp? How accurately do you model the skin
effect and resistive losses? How about the loss tangents of the
dielectric? How close to an ideal transmission line is a pair of wires
side by side? How about the discontinuity between the cable and the
speaker terminals?

In the end, you are making many approximations in such calculations.
Whether to approximate the cable with a lumped model or a transmission
line model is a small difference, compared to the large uncertainties in
the rest of the system.

It will be like trying to find an answer to 5 significant digits, when
all your inputs only have 2 significant digits. You are chasing 4th
order effects when you don't even have second order effects under control.

Besides, looking at transmission line loading is the wrong way to
analyze such amps. Much better to vary the output loading and make sure
that the amp is stable for all possible loads. And you would model that
with resistive, capacitve and inductive loads.

Trust me, no one analyzes the stability of an audio amp by using
transmission line models of cables.


One can use Maxwell equations and quantum mechanics to analyze speaker
cables, too. Have you thought about doing that?


I think that what you are saying is: one can go too far in analyzing
speaker cables.

Suppose you were writing a circuit analysis program that was to
include a model of typical 12 gage speaker cable (any length).


Why bother? Suppose you were to write a circuit analysis program based
on quantum mechanics?

This
model could be useful for educational perposes, and also would have
some pratical applications.


Not for audio cables or audio amps.

Would you write a speaker cable model that was "dead on" accurate, or
write a (close approximation) component type model?


See all the approximations above.

Bob Stanton


  #403   Report Post  
chung
 
Posts: n/a
Default Distorsion percentage, power or voltage?

Bob-Stanton wrote:

chung wrote in message news:e5d5b$4012a5ca$c247604

.............. Even in that case, a lumped model
is good enough, considering all the approximations one uses in that type
of anslysis.


FET output devices can oscillate at a very high frequency. One would
need a speaker cable model that was accurate at very high frequencies

The lumped element speakercable models have a limited bandwidth. I
doubt that you could write a lumped element model, for a 100 ft of
cable, that would be accurate above 200 KHz.


So consider the approximations. How accurately can you model the power
amp? How about the connection between the output transistors to the
terminals at the back of the amp? How accurately do you model the skin
effect and resistive losses? How about the loss tangents of the
dielectric? How close to an ideal transmission line is a pair of wires
side by side? How about the discontinuity between the cable and the
speaker terminals?

In the end, you are making many approximations in such calculations.
Whether to approximate the cable with a lumped model or a transmission
line model is a small difference, compared to the large uncertainties in
the rest of the system.

It will be like trying to find an answer to 5 significant digits, when
all your inputs only have 2 significant digits. You are chasing 4th
order effects when you don't even have second order effects under control.

Besides, looking at transmission line loading is the wrong way to
analyze such amps. Much better to vary the output loading and make sure
that the amp is stable for all possible loads. And you would model that
with resistive, capacitve and inductive loads.

Trust me, no one analyzes the stability of an audio amp by using
transmission line models of cables.


One can use Maxwell equations and quantum mechanics to analyze speaker
cables, too. Have you thought about doing that?


I think that what you are saying is: one can go too far in analyzing
speaker cables.

Suppose you were writing a circuit analysis program that was to
include a model of typical 12 gage speaker cable (any length).


Why bother? Suppose you were to write a circuit analysis program based
on quantum mechanics?

This
model could be useful for educational perposes, and also would have
some pratical applications.


Not for audio cables or audio amps.

Would you write a speaker cable model that was "dead on" accurate, or
write a (close approximation) component type model?


See all the approximations above.

Bob Stanton


  #404   Report Post  
chung
 
Posts: n/a
Default Distorsion percentage, power or voltage?

Bob-Stanton wrote:

chung wrote in message news:e5d5b$4012a5ca$c247604

.............. Even in that case, a lumped model
is good enough, considering all the approximations one uses in that type
of anslysis.


FET output devices can oscillate at a very high frequency. One would
need a speaker cable model that was accurate at very high frequencies

The lumped element speakercable models have a limited bandwidth. I
doubt that you could write a lumped element model, for a 100 ft of
cable, that would be accurate above 200 KHz.


So consider the approximations. How accurately can you model the power
amp? How about the connection between the output transistors to the
terminals at the back of the amp? How accurately do you model the skin
effect and resistive losses? How about the loss tangents of the
dielectric? How close to an ideal transmission line is a pair of wires
side by side? How about the discontinuity between the cable and the
speaker terminals?

In the end, you are making many approximations in such calculations.
Whether to approximate the cable with a lumped model or a transmission
line model is a small difference, compared to the large uncertainties in
the rest of the system.

It will be like trying to find an answer to 5 significant digits, when
all your inputs only have 2 significant digits. You are chasing 4th
order effects when you don't even have second order effects under control.

Besides, looking at transmission line loading is the wrong way to
analyze such amps. Much better to vary the output loading and make sure
that the amp is stable for all possible loads. And you would model that
with resistive, capacitve and inductive loads.

Trust me, no one analyzes the stability of an audio amp by using
transmission line models of cables.


One can use Maxwell equations and quantum mechanics to analyze speaker
cables, too. Have you thought about doing that?


I think that what you are saying is: one can go too far in analyzing
speaker cables.

Suppose you were writing a circuit analysis program that was to
include a model of typical 12 gage speaker cable (any length).


Why bother? Suppose you were to write a circuit analysis program based
on quantum mechanics?

This
model could be useful for educational perposes, and also would have
some pratical applications.


Not for audio cables or audio amps.

Would you write a speaker cable model that was "dead on" accurate, or
write a (close approximation) component type model?


See all the approximations above.

Bob Stanton


  #405   Report Post  
chung
 
Posts: n/a
Default Distorsion percentage, power or voltage?

Bob-Stanton wrote:

chung wrote in message news:e5d5b$4012a5ca$c247604

.............. Even in that case, a lumped model
is good enough, considering all the approximations one uses in that type
of anslysis.


FET output devices can oscillate at a very high frequency. One would
need a speaker cable model that was accurate at very high frequencies

The lumped element speakercable models have a limited bandwidth. I
doubt that you could write a lumped element model, for a 100 ft of
cable, that would be accurate above 200 KHz.


So consider the approximations. How accurately can you model the power
amp? How about the connection between the output transistors to the
terminals at the back of the amp? How accurately do you model the skin
effect and resistive losses? How about the loss tangents of the
dielectric? How close to an ideal transmission line is a pair of wires
side by side? How about the discontinuity between the cable and the
speaker terminals?

In the end, you are making many approximations in such calculations.
Whether to approximate the cable with a lumped model or a transmission
line model is a small difference, compared to the large uncertainties in
the rest of the system.

It will be like trying to find an answer to 5 significant digits, when
all your inputs only have 2 significant digits. You are chasing 4th
order effects when you don't even have second order effects under control.

Besides, looking at transmission line loading is the wrong way to
analyze such amps. Much better to vary the output loading and make sure
that the amp is stable for all possible loads. And you would model that
with resistive, capacitve and inductive loads.

