[Patrick Turner]

What the best free software download for
audio analysis at least 20Hz to 20kHz, or wider BW?

I tried Steve Ekblad's site with a lot of stuff but nothing I downloaded
and saved
in my program files would set up and run automatically.

I want to set up a dedicated W98 PC to just provide the displayed
response from
speakers in my test room instead of using my slow manual 1/3 octave
filtering
analog unit.

Patrick Turner.

[Kalman Rubinson]

On Mon, 21 May 2007 00:18:40 GMT, Patrick Turner
wrote:

What the best free software download for
audio analysis at least 20Hz to 20kHz, or wider BW?

I tried Steve Ekblad's site with a lot of stuff but nothing I downloaded
and saved
in my program files would set up and run automatically.

I want to set up a dedicated W98 PC to just provide the displayed
response from
speakers in my test room instead of using my slow manual 1/3 octave
filtering
analog unit.

TrueRTA is free for the low-rez version and higher rez options are
inexpensive. It is a basic RTA.

RoomEQWizard is not a spectral analyzer but it is a superior tool
suite for measuring speaker response in-room. It's free, too.

Kal

[maxhifi]

2nd the vote for TrueRTA - it's nice software

BUT

watch out for noise from your sound card contaminating the results,
especially at low levels, and mic calibration is multiplying the errors of
your mic times the errors of your soundcard.

It's much better to get a dedicated meter if you're doing serious work - for
example the Quest Soundpro is a really nice instrument.


"Kalman Rubinson" wrote in message
...
On Mon, 21 May 2007 00:18:40 GMT, Patrick Turner
wrote:

What the best free software download for
audio analysis at least 20Hz to 20kHz, or wider BW?

I tried Steve Ekblad's site with a lot of stuff but nothing I downloaded
and saved
in my program files would set up and run automatically.

I want to set up a dedicated W98 PC to just provide the displayed
response from
speakers in my test room instead of using my slow manual 1/3 octave
filtering
analog unit.

TrueRTA is free for the low-rez version and higher rez options are
inexpensive. It is a basic RTA.

RoomEQWizard is not a spectral analyzer but it is a superior tool
suite for measuring speaker response in-room. It's free, too.

Kal

[west[_4_]]

"maxhifi" wrote in message
news:PJ94i.200007$DE1.48419@pd7urf2no...
2nd the vote for TrueRTA - it's nice software

BUT

watch out for noise from your sound card contaminating the results,
especially at low levels, and mic calibration is multiplying the errors of
your mic times the errors of your soundcard.

It's much better to get a dedicated meter if you're doing serious work -
for
example the Quest Soundpro is a really nice instrument.


"Kalman Rubinson" wrote in message
...
On Mon, 21 May 2007 00:18:40 GMT, Patrick Turner
wrote:

What the best free software download for
audio analysis at least 20Hz to 20kHz, or wider BW?

I tried Steve Ekblad's site with a lot of stuff but nothing I downloaded
and saved
in my program files would set up and run automatically.

I want to set up a dedicated W98 PC to just provide the displayed
response from
speakers in my test room instead of using my slow manual 1/3 octave
filtering
analog unit.

TrueRTA is free for the low-rez version and higher rez options are
inexpensive. It is a basic RTA.

RoomEQWizard is not a spectral analyzer but it is a superior tool
suite for measuring speaker response in-room. It's free, too.

Kal

Hi Kal: What sound card, without breaking the bank, would you recommend
that will be good for most of the audio programs and tests (good S/N, IMD,
freq response, etc.)? Perhaps good enough to enjoy some CDs while working
and transferring LP to CDs should fit the bill.

west

[Yves Monmagnon]

"Patrick Turner" a écrit dans le message de news:
...
What the best free software download for
audio analysis at least 20Hz to 20kHz, or wider BW?

I tried Steve Ekblad's site with a lot of stuff but nothing I downloaded
and saved
in my program files would set up and run automatically.

I want to set up a dedicated W98 PC to just provide the displayed
response from
speakers in my test room instead of using my slow manual 1/3 octave
filtering
analog unit.

Patrick Turner.

Hi Patrick,

Look at
http://sillanumsoft.altervista.org/index.htm

There is an english page too.

I was pleased using the previus (8.30.21) version and a "TRANSIT USB"
external soundboard from M-AUDIO:
http://www.m-audio.com/index.php?do=...CESTATICPAGE=1

Yves.

[Iain Churches]

"west" wrote in message
news:JCa4i.4904$qp5.1653@trnddc03...

"maxhifi" wrote in message
news:PJ94i.200007$DE1.48419@pd7urf2no...
2nd the vote for TrueRTA - it's nice software

BUT

watch out for noise from your sound card contaminating the results,
especially at low levels, and mic calibration is multiplying the errors
of
your mic times the errors of your soundcard.

It's much better to get a dedicated meter if you're doing serious work -
for
example the Quest Soundpro is a really nice instrument.


"Kalman Rubinson" wrote in message
...
On Mon, 21 May 2007 00:18:40 GMT, Patrick Turner
wrote:

What the best free software download for
audio analysis at least 20Hz to 20kHz, or wider BW?

I tried Steve Ekblad's site with a lot of stuff but nothing I
downloaded
and saved
in my program files would set up and run automatically.

