Serge Auckland wrote:
"Rob" wrote in message
...
[snip to bit about measurement]

1. Listening. The physical swapping out is doable. I would find/have found
level matching incredibly difficult. I tried this some months ago with a
20-20k hz cd, taking levels at various points. I couldn't get close to
level matching across the range - I'd got weird 'spikes' at points in the
frequency range, around 2000hz IIRC. Even using the same amp! Anyway, this
sort-of-obviously means I need some more reliable measuring equipment and
techniques - any pointers here would be appreciated.

This is quite normal:- If you are level-matching by using a SPL meter
in-room, then what you are measuring is the sound pressure at the point at
which you are holding the meter (so much is obvious :-) ) At mid-high
frequencies, the measurement you make will be very influenced by local
reflections, even from your own body. If I put my SPL meter on a
photographic tripod, then move around even only by a few centimetres, the
reading on the meter will vary by a dB or more at mid-high frequencies. At
low frequencies, it is much more stable, being dependent on room
reflections, and hence positioning in the room, much less influenced by body
movements as the wavelength of the sound becomes larger than my body
dimensions (even in my current less than sylph-like state)


Yes, I'd realised this - if not strictly taken it into account. Quite
how I clamp the listener's head, and avoid ear twitching, is an issue
for a later day :-)

You should level-match by measuring across the 'speaker terminals ideally by
using a high impedance audio millivoltmeter (once called a valve-voltmeter).
I have found that my normal inexpensive multimeter is accurate enough at low
audio frequencies. Mine actually work fine up to 20kHz, but if you use a
100Hz tone for level matching you should be fine. Use your SPL meter to make
sure the volume level is around 85dBC (say 80dBA) at 100Hz and level-match
with the multimeter at that loudness. You should fine it easy to level-match
to better than 0.5dB even with a multimeter.


Ak! I can't have that I'm afraid. I think I understand what you're
saying, but what I'm saying is that loudspeakers are non-linear loads
and something (I know not what) might be happening between the signal
and the sound. I suppose my thinking here comes from tests of kit I've
read in magazines, and the suggestion that different amplifiers do not
do the 'wire with gain' thing in a linear way. But I'm afraid I can't
remember the details - I don't even have a hifi magazine in the house,
having given up on them many years ago.

Now, if you're saying to me that any amplifier that has the
specification you've detailed will interact with a given loudspeaker
('normal'/domestic) in exactly the same way within the specification you
detailed then I'll do my best to satisfy myself that is in fact the case
(if I can), and adopt your test. You just have to say 'it is the case' -
I appreciate your patience thus far. In addition, I would expect the
level matching to apply between 20 -20k hz - just my little quirk :-)


2. If listening using level matched DBT does reveal difference, the
spotlight then turns on the amps. I would then have to measure the amps
and see what variation there is. Again - a pointer would be useful - even
a book.

To make meaningful measurements on an amplifier you don't need a lot of test
equipment, just a good soundcard that samples at 192kHz and some software. I
prefer to use individual physical instruments, but that's because I'm both
an old fart and happen to have them. If I wasn't the first and didn't have
the second, then I would probably use the PC method totally. I have the
RightMark audio analyser software which is freeware, and from what I can see
works beatifully. http://audio.rightmark.org If you are going to make
useful THD and frequency response measurements, you need a 192k sampling
card to give you some 85 kHz of measuring bandwidth.

You *will* need a good dummy load to run the amps into. I have four 50 watt
4 ohm resistors mounted on a large heatsink, each with a flying lead and
croc clips. I can thus set up 4 x 4 ohms @ 50 watt, 2x8 ohms @ 100 watt, 2x
2 ohms @ 100 watt, 1x4 ohms @ 200 watt and other combinations.


That's very interesting, thanks. I'm sure I have the bits and pieces to
manage that, and it'd be a useful means of checking what I've got.

As for books, can't recommend anything specific, as my training was pretty
much continuous since I was 16, in the days when a 100kHz 'scope was the
best my school had. There are a number of good text books about, but they
tend to be rather expensive. There's usually a bookshop at the pro-audio
exhibitions, so if you get a chance to go to the AES, or IBC in Amsterdam or
NAB in Las Vegas, or know anyone else going, they may be able to find one or
two for you.


Thanks again. Really, this doesn't come easy to me. I tried reading a
book on valve amplifier design - quite basic by all accounts. I got
about a third of the way through and got stuck on one paragraph. I
simply couldn't accept on face value what was said. I could make the
effort one day (wilful ignorance,as Jim might say!), and post to one of
the tech groups. I've never got stuck in the same way with some quite
complicated political and social theory. And I would add that i have a
day job ;-)


Once differences are identified and quantified, then any audible
differences are soon accounted for. What I am saying is that with modern
SS amplifiers, it is easy except at the very cheapest end for the above
criteria to be met, consequently any but the cheapest amps will all sound
the same when played at the same volume into the same (sensible) load.
I experience four areas of difference - bass, 'soundstage', voices and
'top end'. Bass is easiest (A NAD 3020 was quite 'soft' compared to the
clearly defined bass lines of a Rotel integrated), a Quad 405 is
noticeably sibilant, and a Roksan Kandy I had a while ago was plain
shrill. I'm using a Cambridge AV amp at the moment, and I can't detect a
difference except at very high sound levels between its built in power
amp, and a Rose power amp, and a Behringer power amp. I'm pretty pleased
with the Cambridge for casual listening.

Just to give you some idea of how difficult it is to assess equipment
subjectively, let me quote from the Hi-Fi Choice reviews for the NAD
3020/3120 and Quad 405:-

NAD. "Bass showed a touch of boom while the mid seemed a little hard tonally
and the treble was mildly grainy."

Quad. " The treble was still showing some mild "feathery muzziness" while
the bass could have offered more extension and impact."

Both these amplifiers are flat to +- 1dB between 20Hz and 20kHz


And I have to take into account I've probably read similar reviews, and
they could affect what I'm hearing. I don't doubt that.

The same applies to CD
players and anything else that meets the criteria. Items that don't meet
the criteria (rarely or never) are transducers and consequently
microphones, pick-up cartridges and loudspeakers will all sound different
for easily identified and measurable reasons.