Trust me, no one analyzes the stability of an audio amp by using
transmission line models of cables.


One can use Maxwell equations and quantum mechanics to analyze speaker
cables, too. Have you thought about doing that?


I think that what you are saying is: one can go too far in analyzing
speaker cables.

Suppose you were writing a circuit analysis program that was to
include a model of typical 12 gage speaker cable (any length).


Why bother? Suppose you were to write a circuit analysis program based
on quantum mechanics?

This
model could be useful for educational perposes, and also would have
some pratical applications.


Not for audio cables or audio amps.

Would you write a speaker cable model that was "dead on" accurate, or
write a (close approximation) component type model?


See all the approximations above.

Bob Stanton




  #406   Report Post  
Bazza
 
Posts: n/a
Default Distorsion percentage, power or voltage?

On Sun, 25 Jan 2004 20:33:59 GMT, gecwhite wrote:

As one who makes his living in the RF field, I say you are flatly
wrong. Distortion is an acute matter of concern in modern digital
radios that require high linearity. For example, massive amounts of
money are spent in feedforward power amplifiers for cell base stations.
For example, PA non-linearity is one of the major limiting factors in
WLAN OFDM transmitter link budgets (52 subcarriers and up to 64-QAM on
each subcarrier). And it goes on and on.


My posts have been in accordance with "the generalities" of dB's
and dBm's and was prompted in response to a particular posting.
With this in mind, it is possible to find exceptions within disciplines.
CATV was merely one of these. So is yours. You get no argument
from me in this respect.



A bandpass filter could be used to reduce out of band signals, sometimes
lowpass, even less frequently, highpass. This would be especially the case for
RF amplifiers, broadband RF amps and to some extent transmitters etc.


BP filters can't remove odd-order intermod products, and no one finds
sine waves, by themselves, particularly interesting, even music
listeners. Because CATV equipment is so wide-banded, even-order
distortion is also major concern. That is why balanced amplifiers
dominate that field.

No one has argued or proposed that they do.
Read what I have said about CATV.

However, we are biginning to stray from the original post. It was your
second-line statement, above, to which I was makeing an exception

I agree. They use dB as a relative level, ...


A dB, by itself, it always just "relative," by definition. It is a
ratio and nothing else.


No one has said otherwise. Thanks


...when it comes to specifying
distortion. When I wrote "distortion in -dB(power)", I ment the
distortion power *relative* to the signal power.


Which I believe should almost go without needing to say so.

Hmmm. I checked my posts.
I believe you may be quoting someone else.


CATV's are outside my experience. Nor did I particularly single them out for
distinction. My point was that most people in the side industries (non-audio)
which you cited, still use and properly so, dB's as a relative term...


It can't be used in any way _but_ relative, because no absolute
reference point can be gleaned from it in the absence of other
information!

Precisely.
So we agree even though my points are understated rather than emphatic.


... and dBm's
when implying a 'standard' level.


Sure it is a "standard" notation, but more importantly it refers to an
_absolute_ power level.

Precisely.
So we agree even though my points are understated rather than emphatic.


Now. This really ends my interest in this thread.
  #407   Report Post  
Bazza
 
Posts: n/a
Default Distorsion percentage, power or voltage?

On Sun, 25 Jan 2004 20:33:59 GMT, gecwhite wrote:

As one who makes his living in the RF field, I say you are flatly
wrong. Distortion is an acute matter of concern in modern digital
radios that require high linearity. For example, massive amounts of
money are spent in feedforward power amplifiers for cell base stations.
For example, PA non-linearity is one of the major limiting factors in
WLAN OFDM transmitter link budgets (52 subcarriers and up to 64-QAM on
each subcarrier). And it goes on and on.


My posts have been in accordance with "the generalities" of dB's
and dBm's and was prompted in response to a particular posting.
With this in mind, it is possible to find exceptions within disciplines.
CATV was merely one of these. So is yours. You get no argument
from me in this respect.



A bandpass filter could be used to reduce out of band signals, sometimes
lowpass, even less frequently, highpass. This would be especially the case for
RF amplifiers, broadband RF amps and to some extent transmitters etc.


BP filters can't remove odd-order intermod products, and no one finds
sine waves, by themselves, particularly interesting, even music
listeners. Because CATV equipment is so wide-banded, even-order
distortion is also major concern. That is why balanced amplifiers
dominate that field.

No one has argued or proposed that they do.
Read what I have said about CATV.

However, we are biginning to stray from the original post. It was your
second-line statement, above, to which I was makeing an exception

I agree. They use dB as a relative level, ...


A dB, by itself, it always just "relative," by definition. It is a
ratio and nothing else.


No one has said otherwise. Thanks


...when it comes to specifying
distortion. When I wrote "distortion in -dB(power)", I ment the
distortion power *relative* to the signal power.


Which I believe should almost go without needing to say so.

Hmmm. I checked my posts.
I believe you may be quoting someone else.


CATV's are outside my experience. Nor did I particularly single them out for
distinction. My point was that most people in the side industries (non-audio)
which you cited, still use and properly so, dB's as a relative term...


It can't be used in any way _but_ relative, because no absolute
reference point can be gleaned from it in the absence of other
information!

Precisely.
So we agree even though my points are understated rather than emphatic.


... and dBm's
when implying a 'standard' level.


Sure it is a "standard" notation, but more importantly it refers to an
_absolute_ power level.

Precisely.
So we agree even though my points are understated rather than emphatic.


Now. This really ends my interest in this thread.
  #408   Report Post  
Bazza
 
Posts: n/a
Default Distorsion percentage, power or voltage?

On Sun, 25 Jan 2004 20:33:59 GMT, gecwhite wrote:

As one who makes his living in the RF field, I say you are flatly
wrong. Distortion is an acute matter of concern in modern digital
radios that require high linearity. For example, massive amounts of
money are spent in feedforward power amplifiers for cell base stations.
For example, PA non-linearity is one of the major limiting factors in
WLAN OFDM transmitter link budgets (52 subcarriers and up to 64-QAM on
each subcarrier). And it goes on and on.


My posts have been in accordance with "the generalities" of dB's
and dBm's and was prompted in response to a particular posting.
With this in mind, it is possible to find exceptions within disciplines.
CATV was merely one of these. So is yours. You get no argument
from me in this respect.



A bandpass filter could be used to reduce out of band signals, sometimes
lowpass, even less frequently, highpass. This would be especially the case for
RF amplifiers, broadband RF amps and to some extent transmitters etc.


BP filters can't remove odd-order intermod products, and no one finds
sine waves, by themselves, particularly interesting, even music
listeners. Because CATV equipment is so wide-banded, even-order
distortion is also major concern. That is why balanced amplifiers
dominate that field.

No one has argued or proposed that they do.
Read what I have said about CATV.

