Dunlavy SC-IV/A HELP
forp
familiar with the rules of the group. I assume this is a suitable
place to ask for help.
I just bought a pair of Dunlavy SC-IV/A cabinets that are empty (no
drivers) in hopes I could emulate the production line for much less $$
than they would cost new, or at least have some good sounding
speakers.
Dunlavy is out business and I just got these cabs on Saturday so I
have not had a business day to try to contact anyone from the company,
but I fear that I will not be able to get a hold of any one.
What it is:
The configuration for each speaker is 2-10"s 2-5"s and one tweet. I
want to try to match the drivers and crossovers in my system to what
came out of the factory. There are two chambers in each cabinet. The
2-10"s share 5.5 cu Ft, and the 2-5"s share .98 cu Ft. All chambers
are sealed.
What I know:
I know that the "A" series was an upgrade to the SC-IV line that
included a change from Vifa to Scanspeak, and some cabinet and
crossover changes as well. The crossover uses a 6db/oct slope that is
matched to the drivers.
What I found out:
Doing some research there is only one 10" woofer made by Scanspeak the
25W/8565-01 that seems to work in a 2-2.5 cu enclosure depending on
where you look (2.5 using AJ Sealed Designer and 1.45cu looking at a
chart in the Parts Express catalog)or what software you use to
calculate that spec. Other choices are the Hi-Vi D10G (which I really
like)and the Morel MW-1075.
As for the 5"s I have noticed that the Scanspeak 13M/8636 and the
13M/8640 seem to be ok with a .44 cu(Parts express) enclosure.
There are two more the 15W/8530K-00 and the 15S8530K-01 (.32cu)
As for the tweet they (scan speak) make a bunch to choose from.
Any help that you could provide would be greatly appreciated.
Thank you for listening.
-Forp
Rhgoo7
BEAR
> John Dunlavy can be contacted at John...@aol.com or 719-593-0313. Good luck.
>
> Richard
>
> >Hi, This is the first time I have posted in this group, so I am not
> >familiar with the rules of the group. I assume this is a suitable
> >place to ask for help.
> >
> >I just bought a pair of Dunlavy SC-IV/A cabinets that are empty (no
> >drivers) in hopes I could emulate the production line for much less $$
> >than they would cost new, or at least have some good sounding
> >speakers.
> >
I composed a response to this query but my wonderfully reliable Win98 crashbox
decided to crash exactly when I was sending it out. Nice.
Anyhow, The place to start is with some understanding of both basic loudspeaker
theory and Dunlavy's design concepts.
For the former, Vance Dickason's Loudspeaker Cookbook will get you on the map,
along with some websites that you can search for on speaker design.
For the latter, search the patent database and see if JD patented any of his
speaker
designs - I'm not sure if he did.
The next step is to acquire some computer based measurement equipment so that
you can measure your results. There is freeware that will do the job, as well as
for
sale software/hardware combos. You'll need to be able to measure the results.
Also, for the bass and mid sections you'll want some decent simulation software
to help determine which drivers have a shot at working in the available volume
(of course you can always *reduce* the volume) given the cabinet.
The Dunlavy design was able to reproduce a credible square wave over a fairly wide
frequency range on axis. This was one of its main features.
If you want to replicate a Dunlavy, you'll need to be able to acheive this same
response. Suddenly, it's not such a simple design.
While the *acoustic* response of the slopes used for crossing over the drivers
was first order (6dB/oct), they were almost certainly not merely simple first
order electrical filters.
What you can do, of course, is to just find some high quality drivers that
mechanically
fit into the same holes, and also perform appropriately with the volumes they
"see" - and have good frequency response - and just build up a crossover of your
choice that gives you a good measured response. It won't be a "Dunlavy" but
it could be a pretty good speaker if you are careful.
Regarding Dunlavy's response, I'd be surprised if he would be interested in
helping
someone clone his work, especially given the way the demise of his company
came about.
For specific design questions, there are some people who participate here who
may be able to field them, and there are some forums and listerves out on the
web that deal with speaker design as well.
Regards,
_-_-bear
--
_-_-bearlabs
http://www.bearlabsUSA.com
- Silver Lightning Interconnects & Custom Mods -
Johnd1001
that possesses the appropriate infrastructure, etc., required to design and
manufacture "truly accuate" loudspeakers.
So - stay tuned for the re-activation of DAL at a location that will permit the
design and manufacture of even more accurate loudspeakers than was possible in
Colorado Springs - with limited investment capital and a dearth of available
technical and engineering personnel at affordable salaries.
Meanwhile, if any owners require replacement parts (under waranty, etc.), feel
free to contact me at: www.jo...@aol.com.
Best of listening,
John Dunlavy
forp
Hi, I wanted to know if you had any information regarding my original
post...
Hi, This is the first time I have posted in this group, so I am not
>familiar with the rules of the group. I assume this is a suitable
>place to ask for help.
>
>I just bought a pair of Dunlavy SC-IV/A cabinets that are empty (no
>drivers) in hopes I could emulate the production line for much less $$
>than they would cost new, or at least have some good sounding
>speakers.
>
Johnd1001
surfaces completly covered with large irregular shaped pieces of high-density,
open-cell polyurethane foam, the design of truly accurate loudspeakers is
virtually impossible.
Anyone stating that it can be done within a large, non-anechoic room by merely
using an FFT program like MLSSA running on a fast computer, is not speaking
from a position of relevant experience.
Designing a loudspeaker the size of the SC-IV/A (that exhibits true plus/minus
2 dB response and near perfect impulse and step response), is not an exercise
to be taken lightly.
Sorry for the bad news - but I believe in telling the truth and not misleading
people with regard to the difficulty associated with building loudspeakers that
exhibit the properties of the SC-IV/A.
And merely building an SC-IV/A "clone" would only yield a speaker with a
plus/minus 5 dB (or worse) performance.
All the best - keep in touch.
François Yves Le Gal
>Anyone stating that it can be done within a large, non-anechoic room by merely
>using an FFT program like MLSSA running on a fast computer, is not speaking
>from a position of relevant experience.
Quite a surprising statement, Mr. Dunlavy, particularly considering that
most speaker manufacturers use MLSSA or similar programs in their design
and voicing process.
Would you be so kind as to explain why approaches such as MLSSA are
flawed?
Brock Hannibal
Well, he didn't say that. He specifically said what is quoted above. Pay
particular attention to the words "non-anechoic" and "merely using".
He never claimed that MLSSA is flawed in and of itself, though it most
likely does have many limitations.
hth
--
Brock
"We also know how cruel the truth often is, and we wonder whether
delusion is not more consoling." -- Heinri Poincaro
forp
any technical information what so ever regarding speaker selection
(model #'s) and crossover layouts (component values and schematic
information) or what ever you would feel comfortable giving out. I
understand that this is your work and do not mean to trivialize it in
any way. It's just that I have these cabinets and want to make them
sound as good as I can. I do have some idea of what to do, however
since this design was in production and the specifications exist, and
it would be helpful to know what they are. Of course this raises
issues with intellectual property and the many hours of hard work that
were put into acquiring this information and I understand why you
might not want to tell me exactly what the specifications are in
public. I would be happy to talk with you about this in person (I
think I live close to you) or on the phone tell you a bit about my
self and my intentions for undertaking this project and why it is
important to me. Lastly, thank you for taking the time to respond to
my message.
Regards
On 5 Feb 2003 16:17:02 GMT, john...@aol.com (Johnd1001) wrote:
Mark Wilkinson
news:Izd0a.182484$6G4.18227@sccrnsc02...
Not to try to speak for Mr. Dunlavy, who no doubt has forgot more than I
know .... but my experience is that none of the FFT programs work with a
damn handling low frequency info indoors. If you open up the window long
enough to read low frequencies, you let in reflections. If you try to close
the window to gate out reflections, you don't get enough info for valid
analysis. It's the simple nature of capturing long wavelengths. Mark
François Yves Le Gal
wrote:
>Well, he didn't say that. He specifically said what is quoted above. Pay
>particular attention to the words "non-anechoic" and "merely using".
Oh, I did pay attention to what Mr. Dunlavy wrote. Using a software
approach such as MLSSA allows designers to dispense of anechoic rooms,
and most designers only use MLSSA-like tools in their design/voicing
process (at least for the measurement part).
Let me rephrase, then :
>> >Anyone stating that it can be done within a large, non-anechoic room by merely
>> >using an FFT program like MLSSA running on a fast computer, is not speaking
>> >from a position of relevant experience.