The *real* point I'd like to explore would be the notion that conventional
measurements are not a reliable guide to sound experienced. Of course
(again) such a statement comes across to some as something between
witchcraft, homoeopathy and astrology, but I set it out here just so you
can categorise my comments properly.

Rob

Yes, witchcraft, homeopathy, astrology *and* subjective hi-fi reviewing are
all characterised by believing things that do not show up under any sort of
scientific scrutiny.

Conventional measurements, *if applied correctly* can characterise
completely the operation of a piece of audio equipment.

That they can is not in doubt. It's possible that some, on learning that
they measure within limits, subsequently hear no difference. And some,
on reading a review that is half-decent, hear difference. It's
bothersome, I know.

What they can't do
is to characterised your reaction to that piece of equipment. What I mean by
this is that we all are conditioned by magazines, friends, received wisdom
etc, and that we characterise the sound we hear according to our prejudices.
When these prejudices are not able to operate, as in unsighted testing, then
many of the previously-held views dissappear.

So far, no-one has been able to come up with realiable evidence that there
are some aspects of audio performance that we have not yet been able to
measure. The closest I suppose is the performance of bit reduced digital
encoders of the psychacoustic type. Conventional test measurements don't
show up the artefacts we all claim to hear, although I have surprised myself
as to just how good MPEG encoders are at low bit rates when listened to
blind. When I was last working professionally with encoders, there was no
standard test signal which would correlate with what we could hear, and as
far as I can recall, every customer had a favourite CD or two which they
used to evaluate audio quality. Possibly someone on this group may have more
recent information on testing audio codecs.


I'm archiving about 300 CDs to disc before I get rid of them. I compared
the wav rips to some 192kbs mp3s last night, and while I could hear a
difference (just, I got caught out from time to time) it wasn't
significant, and made *preference* very hard to establish.

On a tangent, I am interested in learning why my valve amplifier sounds
so good. Put simply, if all of this can be measured, why can't some sort
of (distortion?!) filter be used to recreate the sound? The novelty of
the glow has more or less worn off, and if such a thing existed I'd give
it a go.

All other audio equipment is now so well understood and characterised by
conventional measurements that when audio differences do show up, they are
easily dealt with. Note however, that some products are deliberately
designed to sound different, for marketing reasons. Linn realised this as
long ago as the 80s with their Kan loudspeaker which was highly coloured and
with an appalling frequency response. I can't believe that a company of
Linn's engineering abilities did this by accident or incompetence, so it
must have been deliberate to stand out in demos as sounding different to the
rest of their competitors.

Speakers really are the sharp end of things IMO. My guess with Linn is
that they had in mind a speaker that could (somehow) sound accurate (or
their version of it) given domestic settings. I'm very happy with
Dynaudio speakers, but I'm intrigued by Keith's various forays -
impressive and disarming at once, and something I mean to try one day.

More recently, popularity of SET amplifiers and
high-efficiency horns points to the desire of listeners to have something
that's different to the prevailing norm.


Agreed - and in the scheme of things does no harm.

http://audiopages.googlepages.com

A good read, and a splendid system by the looks of it. A valve cartridge
pre-amp wouldn't go amiss :-)

Rob

--
http://audiopages.googlepages.com
"Rob" wrote in message
...
Serge Auckland wrote:
"Rob" wrote in message
...
[snip to bit about measurement]

1. Listening. The physical swapping out is doable. I would find/have
found level matching incredibly difficult. I tried this some months ago
with a 20-20k hz cd, taking levels at various points. I couldn't get
close to level matching across the range - I'd got weird 'spikes' at
points in the frequency range, around 2000hz IIRC. Even using the same
amp! Anyway, this sort-of-obviously means I need some more reliable
measuring equipment and techniques - any pointers here would be
appreciated.

This is quite normal:- If you are level-matching by using a SPL meter
in-room, then what you are measuring is the sound pressure at the point
at which you are holding the meter (so much is obvious :-) ) At mid-high
frequencies, the measurement you make will be very influenced by local
reflections, even from your own body. If I put my SPL meter on a
photographic tripod, then move around even only by a few centimetres, the
reading on the meter will vary by a dB or more at mid-high frequencies.
At low frequencies, it is much more stable, being dependent on room
reflections, and hence positioning in the room, much less influenced by
body movements as the wavelength of the sound becomes larger than my body
dimensions (even in my current less than sylph-like state)


Yes, I'd realised this - if not strictly taken it into account. Quite how
I clamp the listener's head, and avoid ear twitching, is an issue for a
later day :-)

You should level-match by measuring across the 'speaker terminals ideally
by using a high impedance audio millivoltmeter (once called a
valve-voltmeter). I have found that my normal inexpensive multimeter is
accurate enough at low audio frequencies. Mine actually work fine up to
20kHz, but if you use a 100Hz tone for level matching you should be fine.
Use your SPL meter to make sure the volume level is around 85dBC (say
80dBA) at 100Hz and level-match with the multimeter at that loudness. You
should fine it easy to level-match to better than 0.5dB even with a
multimeter.


Ak! I can't have that I'm afraid. I think I understand what you're saying,
but what I'm saying is that loudspeakers are non-linear loads and
something (I know not what) might be happening between the signal and the
sound. I suppose my thinking here comes from tests of kit I've read in
magazines, and the suggestion that different amplifiers do not do the
'wire with gain' thing in a linear way. But I'm afraid I can't remember
the details - I don't even have a hifi magazine in the house, having given
up on them many years ago.

Loudspeakers are *not* in any way non-linear loads. The impedance of the
'speaker system varies with frequency, but (except for some electrostatics)
there is no variation with level, consequently they *are* linear. Amplifiers
are also linear to the limits of their distortion characteristics, which is
why amplifiers need to have low distortion, maintained into the minimum load
presented to them. What you may be referring to is that loudspeakers have
impedances that drop well below (and raise well above) their nominal figure.
Some amplifiers may change their sound at the impedance minima. This is
true, but this is why I say that two amplifiers being compared have to be
used within their design limits. If one amp is designed for 4-8 ohms, and
another amp is designed for 3-8 ohms, then both should sound identical into
a nominally 8 ohm 'speaker, but could well sound different into a nominally
4 ohm 'speaker as the minimum load could be 3.2 ohms (the minimum figure to
qualify for a 4 ohm rating).