However, we are biginning to stray from the original post. It was your
second-line statement, above, to which I was makeing an exception

I agree. They use dB as a relative level, ...


A dB, by itself, it always just "relative," by definition. It is a
ratio and nothing else.


No one has said otherwise. Thanks


...when it comes to specifying
distortion. When I wrote "distortion in -dB(power)", I ment the
distortion power *relative* to the signal power.


Which I believe should almost go without needing to say so.

Hmmm. I checked my posts.
I believe you may be quoting someone else.


CATV's are outside my experience. Nor did I particularly single them out for
distinction. My point was that most people in the side industries (non-audio)
which you cited, still use and properly so, dB's as a relative term...


It can't be used in any way _but_ relative, because no absolute
reference point can be gleaned from it in the absence of other
information!

Precisely.
So we agree even though my points are understated rather than emphatic.


... and dBm's
when implying a 'standard' level.


Sure it is a "standard" notation, but more importantly it refers to an
_absolute_ power level.

Precisely.
So we agree even though my points are understated rather than emphatic.


Now. This really ends my interest in this thread.
  #409   Report Post  
Bazza
 
Posts: n/a
Default Distorsion percentage, power or voltage?

On Sun, 25 Jan 2004 20:33:59 GMT, gecwhite wrote:

As one who makes his living in the RF field, I say you are flatly
wrong. Distortion is an acute matter of concern in modern digital
radios that require high linearity. For example, massive amounts of
money are spent in feedforward power amplifiers for cell base stations.
For example, PA non-linearity is one of the major limiting factors in
WLAN OFDM transmitter link budgets (52 subcarriers and up to 64-QAM on
each subcarrier). And it goes on and on.


My posts have been in accordance with "the generalities" of dB's
and dBm's and was prompted in response to a particular posting.
With this in mind, it is possible to find exceptions within disciplines.
CATV was merely one of these. So is yours. You get no argument
from me in this respect.



A bandpass filter could be used to reduce out of band signals, sometimes
lowpass, even less frequently, highpass. This would be especially the case for
RF amplifiers, broadband RF amps and to some extent transmitters etc.


BP filters can't remove odd-order intermod products, and no one finds
sine waves, by themselves, particularly interesting, even music
listeners. Because CATV equipment is so wide-banded, even-order
distortion is also major concern. That is why balanced amplifiers
dominate that field.

No one has argued or proposed that they do.
Read what I have said about CATV.

However, we are biginning to stray from the original post. It was your
second-line statement, above, to which I was makeing an exception

I agree. They use dB as a relative level, ...


A dB, by itself, it always just "relative," by definition. It is a
ratio and nothing else.


No one has said otherwise. Thanks


...when it comes to specifying
distortion. When I wrote "distortion in -dB(power)", I ment the
distortion power *relative* to the signal power.


Which I believe should almost go without needing to say so.

Hmmm. I checked my posts.
I believe you may be quoting someone else.


CATV's are outside my experience. Nor did I particularly single them out for
distinction. My point was that most people in the side industries (non-audio)
which you cited, still use and properly so, dB's as a relative term...


It can't be used in any way _but_ relative, because no absolute
reference point can be gleaned from it in the absence of other
information!

Precisely.
So we agree even though my points are understated rather than emphatic.


... and dBm's
when implying a 'standard' level.


Sure it is a "standard" notation, but more importantly it refers to an
_absolute_ power level.

Precisely.
So we agree even though my points are understated rather than emphatic.


Now. This really ends my interest in this thread.
  #410   Report Post  
Bob-Stanton
 
Posts: n/a
Default Distorsion percentage, power or voltage?

chung wrote in messag ...

How accurately do you model the skin
effect and resistive losses? How about the loss tangents of the
dielectric? How close to an ideal transmission line is a pair of wires
side by side?


An S-parameter based cable model, would include all of the above factors and
more.





In the end, you are making many approximations in such calculations.
Whether to approximate the cable with a lumped model or a transmission
line model is a small difference, compared to the large uncertainties in
the rest of the system.


You say that the uncentaninties in the rest of the system are greater
than the differences between a transmission line model, and a crude
components cable model. That is true, in the application you suggested, but
that is not the only possible application for a cable model.





It will be like trying to find an answer to 5 significant digits, when
all your inputs only have 2 significant digits. You are chasing 4th
order effects when you don't even have second order effects under control.
...
Besides, looking at transmission line loading is the wrong way to
analyze such amps.



It was *you*, who in a previous message, suggested using a cable model,
for testing amplifier stability.


Much better to vary the output loading and make sure
that the amp is stable for all possible loads. And you would model that
with resistive, capacitve and inductive loads.

Trust me, no one analyzes the stability of an audio amp by using
transmission line models of cables.


You are the one who suggested using a cable model for testing amplifier
stability.




Why bother?


Because you can't think of any application that requires an accurate cable
model, you think one is not needed. You are being like the man in the
1880's, who suggested closing the US Patent Office, because "all possible
inventions" had already been made.





Suppose you were to write a circuit analysis program based
on quantum mechanics?


It is not necessary. An S-parameter based cable model, already takes into
account *all the factors* that cause cable loss, including quantum effects,
if any.





This
model could be useful for educational perposes, and also would have
some pratical applications.


Not for audio cables or audio amps.


Again, you are being very short sighted.






Would you write a speaker cable model that was "dead on" accurate, or
write a (close approximation) component type model?


See all the approximations above.


The "approximations above" are only for a limited case. They does not prove
there will never be a need for an accurate cable model.

Bob Stantone news:c0439$40143681$c247604


  #411   Report Post  
Bob-Stanton
 
Posts: n/a
Default Distorsion percentage, power or voltage?

chung wrote in messag ...

How accurately do you model the skin
effect and resistive losses? How about the loss tangents of the
dielectric? How close to an ideal transmission line is a pair of wires
side by side?


An S-parameter based cable model, would include all of the above factors and
more.





In the end, you are making many approximations in such calculations.
Whether to approximate the cable with a lumped model or a transmission
line model is a small difference, compared to the large uncertainties in
the rest of the system.


You say that the uncentaninties in the rest of the system are greater
than the differences between a transmission line model, and a crude
components cable model. That is true, in the application you suggested, but
that is not the only possible application for a cable model.





It will be like trying to find an answer to 5 significant digits, when
all your inputs only have 2 significant digits. You are chasing 4th
order effects when you don't even have second order effects under control.
...
Besides, looking at transmission line loading is the wrong way to
analyze such amps.



It was *you*, who in a previous message, suggested using a cable model,
for testing amplifier stability.


Much better to vary the output loading and make sure
that the amp is stable for all possible loads. And you would model that
with resistive, capacitve and inductive loads.

Trust me, no one analyzes the stability of an audio amp by using
transmission line models of cables.


You are the one who suggested using a cable model for testing amplifier
stability.




Why bother?


Because you can't think of any application that requires an accurate cable
model, you think one is not needed. You are being like the man in the
1880's, who suggested closing the US Patent Office, because "all possible
inventions" had already been made.