Quite a surprising statement, Mr. Dunlavy, particularly considering that
most speaker manufacturers use large, non-anechoic romms in conjunction
with MLSSA or similar programs in their design and voicing process.
Would you be so kind as to explain why such approaches are flawed?
François Yves Le Gal
wrote:
>but my experience is that none of the FFT programs work with a
>damn handling low frequency info indoors. If you open up the window long
>enough to read low frequencies, you let in reflections. If you try to close
>the window to gate out reflections, you don't get enough info for valid
>analysis.
Sure, if you're using a "midfield" approach. But anyone familiar with
MLSSA and similar tools use a pressure approach for LF measurements,
thus avoiding all room related problems.
Brock Hannibal
> On 6 Feb 2003 01:51:27 GMT, Brock Hannibal <hami...@pacifier.com>
> wrote:
>
> >Well, he didn't say that. He specifically said what is quoted above. Pay
> >particular attention to the words "non-anechoic" and "merely using".
>
> Oh, I did pay attention to what Mr. Dunlavy wrote. Using a software
> approach such as MLSSA allows designers to dispense of anechoic rooms,
No it doesn't, because the time-domain information can be incorrect
because of the reflections. The FFT programs can't help you sort that
out if the edges' reflections happen to arrive at the test mics
coincident with edges of the direct signal.
> and most designers only use MLSSA-like tools in their design/voicing
> process (at least for the measurement part).
Well most designers aren't after time domain fidelity, only frequency
domain.
> Let me rephrase, then :
>
> >> >Anyone stating that it can be done within a large, non-anechoic room =
by merely
> >> >using an FFT program like MLSSA running on a fast computer, is not sp=
eaking
> >> >from a position of relevant experience.
>
> Quite a surprising statement, Mr. Dunlavy, particularly considering that
> most speaker manufacturers use large, non-anechoic romms in conjunction
> with MLSSA or similar programs in their design and voicing process.
>
> Would you be so kind as to explain why such approaches are flawed?
See, above. If pulse fidelity is the goal, which Dunlavy has more than
hinted at, FFT programs won't get you there.
François Yves Le Gal
wrote:
>No it doesn't, because the time-domain information can be incorrect
>because of the reflections.
One word : windowing.
>The FFT programs can't help you sort that
>out if the edges' reflections happen to arrive at the test mics
>coincident with edges of the direct signal.
Oh, have you used MLSSA or similar tools ? And, no, a TDS-210 with FFT
option doesn't qualify here.
:-)
>Well most designers aren't after time domain fidelity, only frequency
>domain.
Well, it's your opinion. There have been numerous time coherent
loudspeakers designed and voiced without large anechoic rooms (or MLS
gear for that matter), beginning with products from Elipson, Charlin,
Kef, B&W and numerous other companies.
And, as nobody listens in an anechoic room, why bother? In room,
real-world responses are significant, anechoic resposnes are only useful
in the design process.
>See, above. If pulse fidelity is the goal, which Dunlavy has more than
>hinted at, FFT programs won't get you there.
If you want to measure square waves, you just need a mike, suitable
electronics and a scope, no MLS or FFT needed here ! OK, this is the
quick'n dirty approach, but it works very nicely, thank you.
Arny Krueger
news:b1uno...@enews1.newsguy.com
> On Thu, 6 Feb 2003, Fran=C3=A7ois Yves Le Gal wrote:
>> On 6 Feb 2003 01:51:27 GMT, Brock Hannibal <hami...@pacifier.com>
>> wrote:
>> Oh, I did pay attention to what Mr. Dunlavy wrote. Using a software
>> approach such as MLSSA allows designers to dispense of anechoic
>> rooms,
> No it doesn't, because the time-domain information can be incorrect
> because of the reflections. The FFT programs can't help you sort that
> out if the edges' reflections happen to arrive at the test mics
> coincident with edges of the direct signal.
For that to happen wouldn't a straight line NOT be the shortest distance
from the speaker to the microphone?
There is a reason why MLSSA does not allow designers of wide-range speakers
to entirely dispense with anechoic rooms. Basically, MLS approaches presume
device linearity and require adequate SNR, neither of which may be true in
ordinary rooms at the lowest audio frequencies.
BEAR
> On Thu, 6 Feb 2003, Fran=C3=A7ois Yves Le Gal wrote:
>
> > On 6 Feb 2003 01:51:27 GMT, Brock Hannibal <hami...@pacifier.com>
> > wrote:
> >
> > >Well, he didn't say that. He specifically said what is quoted above. Pay
> > >particular attention to the words "non-anechoic" and "merely using".
>
<snipp'd>
>
>
> See, above. If pulse fidelity is the goal, which Dunlavy has more than
> hinted at, FFT programs won't get you there.
But, if edge refractions are still present due to your cabinet, then
eliminating them in the test room gives you a *false* view of the
product, eh?
_-_-bear
>
>
> --
> Brock
François Yves Le Gal
wrote:
>There is a reason why MLSSA does not allow designers of wide-range speakers
>to entirely dispense with anechoic rooms.
>Basically, MLS approaches presume
>device linearity and require adequate SNR, neither of which may be true in
>ordinary rooms at the lowest audio frequencies.
Oh, and would you care to explain what you mean by "MLS approaches
presume device linearity and require adequate SNR, neither of which may
be true in ordinary rooms at the lowest audio frequencies"?
Richard D Pierce
Mark Wilkinson <captg...@direcway.com> wrote:
>"François Yves Le Gal" <fle...@aingeal.com> wrote in message
That's all true, but an anaechoic chamber large enought NOT to
have such low frequency artifacts is ALSO large enought to open
up a window long enough NOT to have low frequency reflections.
An anechoic chamber that is truly anechoic to 20 Hz has to have
a volume sufficient to support absorbtive wedges some 12 feet
long in addition to allowing placement of the test setup some 25
feet from thee boundary thus formed. That's a BIG room. Take the
same room, remove the wedges, and you can now do FFT
measurements windows sufficiently long to get NO reflections.
And basic signal theory says they two are capable of the SAME
accuracy in both the time and frequency domains.
--
| Dick Pierce |
| Professional Audio Development |
| 1-781/826-4953 Voice and FAX |
| DPi...@world.std.com |
Brock Hannibal
> "Brock Hannibal" <hami...@pacifier.com> wrote in message
> news:b1uno...@enews1.newsguy.com
>
> > On Thu, 6 Feb 2003, Fran=C3=A7ois Yves Le Gal wrote:
>
> >> On 6 Feb 2003 01:51:27 GMT, Brock Hannibal <hami...@pacifier.com>
> >> wrote:
>
> >> Oh, I did pay attention to what Mr. Dunlavy wrote. Using a software
> >> approach such as MLSSA allows designers to dispense of anechoic
> >> rooms,
>
> > No it doesn't, because the time-domain information can be incorrect
> > because of the reflections. The FFT programs can't help you sort that
> > out if the edges' reflections happen to arrive at the test mics
> > coincident with edges of the direct signal.
>
> For that to happen wouldn't a straight line NOT be the shortest distance
> from the speaker to the microphone?
If everything was perfect only the direct sound would hit the
microphone. Unfortunately reflections can arrive in time at different
multiples of time related to the room dimensions and wavelength of the
harmonics, such as 1/4, 1/2 and so forth. Sorting these out with the
typical windowing in FFT programs istough. Theoretically it can be done,
but I have my doubts.
> There is a reason why MLSSA does not allow designers of wide-range speakers
> to entirely dispense with anechoic rooms. Basically, MLS approaches presume
> device linearity and require adequate SNR, neither of which may be true in
> ordinary rooms at the lowest audio frequencies.
I don't think that's the reason. However, non-anehoic rooms do have a
lot of problems with varying amplitudes of the reflections and resonance
modes.
Brock Hannibal
> Brock Hannibal wrote:
>
> > On Thu, 6 Feb 2003, Fran=C3=A7ois Yves Le Gal wrote:
> >
> > > On 6 Feb 2003 01:51:27 GMT, Brock Hannibal <hami...@pacifier.com>
> > > wrote:
> > >
> > > >Well, he didn't say that. He specifically said what is quoted above. Pay
> > > >particular attention to the words "non-anechoic" and "merely using".
> >
>
> <snipp'd>
>
> >
> >
> > See, above. If pulse fidelity is the goal, which Dunlavy has more than
> > hinted at, FFT programs won't get you there.