I want to set up a dedicated W98 PC to just provide the displayed
response from
speakers in my test room instead of using my slow manual 1/3 octave
filtering
analog unit.

TrueRTA is free for the low-rez version and higher rez options are
inexpensive. It is a basic RTA.

RoomEQWizard is not a spectral analyzer but it is a superior tool
suite for measuring speaker response in-room. It's free, too.

Kal

Hi Kal: What sound card, without breaking the bank, would you recommend
that will be good for most of the audio programs and tests (good S/N, IMD,
freq response, etc.)? Perhaps good enough to enjoy some CDs while working
and transferring LP to CDs should fit the bill.

Hi West. You need to be able to see and measure what is happening
way beyond 22kHz, particularly if you are concerned with phase
shift and stability issues. That's why so many workshops seem still to
use dedicated instruments.

Cordially
Iain

[Kalman Rubinson]

On Mon, 21 May 2007 06:03:21 GMT, "west" wrote:

Hi Kal: What sound card, without breaking the bank, would you recommend
that will be good for most of the audio programs and tests (good S/N, IMD,
freq response, etc.)? Perhaps good enough to enjoy some CDs while working
and transferring LP to CDs should fit the bill.

I use the M-Audio MobilePre with a calibrated microphone for audio
measurements (in addition to my TEF setup).

Kal

[Patrick Turner]

Iain Churches wrote:

"west" wrote in message
news:JCa4i.4904$qp5.1653@trnddc03...

"maxhifi" wrote in message
news:PJ94i.200007$DE1.48419@pd7urf2no...
2nd the vote for TrueRTA - it's nice software

BUT

watch out for noise from your sound card contaminating the results,
especially at low levels, and mic calibration is multiplying the errors
of
your mic times the errors of your soundcard.

It's much better to get a dedicated meter if you're doing serious work -
for
example the Quest Soundpro is a really nice instrument.


"Kalman Rubinson" wrote in message
...
On Mon, 21 May 2007 00:18:40 GMT, Patrick Turner
wrote:

What the best free software download for
audio analysis at least 20Hz to 20kHz, or wider BW?

I tried Steve Ekblad's site with a lot of stuff but nothing I
downloaded
and saved
in my program files would set up and run automatically.

I want to set up a dedicated W98 PC to just provide the displayed
response from
speakers in my test room instead of using my slow manual 1/3 octave
filtering
analog unit.

TrueRTA is free for the low-rez version and higher rez options are
inexpensive. It is a basic RTA.

RoomEQWizard is not a spectral analyzer but it is a superior tool
suite for measuring speaker response in-room. It's free, too.

Kal

Hi Kal: What sound card, without breaking the bank, would you recommend
that will be good for most of the audio programs and tests (good S/N, IMD,
freq response, etc.)? Perhaps good enough to enjoy some CDs while working
and transferring LP to CDs should fit the bill.

Hi West. You need to be able to see and measure what is happening
way beyond 22kHz, particularly if you are concerned with phase
shift and stability issues. That's why so many workshops seem still to
use dedicated instruments.

Cordially
Iain

Thankyou everybody for the response to my question..

I only want to measure speaker behaviour with the software,
to save precious time.

Presently, I use old fashioned analog gear I built myself.

This gear has a pink noise source, with flat distribution of energy
from 20Hz to 30kHz when filtering with a bandpass filter with same Q =
12 along the band,
and the filter I built is an opamp with bridged RC T filter in the NFB
path and some slight
positive FB to raise the Q, but give the same level and shape of
selectivity acoss the
band.

The band is divided into 3 sub bands 20-200, 200-2k, 2k-20k,
and the filter R elements are switched to give filter F0
at 20, 25, 32, 40, 50, 60, 70, 85, 100, 130, 160, 200 for bass
and x10 for mids, and x 100 for treble, simple.

Si I have 33 rest F for the whole audio band and I find it
just tells me enough.

Microphone is a home made electret with guranteed response from Hi-Q
International in Melbourne.

Mic signal is amplified and sent to the filter, and a peak reading amp
with slow
decay detects the noise signal and converts it to a slow decaying dc
voltage.
This feeds another opamp with logarithmic transfer and this powers a
125mm wide
analog meter which is zeroed by level to the centre position.
Voltage levels from the mic are displayed in db steps, to the meter then
easily
reads +/- 20dB levels each side of centre, ideal for audio measurements
because
many speakers I test after working on them have large unwanted response
variations
with first attempts to get them flat after the first trial of the
theoretical crossover.

It would make things a lot easier if I could just move the mic around
and get say
6 quick responses within a metre of the listening position,
then average them all.
Making 6 response tests with each one taking 15 minutes by using the
switchable
bands and F switch takes a lot of time if I take 6 tests.

Rooms are not perfect, but I have some speakers which measure quite well
in my rather nice large room. My efforts as a builder 30 years ago
left a valuable enduring legacy.
Getting any speaker to give a +/- 2dB response at 4 metres
is doing alright; many high brow brands measure +/- 6dB.

At the moment, I have got the ER Audio ESLIIIB panel ready again for
another test.
Its the fifth time I have stretched a membrane, and this time with over
twice the
0.8Kg recommended membrane tension.

I also have coated the stators with about 0.8mm thick layer of silicone
to prevent the membrane getting too
close to the stators and remaining stuck, as soon as the bass LF signals
swing the membrane
over what seems to be a critical distance.