Suppose you were to write a circuit analysis program based
on quantum mechanics?


It is not necessary. An S-parameter based cable model, already takes into
account *all the factors* that cause cable loss, including quantum effects,
if any.





This
model could be useful for educational perposes, and also would have
some pratical applications.


Not for audio cables or audio amps.


Again, you are being very short sighted.






Would you write a speaker cable model that was "dead on" accurate, or
write a (close approximation) component type model?


See all the approximations above.


The "approximations above" are only for a limited case. They does not prove
there will never be a need for an accurate cable model.

Bob Stantone news:c0439$40143681$c247604
  #412   Report Post  
Bob-Stanton
 
Posts: n/a
Default Distorsion percentage, power or voltage?

chung wrote in messag ...

How accurately do you model the skin
effect and resistive losses? How about the loss tangents of the
dielectric? How close to an ideal transmission line is a pair of wires
side by side?


An S-parameter based cable model, would include all of the above factors and
more.





In the end, you are making many approximations in such calculations.
Whether to approximate the cable with a lumped model or a transmission
line model is a small difference, compared to the large uncertainties in
the rest of the system.


You say that the uncentaninties in the rest of the system are greater
than the differences between a transmission line model, and a crude
components cable model. That is true, in the application you suggested, but
that is not the only possible application for a cable model.





It will be like trying to find an answer to 5 significant digits, when
all your inputs only have 2 significant digits. You are chasing 4th
order effects when you don't even have second order effects under control.
...
Besides, looking at transmission line loading is the wrong way to
analyze such amps.



It was *you*, who in a previous message, suggested using a cable model,
for testing amplifier stability.


Much better to vary the output loading and make sure
that the amp is stable for all possible loads. And you would model that
with resistive, capacitve and inductive loads.

Trust me, no one analyzes the stability of an audio amp by using
transmission line models of cables.


You are the one who suggested using a cable model for testing amplifier
stability.




Why bother?


Because you can't think of any application that requires an accurate cable
model, you think one is not needed. You are being like the man in the
1880's, who suggested closing the US Patent Office, because "all possible
inventions" had already been made.





Suppose you were to write a circuit analysis program based
on quantum mechanics?


It is not necessary. An S-parameter based cable model, already takes into
account *all the factors* that cause cable loss, including quantum effects,
if any.





This
model could be useful for educational perposes, and also would have
some pratical applications.


Not for audio cables or audio amps.


Again, you are being very short sighted.






Would you write a speaker cable model that was "dead on" accurate, or
write a (close approximation) component type model?


See all the approximations above.


The "approximations above" are only for a limited case. They does not prove
there will never be a need for an accurate cable model.

Bob Stantone news:c0439$40143681$c247604
  #413   Report Post  
Bob-Stanton
 
Posts: n/a
Default Distorsion percentage, power or voltage?

chung wrote in messag ...

How accurately do you model the skin
effect and resistive losses? How about the loss tangents of the
dielectric? How close to an ideal transmission line is a pair of wires
side by side?


An S-parameter based cable model, would include all of the above factors and
more.





In the end, you are making many approximations in such calculations.
Whether to approximate the cable with a lumped model or a transmission
line model is a small difference, compared to the large uncertainties in
the rest of the system.


You say that the uncentaninties in the rest of the system are greater
than the differences between a transmission line model, and a crude
components cable model. That is true, in the application you suggested, but
that is not the only possible application for a cable model.





It will be like trying to find an answer to 5 significant digits, when
all your inputs only have 2 significant digits. You are chasing 4th
order effects when you don't even have second order effects under control.
...
Besides, looking at transmission line loading is the wrong way to
analyze such amps.



It was *you*, who in a previous message, suggested using a cable model,
for testing amplifier stability.


Much better to vary the output loading and make sure
that the amp is stable for all possible loads. And you would model that
with resistive, capacitve and inductive loads.

Trust me, no one analyzes the stability of an audio amp by using
transmission line models of cables.


You are the one who suggested using a cable model for testing amplifier
stability.




Why bother?


Because you can't think of any application that requires an accurate cable
model, you think one is not needed. You are being like the man in the
1880's, who suggested closing the US Patent Office, because "all possible
inventions" had already been made.





Suppose you were to write a circuit analysis program based
on quantum mechanics?


It is not necessary. An S-parameter based cable model, already takes into
account *all the factors* that cause cable loss, including quantum effects,
if any.





This
model could be useful for educational perposes, and also would have
some pratical applications.


Not for audio cables or audio amps.


Again, you are being very short sighted.






Would you write a speaker cable model that was "dead on" accurate, or
write a (close approximation) component type model?


See all the approximations above.


The "approximations above" are only for a limited case. They does not prove
there will never be a need for an accurate cable model.

Bob Stantone news:c0439$40143681$c247604
  #414   Report Post  
chung
 
Posts: n/a
Default Distorsion percentage, power or voltage?

Bob-Stanton wrote:
chung wrote in messag ...

How accurately do you model the skin
effect and resistive losses? How about the loss tangents of the
dielectric? How close to an ideal transmission line is a pair of wires
side by side?


An S-parameter based cable model, would include all of the above factors and
more.


How about the rest of the uncertainties I listed which you snipped? How
about the time it would take to carefully measure those parameters of
the cable? How closely you think the speaker cables resemble a
transmission line?





In the end, you are making many approximations in such calculations.
Whether to approximate the cable with a lumped model or a transmission
line model is a small difference, compared to the large uncertainties in
the rest of the system.


You say that the uncentaninties in the rest of the system are greater
than the differences between a transmission line model, and a crude
components cable model. That is true, in the application you suggested, but
that is not the only possible application for a cable model.



Like I said before, that application of determining stability at high
frequencies is the only application for using transmission line models
for speaker cables in audio application that I can remotely justify. Of
course, the transmission line model is very useful in other areas of
electronics. But that is neither here nor there in the present discussion.




It will be like trying to find an answer to 5 significant digits, when
all your inputs only have 2 significant digits. You are chasing 4th
order effects when you don't even have second order effects under control.
...
Besides, looking at transmission line loading is the wrong way to
analyze such amps.



It was *you*, who in a previous message, suggested using a cable model,
for testing amplifier stability.


Yes, and I was trying really hard to come up with a case where you might
want to use the transmission line model in audio.



Much better to vary the output loading and make sure
that the amp is stable for all possible loads. And you would model that
with resistive, capacitve and inductive loads.

Trust me, no one analyzes the stability of an audio amp by using
transmission line models of cables.


You are the one who suggested using a cable model for testing amplifier
stability.


Yes, and I was trying really hard to come up with a case where you might
want to use the transmission line model in audio.



Why bother?


Because you can't think of any application that requires an accurate cable
model, you think one is not needed. You are being like the man in the
1880's, who suggested closing the US Patent Office, because "all possible
inventions" had already been made.


