On 2007-10-02, Don Pearce wrote:
On Tue, 2 Oct 2007 09:51:32 +0300, "Iain Churches"
wrote:

It seems that the THD itself is not so important as the way
in which that total is made up, i.e. the relationship of
2H,3H,4H,5H etc that is important. It may be this which
gives each amp its sonic signature.

Harmonic distortion itself is really not so important for music, which
is loaded with harmonics anyway. What matters is that distortion - any
distortion, even or odd - produces intermods. These products are
non-harmonic and certainly for music with any degree of complexity
will be dissonant. If the distortion level is moderately low, the
result will just be a sort of "thickening" of the sound, which may
indeed be more interesting than the unadorned music.

I have sometimes pondered (without real cause, I admit) about whether the
"usual" IMD tests are good enough as a predictor of audible IMD effects.

Do you think the normal two-tone IMD tests (from SMPTE, DIN, IEC,
etc.) provide an adequate basis for testing something where (I postulate)
more complex IMDs may be the more audible effect?

--
John Phillips

On 02 Oct 2007 12:33:43 GMT, John Phillips
wrote:

On 2007-10-02, Don Pearce wrote:
On Tue, 2 Oct 2007 09:51:32 +0300, "Iain Churches"
wrote:

It seems that the THD itself is not so important as the way
in which that total is made up, i.e. the relationship of
2H,3H,4H,5H etc that is important. It may be this which
gives each amp its sonic signature.

Harmonic distortion itself is really not so important for music, which
is loaded with harmonics anyway. What matters is that distortion - any
distortion, even or odd - produces intermods. These products are
non-harmonic and certainly for music with any degree of complexity
will be dissonant. If the distortion level is moderately low, the
result will just be a sort of "thickening" of the sound, which may
indeed be more interesting than the unadorned music.

I have sometimes pondered (without real cause, I admit) about whether the
"usual" IMD tests are good enough as a predictor of audible IMD effects.

Do you think the normal two-tone IMD tests (from SMPTE, DIN, IEC,
etc.) provide an adequate basis for testing something where (I postulate)
more complex IMDs may be the more audible effect?

Depends what you mean by testing. From a single harmonic distortion
measurement I can predict two, three, four tone intermodulation
performance accurately, given the right data from the harmonic
measurement.

Harmonics and intermodulation are simply the inevitable and calculable
results of a crooked transfer function. So is a normal two-tone IM
measurement sufficient? I would have to say yes, more than enough.

The published result (in % or dB) of a harmonic distortion measurement
is not adequate in itself.

d

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

On 2007-10-02, Don Pearce wrote:
On 02 Oct 2007 12:33:43 GMT, John Phillips
wrote:
I have sometimes pondered (without real cause, I admit) about whether the
"usual" IMD tests are good enough as a predictor of audible IMD effects.

Do you think the normal two-tone IMD tests (from SMPTE, DIN, IEC,
etc.) provide an adequate basis for testing something where (I postulate)
more complex IMDs may be the more audible effect?

Depends what you mean by testing. From a single harmonic distortion
measurement I can predict two, three, four tone intermodulation
performance accurately, given the right data from the harmonic
measurement.

Harmonics and intermodulation are simply the inevitable and calculable
results of a crooked transfer function. So is a normal two-tone IM
measurement sufficient? I would have to say yes, more than enough.

The published result (in % or dB) of a harmonic distortion measurement
is not adequate in itself.

Yes - that was a concern I had and FWIW I think you must be right.

I was also concerned over checking properly the ability of a system to
accurately reproduce low-level signals in the presence of high-level
signals and whether this result could be deduced from simple two-tone
IMD tests (but with full graphic results - not single summed figures).

--
John Phillips

On 02 Oct 2007 13:16:53 GMT, John Phillips
wrote:

On 2007-10-02, Don Pearce wrote:
On 02 Oct 2007 12:33:43 GMT, John Phillips
wrote:
I have sometimes pondered (without real cause, I admit) about whether the
"usual" IMD tests are good enough as a predictor of audible IMD effects.

Do you think the normal two-tone IMD tests (from SMPTE, DIN, IEC,
etc.) provide an adequate basis for testing something where (I postulate)
more complex IMDs may be the more audible effect?

Depends what you mean by testing. From a single harmonic distortion
measurement I can predict two, three, four tone intermodulation
performance accurately, given the right data from the harmonic
measurement.

Harmonics and intermodulation are simply the inevitable and calculable
results of a crooked transfer function. So is a normal two-tone IM
measurement sufficient? I would have to say yes, more than enough.

The published result (in % or dB) of a harmonic distortion measurement
is not adequate in itself.

Yes - that was a concern I had and FWIW I think you must be right.