I will be having more to say soon in another post about the progress
with the ER Audio ESL III3 kitset that
I have been really struggling to assemble.

But to facilitate the measuring process, and maybe even try measuring
amplifiers, I will find a good PC program set up in a spare W98 based PC
to be quite useful.


I will try to dowload and instal what some of you have so kindly
suggested.

Iain mentioned pc software for audio analysis had limited bandwidth, and
I'd agree;
one needs more than 20Hz to 20kHz for amps; more like 1Hz to 1MHz,
so until I do get something more soopery doopery to do what the free
download pc software cannot,
the reliable methods with CRO and meters and some home spun THD/IMD
measuring kit
will continue to play the major role.

Patrick Turner.

[Arny Krueger]

"Patrick Turner" wrote in message


What the best free software download for
audio analysis at least 20Hz to 20kHz, or wider BW?

Audio Rightmark

http://www.rightmark.org/

Your first task is to obtain the residuals of your PC's audio interface with
it.

[Kalman Rubinson]

On Tue, 22 May 2007 00:34:16 +0200, François Yves Le Gal
wrote:

On Mon, 21 May 2007 07:45:06 -0400, Kalman Rubinson wrote:

I use the M-Audio MobilePre with a calibrated microphone for audio
measurements

Yup, the MobilePre is an excellent option if 16 bits/48 Khz are enough. I
prefer 24/96 or even 24/192, even for acoustical measurements limited to 22
KHz or so.

Right. OTOH, my resolution requirements at those frequencies are not
as rigorous as in the lower frequencies. At long as the response is
reasonably flat for mic and pre, it works for me.

Kal

[Rudy]

"Patrick Turner" wrote in message
...


Iain Churches wrote:

"west" wrote in message
news:JCa4i.4904$qp5.1653@trnddc03...

"maxhifi" wrote in message
news:PJ94i.200007$DE1.48419@pd7urf2no...
2nd the vote for TrueRTA - it's nice software

BUT

watch out for noise from your sound card contaminating the results,
especially at low levels, and mic calibration is multiplying the
errors
of
your mic times the errors of your soundcard.

It's much better to get a dedicated meter if you're doing serious
work -
for
example the Quest Soundpro is a really nice instrument.


"Kalman Rubinson" wrote in message
...
On Mon, 21 May 2007 00:18:40 GMT, Patrick Turner
wrote:

What the best free software download for
audio analysis at least 20Hz to 20kHz, or wider BW?

I tried Steve Ekblad's site with a lot of stuff but nothing I
downloaded
and saved
in my program files would set up and run automatically.

I want to set up a dedicated W98 PC to just provide the displayed
response from
speakers in my test room instead of using my slow manual 1/3 octave
filtering
analog unit.

TrueRTA is free for the low-rez version and higher rez options are
inexpensive. It is a basic RTA.

RoomEQWizard is not a spectral analyzer but it is a superior tool
suite for measuring speaker response in-room. It's free, too.

Kal

Hi Kal: What sound card, without breaking the bank, would you
recommend
that will be good for most of the audio programs and tests (good S/N,
IMD,
freq response, etc.)? Perhaps good enough to enjoy some CDs while
working
and transferring LP to CDs should fit the bill.

Hi West. You need to be able to see and measure what is happening
way beyond 22kHz, particularly if you are concerned with phase
shift and stability issues. That's why so many workshops seem still to
use dedicated instruments.

Cordially
Iain

Thankyou everybody for the response to my question..

I only want to measure speaker behaviour with the software,
to save precious time.

Presently, I use old fashioned analog gear I built myself.

This gear has a pink noise source, with flat distribution of energy
from 20Hz to 30kHz when filtering with a bandpass filter with same Q =
12 along the band,
and the filter I built is an opamp with bridged RC T filter in the NFB
path and some slight
positive FB to raise the Q, but give the same level and shape of
selectivity acoss the
band.

The band is divided into 3 sub bands 20-200, 200-2k, 2k-20k,
and the filter R elements are switched to give filter F0
at 20, 25, 32, 40, 50, 60, 70, 85, 100, 130, 160, 200 for bass
and x10 for mids, and x 100 for treble, simple.

Si I have 33 rest F for the whole audio band and I find it
just tells me enough.

Microphone is a home made electret with guranteed response from Hi-Q
International in Melbourne.

Mic signal is amplified and sent to the filter, and a peak reading amp
with slow
decay detects the noise signal and converts it to a slow decaying dc
voltage.
This feeds another opamp with logarithmic transfer and this powers a
125mm wide
analog meter which is zeroed by level to the centre position.
Voltage levels from the mic are displayed in db steps, to the meter then
easily
reads +/- 20dB levels each side of centre, ideal for audio measurements
because
many speakers I test after working on them have large unwanted response
variations
with first attempts to get them flat after the first trial of the
theoretical crossover.

It would make things a lot easier if I could just move the mic around
and get say
6 quick responses within a metre of the listening position,
then average them all.
Making 6 response tests with each one taking 15 minutes by using the
switchable
bands and F switch takes a lot of time if I take 6 tests.

Rooms are not perfect, but I have some speakers which measure quite well
in my rather nice large room. My efforts as a builder 30 years ago
left a valuable enduring legacy.
Getting any speaker to give a +/- 2dB response at 4 metres
is doing alright; many high brow brands measure +/- 6dB.