Now, if you're saying to me that any amplifier that has the specification
you've detailed will interact with a given loudspeaker ('normal'/domestic)
in exactly the same way within the specification you detailed then I'll do
my best to satisfy myself that is in fact the case (if I can), and adopt
your test. You just have to say 'it is the case' - I appreciate your
patience thus far. In addition, I would expect the level matching to apply
between 20 -20k hz - just my little quirk :-)


2. If listening using level matched DBT does reveal difference, the
spotlight then turns on the amps. I would then have to measure the amps
and see what variation there is. Again - a pointer would be useful -
even a book.

To make meaningful measurements on an amplifier you don't need a lot of
test equipment, just a good soundcard that samples at 192kHz and some
software. I prefer to use individual physical instruments, but that's
because I'm both an old fart and happen to have them. If I wasn't the
first and didn't have the second, then I would probably use the PC method
totally. I have the RightMark audio analyser software which is freeware,
and from what I can see works beatifully. http://audio.rightmark.org If
you are going to make useful THD and frequency response measurements, you
need a 192k sampling card to give you some 85 kHz of measuring bandwidth.

You *will* need a good dummy load to run the amps into. I have four 50
watt 4 ohm resistors mounted on a large heatsink, each with a flying lead
and croc clips. I can thus set up 4 x 4 ohms @ 50 watt, 2x8 ohms @ 100
watt, 2x 2 ohms @ 100 watt, 1x4 ohms @ 200 watt and other combinations.


That's very interesting, thanks. I'm sure I have the bits and pieces to
manage that, and it'd be a useful means of checking what I've got.

As for books, can't recommend anything specific, as my training was
pretty much continuous since I was 16, in the days when a 100kHz 'scope
was the best my school had. There are a number of good text books about,
but they tend to be rather expensive. There's usually a bookshop at the
pro-audio exhibitions, so if you get a chance to go to the AES, or IBC in
Amsterdam or NAB in Las Vegas, or know anyone else going, they may be
able to find one or two for you.


Thanks again. Really, this doesn't come easy to me. I tried reading a book
on valve amplifier design - quite basic by all accounts. I got about a
third of the way through and got stuck on one paragraph. I simply couldn't
accept on face value what was said. I could make the effort one day
(wilful ignorance,as Jim might say!), and post to one of the tech groups.
I've never got stuck in the same way with some quite complicated political
and social theory. And I would add that i have a day job ;-)


Once differences are identified and quantified, then any audible
differences are soon accounted for. What I am saying is that with
modern SS amplifiers, it is easy except at the very cheapest end for
the above criteria to be met, consequently any but the cheapest amps
will all sound the same when played at the same volume into the same
(sensible) load.
I experience four areas of difference - bass, 'soundstage', voices and
'top end'. Bass is easiest (A NAD 3020 was quite 'soft' compared to the
clearly defined bass lines of a Rotel integrated), a Quad 405 is
noticeably sibilant, and a Roksan Kandy I had a while ago was plain
shrill. I'm using a Cambridge AV amp at the moment, and I can't detect a
difference except at very high sound levels between its built in power
amp, and a Rose power amp, and a Behringer power amp. I'm pretty pleased
with the Cambridge for casual listening.

Just to give you some idea of how difficult it is to assess equipment
subjectively, let me quote from the Hi-Fi Choice reviews for the NAD
3020/3120 and Quad 405:-

NAD. "Bass showed a touch of boom while the mid seemed a little hard
tonally and the treble was mildly grainy."

Quad. " The treble was still showing some mild "feathery muzziness" while
the bass could have offered more extension and impact."

Both these amplifiers are flat to +- 1dB between 20Hz and 20kHz


And I have to take into account I've probably read similar reviews, and
they could affect what I'm hearing. I don't doubt that.

The same applies to CD
players and anything else that meets the criteria. Items that don't
meet the criteria (rarely or never) are transducers and consequently
microphones, pick-up cartridges and loudspeakers will all sound
different for easily identified and measurable reasons.

The *real* point I'd like to explore would be the notion that
conventional measurements are not a reliable guide to sound experienced.
Of course (again) such a statement comes across to some as something
between witchcraft, homoeopathy and astrology, but I set it out here
just so you can categorise my comments properly.

Rob

Yes, witchcraft, homeopathy, astrology *and* subjective hi-fi reviewing
are all characterised by believing things that do not show up under any
sort of scientific scrutiny.

Conventional measurements, *if applied correctly* can characterise
completely the operation of a piece of audio equipment.

That they can is not in doubt. It's possible that some, on learning that
they measure within limits, subsequently hear no difference. And some, on
reading a review that is half-decent, hear difference. It's bothersome, I
know.

What they can't do
is to characterised your reaction to that piece of equipment. What I mean
by this is that we all are conditioned by magazines, friends, received
wisdom etc, and that we characterise the sound we hear according to our
prejudices. When these prejudices are not able to operate, as in
unsighted testing, then many of the previously-held views dissappear.

So far, no-one has been able to come up with realiable evidence that
there are some aspects of audio performance that we have not yet been
able to measure. The closest I suppose is the performance of bit reduced
digital encoders of the psychacoustic type. Conventional test
measurements don't show up the artefacts we all claim to hear, although I
have surprised myself as to just how good MPEG encoders are at low bit
rates when listened to blind. When I was last working professionally with
encoders, there was no standard test signal which would correlate with
what we could hear, and as far as I can recall, every customer had a
favourite CD or two which they used to evaluate audio quality. Possibly
someone on this group may have more recent information on testing audio
codecs.


I'm archiving about 300 CDs to disc before I get rid of them. I compared
the wav rips to some 192kbs mp3s last night, and while I could hear a
difference (just, I got caught out from time to time) it wasn't
significant, and made *preference* very hard to establish.