Note the context here. In audio, there is no need to use a transmission
line model for speaker cables. In other areas of electronics,
transmission line models are used for the study of, well, transmission
lines.


Why don't you come up with some reason to use that model in audio?





Suppose you were to write a circuit analysis program based
on quantum mechanics?


It is not necessary. An S-parameter based cable model, already takes into
account *all the factors* that cause cable loss, including quantum effects,
if any.


No, I was referring to circuit analysis in general. You know,
transisitors, resistors, capacitors and so on?




This
model could be useful for educational perposes, and also would have
some pratical applications.


Not for audio cables or audio amps.


Again, you are being very short sighted.


Give your examples, then. You are looking like someone who has a certain
tool and want to solve every problem with that tool.






Would you write a speaker cable model that was "dead on" accurate, or
write a (close approximation) component type model?


See all the approximations above.


The "approximations above" are only for a limited case. They does not prove
there will never be a need for an accurate cable model.


Look at those approximations again. Those are realistic modelling issues
in any application where speaker cables are involved.


Bob Stantone news:c0439$40143681$c247604


  #415   Report Post  
chung
 
Posts: n/a
Default Distorsion percentage, power or voltage?

Bob-Stanton wrote:
chung wrote in messag ...

How accurately do you model the skin
effect and resistive losses? How about the loss tangents of the
dielectric? How close to an ideal transmission line is a pair of wires
side by side?


An S-parameter based cable model, would include all of the above factors and
more.


How about the rest of the uncertainties I listed which you snipped? How
about the time it would take to carefully measure those parameters of
the cable? How closely you think the speaker cables resemble a
transmission line?





In the end, you are making many approximations in such calculations.
Whether to approximate the cable with a lumped model or a transmission
line model is a small difference, compared to the large uncertainties in
the rest of the system.


You say that the uncentaninties in the rest of the system are greater
than the differences between a transmission line model, and a crude
components cable model. That is true, in the application you suggested, but
that is not the only possible application for a cable model.



Like I said before, that application of determining stability at high
frequencies is the only application for using transmission line models
for speaker cables in audio application that I can remotely justify. Of
course, the transmission line model is very useful in other areas of
electronics. But that is neither here nor there in the present discussion.




It will be like trying to find an answer to 5 significant digits, when
all your inputs only have 2 significant digits. You are chasing 4th
order effects when you don't even have second order effects under control.
...
Besides, looking at transmission line loading is the wrong way to
analyze such amps.



It was *you*, who in a previous message, suggested using a cable model,
for testing amplifier stability.


Yes, and I was trying really hard to come up with a case where you might
want to use the transmission line model in audio.



Much better to vary the output loading and make sure
that the amp is stable for all possible loads. And you would model that
with resistive, capacitve and inductive loads.

Trust me, no one analyzes the stability of an audio amp by using
transmission line models of cables.


You are the one who suggested using a cable model for testing amplifier
stability.


Yes, and I was trying really hard to come up with a case where you might
want to use the transmission line model in audio.



Why bother?


Because you can't think of any application that requires an accurate cable
model, you think one is not needed. You are being like the man in the
1880's, who suggested closing the US Patent Office, because "all possible
inventions" had already been made.


Note the context here. In audio, there is no need to use a transmission
line model for speaker cables. In other areas of electronics,
transmission line models are used for the study of, well, transmission
lines.


Why don't you come up with some reason to use that model in audio?





Suppose you were to write a circuit analysis program based
on quantum mechanics?


It is not necessary. An S-parameter based cable model, already takes into
account *all the factors* that cause cable loss, including quantum effects,
if any.


No, I was referring to circuit analysis in general. You know,
transisitors, resistors, capacitors and so on?




This
model could be useful for educational perposes, and also would have
some pratical applications.


Not for audio cables or audio amps.


Again, you are being very short sighted.


Give your examples, then. You are looking like someone who has a certain
tool and want to solve every problem with that tool.






Would you write a speaker cable model that was "dead on" accurate, or
write a (close approximation) component type model?


See all the approximations above.


The "approximations above" are only for a limited case. They does not prove
there will never be a need for an accurate cable model.


Look at those approximations again. Those are realistic modelling issues
in any application where speaker cables are involved.


Bob Stantone news:c0439$40143681$c247604




  #416   Report Post  
chung
 
Posts: n/a
Default Distorsion percentage, power or voltage?

Bob-Stanton wrote:
chung wrote in messag ...

How accurately do you model the skin
effect and resistive losses? How about the loss tangents of the
dielectric? How close to an ideal transmission line is a pair of wires
side by side?


An S-parameter based cable model, would include all of the above factors and
more.


How about the rest of the uncertainties I listed which you snipped? How
about the time it would take to carefully measure those parameters of
the cable? How closely you think the speaker cables resemble a
transmission line?





In the end, you are making many approximations in such calculations.
Whether to approximate the cable with a lumped model or a transmission
line model is a small difference, compared to the large uncertainties in
the rest of the system.


You say that the uncentaninties in the rest of the system are greater
than the differences between a transmission line model, and a crude
components cable model. That is true, in the application you suggested, but
that is not the only possible application for a cable model.



Like I said before, that application of determining stability at high
frequencies is the only application for using transmission line models
for speaker cables in audio application that I can remotely justify. Of
course, the transmission line model is very useful in other areas of
electronics. But that is neither here nor there in the present discussion.




It will be like trying to find an answer to 5 significant digits, when
all your inputs only have 2 significant digits. You are chasing 4th
order effects when you don't even have second order effects under control.
...
Besides, looking at transmission line loading is the wrong way to
analyze such amps.



It was *you*, who in a previous message, suggested using a cable model,
for testing amplifier stability.


Yes, and I was trying really hard to come up with a case where you might
want to use the transmission line model in audio.



Much better to vary the output loading and make sure
that the amp is stable for all possible loads. And you would model that
with resistive, capacitve and inductive loads.

Trust me, no one analyzes the stability of an audio amp by using
transmission line models of cables.


You are the one who suggested using a cable model for testing amplifier
stability.


Yes, and I was trying really hard to come up with a case where you might
want to use the transmission line model in audio.



Why bother?


Because you can't think of any application that requires an accurate cable
model, you think one is not needed. You are being like the man in the
1880's, who suggested closing the US Patent Office, because "all possible
inventions" had already been made.


Note the context here. In audio, there is no need to use a transmission
line model for speaker cables. In other areas of electronics,
transmission line models are used for the study of, well, transmission
lines.


Why don't you come up with some reason to use that model in audio?





Suppose you were to write a circuit analysis program based
on quantum mechanics?


It is not necessary. An S-parameter based cable model, already takes into
account *all the factors* that cause cable loss, including quantum effects,
if any.


No, I was referring to circuit analysis in general. You know,
transisitors, resistors, capacitors and so on?




This
model could be useful for educational perposes, and also would have
some pratical applications.


Not for audio cables or audio amps.