At the moment, I have got the ER Audio ESLIIIB panel ready again for
another test.
Its the fifth time I have stretched a membrane, and this time with over
twice the
0.8Kg recommended membrane tension.

I also have coated the stators with about 0.8mm thick layer of silicone
to prevent the membrane getting too
close to the stators and remaining stuck, as soon as the bass LF signals
swing the membrane
over what seems to be a critical distance.

I will be having more to say soon in another post about the progress
with the ER Audio ESL III3 kitset that
I have been really struggling to assemble.

But to facilitate the measuring process, and maybe even try measuring
amplifiers, I will find a good PC program set up in a spare W98 based PC
to be quite useful.


I will try to dowload and instal what some of you have so kindly
suggested.

Iain mentioned pc software for audio analysis had limited bandwidth, and
I'd agree;
one needs more than 20Hz to 20kHz for amps; more like 1Hz to 1MHz,
so until I do get something more soopery doopery to do what the free
download pc software cannot,
the reliable methods with CRO and meters and some home spun THD/IMD
measuring kit
will continue to play the major role.

Patrick Turner.


Have a look he
http://www.fesb.hr/~mateljan/arta/download.htm

Dr. Ivo Mateljan, of Kroatia-Split University developed a set of programs
for his students to use - originally. Shareware version, no file save/load
available,
personal license 80 euro's, commercial license 150 euro's.
That is extremely affordable, as it can do all MLSSA can do and then some
:-)

Basically, you use a 96/24 fullduplex soundcard, an outboard 5W amp,
a calibrated mic and, if necessary, a pre-amp.
see at http://www.fesb.hr/~mateljan/arta/support.htm
AP4_FreeField-Rev03eng.pdf over there will tell you how it works.

Rudy

[Patrick Turner]

Have a look he
http://www.fesb.hr/~mateljan/arta/download.htm

Dr. Ivo Mateljan, of Kroatia-Split University developed a set of programs
for his students to use - originally. Shareware version, no file save/load
available,
personal license 80 euro's, commercial license 150 euro's.
That is extremely affordable, as it can do all MLSSA can do and then some
:-)

Basically, you use a 96/24 fullduplex soundcard, an outboard 5W amp,
a calibrated mic and, if necessary, a pre-amp.
see at http://www.fesb.hr/~mateljan/arta/support.htm
AP4_FreeField-Rev03eng.pdf over there will tell you how it works.

Rudy

Thanks Rudy, I already have 3 things that look OK but will get what you
have also.

Patrick Turner.

[Rudy]

"Patrick Turner" wrote in message
...



Have a look he
http://www.fesb.hr/~mateljan/arta/download.htm

Dr. Ivo Mateljan, of Kroatia-Split University developed a set of
programs
for his students to use - originally. Shareware version, no file
save/load
available,
personal license 80 euro's, commercial license 150 euro's.
That is extremely affordable, as it can do all MLSSA can do and then
some
:-)

Basically, you use a 96/24 fullduplex soundcard, an outboard 5W amp,
a calibrated mic and, if necessary, a pre-amp.
see at http://www.fesb.hr/~mateljan/arta/support.htm
AP4_FreeField-Rev03eng.pdf over there will tell you how it works.

Rudy

Thanks Rudy, I already have 3 things that look OK but will get what you
have also.

Patrick Turner.


.......
For those who are too lazy to read the articles on that site, but are
nevertheless interested in getting accurate lsp measurements done,
i'll describe the method used in some more dtail:

to get a full 20Hz-20kHz response measurement at 1m or even further
from a speaker is impossible (unless you have an enormous anechoic room
available, the early room reflections will create a combfiltering effect,
easily creating +/- 5 dB deviations through summing/cancellation at
various frequencies in a typical room@home.
to get the halfroom or 2pi space angle response, a combination of
measurements is done:

1) farfield, at 1m from the speaker
by placing the speaker at some elevation in order to maximize the
path length for early reflections to reach the microphone from the
floor (a high ceiling helps) and ceiling, and obviously well away from any
walls, there will be a short time interval after setting the speaker into
action where only direct sound will be captured.
the software generates a short pulse, from the response measurement
the onset of early reflections can be easily seen, the measurements
will then be set to measure only within such a window.
say this is 7 ms wide, this means that measurements below some
150 Hz will be off as there wasn't time for a completed sine.

2) nearfield to establish the lowend response qualitatively.
by measuring very close to the woofer, the early reflections
and following room reverb tail don't matter as their level
is far below the direct SPL at the mic.
speakers with multiple woofers and/or having a TL or
bassreflex port require measuring of each of these sources.
now by summing and correcting for enclosure, you'll have
a low end response. matching it to the farfield measurements
somewhat above the suspect range, in this case 150 Hz,
will create the full response. (the summation and correction
algorithms used work very well at low f).
....
a setup with a laptop is quite nice, the M-Audio USB 24/96
costs some 80 USD and measured over 110 dB s/n+thd
(0.0069 %) , fr 20-20K flat within 0.1 dB with Arta,
quite ok ;-). you can use it with an old PC running W98 SE
too.
getting a calibrated mic is not as expensive as you'd think.
(ok, Patrick has one already)
the Panasonic MCE2000 capsule, apparently used in the
Monacor PM60, costs some 25 euro's.

by building a small pressure chamber it can be calibrated for
the low range, to get a calibrating curve for the rest of the
spectrum, take a good tweeter with a measured and published
response and measure it - apparently, manufacturing tolerances
and variations are quite small and this will get you to within 1
dB. this curve is then used to compensate measurements
you'll make.