On a tangent, I am interested in learning why my valve amplifier sounds so
good. Put simply, if all of this can be measured, why can't some sort of
(distortion?!) filter be used to recreate the sound? The novelty of the
glow has more or less worn off, and if such a thing existed I'd give it a
go.

There are plenty of valve simulators available on the pro market for DAW
plug-ins. They add level-dependent harmonic distortion, some low-level noise
and soft clipping. Some people pass their mix or part of it through a
reel-to-reel recorder to get a similar effect.

All other audio equipment is now so well understood and characterised by
conventional measurements that when audio differences do show up, they
are easily dealt with. Note however, that some products are deliberately
designed to sound different, for marketing reasons. Linn realised this as
long ago as the 80s with their Kan loudspeaker which was highly coloured
and with an appalling frequency response. I can't believe that a company
of Linn's engineering abilities did this by accident or incompetence, so
it must have been deliberate to stand out in demos as sounding different
to the rest of their competitors.

Speakers really are the sharp end of things IMO. My guess with Linn is
that they had in mind a speaker that could (somehow) sound accurate (or
their version of it) given domestic settings. I'm very happy with Dynaudio
speakers, but I'm intrigued by Keith's various forays - impressive and
disarming at once, and something I mean to try one day.

I am perhaps a bit more cynical of Linn's motives, as the Kan wasn't in any
sense of the word accurate. Dynamic, aggressive, punchy, loud, but not
accurate...I also like Dynaudio 'speakers, found them very pleasant. Sort of
'speaker you don't tire of.


More recently, popularity of SET amplifiers and
high-efficiency horns points to the desire of listeners to have something
that's different to the prevailing norm.


Agreed - and in the scheme of things does no harm.

http://audiopages.googlepages.com

A good read, and a splendid system by the looks of it. A valve cartridge
pre-amp wouldn't go amiss :-)

Thanks. The two turntables both have built-in RIAA pre-amps, which are quiet
and accurate in their EQ, so I have no real incentive to change.
Valve preamps for MC cartridges have great overload figures, but tend to be
a bit noisy, so I prefer SS.

S.

http://audiopages.googlepages.com

Serge Auckland wrote:

-- http://audiopages.googlepages.com "Rob"
wrote in message
...
Serge Auckland wrote:
"Rob" wrote in message
...
[snip to bit about measurement]

1. Listening. The physical swapping out is doable. I would
find/have
found level matching incredibly difficult. I tried this some
months ago
with a 20-20k hz cd, taking levels at various points. I
couldn't get
close to level matching across the range - I'd got weird
'spikes' at
points in the frequency range, around 2000hz IIRC. Even using
the same
amp! Anyway, this sort-of-obviously means I need some more
reliable
measuring equipment and techniques - any pointers here would be
appreciated.

This is quite normal:- If you are level-matching by using a SPL
meter
in-room, then what you are measuring is the sound pressure at the
point
at which you are holding the meter (so much is obvious :-) ) At
mid-high
frequencies, the measurement you make will be very influenced by
local
reflections, even from your own body. If I put my SPL meter on a
photographic tripod, then move around even only by a few
centimetres, the
reading on the meter will vary by a dB or more at mid-high
frequencies.
At low frequencies, it is much more stable, being dependent on room
reflections, and hence positioning in the room, much less
influenced by
body movements as the wavelength of the sound becomes larger than
my body
dimensions (even in my current less than sylph-like state)


Yes, I'd realised this - if not strictly taken it into account.
Quite how
I clamp the listener's head, and avoid ear twitching, is an issue
for a
later day :-)

You should level-match by measuring across the 'speaker terminals
ideally
by using a high impedance audio millivoltmeter (once called a
valve-voltmeter). I have found that my normal inexpensive
multimeter is
accurate enough at low audio frequencies. Mine actually work fine
up to
20kHz, but if you use a 100Hz tone for level matching you should
be fine.
Use your SPL meter to make sure the volume level is around 85dBC
(say
80dBA) at 100Hz and level-match with the multimeter at that
loudness. You
should fine it easy to level-match to better than 0.5dB even with a
multimeter.


Ak! I can't have that I'm afraid. I think I understand what you're
saying,
but what I'm saying is that loudspeakers are non-linear loads and
something (I know not what) might be happening between the signal
and the
sound. I suppose my thinking here comes from tests of kit I've read in
magazines, and the suggestion that different amplifiers do not do the
'wire with gain' thing in a linear way. But I'm afraid I can't
remember
the details - I don't even have a hifi magazine in the house,
having given
up on them many years ago.

Loudspeakers are *not* in any way non-linear loads. The impedance of the
'speaker system varies with frequency, but (except for some electrostatics)
there is no variation with level, consequently they *are* linear.
Amplifiers
are also linear to the limits of their distortion characteristics, which is
why amplifiers need to have low distortion, maintained into the minimum
load
presented to them.

***** That being the case, it's very likely that you're right - most
modern amplifiers sound identical. I suspect we're at odds about the
definition of linear, and variables under consideration, and hence the
maintenance of linearity across all variables. It makes *sense* to me to
measure the sound and not the voltage, but I think you're saying that
has no sense - the two things come to the same in a linear way, and it's
a waste of time measuring sound when there's a far easier method.


What you may be referring to is that loudspeakers have
impedances that drop well below (and raise well above) their nominal
figure.
Some amplifiers may change their sound at the impedance minima. This is
true, but this is why I say that two amplifiers being compared have to be
used within their design limits. If one amp is designed for 4-8 ohms, and
another amp is designed for 3-8 ohms, then both should sound identical into
a nominally 8 ohm 'speaker, but could well sound different into a nominally
4 ohm 'speaker as the minimum load could be 3.2 ohms (the minimum figure to
qualify for a 4 ohm rating).

****** That's OK - I don't have any speakers below 3.5 ohms, and I'm
pretty sure all the amps are OK to 4 ohms. And yes, I think that's where
I get the non-linearity thing from, at least in part.

Rob

In article , Serge Auckland
wrote:

"Rob" wrote in message
...