>
> But, if edge refractions are still present due to your cabinet, then
> eliminating them in the test room gives you a *false* view of the
> product, eh?
I don't think edge diffraction from your cabinet or the interface
between driver and cabinet is eliminated by an anechoic chamber. Why
would it be?
Johnd1001
Since the delay in arrival-time of energy diffracted from the edges of almost
all loudspeaker enclosures possessing reasonable physical dimensions, it is
likely that the audible effect of such "diffraction" will consist of two
components - those that affect the amplitude/frequency response and those that
affect the impulse/step response (ampl.vs time).
While measurement systems such as MLSSA are valuable and important tools for
designers of accurate loudspeakers - they have fundamental limitations
introduced by the late arrival-time of energy reflected or diffracted by nearby
reflective surfaces.
Designers of loudspeakers wishing to take accurate measurements might want to
look at technology used in the measurement of antennas possessing physical
dimensions that are relatively large with respect to a wavelength within their
intended operating range (especially at relatively low frequencies - where many
of the same problems exist as with loudspeakers.
Having begun my professional engineering career in the field of antennas back
in the mid-1940's, I quickly encountered many of the same problems asociated
with obtaining accurate measurements as are now current in the audio field.
Along with other researchers in the antenna field, I discovered it was possible
to take measurements at three different distances and use simple mathematical
means (3-equations with 3-unknowns) to achieve results with useful accuracy.
Those interested in this work might wish to research archives of the IEEE, IRE,
IREE, the Antenna and Propagation Section of the IEEE, etc. (In this regard,
high-performance antennas are, generally, far more complex with respect to
their design, performance properties and measurement than are audio
loudspeakers.)
John Dunlavy
Johnd1001
Research my "loudspeaker edge-diffraction patent" at the U.S. Patent And
Trademark" site on the Internet at WWW....@USGOV.COM (if my memory is
correct).
Search and downloading is free.
If the above does not work, try a simple search for "U.S. Patent Office".
Let me know if you encounter any difficulty.
John Dunlavy (John...@AOL.Com)
François Yves Le Gal
>Research my "loudspeaker edge-diffraction patent"
There is no such patent, Mr. Dunlavy. The closest would be US patent
4,167,985, granted September 18, 1979 and called "Speaker system"
"Abstract
A speaker system in which at least one driver is mounted on an enclosure
and is adapted to radiate sound waves outwardly from said enclosure in
response to an input signal. A sound absorbing material is disposed on
at least a portion of the outer surface area of said enclosure to reduce
the effect of diffractions and reflections of said sound waves relative
to said enclosure."
>at the U.S. Patent And
>Trademark" site on the Internet at WWW....@USGOV.COM (if my memory is
>correct).
Well, http://www.uspto.gov will work better.
:-)
Richard D Pierce
news:b1uno...@enews1.newsguy.com
> On Thu, 6 Feb 2003, Fran=C3=A7ois Yves Le Gal wrote:
>> On 6 Feb 2003 01:51:27 GMT, Brock Hannibal <hami...@pacifier.com>
>> wrote:
>> Oh, I did pay attention to what Mr. Dunlavy wrote. Using a software
>> approach such as MLSSA allows designers to dispense of anechoic
>> rooms,
>
> No it doesn't, because the time-domain information can be incorrect
> because of the reflections. The FFT programs can't help you sort that
> out if the edges' reflections happen to arrive at the test mics
> coincident with edges of the direct signal.
Well, if you sit down and think about it, you'll quickly come to
the conclusion that this is physically impossible. For this to
occur, the reflected path length must be equal to the direct
path length. Can you suggest a physically realizable situation
where this is possible?
It is almost axiomatic the low frequency limit of measurement is
inversely proportional to the size of the room in which the
measurement is taken, and it makes little difference if the
chamber is anechoic with non-gated measurements or in a room
with gated or windowed measurements. Both set the same limit in
terms of the time between the direct arrival and first
reflection arrival times, which limits the low frequency in the
same way.
And once you realize that it is simply NOT possible for the
first reflection and direct arrival times to be the same, then
any FFT-based measurement, properly conducted, will within that
time limit, give you no more or less accurate results than any
other measurements in the same volume.
BEAR
That would be patent#: 5731555
_-_-bear
Richard D Pierce
Arny Krueger <ar...@hotpop.com> wrote:
>There is a reason why MLSSA does not allow designers of wide-range speakers
>to entirely dispense with anechoic rooms. Basically, MLS approaches presume
>device linearity and require adequate SNR, neither of which may be true in
>ordinary rooms at the lowest audio frequencies.
Well, hold on a second. If might be true that:
"FFT approaches pesume device linearity and require
adequate SNR, ..."
But there is a clear confusion between "MLS" and "FFT" in this
discussion. They are NOT equivalent or interchangeable.
Systems such as MLSSA and CLIO's MLS mode us maximal length
sequences as the stimulus, aloowing them to emply a fast
Hadamard transform to derive the system's impulse response
via cross-correlation.
Now, once that impulse response is in hand, one then applies the
FFT to derive the complex frequency transfer function and, from
that, the magnitude vs frequency response, the phase response,
and so forth.
In that sense, when you talk about "FFT" systems, this INCLUDES
MLS-based (MLSSA, CLIO, etc) and impulse (B&K and others). The
only difference in in the stimulus used and the preprocessing to
get the impulse response.
The BIG difference lies in the fact that using MLS sequences has
the advantage of high noise immunity, low crest factor and a FAR
lower chance of non-linear operation of the transducer.
And if S/N is an issue, averaging is a pretty quick and
effective means of increasing the S/N over non-correlated
noise.
Indeed, using MLS-based systems, I routinely do measurements in
large rooms that are essentially identical to those obtained in
very large, very expensive anechoic chambers. And there's not
the slightest bit of evidence to suggest they are any less
accurate in determining the transient behavior of the system
because, indeed, that is the very nature of the system itself.
I also, once again, question the seeming strong dependence on
square wave response as a figure-of-merit for perforemance. As I
illustrated some months ago, one can construct systems with
essentially perfect square wave performance that sound
completely awful.
Performing good, dependable measurements is NOT a trivial
excercise nor does a single measurement such as an anechoic
response nor a picture of square waves even remotely comprise a
comprehensive or even telling suite of measurements. Measuring
loudspeakers is substantially more difficult than other
components in the reproduction chain, even given the fact that
their measurable anomolies are orders of magnitude worse. Most
people measuring speakers get it wrong.
François Yves Le Gal
wrote:
>Indeed, using MLS-based systems, I routinely do measurements in
>large rooms that are essentially identical to those obtained in
>very large, very expensive anechoic chambers. And there's not
>the slightest bit of evidence to suggest they are any less
>accurate in determining the transient behavior of the system
>because, indeed, that is the very nature of the system itself.
Amen.
And I suppose that you do pressure measurements below 150 Hz or so, in
order to nearly totally take the room out of the picture.
>I also, once again, question the seeming strong dependence on
>square wave response as a figure-of-merit for perforemance. As I
>illustrated some months ago, one can construct systems with
>essentially perfect square wave performance that sound
>completely awful.
Yes, but some really nice sounding systems do also happen to be time
coherent and are able to reproduce decent facsimiles of square waves.
François Yves Le Gal
>That would be patent#: 5731555
Ahem, no. Patent 5,731,555 has been granted to Sony, not to Mr. Dunlavy,
whoose patent 4,167,985 is only listed in the references.
"Loudspeaker enclosure having a low reflection/low diffraction baffle
Abstract
A loudspeaker enclosure having a low reflection/low diffraction baffle
for reducing acoustic reflections and diffraction. The baffle of the
loudspeaker enclosure has a substantially entire front surface covered
by a first layer of optimized acoustical foam for reducing acoustic
reflections off of the front surface of the baffle and for reducing
diffraction around the edges of the enclosure. A second layer of
optimized acoustical foam is secured over the mounting flanges of the
transducers, which are mounted to the baffle. The first and second foam
layers are formed of a thin sheet casted, polyether urethane foam. A
front surface of the first layer of foam is flush with a front surface
of the second layer of foam. An interference fit is provided between the
first and second layers of foam to prevent any gaps from being formed
between the first and second layers. The first layer of foam is thicker
than the second layer of foam. An inside diameter of the second layer of
foam has a chamfer that tapers outwardly away from the transducer to
maximize output dispersion characteristics of the transducer.