Arta uses both input channels during measurement, the attenuated
amp output and the mic output and can thus give the real loudspeaker
distortion curves.

Rudy

[Patrick Turner]

Rudy wrote:

"Patrick Turner" wrote in message
...



Have a look he
http://www.fesb.hr/~mateljan/arta/download.htm

Dr. Ivo Mateljan, of Kroatia-Split University developed a set of
programs
for his students to use - originally. Shareware version, no file
save/load
available,
personal license 80 euro's, commercial license 150 euro's.
That is extremely affordable, as it can do all MLSSA can do and then
some
:-)

Basically, you use a 96/24 fullduplex soundcard, an outboard 5W amp,
a calibrated mic and, if necessary, a pre-amp.
see at http://www.fesb.hr/~mateljan/arta/support.htm
AP4_FreeField-Rev03eng.pdf over there will tell you how it works.

Rudy

Thanks Rudy, I already have 3 things that look OK but will get what you
have also.

Patrick Turner.

......
For those who are too lazy to read the articles on that site, but are
nevertheless interested in getting accurate lsp measurements done,
i'll describe the method used in some more dtail:

to get a full 20Hz-20kHz response measurement at 1m or even further
from a speaker is impossible (unless you have an enormous anechoic room
available, the early room reflections will create a combfiltering effect,
easily creating +/- 5 dB deviations through summing/cancellation at
various frequencies in a typical room@home.
to get the halfroom or 2pi space angle response, a combination of
measurements is done:

1) farfield, at 1m from the speaker
by placing the speaker at some elevation in order to maximize the
path length for early reflections to reach the microphone from the
floor (a high ceiling helps) and ceiling, and obviously well away from any
walls, there will be a short time interval after setting the speaker into
action where only direct sound will be captured.
the software generates a short pulse, from the response measurement
the onset of early reflections can be easily seen, the measurements
will then be set to measure only within such a window.
say this is 7 ms wide, this means that measurements below some
150 Hz will be off as there wasn't time for a completed sine.

2) nearfield to establish the lowend response qualitatively.
by measuring very close to the woofer, the early reflections
and following room reverb tail don't matter as their level
is far below the direct SPL at the mic.
speakers with multiple woofers and/or having a TL or
bassreflex port require measuring of each of these sources.
now by summing and correcting for enclosure, you'll have
a low end response. matching it to the farfield measurements
somewhat above the suspect range, in this case 150 Hz,
will create the full response. (the summation and correction
algorithms used work very well at low f).
...
a setup with a laptop is quite nice, the M-Audio USB 24/96
costs some 80 USD and measured over 110 dB s/n+thd
(0.0069 %) , fr 20-20K flat within 0.1 dB with Arta,
quite ok ;-). you can use it with an old PC running W98 SE
too.
getting a calibrated mic is not as expensive as you'd think.
(ok, Patrick has one already)
the Panasonic MCE2000 capsule, apparently used in the
Monacor PM60, costs some 25 euro's.

by building a small pressure chamber it can be calibrated for
the low range, to get a calibrating curve for the rest of the
spectrum, take a good tweeter with a measured and published
response and measure it - apparently, manufacturing tolerances
and variations are quite small and this will get you to within 1
dB. this curve is then used to compensate measurements
you'll make.

Arta uses both input channels during measurement, the attenuated
amp output and the mic output and can thus give the real loudspeaker
distortion curves.

Rudy

I'd have to agree with most of what he says, and wonder why he's left
out a few things.

With continuous sine wave signals fed to an amp, the measured response
can be more than
+/- 5dB in a room. I tried all that and got meaningless response
measurements with up to
+/-12dB peaks and nulls, and a Mr Moir published similar responses for
Wireless World back in the
1970s to illustrate the pointlessness of room response measuring using
sine waves.

So ppl started varying the frequency of the sine waves chosen in such a
way to cause the
reflected wave tp maybe cancel of maybe add to the direct path sound,
and the
worst of peaks and nulls was avoided. The signal was produced by what
was called a wobbulator.
you could wobble 1kHz +/- 100Hz about 20 times a second, and
this made refelections easier to live with.

Someone got the idea also of using pink noise; RDH4 referes to this
signal being to only
one worth using in rooms to get some idea of acoustic responses.

Pink noise is very like the sound of Niagra Falls, and pink noise signal
has randomly changing amplitude, phase, and frequency
of all frequencies.
So if you have a narrow bandpass filter, one can filter out
F between say 975 and 1025 and such a filter has a Q = 20, and there is
more than 20dB rejection
of 500Hz or less, and 2kHz and more.
Displayed on the CRO, we see a jittery sine wave amplitude but the F,
phase,
and amplitude are all randomly varying.
The ablility of refelections to perfectly add or null direct waves is
very much reduced.
I use a filter with Q = 12.
The filter is switched, and its not critical which F one chooses, its
not important
for the selected F to be exact increments along the band.