1. Listening. The physical swapping out is doable. I would find/have
found level matching incredibly difficult. I tried this some months
ago with a 20-20k hz cd, taking levels at various points. I couldn't
get close to level matching across the range - I'd got weird 'spikes'
at points in the frequency range, around 2000hz IIRC. Even using the
same amp! Anyway, this sort-of-obviously means I need some more
reliable measuring equipment and techniques - any pointers here would
be appreciated.

This is quite normal:- If you are level-matching by using a SPL meter
in-room, then what you are measuring is the sound pressure at the point
at which you are holding the meter (so much is obvious :-) ) At
mid-high frequencies, the measurement you make will be very influenced
by local reflections, even from your own body.

This is another reason why 'casual' listening tests can be so unreliable.
Small changes of the listening position, etc, can affect the sound reaching
the ears. Thus this is one of many factors which can produce 'differences'
even if the amplifiers are functionally identical.



You *will* need a good dummy load to run the amps into. I have four 50
watt 4 ohm resistors mounted on a large heatsink, each with a flying
lead and croc clips. I can thus set up 4 x 4 ohms @ 50 watt, 2x8 ohms
@ 100 watt, 2x 2 ohms @ 100 watt, 1x4 ohms @ 200 watt and other
combinations.

If you can stand the 'noise' it can also be revealing to do some
measurements using the speakers as loads. This may expose effects that
don't show up with resistive loads. However as well as the row, this can
risk speaker damage unless you take care.

As for books, can't recommend anything specific, as my training was
pretty much continuous since I was 16, in the days when a 100kHz 'scope
was the best my school had. There are a number of good text books
about, but they tend to be rather expensive. There's usually a bookshop
at the pro-audio exhibitions, so if you get a chance to go to the AES,
or IBC in Amsterdam or NAB in Las Vegas, or know anyone else going,
they may be able to find one or two for you.

I can't off-hand think of a book suitable for 'beginners' that explains how
they can do a sensible set of tests on something like an amplifier. Can
anyone suggest one? I'd agree that these days what would be 'best' would
probably be one that allows the user to make use of their computer with an
audio input as the test instrument, and to use a test CD as the source.

FWIW I've recently been writing a series of short articles for a computing
mag which deals with how people can generate and analyse sound files. It
would be useful to know if there is a decent book dealing with amp testing,
or indeed testing other items of audio gear, which would be suitable for
enthusuasts who don't already have experience and know how such things are
done, and why.



Just to give you some idea of how difficult it is to assess equipment
subjectively, let me quote from the Hi-Fi Choice reviews for the NAD
3020/3120 and Quad 405:-

NAD. "Bass showed a touch of boom while the mid seemed a little hard
tonally and the treble was mildly grainy."

Quad. " The treble was still showing some mild "feathery muzziness"
while the bass could have offered more extension and impact."

Both these amplifiers are flat to +- 1dB between 20Hz and 20kHz

The problem with such statements in 'reviews' is that we not only have no
idea if they were based on a genuine difference. We also can't tell if they
were due to something like the reviewer moving their head an inch, or
opening the room door, or...

A number of controlled tests have been done where the amps A and B were
played in randomised pairings. i.e. in each case A was played then A, or A
then B, or B then B, or B then A. The listeners were asked in each case if
the sound for each example in a pair was the 'same' or 'different'. This
exposed a significant (statistically) tendency for people to say that AA or
BB were 'different' rather than 'same'. Thus showing that people report
differences when the amp is unchanged and the system has not been altered.


Yes, witchcraft, homeopathy, astrology *and* subjective hi-fi reviewing
are all characterised by believing things that do not show up under any
sort of scientific scrutiny.

That is perhaps a little unfair as some reports may well be reliable. The
snag is distinguishing them from the unreliable claims. Alas, if you have
no way of telling if an individual claim is reliable or not, and the
authors are well-known for making claims which can't be backed up, then
their claims become worthless. Might as well toss coins to make a decision
as to use the reviews as a basis.

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

In article , Rob
wrote:
Serge Auckland wrote:
"Rob" wrote in message
...
[snip to bit about measurement]



Yes, I'd realised this - if not strictly taken it into account. Quite
how I clamp the listener's head, and avoid ear twitching, is an issue
for a later day :-)

You don't need to do that. Indeed it would be futile as there will be other
variables which will be likely to change in an unpredictable way and which
you won't be monitoring.

The appropriate response is to make *many* tests and allow your head to
move as you please in each case. This will tend over time to average away
the effect of the uncontrolled variable and systematic effects then show
through the statistical analysis.

Of course, if the uncontrolled/unknown variables are so pervasive and
serious, then it may be daft to waste time trying to see if you can find
any amplifier 'differences' which they have swamped. Indeed, this is often
my experience, and has for years seemed to me to make following this topic
beyond a given point rather a waste of time. However if others wish to
struggle to find evidence for amp 'differences' which seem to vanish in
simply comparison tests I guess I should just allow them to waste their own
time. If they ever find anything I'd be interested. :-)

You should level-match by measuring across the 'speaker terminals
ideally by using a high impedance audio millivoltmeter (once called a
valve-voltmeter). I have found that my normal inexpensive multimeter
is accurate enough at low audio frequencies. Mine actually work fine
up to 20kHz, but if you use a 100Hz tone for level matching you
should be fine. Use your SPL meter to make sure the volume level is
around 85dBC (say 80dBA) at 100Hz and level-match with the multimeter
at that loudness. You should fine it easy to level-match to better
than 0.5dB even with a multimeter.


Ak! I can't have that I'm afraid. I think I understand what you're
saying, but what I'm saying is that loudspeakers are non-linear loads
and something (I know not what) might be happening between the signal
and the sound.

Unfortunately vague feelings you have that something might be happening,
but you've no idea what/how/why/etc are in themselves worthless. The
requirement is quite simple in principle. You form an idea as to a 'cause'
and then devise a test for it, then do the test and evaluate the evidence.

The snag is that a suitable test requires some knowledge and understanding
to devise. And that then doing the test and evaluating the results requires
more understanding and some sweat. The cost of mining understanding from a
flood of formless fears and delusions, I'm afraid.

I suppose my thinking here comes from tests of kit I've
read in magazines, and the suggestion that different amplifiers do not
do the 'wire with gain' thing in a linear way. But I'm afraid I can't
remember the details - I don't even have a hifi magazine in the house,
having given up on them many years ago.