--------------------------------------------------------------------------------
Inventors: Anagnos; Daniel P. (Grandview, NY)
Assignee: Sony Corporation (JP); Sony Electronics, Inc. (US) "
Johnd1001
Even if one attempts to take MLSSA neasurements within a relatively large room,
there still remain the floor and ceiling surfaces, reflections from which would
place a limit on the lowest frequency at which accurate measurements could be
made.
Certainly, I would not have been foolish enough to spend several tens of
thousands of dollars to construct two large anechoic chambers had viable
alternatives been available.
John Dunlavy
François Yves Le Gal
>Even if one attempts to take MLSSA neasurements within a relatively large room,
>there still remain the floor and ceiling surfaces, reflections from which would
>place a limit on the lowest frequency at which accurate measurements could be
>made.
Pressure measurements allow to get rid of the room influence.
Richard D Pierce
Johnd1001 <john...@aol.com> wrote:
>Whoa!
>
>Since the delay in arrival-time of energy diffracted from the edges of almost
>all loudspeaker enclosures possessing reasonable physical dimensions, it is
>likely that the audible effect of such "diffraction" will consist of two
>components - those that affect the amplitude/frequency response and those that
>affect the impulse/step response (ampl.vs time).
Well, I would actually disagree. There's no suggestion that such
diffraction effects, by themselves, are anything but minimum-
phase. Thus, they will NOT have "two components,", i.e., a
sparate amplitude/frequency and impulse response. Rather, in any
minimum phase response, the impulse reponse and amplitude
frequency response will ALWAYS be unique duals of one another.
>While measurement systems such as MLSSA are valuable and important tools for
>designers of accurate loudspeakers - they have fundamental limitations
>introduced by the late arrival-time of energy reflected or diffracted by nearby
>reflective surfaces.
Come on, John, this is simply untrue and claiming so suggests
the fundamental time-frequency uncertainty relation is violable.
And realms of study pursued with FAR more rigor that high-end
loudspeakers have not only failed to suggest such violability,
but have strengthened the basic premise to the point of axiom
The basic point that is being avoided is that if you want to
measure to a frequency of F, you MUST ensure that you have no
interference for a time of AT LEAST 1/2F. And that tells you the
minimum reflected path length allowable NO MATTER HOW YOU DO THE
MEASUREMENT. That would imply that, for a 20 Hz measurement to
be accurate, the first reflection must be delayed by at least
1/(2*20 Hz) or 25 mSec. That thus says the nearest boundary MUST
BE AT LEAST 25 FEET AWAY, suggesting a miniumum measurement
volume equvalent to a 50 foot cube.
Now, it makes NO DIFFERENCE AT ALL whether that 50 foot cube is
a gymnasium or the free volume of an anechoice chamber, the same
fundamental limitation inflicts BOTH paradigms. There is NO way
of getting around the fundamental time/frequency relationship.
There is no way to fool mother nature.
Under thus similar limitations of free time before reflection,
Impulse based FFT measurements are capable of the SAME accuracy
and the SAME informartion as those performed under anechoic
conditions BECAUSE BOTH ARE PERFORMED, BY DEFINITION, UNDER
ANECHOIC CONDITIONS. The window data is free from reflections,
it is therefore, anechoic, whether there are big fancy soft
wedged present or not.
To claim, a priori, otherwise, is to deny the dicatets of the
fundamental time-frequency uncertainty axiom. Are we willing to
do that?
>Designers of loudspeakers wishing to take accurate measurements might want to
>look at technology used in the measurement of antennas possessing physical
>dimensions that are relatively large with respect to a wavelength within their
>intended operating range (especially at relatively low frequencies - where many
>of the same problems exist as with loudspeakers.
And MANY are different. Polarization phenomenon, for one, are
utterly irrelevant in loudspeakers but crucial, as I am sure you
are aware, in antennae.
>Those interested in this work might wish to research archives of the IEEE, IRE,
>IREE, the Antenna and Propagation Section of the IEEE, etc. (In this regard,
>high-performance antennas are, generally, far more complex with respect to
>their design, performance properties and measurement than are audio
>loudspeakers.)
And those interested to research themore relevant work in
loudspeakers might wish to research the works of Fincham, Rife,
Lipshitz, Vanderkooy and others in the AES, along with the basic
tenants of signal properties, notable time/frequency incertainty
and the limitations of knowable information.
Richard D Pierce
Johnd1001 <john...@aol.com> wrote:
>Mark is, of course, quite right.
>
>Even if one attempts to take MLSSA neasurements within a relatively large room,
>there still remain the floor and ceiling surfaces, reflections from which would
>place a limit on the lowest frequency at which accurate measurements could be
>made.
Really, John, this is something of a strawman. It's clear to
ANYONE that the issue of obtaining free-field response is
directly related to the path lengths to the nearest boudary,
whether that's done with windowed measurements in a live room or
non-windowed measurements in an anechoic chamber. Both are beset
by the same issues and the same physics.
To make the suggestion that one simply pushes the walls away
without dealing with floor and ceiling reflections might lead
one to believe that a strawman argument is being constructed.
If you want anechoic measurements to 20 Hz, you need a free
volume equivalent to a 50 foot diameter space, whether it's in a
room with windowed impulse measurements or in an anechoic
chamber.
>Certainly, I would not have been foolish enough to spend several tens of
>thousands of dollars to construct two large anechoic chambers had viable
>alternatives been available.
We can't speak to what was available when the decision was made
to make a purchase.
What we can speak to is the fundamental physical principles at
play, and those are generally ignorant and uncaring of purchase
decisions. and those principles say that in a similar free
volume, the limitations of the measurements are set
fundamentally by the effective volume, NOT by the measurement
t4echnique employed. In a 50-foot cube, ALL systems will be
limited to approximately 20 Hz for free-field measurements. ALL
systems.
Richard D Pierce
François Yves Le Gal <fle...@aingeal.com> wrote:
>On 8 Feb 2003 19:25:16 GMT, DPi...@TheWorld.com (Richard D Pierce)
>wrote:
>
>>Indeed, using MLS-based systems, I routinely do measurements in
>>large rooms that are essentially identical to those obtained in
>>very large, very expensive anechoic chambers. And there's not
>>the slightest bit of evidence to suggest they are any less
>>accurate in determining the transient behavior of the system
>>because, indeed, that is the very nature of the system itself.
>
>Amen.
>
>And I suppose that you do pressure measurements below 150 Hz or so, in
>order to nearly totally take the room out of the picture.
ONly if its convenient and expedient. I, in fact, measure in
large spaces if I really need to know the free0field response at
low frequencies.
>>I also, once again, question the seeming strong dependence on
>>square wave response as a figure-of-merit for perforemance. As I
>>illustrated some months ago, one can construct systems with
>>essentially perfect square wave performance that sound
>>completely awful.
>
>Yes, but some really nice sounding systems do also happen to be time
>coherent and are able to reproduce decent facsimiles of square waves.
The point being that, speacificallt, flat response and linear
phase ensure decent square waves, but the converse does NOT
necessarily hold. My example from last year of a system
consiting of a direct path and a unity-gain delay of 10 mS
summed result in a system that will reproduce 100 Hz square
waves perfectly, but sound very bizarre. It's an existance proof
of a system with very good square wave performance but very poor
"time coherence."
François Yves Le Gal
Pierce) wrote:
>ONly if its convenient and expedient.
It is most of the time.
>I, in fact, measure in
>large spaces if I really need to know the free0field response at
>low frequencies.
Yeah, I've got fond memories of free field measurements with a 100 lb
speaker sitting atop some kind of a pole. And guess what: as soon as
everything was neatly setup, the breeze became quite a gusty wind.
>The point being that, speacificallt, flat response and linear
>phase ensure decent square waves, but the converse does NOT
>necessarily hold.
Agreed, of course.
>It's an existance proof
>of a system with very good square wave performance but very poor
>"time coherence."
Same as the disto box which behaves very nicely on steady state signals
and distorts everything else.
Brock Hannibal
> "Brock Hannibal" <hami...@pacifier.com> wrote in message
> news:b1uno...@enews1.newsguy.com
> > On Thu, 6 Feb 2003, Fran=C3=A7ois Yves Le Gal wrote:
> >> On 6 Feb 2003 01:51:27 GMT, Brock Hannibal <hami...@pacifier.com>
> >> wrote:
> >> Oh, I did pay attention to what Mr. Dunlavy wrote. Using a software
> >> approach such as MLSSA allows designers to dispense of anechoic
> >> rooms,
> >
> > No it doesn't, because the time-domain information can be incorrect
> > because of the reflections. The FFT programs can't help you sort that
> > out if the edges' reflections happen to arrive at the test mics
> > coincident with edges of the direct signal.