So if you take six response measurements randomly chosen with a metre of
where one want to sit down
to Motzart or the Stones, and you average the dB at 33 points of
frequency along the band,
you do get a remarkable indication of what the speaker is doing, all
without
spending a fortune on hiring an anechoic chamber.

I've also thought of filtering the pink noise source first, the applying
the filtered
signal with its noise perturbations to the speaker via the amp,
rather than applying pink noise, then filtering the sound of the pink
noise.
And presumably the response for the centre frequency would be meterable.

But there are other pulse methods using PC software which I don't
understand very well.


I find a strong corelation between the enjoyability of good music
when i can cajole the speaker and its crossover filters to produce a
response
+/- 2dB for any of say 4 positions around near where I sit.

Very few speakers ever achieve this, even the ones which have had the
benefit
of a factory spending the time and money to perform the tests
several times during development.
I doubt that any of the budget speakers in the shops are made using
a-chambers; mostly they deliberately tailor the sound for boosted bass
and strident treble;
ie, a loudness contour is fitted whether you like it or not.
Trouble is the contour remains at high levels.
An amp with loudness facility doesn't produce a loudness contour above
about
25% rotation of the volume control; loudness is for low level use.
I used to like it, but not now.

Where I find the pink noise is magical is with crossover design and
phase shift
effects.
You can take a response at 4M, or 1M, and the transition from bass to
midrange and midrange
to treble is very well portayed, and nulls or peaks due
to phase cohherence is very easily checked.
The changes one makes to a crossover as a result are valid when you move
the
speaker to a new room position, and the mic,
so the results do not tend to contradict each other.

What I do find with pink noise is that if you test speaker at 1M, then
at 4M, in a good room,
then you find the treble intensity reduces more than bass intensity, so
rooms tend to
hold their bass with distance away from the speaker, but treble falls
more.

This means one has to choose just where you measure your speaker.

I never listen at 1M, so I set the response to be right for where I
think the speaker
will be used, and 90% of ppl have their speakers at 3M to 5M and they
sit,
so the response at 4M and 0.8M above floor level is the one I like to
get right.

If ppl want it different, they always can press their loudness
button of play with tone controls but most usually they never need to.
I have re-engineered lots of speakers to this recipe, and
found people like the sound, often saying its better than when they
bought them, and they don't need to do any eq.


Having said that I read the manual pdf for the www.fesb.hr site,
and there is quite a bit about noise, and much more than I could absorb
fast, or talk about here.

There is a point where one cannot engineer out all of what one doesn't
like
about a given speaker.
Dynamic drivers cannot be changed much; they just do what they do,
and have whatever response they were born with so to speak,
and the process of getting good music with acceptable response
is to optimise what can be wrung out of well chosen drivers of
sufficiently
good quality.
I once used asian made el-cheapo drivers, and they were hard to get
right,
but when I switched to nth Euro drivers from SEAS, the process became
easier,
and the sound was a lot better.

The ERA ESL kit speakers I have been building are expected to give as
flat a response
as my dynamics, and this week i managed to get them to do it and with
very simple networking. They are insensitive, but that's another issue,
and a more voltage capable amp will be used tomorrow to
see how they go.

How they sound with music is the ultimate test once I have the basic
numbers right.

Patrick Turner.

[Yves Monmagnon]

"Patrick Turner" a écrit dans le message de news:
...
[snip]

Someone got the idea also of using pink noise; RDH4 referes to this
signal being to only
one worth using in rooms to get some idea of acoustic responses.

Pink noise is very like the sound of Niagra Falls, and pink noise signal
has randomly changing amplitude, phase, and frequency
of all frequencies.
So if you have a narrow bandpass filter, one can filter out
F between say 975 and 1025 and such a filter has a Q = 20, and there is
more than 20dB rejection
of 500Hz or less, and 2kHz and more.
Displayed on the CRO, we see a jittery sine wave amplitude but the F,
phase,
and amplitude are all randomly varying.
The ablility of refelections to perfectly add or null direct waves is
very much reduced.
I use a filter with Q = 12.
The filter is switched, and its not critical which F one chooses, its
not important
for the selected F to be exact increments along the band.

Yeap, pink or white noise, visualized on a spectrum analyser with some
averaging gives directly the frequency response of anything.
Just be aware that average level is much lower than with pure sine (some 40
dB below) and keep care to do not overload the device under test.

For speakers measurement, I've no anechoïd room at hand, but I've a large
garden ! ! !
Put the LSP on a pair of stand, 1 meter above a grassy ground and aiming to
the sky.
Then suspend the meusuring mike above ;)

Yves.

[Patrick Turner]

Yves Monmagnon wrote:

"Patrick Turner" a écrit dans le message de news:
...
[snip]

Someone got the idea also of using pink noise; RDH4 referes to this
signal being to only
one worth using in rooms to get some idea of acoustic responses.

Pink noise is very like the sound of Niagra Falls, and pink noise signal
has randomly changing amplitude, phase, and frequency
of all frequencies.
So if you have a narrow bandpass filter, one can filter out
F between say 975 and 1025 and such a filter has a Q = 20, and there is
more than 20dB rejection
of 500Hz or less, and 2kHz and more.
Displayed on the CRO, we see a jittery sine wave amplitude but the F,
phase,
and amplitude are all randomly varying.
The ablility of refelections to perfectly add or null direct waves is
very much reduced.
I use a filter with Q = 12.
The filter is switched, and its not critical which F one chooses, its
not important
for the selected F to be exact increments along the band.