Now, if you're saying to me that any amplifier that has the
specification you've detailed will interact with a given loudspeaker
('normal'/domestic) in exactly the same way within the specification you
detailed then I'll do my best to satisfy myself that is in fact the
case (if I can), and adopt your test. You just have to say 'it is the
case' - I appreciate your patience thus far. In addition, I would
expect the level matching to apply between 20 -20k hz - just my little
quirk :-)

The key point is that the loudspeaker has specifically limited info on
what amplifer you are using. It can't read the name badge, nor the reviews.
It can't look and see if you have valves or transistors. It just finds that
a pattern of voltages is being asserted on its input terminals.

If two amplifiers are able to assert the same pattern and level of voltage,
they will be able to supply the same current and power patterns. Either the
speaker then delivers the same results in both cases or it does not.

If it does not, then any 'difference' is due to the speakers, not the amp.

If it does, then the amps are functionally the same so far as the test
system is concerned.



Thanks again. Really, this doesn't come easy to me. I tried reading a
book on valve amplifier design - quite basic by all accounts. I got
about a third of the way through and got stuck on one paragraph. I
simply couldn't accept on face value what was said. I could make the
effort one day (wilful ignorance,as Jim might say!), and post to one of
the tech groups. I've never got stuck in the same way with some quite
complicated political and social theory. And I would add that i have a
day job ;-)

It is not 'wilful ignorance' if you are trying as and when you can, despite
finding it hard going (or inconvenient).

These matters are *not* either trivial or self-evident. People have spent
many hard years working on them to develop an understanding. The 'wilful
ignorance' is when people can't be bothered to study and understand what
has been done and airly dismiss any evidence-based results which do not
suit their personal fancies.


Just to give you some idea of how difficult it is to assess equipment
subjectively, let me quote from the Hi-Fi Choice reviews for the NAD
3020/3120 and Quad 405:-

NAD. "Bass showed a touch of boom while the mid seemed a little hard
tonally and the treble was mildly grainy."

Quad. " The treble was still showing some mild "feathery muzziness"
while the bass could have offered more extension and impact."

Both these amplifiers are flat to +- 1dB between 20Hz and 20kHz


And I have to take into account I've probably read similar reviews, and
they could affect what I'm hearing. I don't doubt that.

Please bear in mind that a great deal of what appears in magazine reviews
may well be tosh. This can apply to the 'measured results' in some cases as
well as the impressive purple prose about the 'sound'. Treat with great
caution.


I'm archiving about 300 CDs to disc before I get rid of them. I compared
the wav rips to some 192kbs mp3s last night, and while I could hear a
difference (just, I got caught out from time to time) it wasn't
significant, and made *preference* very hard to establish.

Erm... what will you do when the HD fails without warning?

On a tangent, I am interested in learning why my valve amplifier sounds
so good. Put simply, if all of this can be measured, why can't some sort
of (distortion?!) filter be used to recreate the sound? The novelty of
the glow has more or less worn off, and if such a thing existed I'd give
it a go.

1) Don't assume any difference is for the reason you give. :-)

2) Look carefully at reviews of valve amps. Note when you can what they say
about 'output impedance' or 'damping factor'.

3) Note than many reviews and makers specs simply fail to mention various
factors which may well affect the results - particularly with the common
types of valve amp design.

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

In article , Serge Auckland
wrote:

You put everything into the 'sig' of the posting so I had to fiddle about
the rescue it for the following:


Ak! I can't have that I'm afraid. I think I understand what you're
saying, but what I'm saying is that loudspeakers are non-linear loads
and something


Loudspeakers are *not* in any way non-linear loads.

Erm... IIUC that isn't correct. 'Dynamic' (i.e. coil and magnet) speakers
have distinct nonlinearities. e.g. the inductance of the coil is affected
by its surroundings and the interaction varies with displacement. I have a
feeling there are JAES papers on this stemming from ideas about changing
from voltage drive to current drive. IIRC There is also a similar effect in
terms of the back EMF varying with the changing field coupling. I think
this can be measured as current waveform effects.

However that does not change the basis of using the voltage input to the
speaker for setting levels when doing an amp comparison as any speaker
nonlinearity will be 'common mode'. If you are using the same speaker to
compare two amps, either they are (un)affected by the above in the
(audibly) same way, or not.



On a tangent, I am interested in learning why my valve amplifier sounds
so good. Put simply, if all of this can be measured, why can't some
sort of (distortion?!) filter be used to recreate the sound? The
novelty of the glow has more or less worn off, and if such a thing
existed I'd give it a go.

There are plenty of valve simulators available on the pro market for DAW
plug-ins. They add level-dependent harmonic distortion, some low-level
noise and soft clipping. Some people pass their mix or part of it
through a reel-to-reel recorder to get a similar effect.

That does. however, omit the frequency response changes which arise when a
valve amp design has a high output impedance and the speaker has a load
impedance that varies a lot with frequency.

Some recent reviews I've seen list 'samping factors' down to about 2! I'd
be amazed if that didn't change the results quite audibly with many
speakers. Indeed, it seemed to me to make a subjective review worthless
unless you are going to use the same speakers as the reviewer. (Alas, they
sometimes don't bother to say what speakers they used, which doesn't help
much!)

The above point are examples of why I've personally always been a fan of
people taking o/p impedance measurements more seriously than is normal in
audio mags.

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

Jim Lesurf wrote:
In article , Rob
wrote:
Jim Lesurf wrote:


Is your concern simply that the evidence is based only on all the
cases tried? If so, it is open to you or anyone else to collect more
evidence. i.e. to carry out more comparison tests of appropriate kinds.


My concern (such as it is) is simply this: I maintain different modern
SS amplifiers can sound different.

It is essentially self-evident that they 'can'. Indeed, you can choose to
design two amps so that they do. However when people have done tests on
amps that were not specifically designed to do this and which were designed
to amplify, then they tend to be able to distinguishg one from another.
i.e. they are found not to 'sound different'.

But if you choose a design which is - either by deliberate choice or
incompetence - making the output sufficiently different from a scaled
version of the input then it may well 'sound different'.