>
> Well, if you sit down and think about it, you'll quickly come to
> the conclusion that this is physically impossible. For this to
> occur, the reflected path length must be equal to the direct
> path length. Can you suggest a physically realizable situation
> where this is possible?
I'm sorry, I wasn't clear. Your paragraph below sums it up quite
nicely.
> It is almost axiomatic the low frequency limit of measurement is
> inversely proportional to the size of the room in which the
> measurement is taken, and it makes little difference if the
> chamber is anechoic with non-gated measurements or in a room
> with gated or windowed measurements. Both set the same limit in
> terms of the time between the direct arrival and first
> reflection arrival times, which limits the low frequency in the
> same way.
Yes, because of interference of the first reflected wave with the
directly arriving wave for those low frequencies.
Mark Wilkinson
news:b28j2...@enews3.newsguy.com...
Mr Pierce, with all due respect, as it is obvious to me that you too are
someone who has forgotten more than I know ..... how realistic is it to
assume tests are actully conducted inside a building that has a 50 ft or
larger interior cube, with the speaker and mic placed in the center of the
cube? I don't think this is reality, and is the whole point of the claim
MLSSA doesn't work for low frequencies. I mean think about it, 5 stories of
open interior space dedicated to loudspeaker testing ......
And Mr Le Gal, I'm sorry, but I don't know what you mean by "pressure
approach". Is it a technique that doesn't use a mic? Or doesn't require
the necessary time window being discussed? Mark
Ed Seedhouse
<captg...@direcway.com> wrote:
>Mr Pierce, with all due respect, as it is obvious to me that you too are
>someone who has forgotten more than I know ..... how realistic is it to
>assume tests are actully conducted inside a building that has a 50 ft or
>larger interior cube, with the speaker and mic placed in the center of the
>cube? I don't think this is reality, and is the whole point of the claim
>MLSSA doesn't work for low frequencies. I mean think about it, 5 stories of
>open interior space dedicated to loudspeaker testing ......
Well, I think it would be somewhat more realistic than taking a
similar sized room and covering all the walls with 12 foot long wedges
of foam.
IF you can get the same accuracy with the bare room and a time gated
measurement system, why pay out all the money for the wedges?
Ed Seedhouse
"I'm on my second cup of coffee
and I still can't face the day"
Mark Wilkinson
news:ts_1a.69138$Ec4....@rwcrnsc52.ops.asp.att.net...
Yea, if you can get the same accuracy, you'd of course be crazy to spend
money for nothing But people don't build these chambers for nothing, and
MLS, time delay spectrometry, dual FFT have been around awhile. Anybody who
has the dough and the need for a chamber is bound to know of these
techniques and their limitations. So you gotta believe S/N is waay increased
through use of a chamber.
Anyway, the point I've been trying to make is not about the merits of
anechoic chambers vs large rooms, but simply that low freq readings are
difficult at best indoors.
And as Dick pointed out, low freq physics are the same for either type room.
It just comes down to dimensions, and my bet is virtually no rooms, be they
anechoic or large, have em big enough to measure the low end. For
instance, I know at least one manufacturer who has a chamber, Meyersound,
who says they still go outside to test below 80hz. Mark
Brent
I hope John provides you all the driver/crossover/wires configuration
and you would be home free....I hate to see a good box goes to
waste...The only problem is if the drivers were designed exclusively
for DAL.....
forp <for...@yahoo.com> wrote in message news:<b1u0l...@enews4.newsguy.com>...
> Thank you for your input. I would like to know if you would give me
> any technical information what so ever regarding speaker selection
> (model #'s) and crossover layouts (component values and schematic
> information) or what ever you would feel comfortable giving out. I
> understand that this is your work and do not mean to trivialize it in
> any way. It's just that I have these cabinets and want to make them
> sound as good as I can. I do have some idea of what to do, however
> since this design was in production and the specifications exist, and
> it would be helpful to know what they are. Of course this raises
> issues with intellectual property and the many hours of hard work that
> were put into acquiring this information and I understand why you
> might not want to tell me exactly what the specifications are in
> public. I would be happy to talk with you about this in person (I
> think I live close to you) or on the phone tell you a bit about my
> self and my intentions for undertaking this project and why it is
> important to me. Lastly, thank you for taking the time to respond to
> my message.
>
> Regards
>
> On 5 Feb 2003 16:17:02 GMT, john...@aol.com (Johnd1001) wrote:
>
> >Without an anechoic chamber of appropriate size, outfitted with all internal
> >surfaces completly covered with large irregular shaped pieces of high-density,
> >open-cell polyurethane foam, the design of truly accurate loudspeakers is
> >virtually impossible.
> >
> >Anyone stating that it can be done within a large, non-anechoic room by merely
> >using an FFT program like MLSSA running on a fast computer, is not speaking
> >from a position of relevant experience.
> >
> >Designing a loudspeaker the size of the SC-IV/A (that exhibits true plus/minus
> >2 dB response and near perfect impulse and step response), is not an exercise
> >to be taken lightly.
> >
> >Sorry for the bad news - but I believe in telling the truth and not misleading
> >people with regard to the difficulty associated with building loudspeakers that
> >exhibit the properties of the SC-IV/A.
> >
> >And merely building an SC-IV/A "clone" would only yield a speaker with a
> >plus/minus 5 dB (or worse) performance.
> >
> >All the best - keep in touch.
> >
> >John...@aol.com
François Yves Le Gal
wrote:
>For
>instance, I know at least one manufacturer who has a chamber, Meyersound,
>who says they still go outside to test below 80hz.
Meyersound does *indoors* boundary/half plane measurements and has even
published a number of white papers on this method. I've the supplied a
link to one of them in an earlier post. Here it is again
:http://www.meyersound.com/products/software/mapponline/low_frequency_polar_data.htm
François Yves Le Gal
>I hope John provides you all the driver/crossover/wires configuration
>and you would be home free....
Dunlavy used some faily basic first order xovers. But basic doesn't mean
easy to reproduce: reaching simplicity is an art.
>The only problem is if the drivers were designed exclusively
>for DAL.....
AFAICT, DAL always used off the shelf drivers from a couple of
Scandinavian manufacturers now part of DST.
Mark Wilkinson
news:d%w2a.87153$be.64904@rwcrnsc53...
Sure Meyer does indoor measurements -- their SIM system may be the best out
there. But how does that challenge my assertion that indoor low freq
measurement is difficult AT BEST for MLSSA, FFTs, etc.?
It would seem to me the link you provided proves my point. To quote from
the Meyer link:
"However, the wedges used in our anechoic chamber are only rated to 100 Hz.
Below 100 Hz, the data in our anechoic chamber, while "pretty good", is not
accurate enough to give good results in MAPP Online. For this reason, until
now, we have not allowed any predictions in MAPP Online below 100 Hz."
The article is about their acoustic simulation program MAPP, and the attempt
to seed it with loudspeaker outputs taken at fine angular increments. The
crux of the article ends up being how to overcome the inability to measure
below 100hz indoors (in the anechoic chamber). Please note that the
solution, the fancy Boundary Element mathematics are computed from a
starting point of a 4m OUTDOOR half plane reading. Nuff said.
And I must ask again, what do you mean by pressure approach? Does it use a
mic? Can it measure faster than a 1/2 wavelength? Mark
François Yves Le Gal
wrote:
>But how does that challenge my assertion that indoor low freq
>measurement is difficult AT BEST for MLSSA, FFTs, etc.?
Well,
>It would seem to me the link you provided proves my point. To quote from
>the Meyer link:
>"However, the wedges used in our anechoic chamber are only rated to 100 Hz.
>Below 100 Hz, the data in our anechoic chamber, while "pretty good", is not
>accurate enough to give good results in MAPP Online. For this reason, until
>now, we have not allowed any predictions in MAPP Online below 100 Hz."
That's the irrelevant part. Here's the relevant one: "The figure below
shows an on-axis frequency response measurement of a single Meyer Sound
M2D loudspeaker at 4 meters, measured on a flat surface with the
microphone placed directly on the ground. The ground is known as a half
plane, and on axis to a single loudspeaker it creates a very accurate
measurement of the frequency response compared with a single loudspeaker
measured in an anechoic chamber. The half plane causes the magnitude of
the response to increase by 6 dB compared with the anechoic chamber
measurement.