Yeap, pink or white noise, visualized on a spectrum analyser with some
averaging gives directly the frequency response of anything.
Just be aware that average level is much lower than with pure sine (some 40
dB below) and keep care to do not overload the device under test.

Pink noise appears to me to very similar to a recording made
of 3 heavy metal bands all blasting away together out of key.

The dynamic range is low due to compression, and average level appears
to
be much higher than -40dB below V maximum sine waves.

Its easy to calibrate a CRO with a sine wave to make an amp clip
so that the clipping just fills the screen.
Then if the pink noise is turned up so that peaks in it just begin to
clip
occasionally or the rock and roll / heavy metal band music
does the same thing, then the load power will average maybe 1/4 of the
max sine wave power,
and hence the pink noise appears quite OK to use.

A class AB tube amp can become seriously overloaded
or mis-biased at clipping with sine waves due to cathode voltage
movement
or B+rail sag.
With rock music or pink noise without much clipping, the same amp does
suffer much dc bias changes.

However, with classical music the dynamic range varies more widely.


For speakers measurement, I've no anechoïd room at hand, but I've a large
garden ! ! !
Put the LSP on a pair of stand, 1 meter above a grassy ground and aiming to
the sky.
Then suspend the meusuring mike above ;)


I could do the same thing in my yard, but I'd rather
find out what the speaker is like a room, and when compared to my
reference
speakers.

We don't listen to speakers while standing on a step ladder looking down
at speakers in a yard,
and nor do we listen in an anechoic chamber, and my customers all have
rooms which will give a similar response I measure in my room.
After trying out this simple rationale, I know it works well,
and sure I could get a lot more technical about it all if I had time for
it
But after measuring and listening to the efforts of those speakers which
may have
had the benefits of greater technical knowhow, judging by the sales
talk,
I know I must be doing OK.

Patrick Turner.




Yves.

[Yves Monmagnon]

"Patrick Turner" a écrit dans le message de news:
...


Yves Monmagnon wrote:

"Patrick Turner" a écrit dans le message de
news:
...
[snip]

Someone got the idea also of using pink noise; RDH4 referes to this
signal being to only
one worth using in rooms to get some idea of acoustic responses.

Pink noise is very like the sound of Niagra Falls, and pink noise
signal
has randomly changing amplitude, phase, and frequency
of all frequencies.
So if you have a narrow bandpass filter, one can filter out
F between say 975 and 1025 and such a filter has a Q = 20, and there is
more than 20dB rejection
of 500Hz or less, and 2kHz and more.
Displayed on the CRO, we see a jittery sine wave amplitude but the F,
phase,
and amplitude are all randomly varying.
The ablility of refelections to perfectly add or null direct waves is
very much reduced.
I use a filter with Q = 12.
The filter is switched, and its not critical which F one chooses, its
not important
for the selected F to be exact increments along the band.

Yeap, pink or white noise, visualized on a spectrum analyser with some
averaging gives directly the frequency response of anything.
Just be aware that average level is much lower than with pure sine (some
40
dB below) and keep care to do not overload the device under test.

Pink noise appears to me to very similar to a recording made
of 3 heavy metal bands all blasting away together out of key.

The dynamic range is low due to compression, and average level appears
to
be much higher than -40dB below V maximum sine waves.

Its easy to calibrate a CRO with a sine wave to make an amp clip
so that the clipping just fills the screen.
Then if the pink noise is turned up so that peaks in it just begin to
clip
occasionally or the rock and roll / heavy metal band music
does the same thing, then the load power will average maybe 1/4 of the
max sine wave power,
and hence the pink noise appears quite OK to use.

A class AB tube amp can become seriously overloaded
or mis-biased at clipping with sine waves due to cathode voltage
movement
or B+rail sag.
With rock music or pink noise without much clipping, the same amp does
suffer much dc bias changes.

However, with classical music the dynamic range varies more widely.

True !
With such software tools you may use "burst noise".
That is feeding the device under test with e..g. 100ms on and 100ms off
noise (I do prefer white when testing amps).
It's not so difficult to syncronise the CRO to show how a sudent change of
level affects bias and more generally the ability of the PSU to cope with
this "simulated transients".



For speakers measurement, I've no anechoïd room at hand, but I've a large
garden ! ! !
Put the LSP on a pair of stand, 1 meter above a grassy ground and aiming
to
the sky.
Then suspend the meusuring mike above ;)


I could do the same thing in my yard, but I'd rather
find out what the speaker is like a room, and when compared to my
reference
speakers.

We don't listen to speakers while standing on a step ladder looking down
at speakers in a yard,

Nor me !
But it's like measuring an amplifier loaded with a pure ohmic load.
Just an help to trace "gross" faults !

and nor do we listen in an anechoic chamber, and my customers all have
rooms which will give a similar response I measure in my room.
After trying out this simple rationale, I know it works well,
and sure I could get a lot more technical about it all if I had time for
it
But after measuring and listening to the efforts of those speakers which
may have
had the benefits of greater technical knowhow, judging by the sales
talk,
I know I must be doing OK.

Patrick Turner.




Yves.

[Patrick Turner]

Yves Monmagnon wrote:

"Patrick Turner" a écrit dans le message de news:
...