I wouldn't mean to use or buy an incompetent amplifier. I take it you'd
consider anything from the big names (Sony, Cambridge, Rotel, NAD, Quad,
Behringer etc) to be competent - in the sense that if they did turn out
to be incompetent you'd be surprised.


They are particularly affected by load (speakers) in my extraordinarily
humble opinion. Others (on this NG, not you explicitly so far as I can
tell) maintain this is highly unlikely. For example, I'm pretty sure I
could hear the difference between a Cambridge audio power amp, a
semi-pro power amp, and a Quad 405 power amp.

Well you may be "pretty sure", but the problem is that many people in the
past have been "pretty sure" of similar claims - but then failed to be able
to do what they were "pretty sure" of when tested on the basis of the
sounds in a matter that excluded well known and uncontentious sources of
differences. Thus your belief is simply a statement of faith at this point,
not evidence. Given all the previous failures it isn't clear why anyone
would be wise to take your belief seriously *unless* you put it to such a
test and showed you can do what you believe. Until then...


Well, it's the result of experience. Many people claim similar
experiences. As 'evidence' of 'fact' I'd agree that it's flimsy. Faith?
Belief with reason? Yes, we agree.

I don't think there's much point going over this again. You may not
(hopefully!) remember, but this comes up every six months or so. I'm
simply trying to get across that it's a hypothesis worth testing, and
Serge has kindly offered some practical tips to test the notion. The
problem I have is that technocrats *seem* on occasion rather
condescending towards the opinions of others, without offering a
reasoned explanation for their experience.

Perhaps if I explain it this way - it's not what people think and say,
but why they think and say it, which is of most importance. Do you
*know* (or even have the vaguest idea) why people (like me) experience
these 'physical world anomalies'? I certainly don't *know*.


To do this you would have to define what you mean by "most". i.e. do
you mean more than half of all amps in use, or more than half the
designs, or more than half of those ever used, etc? e.g. can we ignore
the amps in TVs and portable radios and cheap 'music centers'? If so,
we have to define the line that rules in/out a given amp, and give a
plausible basis for doing so.


Again, Serge guided me on this. This issue has been mentioned many
times. 'Most' means virtually all modern SS amps that meet certain
criteria (1).

On such a basis the evidence supports what Serge has been saying. If you
wish to contend with that you will need relevant assessable evidence for
people to take you seriously. Given the history of this topic people will
regard claims of what you are "pretty sure" of as no more than an
unsupported belief of the kind which has in the past repeatedly been shown
to have no foundation.


Again, we've trawled through all of this. It *is* supported, albeit not
with the scientific test *you* require.



I can only point out that "most" (indeed all) the relevant tests I
know of showed no sign that those who listened could distinguish one
amp from another - given a fairly basic set of requirements like level
matching.

I find this virtually impossible to do, even (or because of?) the crude
sound meter I have. I assume the idea is to match levels between 20-20K
Hz?

You don't require a sound pressure meter if you are using the same speakers
and listening position, etc, throughout. You can then use a meter to check
the input voltages to the speakers and ensure they are about the same for
one amp as for the other. If you find differences of the order of 1dB or
more you can expect that to be audible. But if the differences are much
smaller - e.g. around 0.1dB - then that is unlikely to produce an audible
difference. Hence the aim is to make these as small as feasible, and well
below 1dB.


Yes, thanks, Serge helped me on that one as well.

Rob

"Rob" wrote in message


I don't think there's much point going over this again.

Agreed, but that doesn't seem to stop you, Rob.

You may not (hopefully!) remember, but this comes up
every six months or so.

It is so trite...

I'm simply trying to get across
that it's a hypothesis worth testing,

It is a hyposthesis that at this point has been tested to death. The results
are well-known among those who have done a craftsmanlike job of testing it.
If you can competently search the web, the resources and evidence are all
there.

In article , Rob
wrote:
Jim Lesurf wrote:
In article , Rob
wrote:
Jim Lesurf wrote:

My concern (such as it is) is simply this: I maintain different
modern SS amplifiers can sound different.

It is essentially self-evident that they 'can'. Indeed, you can choose
to design two amps so that they do. However when people have done
tests on amps that were not specifically designed to do this and which
were designed to amplify, then they tend to be able to distinguishg
one from another. i.e. they are found not to 'sound different'.

But if you choose a design which is - either by deliberate choice or
incompetence - making the output sufficiently different from a scaled
version of the input then it may well 'sound different'.


I wouldn't mean to use or buy an incompetent amplifier. I take it you'd
consider anything from the big names (Sony, Cambridge, Rotel, NAD, Quad,
Behringer etc) to be competent - in the sense that if they did turn out
to be incompetent you'd be surprised.

Actually, I have long ceased being 'surprised' by what makers do at times.
:-) Hence I'd wary of placing too much reliance on the name badge.
Particularly now that various brands are no longer owned by the original
people who built them up to have a reputation for quality.

e.g. Despite Sony being a regarded name I recently tried a DVD recorder
from them. It had a cooling fan that was so loud that it was distracting
when listening to dialogue on items recorded. This is absurd and needless,
yet when I asked a local Sony center their reaction was 'they all do it'.
They seemed not to think it was a problem, and clearly had no idea that
quiet fans can be bought. Perhaps they think everyone is too busy looking
at the pictures to notice the sound of a helicopter accompanying it. ;-

The above is nothing to do with amps per se. But it shows that some
companies end up being run by suits, and have development engineers who
just produce what they are told by the suits and go home with their
paychecks. The spec says nothing about the fan not making a loud noise, so
this cheap one is fine...

I've also repeatedly witnessed companies who bring out 'new' products
distinctly worse than the ones they'd made a year or two before. Different
suits and a different R&D team. No internal communications. No clue.
The relevant suits and engineers are often isolated from the public
and getting contact with them is almost impossible in practice.

[snip]



Well you may be "pretty sure", but the problem is that many people in
the past have been "pretty sure" of similar claims - but then failed
to be able to do what they were "pretty sure" of when tested on the
basis of the sounds in a matter that excluded well known and
uncontentious sources of differences. Thus your belief is simply a
statement of faith at this point, not evidence. Given all the previous
failures it isn't clear why anyone would be wise to take your belief
seriously *unless* you put it to such a test and showed you can do
what you believe. Until then...