./...
Below 100 Hz, as expected, the two traces differ. Since our chamber
wedges are only rated to 100 Hz, we have found that ground plane
measurements are more accurate. Especially visible in the red trace is a
10 dB hole centered at around 80 Hz. This acoustic "hole" in our
anechoic chamber below 100 Hz is the reason that we have not allowed
predictions in MAPP Online below 100 Hz."
Ground plane measurements, done in a normal room are MORE ACCURATE.
>And I must ask again, what do you mean by pressure approach? Does it use a
>mic? Can it measure faster than a 1/2 wavelength?
Well, if you had followed this thread, you wouldn't ask again and again
the same question: I've already answered you and supplied a couple of
references.
Here they are again:
"the pressure technique has been devised by Don Keele, and published in
the Journal of the Audio Engineering Society: "Low-Frequency Loudspeaker
Assessment by Nearfield Sound-Pressure Measurement", JAES 04/74,
reprinted in the AES Loudspeaker anthology.
John Atkinson, editor in chief of Stereophile, has published an
excellent summary of the various techniques as an AES preprint
"Loudspeakers: What Measurements Can Tell Us - And What They Can't Tell
Us!," AES 103, Preprint 4608.
This paper has been reworked and published in Stereophile as the series
"Measuring Loudspeakers", available either in print or on the
Stereophile web site:
http://www.stereophile.com/printarchives.cgi?99 (Pt. I)
http://www.stereophile.com/printarchives.cgi?100 (Pt. II)
http://www.stereophile.com/printarchives.cgi?103 (Pt. III)"
François Yves Le Gal
wrote:
>And Mr Le Gal, I'm sorry, but I don't know what you mean by "pressure
>approach". Is it a technique that doesn't use a mic? Or doesn't require
>the necessary time window being discussed?
There are various techniques allowing for precise measurements w/o
resorting to anechoic rooms of free field.
Pressure measurements are done nearfield, with the mic very close to the
driver/port/..., but you may also locate the mic on the floor and do an
half plane measurement.
Both approaches correlate very well with anechoic measurements. Here is
a web page with some info on half plane measurements:
http://www.meyersound.com/products/software/mapponline/low_frequency_polar_data.htm.
The pressure technique has been devised by Don Keele, and published in
Mark Wilkinson
What you are missing is that the ground plane measurement was taken
outdoors. (Why? Hint, outdoors is more accurate for low freq than indoors
:)
The VERY next sentence states "The blue trace in the figure below shows the
outdoor SIM half plane measurement of a single M2D, minus 6 dB." Please
note outdoor.
>
> >And I must ask again, what do you mean by pressure approach? Does it use
a
> >mic? Can it measure faster than a 1/2 wavelength?
>
> Well, if you had followed this thread, you wouldn't ask again and again
> the same question: I've already answered you and supplied a couple of
> references.
>
Your (previous) reply hit my newserver after this post of yours?? I'll
reply to it. Mark
Mark Wilkinson
> On 11 Feb 2003 01:06:22 GMT, "Mark Wilkinson" <captg...@direcway.com>
> wrote:
>
> >And Mr Le Gal, I'm sorry, but I don't know what you mean by "pressure
> >approach". Is it a technique that doesn't use a mic? Or doesn't require
> >the necessary time window being discussed?
>
> There are various techniques allowing for precise measurements w/o
> resorting to anechoic rooms of free field.
>
> Pressure measurements are done nearfield, with the mic very close to the
> driver/port/..., but you may also locate the mic on the floor and do an
> half plane measurement.
>
Okay, that's what I thought you might mean -- I'm familiar with both
techniques. I find they do help with reflections, but they don't effect the
time window necessary, so they are far from a full cure.
One thing I don't like about either of these techniques is I haven't figured
out how to use them and get a mic anywhere near equidistant to multiple
drivers. Phase info goes whacko.
> Both approaches correlate very well with anechoic measurements. Here is
> a web page with some info on half plane measurements:
>
http://www.meyersound.com/products/software/mapponline/low_frequency_polar_d
ata.htm.
>
No. There is poor coorelation between anechoic and OUTDOOR half plane.
Look at the chart, read the text.
The good coorelation is between anechoic and a MATHEMATICAL SIMULATION
overlayed on an OUTDOOR 4m halfplane. Mark
>
François Yves Le Gal
wrote:
>What you are missing is that the ground plane measurement was taken
>outdoors.
In the context, yes, but such measurements can be done indoors or
outdoors.
>Your (previous) reply hit my newserver after this post of yours??
Yes, this was a server-related problem : one of the 8-bit chars I've
used triggered a brain damaged gateway which transformed my initial post
into Mime (yeeech). The modified version didn't went thru, or so it
seems, and had to be resent. Sorry about that.
François Yves Le Gal
wrote:
>I find they do help with reflections, but they don't effect the
>time window necessary, so they are far from a full cure.
They are fine by any standards AFAICT.
>One thing I don't like about either of these techniques is I haven't figured
>out how to use them and get a mic anywhere near equidistant to multiple
>drivers. Phase info goes whacko.
Good point. Measuring loudpseakers fitted with multiple bass drivers
implies either outddors or anechoic measurements, *or* some creative
processing with multiple measurements "spliced" thus forming a
composite.
>No. There is poor coorelation between anechoic and OUTDOOR half plane.
>Look at the chart, read the text.
Correlation is fine while the anechoic romm doesn't impose it's
signature. As Meyer states: "The blue trace in the figure below shows
the outdoor SIM half plane measurement of a single M2D, minus 6 dB. The
red trace shows the anechoic chamber measurement of a single M2D. Note
the excellent correlation between the two measurements between 100 Hz
and 10 kHz. Above 10 kHz, ground plane measurements are inaccurate due
to the size of the microphone compared with the small wavelengths of the
sound. Above 10 kHz, we have found through extensive testing that our
anechoic chamber is very accurate.
Below 100 Hz, as expected, the two traces differ. Since our chamber
wedges are only rated to 100 Hz, we have found that ground plane
measurements are more accurate. Especially visible in the red trace is a
10 dB hole centered at around 80 Hz. This acoustic "hole" in our
anechoic chamber below 100 Hz is the reason that we have not allowed
predictions in MAPP Online below 100 Hz."
Ground plane measurements are *more* accurate below 100 Hz.
Mark Wilkinson
> On 13 Feb 2003 16:04:58 GMT, "Mark Wilkinson" <captg...@direcway.com>
> wrote:
>
> >I find they do help with reflections, but they don't effect the
> >time window necessary, so they are far from a full cure.
>
> They are fine by any standards AFAICT.
>
> >One thing I don't like about either of these techniques is I haven't
figured
> >out how to use them and get a mic anywhere near equidistant to multiple
> >drivers. Phase info goes whacko.
>
> Good point. Measuring loudpseakers fitted with multiple bass drivers
> implies either outddors or anechoic measurements, *or* some creative
> processing with multiple measurements "spliced" thus forming a
> composite.
>
Was thinking more about the multiple drivers of a full range unit, than say
multiple bass drivers. My experience is that multiple bass drivers couple
acoustically and pose no problem. Indeed, I guess this coupling is what
let's ground plane readings work --- the driver(s) close couple with the
mirror image provided by the ground plane.
> >No. There is poor coorelation between anechoic and OUTDOOR half plane.
> >Look at the chart, read the text.
>
> Correlation is fine while the anechoic romm doesn't impose it's
> signature. As Meyer states: "The blue trace in the figure below shows
> the outdoor SIM half plane measurement of a single M2D, minus 6 dB. The
> red trace shows the anechoic chamber measurement of a single M2D. Note
> the excellent correlation between the two measurements between 100 Hz
> and 10 kHz. Above 10 kHz, ground plane measurements are inaccurate due
> to the size of the microphone compared with the small wavelengths of the
> sound. Above 10 kHz, we have found through extensive testing that our
> anechoic chamber is very accurate.
>
> Below 100 Hz, as expected, the two traces differ. Since our chamber
> wedges are only rated to 100 Hz, we have found that ground plane
> measurements are more accurate. Especially visible in the red trace is a
> 10 dB hole centered at around 80 Hz. This acoustic "hole" in our
> anechoic chamber below 100 Hz is the reason that we have not allowed
> predictions in MAPP Online below 100 Hz."