Again, you are being very short sighted.


Give your examples, then. You are looking like someone who has a certain
tool and want to solve every problem with that tool.






Would you write a speaker cable model that was "dead on" accurate, or
write a (close approximation) component type model?


See all the approximations above.


The "approximations above" are only for a limited case. They does not prove
there will never be a need for an accurate cable model.


Look at those approximations again. Those are realistic modelling issues
in any application where speaker cables are involved.


Bob Stantone news:c0439$40143681$c247604


  #417   Report Post  
chung
 
Posts: n/a
Default Distorsion percentage, power or voltage?

Bob-Stanton wrote:
chung wrote in messag ...

How accurately do you model the skin
effect and resistive losses? How about the loss tangents of the
dielectric? How close to an ideal transmission line is a pair of wires
side by side?


An S-parameter based cable model, would include all of the above factors and
more.


How about the rest of the uncertainties I listed which you snipped? How
about the time it would take to carefully measure those parameters of
the cable? How closely you think the speaker cables resemble a
transmission line?





In the end, you are making many approximations in such calculations.
Whether to approximate the cable with a lumped model or a transmission
line model is a small difference, compared to the large uncertainties in
the rest of the system.


You say that the uncentaninties in the rest of the system are greater
than the differences between a transmission line model, and a crude
components cable model. That is true, in the application you suggested, but
that is not the only possible application for a cable model.



Like I said before, that application of determining stability at high
frequencies is the only application for using transmission line models
for speaker cables in audio application that I can remotely justify. Of
course, the transmission line model is very useful in other areas of
electronics. But that is neither here nor there in the present discussion.




It will be like trying to find an answer to 5 significant digits, when
all your inputs only have 2 significant digits. You are chasing 4th
order effects when you don't even have second order effects under control.
...
Besides, looking at transmission line loading is the wrong way to
analyze such amps.



It was *you*, who in a previous message, suggested using a cable model,
for testing amplifier stability.


Yes, and I was trying really hard to come up with a case where you might
want to use the transmission line model in audio.



Much better to vary the output loading and make sure
that the amp is stable for all possible loads. And you would model that
with resistive, capacitve and inductive loads.

Trust me, no one analyzes the stability of an audio amp by using
transmission line models of cables.


You are the one who suggested using a cable model for testing amplifier
stability.


Yes, and I was trying really hard to come up with a case where you might
want to use the transmission line model in audio.



Why bother?


Because you can't think of any application that requires an accurate cable
model, you think one is not needed. You are being like the man in the
1880's, who suggested closing the US Patent Office, because "all possible
inventions" had already been made.


Note the context here. In audio, there is no need to use a transmission
line model for speaker cables. In other areas of electronics,
transmission line models are used for the study of, well, transmission
lines.


Why don't you come up with some reason to use that model in audio?





Suppose you were to write a circuit analysis program based
on quantum mechanics?


It is not necessary. An S-parameter based cable model, already takes into
account *all the factors* that cause cable loss, including quantum effects,
if any.


No, I was referring to circuit analysis in general. You know,
transisitors, resistors, capacitors and so on?




This
model could be useful for educational perposes, and also would have
some pratical applications.


Not for audio cables or audio amps.


Again, you are being very short sighted.


Give your examples, then. You are looking like someone who has a certain
tool and want to solve every problem with that tool.






Would you write a speaker cable model that was "dead on" accurate, or
write a (close approximation) component type model?


See all the approximations above.


The "approximations above" are only for a limited case. They does not prove
there will never be a need for an accurate cable model.


Look at those approximations again. Those are realistic modelling issues
in any application where speaker cables are involved.


Bob Stantone news:c0439$40143681$c247604


  #418   Report Post  
Harry Lavo
 
Posts: n/a
Default Distorsion percentage, power or voltage?


"Bob-Stanton" wrote in message
om...
chung wrote in message news:eb5e8$400d7230$c247604


snip, not relevant to below


There are a lot of different kinds of "dB's":

dB - plain old power ratio.
dBa - adjusted decibels
dBc - referred to a carrier.
dBd - gain of an antenna
dBi - another gain of an antenna
dBk - dB relative to 1000 Watts
dBm - obscure reference level, often used by military types.
dBm0 - refered to zero transmission level (in dBm)
dBmp - dB measured with psophometer weighting.
dBmr - dB title for men.
dbmV - Very useful CATV reference level, in wide usage.
dBrap - Relative power levels of RAP music songs.
dBrn - dB used by registered nurses.
dBrnc - dB's used by the Republicans?
dBV - Output voltage, referred to 1 V input.
dBW - dB referred to 1 Watt
dBx - Output level, relative to an *unknown* input level.

(I may have gotten some of the above definitions wrong.)

Bob Stanton


LOL!! Thanks.

Harry


  #419   Report Post  
Harry Lavo
 
Posts: n/a
Default Distorsion percentage, power or voltage?


"Bob-Stanton" wrote in message
om...
chung wrote in message news:eb5e8$400d7230$c247604


snip, not relevant to below


There are a lot of different kinds of "dB's":

dB - plain old power ratio.
dBa - adjusted decibels
dBc - referred to a carrier.
dBd - gain of an antenna
dBi - another gain of an antenna
dBk - dB relative to 1000 Watts
dBm - obscure reference level, often used by military types.
dBm0 - refered to zero transmission level (in dBm)
dBmp - dB measured with psophometer weighting.
dBmr - dB title for men.
dbmV - Very useful CATV reference level, in wide usage.
dBrap - Relative power levels of RAP music songs.
dBrn - dB used by registered nurses.
dBrnc - dB's used by the Republicans?
dBV - Output voltage, referred to 1 V input.
dBW - dB referred to 1 Watt
dBx - Output level, relative to an *unknown* input level.

(I may have gotten some of the above definitions wrong.)

Bob Stanton


LOL!! Thanks.

Harry


  #420   Report Post  
Harry Lavo
 
Posts: n/a
Default Distorsion percentage, power or voltage?


"Bob-Stanton" wrote in message
om...
chung wrote in message news:eb5e8$400d7230$c247604


snip, not relevant to below


There are a lot of different kinds of "dB's":

dB - plain old power ratio.
dBa - adjusted decibels
dBc - referred to a carrier.
dBd - gain of an antenna
dBi - another gain of an antenna
dBk - dB relative to 1000 Watts
dBm - obscure reference level, often used by military types.
dBm0 - refered to zero transmission level (in dBm)
dBmp - dB measured with psophometer weighting.
dBmr - dB title for men.
dbmV - Very useful CATV reference level, in wide usage.
dBrap - Relative power levels of RAP music songs.
dBrn - dB used by registered nurses.
dBrnc - dB's used by the Republicans?
dBV - Output voltage, referred to 1 V input.
dBW - dB referred to 1 Watt
dBx - Output level, relative to an *unknown* input level.

(I may have gotten some of the above definitions wrong.)