Well, it's the result of experience. Many people claim similar
experiences. As 'evidence' of 'fact' I'd agree that it's flimsy. Faith?
Belief with reason? Yes, we agree.

Many people claim many things - often contradicting one another in the
process. People believe all kinds of things. And tests in this area have
often shown that what someone believes they can do, they are unable to
show they could do when tested. To the point where many reviewers decided
to stop participating in such tests and instead behave as if they'd never
happened.

There are simply too many ways for someone to decide that a 'difference'
they hear is for one reason when in reality it is for another that hasn't
occurred to them, or they have made a simple mistake. Hence 'evidence'
comes from tests which are designed to deal with these problems and produce
results whose reliability and relevance can be assessed by others who
understand the problems. This means people can decide on the basis of the
evidence returned, and avoids having to accept what they are told in a
claim which may be worthless nonsense.



Perhaps if I explain it this way - it's not what people think and say,
but why they think and say it, which is of most importance. Do you
*know* (or even have the vaguest idea) why people (like me) experience
these 'physical world anomalies'? I certainly don't *know*.

What we (i.e. various engineers and experimental scientists who have
studied these topics seriously and engaged in tests, etc) 'know' (from both
experience and study) is that there are many pitfalls and other effects
that can cause people to ascribe the wrong 'reason' to what they think they
heard. Thus claims by people who don't know and understand the 'history' of
this topic nor the science tend to be greeted with sighs and groans. This
is why some of the reactions you get are dismissive or sharp. It is a loop
some of us have been round countless times for decades.

I get involved partly to try and point out to people that there is a great
deal on these topics which has aready been covered, and that the problems
of claims are also not new. Partly because I am open to the idea that every
now and then someone will come up with a genuinely 'new' idea or discovery.
Alas, this openness seems not to pan out in general... it does mean I
get bored at times at seeing the same old claims with the same lack of
a reliable basis or understanding by the claimant of the problems.

None of the above means 'all amps sound the same'. Nor that 'they all
sound different'. But it does means that claims which are not based
on taking the above into account are generally worthless. I can take
an interest in discussing them here as I do so for the reasons I give.
But most professionals in, say, the AES or IEEE would regard discussions
on this topic as a waste of breath, and expose you to abuse from people
who can't bear the idea that what they claim may simply be wrong. Look
back a few decades and you can find out why...




Again, Serge guided me on this. This issue has been mentioned many
times. 'Most' means virtually all modern SS amps that meet certain
criteria (1).

On such a basis the evidence supports what Serge has been saying. If
you wish to contend with that you will need relevant assessable
evidence for people to take you seriously. Given the history of this
topic people will regard claims of what you are "pretty sure" of as no
more than an unsupported belief of the kind which has in the past
repeatedly been shown to have no foundation.


Again, we've trawled through all of this. It *is* supported, albeit not
with the scientific test *you* require.

Afraid not. 'Support' does not mean that 'many people have the same
opinion'. Afraid you are still thinking in 'social science' terms, not
those of science and engineering. Nor does 'support' mean a result
which may be for many other well-established reasons which were ignored
by the claimant. Distinction between an observation and claiming the
*reason* for it. Someone may hear a 'difference' when they changed one
amp for another - but that does not support the amps being the *reason*
*unless* the other factors which can cause a change are dealt with.

Your approach would mean all views are 'supported' which makes 'support'
worthless since the 'supported' views conflict. Thus the need for boringly
tedious things like an appropriate test method to deal with the various
types of factors which otherwise cause misleading results, and to produce
assessable evidence rather than bald statements of 'faith'. What matters is
support by reliable, assessable, *evidence*. (This means both the test
method and the results.)

I am not talking about one single form of 'scientific test'. I am talking
about the approach being relevant and able to be shown to be capable of
producing reliable results rather than simple, well known, errors. I am
also talking about people knowing about, and understanding, the many
tests and sets of evidence which already exist and have largely settled
these matters for most professional engineers and academic scientists.

The point is that there are many well-known pitfalls and factors which can
produce 'differences' which have nothing to do with changing one amp for
another. Unless a comparison deals with these a claim that a change was due
to one *amp* 'sounding different' to another is worthless. It also is a
waste of time given that the pitfalls, and ways to deal with them, are well
known to those who have studied the topic seriously. Given this it is
perhaps understandable when some become irritable at seeing repeated claims
based on ignorance or dismissal of these factors. It is also why I've come
to expect evidence rather that assertions of faith or confidence.

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

"Jim Lesurf" wrote in message
...
In article , Serge Auckland
wrote:

You put everything into the 'sig' of the posting so I had to fiddle about
the rescue it for the following:


Ak! I can't have that I'm afraid. I think I understand what you're
saying, but what I'm saying is that loudspeakers are non-linear loads
and something


Loudspeakers are *not* in any way non-linear loads.

Erm... IIUC that isn't correct. 'Dynamic' (i.e. coil and magnet) speakers
have distinct nonlinearities. e.g. the inductance of the coil is affected
by its surroundings and the interaction varies with displacement. I have a
feeling there are JAES papers on this stemming from ideas about changing
from voltage drive to current drive. IIRC There is also a similar effect
in
terms of the back EMF varying with the changing field coupling. I think
this can be measured as current waveform effects.

However that does not change the basis of using the voltage input to the
speaker for setting levels when doing an amp comparison as any speaker
nonlinearity will be 'common mode'. If you are using the same speaker to
compare two amps, either they are (un)affected by the above in the
(audibly) same way, or not.

Jim, as befits an academic, is of course correct in saying that there are
non-linearities, but as far as I know these are secondary if not tertiary
effects which are swamped by the mechanical non-linearities of a
loudspeaker. The variation in load impedance with displacement is of such a
low level that, in engineering terms if not in academic terms, can be safely
ignored. Jim, if you have some figures that are at odds with this view, I
would be interested to know them.


S.

--
http://audiopages.googlepages.com