Yves Monmagnon wrote:

"Patrick Turner" a écrit dans le message de
news:
...
[snip]

Someone got the idea also of using pink noise; RDH4 referes to this
signal being to only
one worth using in rooms to get some idea of acoustic responses.

Pink noise is very like the sound of Niagra Falls, and pink noise
signal
has randomly changing amplitude, phase, and frequency
of all frequencies.
So if you have a narrow bandpass filter, one can filter out
F between say 975 and 1025 and such a filter has a Q = 20, and there is
more than 20dB rejection
of 500Hz or less, and 2kHz and more.
Displayed on the CRO, we see a jittery sine wave amplitude but the F,
phase,
and amplitude are all randomly varying.
The ablility of refelections to perfectly add or null direct waves is
very much reduced.
I use a filter with Q = 12.
The filter is switched, and its not critical which F one chooses, its
not important
for the selected F to be exact increments along the band.

Yeap, pink or white noise, visualized on a spectrum analyser with some
averaging gives directly the frequency response of anything.
Just be aware that average level is much lower than with pure sine (some
40
dB below) and keep care to do not overload the device under test.

Pink noise appears to me to very similar to a recording made
of 3 heavy metal bands all blasting away together out of key.

The dynamic range is low due to compression, and average level appears
to
be much higher than -40dB below V maximum sine waves.

Its easy to calibrate a CRO with a sine wave to make an amp clip
so that the clipping just fills the screen.
Then if the pink noise is turned up so that peaks in it just begin to
clip
occasionally or the rock and roll / heavy metal band music
does the same thing, then the load power will average maybe 1/4 of the
max sine wave power,
and hence the pink noise appears quite OK to use.

A class AB tube amp can become seriously overloaded
or mis-biased at clipping with sine waves due to cathode voltage
movement
or B+rail sag.
With rock music or pink noise without much clipping, the same amp does
suffer much dc bias changes.

However, with classical music the dynamic range varies more widely.

True !
With such software tools you may use "burst noise".
That is feeding the device under test with e..g. 100ms on and 100ms off
noise (I do prefer white when testing amps).
It's not so difficult to syncronise the CRO to show how a sudent change of
level affects bias and more generally the ability of the PSU to cope with
this "simulated transients".

Most well designed tube amps initially deal with a burst of frequencies
for a short time such a large drum beat in the same way a fixed bias
amps handle the burst;
ie, without hardly any bias change anywhere to rail voltage or grid
voltage
or charge in coupling caps providimg there is no clipping.

Many drum sounds ate seriously compressed or limited by the recording
gear,
because the unclipped dynamic range won't "fit in to dynamic range
ceilings".
so the beats have some longer duration at clipping levels, but even so
if the duty cycle for the burst of noise energy that is a drum beat
is furiously repeated by some rock and roll git
pounding away often as possible, the noise can't have anything like 100%
of duty cycle.

Old class AB amps with low capacitiances bypassing power rails and
cathode resistors change their biasing
more easily than ones with high value C.
However, with high C values, such as 1,000 uF across cathode R there is
the problem
with rising Ek with transients and sustained high lebels and once the
bias has risen,
and the amp is effectively over biased, the distortion rises hugely,
and when the burst is finished the bias takes a long tome to settle.
But for hi-fi the cathode bias is OK because most of the time the amp
is in class A and there is little bias change.
SE amps are all class A, so put up with transients just fine.

I know I have drifet off topic to biasing from spectral analysis.
But to bias without problems of rising Ek and mis bias
one can use my idea of dynamic biasing of PP output stages
which is fully explained at
http://www.turneraudio.com.au/schem-...tabilizer.html

Where such dynamic biasing is used, there is no substantial rise in EK
due to
non linear tube currents during class AB operation of cathode biased
amplifiers.
The dynamic biasing allows a PP output tube to be biased with a low
current,
but with the regulation offered by the RC cathode bias network.
Excess cathode current increase beyond an amount equal to the
idle current is shunted shunted through a bjt rather than allowed to
charge up a cathode cap
and the output stage acts exactly like a a fixed bias output stage but
without the
drift of bias, or the need to keep bias current high for class A,
and without a dozen adjust pots when you have 12 output tubes as in my
300 watt amps.
So the greenhouse effects caused by such amps is reduced considerably.


For speakers measurement, I've no anechoïd room at hand, but I've a large
garden ! ! !
Put the LSP on a pair of stand, 1 meter above a grassy ground and aiming
to
the sky.
Then suspend the meusuring mike above ;)


I could do the same thing in my yard, but I'd rather
find out what the speaker is like a room, and when compared to my
reference
speakers.

We don't listen to speakers while standing on a step ladder looking down
at speakers in a yard,

Nor me !
But it's like measuring an amplifier loaded with a pure ohmic load.
Just an help to trace "gross" faults !

Well indeed; the groos faults can easily be dealt with, the micro
faults are the difficult things to change...

Patrick Turner.

and nor do we listen in an anechoic chamber, and my customers all have
rooms which will give a similar response I measure in my room.
After trying out this simple rationale, I know it works well,
and sure I could get a lot more technical about it all if I had time for
it
But after measuring and listening to the efforts of those speakers which
may have
had the benefits of greater technical knowhow, judging by the sales
talk,
I know I must be doing OK.

Patrick Turner.




Yves.