I was also concerned over checking properly the ability of a system to
accurately reproduce low-level signals in the presence of high-level
signals and whether this result could be deduced from simple two-tone
IMD tests (but with full graphic results - not single summed figures).

That also comes out of the basic linearity test. Any system that can't
do that will reveal it in an IM test.

Some systems do this deliberately of course. MP3 codecs work this way
using the phenomenon called masking. Even moderately large signals
close in frequency to very large ones can be deleted inaudibly.

d

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

"John Phillips" wrote
in message

I have sometimes pondered (without real cause, I admit)
about whether the "usual" IMD tests are good enough as a
predictor of audible IMD effects.

Not really.

Do you think the normal two-tone IMD tests (from SMPTE,
DIN, IEC, etc.) provide an adequate basis for testing something
where (I postulate) more complex IMDs may be the more
audible effect?

Not really.

Swept 2-tone tests are more to the point.

Multitone-based tests can also work well. Do it right and you can get a
good fast FR test and a good test for nonlinear distortion at all audible
frequencies done with one test, two analyses.

"Arny Krueger" wrote in message
. ..
"John Phillips" wrote
in message

I have sometimes pondered (without real cause, I admit)
about whether the "usual" IMD tests are good enough as a
predictor of audible IMD effects.

Not really.

Agreed. They must be better than nothing, but probably have
little bearing on what might be happening in a complex musical
signal.

Do you think the normal two-tone IMD tests (from SMPTE,
DIN, IEC, etc.) provide an adequate basis for testing something
where (I postulate) more complex IMDs may be the more
audible effect?

Not really.

There seem to be several opinions regarding the
frequencies of the fundamentals used in such tests.
Is there any standard?
I have seen 70Hz and 6kHz mentioned, and
also 19kHz and 20kHz.

Swept 2-tone tests are more to the point.

Much more like music.

Multitone-based tests can also work well.

Multitone is indeed interesting:-)
Are there any recommendation for centre frequencies?

Iain

On 2007-10-02, Iain Churches wrote:
"Arny Krueger" wrote in message
. ..
"John Phillips" wrote
in message

I have sometimes pondered (without real cause, I admit)
about whether the "usual" IMD tests are good enough as a
predictor of audible IMD effects.

Not really.

Agreed. They must be better than nothing, but probably have
little bearing on what might be happening in a complex musical
signal.

That's what has always concerned me. I no longer design and build kit
(for 20 or more years and it was only ever a hobby). Although I got
to understand linear distortions and noise in audio systems I never did
get to grips with understanding non-linear distortion. Of course I
did understand the basics but I always felt the simple IMD tests were
not the complete way to evaluate system non-linearities. However I'm
not really sure.

Do you think the normal two-tone IMD tests (from SMPTE,
DIN, IEC, etc.) provide an adequate basis for testing something
where (I postulate) more complex IMDs may be the more
audible effect?

Not really.

There seem to be several opinions regarding the
frequencies of the fundamentals used in such tests.
Is there any standard?
I have seen 70Hz and 6kHz mentioned, and
also 19kHz and 20kHz.

See, for example, http://www.rane.com/par-i.html (look down for IM/IMD)
for some of the standard IMD test methods.

--
John Phillips

"John Phillips" wrote in message
...

See, for example, http://www.rane.com/par-i.html (look down for IM/IMD)
for some of the standard IMD test methods.

--
John Phillips

Thanks John. I will take a look
Iain

"Iain Churches" wrote in message
i.fi
"Arny Krueger" wrote in message
. ..
"John Phillips" wrote
in message

I have sometimes pondered (without real cause, I admit)
about whether the "usual" IMD tests are good enough as a
predictor of audible IMD effects.

Not really.

Agreed. They must be better than nothing, but probably
have little bearing on what might be happening in a
complex musical signal.

Do you think the normal two-tone IMD tests (from SMPTE,
DIN, IEC, etc.) provide an adequate basis for testing
something where (I postulate) more complex IMDs may be
the more audible effect?

Not really.

There seem to be several opinions regarding the
frequencies of the fundamentals used in such tests.
Is there any standard?
I have seen 70Hz and 6kHz mentioned, and
also 19kHz and 20kHz.

Swept 2-tone tests are more to the point.

Much more like music.

Multitone-based tests can also work well.

Much more like music.

Multitone is indeed interesting:-)

http://www.nti-audio.com/Portals/0/P...nual_v332e.pdf

http://wireless.agilent.com/rfcomms/...audio_desc.php

http://www.tti.co.uk/products-resale/ap/ap-ats2-1.htm

http://www.pcavtech.com/play-rec/ra10436/index.htm

Are there any recommendation for centre frequencies?

Depends... ;-)