Bob Stanton


LOL!! Thanks.

Harry




  #421   Report Post  
Harry Lavo
 
Posts: n/a
Default Distorsion percentage, power or voltage?


"Bob-Stanton" wrote in message
om...
chung wrote in message news:eb5e8$400d7230$c247604


snip, not relevant to below


There are a lot of different kinds of "dB's":

dB - plain old power ratio.
dBa - adjusted decibels
dBc - referred to a carrier.
dBd - gain of an antenna
dBi - another gain of an antenna
dBk - dB relative to 1000 Watts
dBm - obscure reference level, often used by military types.
dBm0 - refered to zero transmission level (in dBm)
dBmp - dB measured with psophometer weighting.
dBmr - dB title for men.
dbmV - Very useful CATV reference level, in wide usage.
dBrap - Relative power levels of RAP music songs.
dBrn - dB used by registered nurses.
dBrnc - dB's used by the Republicans?
dBV - Output voltage, referred to 1 V input.
dBW - dB referred to 1 Watt
dBx - Output level, relative to an *unknown* input level.

(I may have gotten some of the above definitions wrong.)

Bob Stanton


LOL!! Thanks.

Harry


  #422   Report Post  
gwhite
 
Posts: n/a
Default Distorsion percentage, power or voltage?



Glenn Booth wrote:

Hi,

In message , gecwhite
writes

The twenty comes in because the "voltage dB" is essentially derived from
the "power dB," but power is a non-linear function (square law) of
voltage.


Yes, I got that; I just needed to know which came first :-)

By _Bell System Blue Book definition_:

X(dB) = 10*log(P/P_ref)


*That's* what I was looking for; many thanks.



I would like to add that I don't believe the "origin" notes at

http://www.madengineer.com/blunders/decibels.htm.

There are no sources cited and the author fails to decently answer his own
questions: "Why is the log multiplied by twenty? If it's deci, why not multiply
(or divide!) by 10? That is, why the deci?"

It is "deci" because of the '10' multiplying the log. The -bel was "already"
log-based-10; so that would not explain it. The only thing that made it deci-
was the 10 multiplier for the log-base-10. For electrical work, how the '20'
comes in has been demonstrated. I'll leave the '20,' as a sound pressure, as an
exercise for the student, hah hah.



On the other hand, the site

http://www.sizes.com/units/decibel.htm

does cite sources and very old ones at that. I have many engineering texts, and
*every one* defines the original dB as a power ratio. One of my physics texts
(R. Serway) gives the original definition in the chapter on sound waves as

/ I \
x(dB) = 10*log(-----)
\ Io/


where the intensity 'I' of a sound wave is

P_sound_wave
I := --------------.
Area

So the definition is definitely a power (or energy) relationship, and the work
is relevent to that carried in an acoustical wave. 'Io' is defined as the
intensity of a sound wave at the threshold of hearing. Note that the area of
any individual's ear is constant, regardless of wave intensity. Therefore, the
above acoustically related definition has the area simply "drop out," since it
is identical in the numerator and denominator. We are then left with

/ P_sound_wave \
x(dB) = 10*log(------------------).
\P_sound_wave_ref/


This is a _power ratio_ and once it is reduced to that, it makes no difference
"where" the power is. It could be in an acoustical wave, a resistor, a light
wave..... the definition does not then require the statement of any particular
medium, or energy carrying means; although it probably was _sourced_ in the
acoustical world. The basic power definition of the dB is therefore a highly
generalized one. It is doubtful that the original work was concerned with
anything other than sound intensity. So bet 20*log(pressure_ratio), as an
original definition, to be wrong. A good rule is to demand citations.


Is the Bell System Blue
book available?


You might see them on ebay from time to time. They seem to get high prices
($50).

I should note my language is sloppy. The term "Blue Book" is not well defined
and I know Bell labs published more than one. How many were "blue," I don't
know.

My particular "blue book" is

_Transmission Systems for Communications_
3rd Ed, (c) 1964
Bell Telephone Laboratories, Inc.


Still an excellent text, ime.


[Much good stuff snipped to the clipboard for later]

I hope this clears up some of your questions.


All of them. Thanks very much.



You're welcome!
  #423   Report Post  
gwhite
 
Posts: n/a
Default Distorsion percentage, power or voltage?



Glenn Booth wrote:

Hi,

In message , gecwhite
writes

The twenty comes in because the "voltage dB" is essentially derived from
the "power dB," but power is a non-linear function (square law) of
voltage.


Yes, I got that; I just needed to know which came first :-)

By _Bell System Blue Book definition_:

X(dB) = 10*log(P/P_ref)


*That's* what I was looking for; many thanks.



I would like to add that I don't believe the "origin" notes at

http://www.madengineer.com/blunders/decibels.htm.

There are no sources cited and the author fails to decently answer his own
questions: "Why is the log multiplied by twenty? If it's deci, why not multiply
(or divide!) by 10? That is, why the deci?"

It is "deci" because of the '10' multiplying the log. The -bel was "already"
log-based-10; so that would not explain it. The only thing that made it deci-
was the 10 multiplier for the log-base-10. For electrical work, how the '20'
comes in has been demonstrated. I'll leave the '20,' as a sound pressure, as an
exercise for the student, hah hah.



On the other hand, the site

http://www.sizes.com/units/decibel.htm

does cite sources and very old ones at that. I have many engineering texts, and
*every one* defines the original dB as a power ratio. One of my physics texts
(R. Serway) gives the original definition in the chapter on sound waves as

/ I \
x(dB) = 10*log(-----)
\ Io/


where the intensity 'I' of a sound wave is

P_sound_wave
I := --------------.
Area

So the definition is definitely a power (or energy) relationship, and the work
is relevent to that carried in an acoustical wave. 'Io' is defined as the
intensity of a sound wave at the threshold of hearing. Note that the area of
any individual's ear is constant, regardless of wave intensity. Therefore, the
above acoustically related definition has the area simply "drop out," since it
is identical in the numerator and denominator. We are then left with

/ P_sound_wave \
x(dB) = 10*log(------------------).
\P_sound_wave_ref/


This is a _power ratio_ and once it is reduced to that, it makes no difference
"where" the power is. It could be in an acoustical wave, a resistor, a light
wave..... the definition does not then require the statement of any particular
medium, or energy carrying means; although it probably was _sourced_ in the
acoustical world. The basic power definition of the dB is therefore a highly
generalized one. It is doubtful that the original work was concerned with
anything other than sound intensity. So bet 20*log(pressure_ratio), as an
original definition, to be wrong. A good rule is to demand citations.


Is the Bell System Blue
book available?


You might see them on ebay from time to time. They seem to get high prices
($50).

I should note my language is sloppy. The term "Blue Book" is not well defined
and I know Bell labs published more than one. How many were "blue," I don't
know.

My particular "blue book" is

_Transmission Systems for Communications_
3rd Ed, (c) 1964
Bell Telephone Laboratories, Inc.