>
> Ground plane measurements are *more* accurate below 100 Hz.
>
Yes, but *why* has been my point all along. The why is because they were
taken outdoors free from any room reflections. The problem with Meyer's
anechoic chamber is not a problem with the foam wedges or such, it's simply
a problem in that it's a ROOM. (Please remember that I have only been
addressing low freq.)
And I was wrong in a previous post to say that there was good correlation
between their mathematical simualtion overlayed on a 4m outdoor groundplane,
to their anechoic chamber. They say they can't get good readings below
100hz inside, and they can get good readings outside, but the readings would
be tedious and error prone due to the vagaries that would occur outdoors
while trying to take the huge number of readings necessary to generate 1
degree polars. So basically, their MAPP (below 100hz) seems to be pure
simulation seeded by one, or only a few, real readings. IOW, nothing to
correlate to. To me, this is very telling -- a company like Meyer, renowned
for their measurement skills--ends up using a simulation because good low
freq data is hard to get. Mark
François Yves Le Gal
wrote:
>Was thinking more about the multiple drivers of a full range unit
Then the splicing technique also works fine if you want to aim your mike
at the acoustic center of each driver.
>My experience is that multiple bass drivers couple
>acoustically and pose no problem.
I would kindly disagree here: there is a number of dual woofer systems
around with different alignments (or different drivers) used, and
pressure measurements have to be done individually.
>Yes, but *why* has been my point all along. The why is because they were
>taken outdoors free from any room reflections.
Oh, we agree here. We don't share the same views regarding pressure
indoors measurements, that's all.
Dick Pierce
gets its act together.
"Mark Wilkinson" <captg...@direcway.com> wrote in message news:<b29ia...@enews1.newsguy.com>...
> > What we can speak to is the fundamental physical principles at
> > play, and those are generally ignorant and uncaring of purchase
> > decisions. and those principles say that in a similar free
> > volume, the limitations of the measurements are set
> > fundamentally by the effective volume, NOT by the measurement
> > t4echnique employed. In a 50-foot cube, ALL systems will be
> > limited to approximately 20 Hz for free-field measurements. ALL
> > systems.
> >
>
> Mr Pierce, with all due respect, as it is obvious to me that you too are
> someone who has forgotten more than I know ..... how realistic is it to
> assume tests are actully conducted inside a building that has a 50 ft or
> larger interior cube, with the speaker and mic placed in the center of the
> cube?
Then who has the room for an anechoic chmaber that fits in a
50 foot cube?
>I don't think this is reality, and is the whole point of the claim
> MLSSA doesn't work for low frequencies. I mean think about it, 5 stories of
> open interior space dedicated to loudspeaker testing ......
Okay, let's try this once again. If it DOESN'T work for MLSSA, IT DOESN'T
WORK FOR ANYTHING ELSE. ALL systems suffer from the same fundamental
physical limitations. Thus, if you DON'T have space for making measurements
to very low frequencies with MLSSA, YOU DON'T HAVE SPACE FOR AN ANECHOIC
CHAMBER TO DO THE SAME MEASUREMENS
Dick Pierce
> > IF you can get the same accuracy with the bare room and a time gated
> > measurement system, why pay out all the money for the wedges?
> >
>
> Yea, if you can get the same accuracy, you'd of course be crazy to spend
> money for nothing But people don't build these chambers for nothing, and
> MLS, time delay spectrometry, dual FFT have been around awhile.
You know what, people AREN'T BUILDING THESE CHAMBERS ANYMORE. And
there are VERY few acoustic anechoice chambers that are useful to
20 Hz in the world. My guess is there might be 10 or 12. Back in
the 70's when one could go to the yellow pages from a major metro
area and find companies thatr made anechoic chambers, I actually
enquired what it would cost to build a 20 Hz chamber, and I got
quotes approaching $1,000,000, not including the building to put it
in. 100 Hz chambers were going for $20,000.
People AREN'T building these chambers anymore BECAUSE THEY ARE NOT
NEEDED.
Try finding a 20 Hz anechoic chamber within, oh, 500 miles of where
you live.
>Anybody who
> has the dough and the need for a chamber is bound to know of these
> techniques and their limitations. So you gotta believe S/N is waay increased
> through use of a chamber.
No, it is not, because MLS techniques have an inherent high S/N as
an intrinsic property of technique, and averaging buys you S/N for
only the cost of measurement and compute time. I can, for example,
in a 65K length MLS @100 kHz buy another 12 dB of S/N at the cost of
taking 15 seconds to do the measurement.
> Anyway, the point I've been trying to make is not about the merits of
> anechoic chambers vs large rooms, but simply that low freq readings are
> difficult at best indoors.
And that's the point which you seem to want to contradict.
Dick Pierce
> > Pressure measurements are done nearfield, with the mic very close to the
> > driver/port/..., but you may also locate the mic on the floor and do an
> > half plane measurement.
> >
>
> Okay, that's what I thought you might mean -- I'm familiar with both
> techniques. I find they do help with reflections, but they don't effect the
> time window necessary, so they are far from a full cure.
>
> One thing I don't like about either of these techniques is I haven't figured
> out how to use them and get a mic anywhere near equidistant to multiple
> drivers. Phase info goes whacko.
Uhm, no. At the low frequencies where this type of measurement
is done, the phase info does NOT go whacko be cause of distance
problems. Let's say you're distance error is on the order of 6
(that's a HUGE error to make without knowing it). At 50 Hz, that
6 inches corresponds to a "whacko" phase error of less than 8
degrees.
Mark Wilkinson
news:Ekd3a.112001$be.78505@rwcrnsc53...
Completely 100% agree. The discussion with Mr Le Gal re Meyer's chamber and
simulation technique, is a real life illustration of this fact.
Somehow or another you guys took my statement that "MLSSA, FFTs etc don't
work well for low freq indoors" to be a statement that anechoic chamber's
are superior to regular rooms. Uh, uh ... it's a statement that stands on
its own -- low freq measurements are difficult at best indoors. Mark
Penury
<captg...@direcway.com> wrote:
>-- low freq measurements are difficult at best indoors. Mark
Several years ago when several of us bought Tonegen 1259 drivers
(used in the NHT 3.3) and built cabinets of varying size and stuffing,
and wanted to compare the results. We placed a Radio Shack SPL meter
one inch from the dust cap and charted the results of 20Hz to 200Hz
warble tones from the Stereophile Test CD 2.
This gave us a useful comparison of various EQ settings, box sizes
and stuffing via the Net.
This would not be useful for a commercial enterprise, but was very
useful (and informative) for us DIYers. I believe it was Don Keel Jr.
that suggested this method.
-=Bill Eckle=-
ab...@wmeckle.com
Vanity Web pages at:
http://www.wmeckle.com
Mark Wilkinson
As I said to Mr Le Gal, I meant equidistant to multiple drivers in a full
range system . Sorry I wasn't clearer. I didn't figure you guys would
automatically assume multiple low freq drivers. No offense, but it often
seems like both of you are more interested in debate than discussion. Mark
Dick Pierce
Well, if you are familiar with the technique, IT assumes low frequency
only. In fact, all such pressure techniques as described in the relevant
literature have such limitations. Thus, assuming that you're using such
techniques for measuring full range systems essentially violates the
basic requirements of the technique.
For example, close mic'ing a 12" woofer gives very good correlation for
the frequency response up to frequencies whose wavelength starts to be
proximal to the dimensions of the drivers: above that point, all hell
breaks loose as you are now faced with the phase errors due to the
different path-length differences to different parts of the driver.
Similarily, with the ground-plane technique, as soo as the path length
differences approach a sizeable fraction of a wavelength, you run into
problems.
> No offense, but it often
> seems like both of you are more interested in debate than discussion.
Well, no offense taken, but no one was aware of your assumption but you,
and for myself at any rate, having read the relevant literature, I was
aware of what the literature clearly states as the limiting assumptions
of the various techniques. There is no attempt on my part of debate
posturing, rather, you made a statement that clearly violated the basic
known limitations of the techniques being discussed, and I provided an
answer, not being made aware of your assumptions, that essentially
answered your objection within the context of the know limitations.
Unfortunately, I assert that precisely the same kind of hidden
assumptions are at the basis of this particular thread. John Dunlavy
made what I consider an overly sweeping generalization that impulse
and FFT-based techniques simply are not as good as anechoic results.
As a generalization, this statement is technially wrong. Now, if John
had stated, which he has stated to me in private conversations,
something to the tune of:
"Measurements made using impulse-based stimulus in small
rooms using windowed data are not as accurate at low
frequencies as those made in large anechoic chambers,
all other things being equal,"
I would have been in total agreement, suggesting two steps of
generalization to reach the overriding principle at hand:
"The lower the frequency you need to measure, the longer the
time window must be and thus the larger the space in which
the measurement must be made."
And the final step is simply the classical time-frequency un-
certainty principle, where:
delta F * delta T >= 1/2
where delta F is uncertainty in frequency and delta T is un-
certainty in time. In this case, where we want delta F to be
small, i.e. measuring low frequencies, delta T, the time we must
measure, must be large. Simple as that. And since, in acoustics,
time is distance, then it simply means that FUNDAMENTALLY, low
frequency interference-free measurement requires big spaces, no
matter HOW you do it.
No debate for debate's sake, simply revealing yet another
manifestation of the underlying physical principles involved.
Dick Pierce
> > Okay, let's try this once again. If it DOESN'T work for MLSSA, IT DOESN'T
> > WORK FOR ANYTHING ELSE. ALL systems suffer from the same fundamental
> > physical limitations. Thus, if you DON'T have space for making
> measurements
> > to very low frequencies with MLSSA, YOU DON'T HAVE SPACE FOR AN ANECHOIC
> > CHAMBER TO DO THE SAME MEASUREMENS
>
> Completely 100% agree. The discussion with Mr Le Gal re Meyer's chamber and
> simulation technique, is a real life illustration of this fact.
>
> Somehow or another you guys took my statement that "MLSSA, FFTs etc don't
> work well for low freq indoors" to be a statement that anechoic chamber's
> are superior to regular rooms. Uh, uh ... it's a statement that stands on
> its own -- low freq measurements are difficult at best indoors. Mark
How about this:
"Low frequency measurements are difficult."
I'd generalize it even further:
"Measuring loudspeakers correctly is VERY difficult, so much
so that very few people get it right."
Mark Wilkinson
Hi Dick,
Thanks for your latest replies -- they make complete sense, and seem quite
fair, to me.
As an overly interested amatuer, I'm surprised at just how difficult it is
(for me) learning to measure speakers. I've spent the last year or two
playing with dual FFT and MLS, and am just now starting to understand
windowing issues, particularly those that relate to resolution per octave.
And I clearly have holes in my learning. Being the "keep it simple so I can
understand it" kinda guy that I am, I'm prone to discard and forget terms
like "pressure approach", because to me all it's saying is put the mic on
the floor (or up close).
If I may, do you have any tips for those of us trying to learn to measure
speakers? Your posts are often quite informative. Thanks, Mark
Johnd1001
chambers have virtually ignored one of their most important virtues: improving
the S/N ratio of measurements - particularly at low frequencies.
Indeed, S/N limitations are a major "curse" encountered within almost all
indoor environments - and, to say the least, make outdoor measurements nearly
impossible, even on days when the wind can barely be felt, if at all.
Indoor air-conditioners, etc., are another curse that can virtually eliminate
the ability to make indoor measurements at frequencies much below a few hundred
Hertz
Anyway, I doubt that I and my technical/engineering staff could have designed
(and manufactured) loudspeakers with uncondutionally guaranteed,
truly-accurate, performance without access to our two large and one small
chambers.
Best of listening.
John Dunlavy
Johnd1001
was to isolate our measurements from "background noise generated within a
33,000 sq.ft. building that also housed our cabinet shop, shipping & receiving,
etc., all of whic were "noise generating sources" we needed to acoustically
isolate to obtain the level of accuracy we demanded for our products.
Best of listening,
John Dunlavy
Richard D Pierce
Johnd1001 <john...@aol.com> wrote:
>It seems that many of the posts concerned with the attributes of anechoic
>chambers have virtually ignored one of their most important virtues: improving
>the S/N ratio of measurements - particularly at low frequencies.
>
>Indeed, S/N limitations are a major "curse" encountered within almost all
>indoor environments - and, to say the least, make outdoor measurements nearly
>impossible, even on days when the wind can barely be felt, if at all.
I think, while Mr., Dunlavy's comments may well be appropriate
in the realm of sontinuous-tone or swept-tone measurements,
since at least the early 1970's with the advent of techniques
such as impulse based of time domain spectrometry and especially
since the invention of MLS based techniques, the comments
regarding signal to noise issues are quite obsolete.
Let's just examine the simple case of averaging techniques.
Let's examine a real-world case based on a 100 kHz sample rate
and a sample window that's 128K samples long. This gives an
effective measurement window over 1.28 seconds long, an
effective frequency resolution of about 1 Hz over the entire
measurement bandwidth (50 kHz), and requires, similarily, 1.28
seconds to make the measurement.
Let's assume the environment is noisy, with a background of, oh,
65 dB unweighted. Assume the measurements are made at levels of
85 dB. It would seem that a 20 dB S/N is not very good. But,
simply by applying multiple repeated measurements and avergaing,
we gain S/N simply at the expense of time. FOr example, let's do
32 averages. That will take us 40 second and gain us 15 dB os
broadband S/N.
INdeed, 65 dB noise background is severe. I regularily maintain
at least 30 dB MORE than thafor my measurements.
Further, since MLS and other similar techniques use completely
determninstic signals and cross-correlation techniuqes, these
techniques already have a much higher noise immunity than
conventional techniques.
Further FFT processing of the resulting impulse response means
that the broadband noise floor divided amongst the individual
frequency bins. This means, in our example above, that a 65 dB
noise floor with white distribution, is reduced by some 86 dB in
each bin.
>Indoor air-conditioners, etc., are another curse that can virtually eliminate
>the ability to make indoor measurements at frequencies much below a few hundred
>Hertz
Uh, is there a problem with turning off the air conditioners
when making measurements?
>Anyway, I doubt that I and my technical/engineering staff could have designed
>(and manufactured) loudspeakers with uncondutionally guaranteed,
>truly-accurate, performance without access to our two large and one small
>chambers.
Mr. Dunlavy keeps asking us why he and his staff built these
chambers. The question itself is no proof of efficiacy of the
concept. We don't know why people make decisions, especially
when there are a variety of quite viable alternatives
available.
What we do know is that there is absolutely no intrinsic
properties of either techniaue that spell absolute superiority
of one over the other. We do know that by the very fundamentals
of of the properties of signals, by the fundamental axioms of
Fourier analysis, by the high noise immunity that IS an inherent
property of tecniques suuch as Maximal Length Sequence
measurement and the Hadamard transform, that the alledged
advantages thus far advanced for measurement in anechoice
chambers are directly provided also by these more advanced
techniques.
I would close by noting first that Mr. Dunlavy has only revealed
that measurements were made in anechoic chambers without once
revealing the properties of those chambers, i.e., thee size,
their effective absorbtion vs frequency, at what frequency do
they ceases to provide true inverse-square-law behavior, and so
forth. Nor has it been revealed that once inside the chamber,
what measurement system is used to get the data. To be frank,
once the acoustic space in which the measurements take place has
been dealt with, the easy part is done.
In short, I have seen many people do bad measurements under a
wide variety of conditions, and some of the most horrifically
useless measurements have been done in superb anechoic chambers.
While I am sure this is not the case with Mr. Dunlavy, simply
stating that one has anechoic facilities is no assurance of
relaible data.
And, indeed, windowed impulse based measurements in live rooms
IS an anechoic facility.
--
| Dick Pierce |
| Professional Audio Development |
| 1-781/826-4953 Voice and FAX |
| DPi...@world.std.com |
Denis Sbragion
DPi...@TheWorld.com (Richard D Pierce) wrote in
news:b330c...@enews3.newsguy.com:
..
> since the invention of MLS based techniques, the comments
> regarding signal to noise issues are quite obsolete.
..
i think this paper may be of help:
Transfer-Function Measurement with Sweeps
Swen Müller and Paulo Massarani
The abstract may be found at:
http://www.aes.org/journal/toc/AES-June2001TOC.html
It covers both MLS and log sweep measurement methods, which provides even
greater noise immunity then MLS, especially at low frequencies.
Regards,
--
Denis Sbragion
InfoTecna
Tel: +39 0362 805396, Fax: +39 0362 805404
URL: http://www.infotecna.it