Still an excellent text, ime.


[Much good stuff snipped to the clipboard for later]

I hope this clears up some of your questions.


All of them. Thanks very much.



You're welcome!
  #424   Report Post  
gwhite
 
Posts: n/a
Default Distorsion percentage, power or voltage?



Glenn Booth wrote:

Hi,

In message , gecwhite
writes

The twenty comes in because the "voltage dB" is essentially derived from
the "power dB," but power is a non-linear function (square law) of
voltage.


Yes, I got that; I just needed to know which came first :-)

By _Bell System Blue Book definition_:

X(dB) = 10*log(P/P_ref)


*That's* what I was looking for; many thanks.



I would like to add that I don't believe the "origin" notes at

http://www.madengineer.com/blunders/decibels.htm.

There are no sources cited and the author fails to decently answer his own
questions: "Why is the log multiplied by twenty? If it's deci, why not multiply
(or divide!) by 10? That is, why the deci?"

It is "deci" because of the '10' multiplying the log. The -bel was "already"
log-based-10; so that would not explain it. The only thing that made it deci-
was the 10 multiplier for the log-base-10. For electrical work, how the '20'
comes in has been demonstrated. I'll leave the '20,' as a sound pressure, as an
exercise for the student, hah hah.



On the other hand, the site

http://www.sizes.com/units/decibel.htm

does cite sources and very old ones at that. I have many engineering texts, and
*every one* defines the original dB as a power ratio. One of my physics texts
(R. Serway) gives the original definition in the chapter on sound waves as

/ I \
x(dB) = 10*log(-----)
\ Io/


where the intensity 'I' of a sound wave is

P_sound_wave
I := --------------.
Area

So the definition is definitely a power (or energy) relationship, and the work
is relevent to that carried in an acoustical wave. 'Io' is defined as the
intensity of a sound wave at the threshold of hearing. Note that the area of
any individual's ear is constant, regardless of wave intensity. Therefore, the
above acoustically related definition has the area simply "drop out," since it
is identical in the numerator and denominator. We are then left with

/ P_sound_wave \
x(dB) = 10*log(------------------).
\P_sound_wave_ref/


This is a _power ratio_ and once it is reduced to that, it makes no difference
"where" the power is. It could be in an acoustical wave, a resistor, a light
wave..... the definition does not then require the statement of any particular
medium, or energy carrying means; although it probably was _sourced_ in the
acoustical world. The basic power definition of the dB is therefore a highly
generalized one. It is doubtful that the original work was concerned with
anything other than sound intensity. So bet 20*log(pressure_ratio), as an
original definition, to be wrong. A good rule is to demand citations.


Is the Bell System Blue
book available?


You might see them on ebay from time to time. They seem to get high prices
($50).

I should note my language is sloppy. The term "Blue Book" is not well defined
and I know Bell labs published more than one. How many were "blue," I don't
know.

My particular "blue book" is

_Transmission Systems for Communications_
3rd Ed, (c) 1964
Bell Telephone Laboratories, Inc.


Still an excellent text, ime.


[Much good stuff snipped to the clipboard for later]

I hope this clears up some of your questions.


All of them. Thanks very much.



You're welcome!
  #425   Report Post  
gwhite
 
Posts: n/a
Default Distorsion percentage, power or voltage?



Glenn Booth wrote:

Hi,

In message , gecwhite
writes

The twenty comes in because the "voltage dB" is essentially derived from
the "power dB," but power is a non-linear function (square law) of
voltage.


Yes, I got that; I just needed to know which came first :-)

By _Bell System Blue Book definition_:

X(dB) = 10*log(P/P_ref)


*That's* what I was looking for; many thanks.



I would like to add that I don't believe the "origin" notes at

http://www.madengineer.com/blunders/decibels.htm.

There are no sources cited and the author fails to decently answer his own
questions: "Why is the log multiplied by twenty? If it's deci, why not multiply
(or divide!) by 10? That is, why the deci?"

It is "deci" because of the '10' multiplying the log. The -bel was "already"
log-based-10; so that would not explain it. The only thing that made it deci-
was the 10 multiplier for the log-base-10. For electrical work, how the '20'
comes in has been demonstrated. I'll leave the '20,' as a sound pressure, as an
exercise for the student, hah hah.



On the other hand, the site

http://www.sizes.com/units/decibel.htm

does cite sources and very old ones at that. I have many engineering texts, and
*every one* defines the original dB as a power ratio. One of my physics texts
(R. Serway) gives the original definition in the chapter on sound waves as

/ I \
x(dB) = 10*log(-----)
\ Io/


where the intensity 'I' of a sound wave is

P_sound_wave
I := --------------.
Area

So the definition is definitely a power (or energy) relationship, and the work
is relevent to that carried in an acoustical wave. 'Io' is defined as the
intensity of a sound wave at the threshold of hearing. Note that the area of
any individual's ear is constant, regardless of wave intensity. Therefore, the
above acoustically related definition has the area simply "drop out," since it
is identical in the numerator and denominator. We are then left with

/ P_sound_wave \
x(dB) = 10*log(------------------).
\P_sound_wave_ref/


This is a _power ratio_ and once it is reduced to that, it makes no difference
"where" the power is. It could be in an acoustical wave, a resistor, a light
wave..... the definition does not then require the statement of any particular
medium, or energy carrying means; although it probably was _sourced_ in the
acoustical world. The basic power definition of the dB is therefore a highly
generalized one. It is doubtful that the original work was concerned with
anything other than sound intensity. So bet 20*log(pressure_ratio), as an
original definition, to be wrong. A good rule is to demand citations.


Is the Bell System Blue
book available?


You might see them on ebay from time to time. They seem to get high prices
($50).

I should note my language is sloppy. The term "Blue Book" is not well defined
and I know Bell labs published more than one. How many were "blue," I don't
know.

My particular "blue book" is

_Transmission Systems for Communications_
3rd Ed, (c) 1964
Bell Telephone Laboratories, Inc.


Still an excellent text, ime.


[Much good stuff snipped to the clipboard for later]

I hope this clears up some of your questions.


All of them. Thanks very much.



You're welcome!
Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is On
HTML code is Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
System warm-up James Harris Audio Opinions 69 May 19th 04 04:09 AM
Damping Material Question Ron Car Audio 68 April 17th 04 07:55 AM
rec.audio.car FAQ (Part 2/5) Ian D. Bjorhovde Car Audio 0 March 6th 04 07:54 AM
rec.audio.car FAQ (Part 1/5) Ian D. Bjorhovde Car Audio 0 March 6th 04 07:54 AM
FS: SOUNDSTREAM CLOSEOUTS AND MORE!! Nexxon Car Audio 0 November 21st 03 03:59 AM


All times are GMT +1. The time now is 11:46 AM.

Powered by: vBulletin
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 AudioBanter.com.
The comments are property of their posters.
 

About Us

"It's about Audio and hi-fi"