How to measure high freq in curved panels? - diyAudio
Go Back   Home > Forums > Loudspeakers > Planars & Exotics

Planars & Exotics ESL's, planars, and alternative technologies

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 23rd August 2014, 06:21 PM   #1
diyAudio Member
 
Join Date: Jan 2008
Location: Toronto and Delray Beach, FL
Default How to measure high freq in curved panels?

How do you measure the upper treble response of curved panels or arrays?

(I use REW and trusty old lab meters too.)

To illustrate, in my array pictured below, there are 4 circular ESL panels mounted on a curved articulated surface. the two outer panels are roughly a half-inch further from any point in the direction the array is facing. While the treble sound is dramatically wonderful on treble percussion (and measurement of a single ESL in isolation is as swell as I'd want and with dispersion a lot wider than the panel angle), the listening position measurement of the whole array in action gets horrible north of maybe 7 kHz.

Thanks.
Ben
Attached Images
File Type: jpg tweeter.jpg (37.7 KB, 182 views)
__________________
Dennesen ESL tweets, Dayton-Wright ESL (110-3200Hz), mixed-bass Klipschorn (28-110), and giant OB using 1960's Stephens woofer (18-110); Behringer DSP. HiFi aspirations since 1956

Last edited by bentoronto; 23rd August 2014 at 06:52 PM.
  Reply With Quote
Old 24th August 2014, 07:48 PM   #2
diyAudio Member
 
Join Date: Feb 2008
Location: Jackson,michigan
Hi Ben,

I had to think about this for awhile but I think I know what you mean.

One of the easiest ways I do such measurements is by feeding the system with some noise and use a RTA program to see the response in realtime as I move the microphone around.

This method is quicker than moving the microphone and waiting for the sweep test to finish each time.

I use either Visual Analyzer or a program called WaveSpectra for this, there are a few other ones available as well but I forget their names at this time.

Wave spectra seems to be one of the best to use as it gives you a scope display as well.

It is very fast and doesn't glitch as it operates in realtime!

Not to mention that it can be set in a 24bit mode too.

Wave spectra can be found here and it is Free,

efu's page

Here are the links to other programs that I use for such measurements as well,

Software to analyze and equalize frequency response of entire sound system?

The dispersion is a function of each of the panels width vs frequency.

It may be as well that the heavy gas that is used in your DW's could also be an issue as well.
I have read this mentioned in a discussion about the DW's before, and/or a combination of having to have the sound go through it and the extra membrane (bag) that keeps it encapsulated.

It has been shown that thicker material for the diaphragms can have a pretty good effect on the high frequency output that a panel can produce without the use of some sort of equalization.

So the compromise was loss of High Frequency efficiency vs overall output

I have experienced this some what with different thickness diaphragm materials and dirt that has an added mass as well, I mentioned this in the coatings thread.

Also, I tried to make a pop and spit filter once out of some common food wrap (saran) for my very expensive ($1000) Audio Technica 4050 microphone and even this seemed to have a bit of a slight effect on the purity of the sound so we never used it.

As far as the dispersion is concerned, Once the Frequency being produced has a wavelength equal to the width of the panel that is emitting it, the dispersion resembles that of a circle off of the front of the diaphragm.

A figure eight pattern if you will, and as the frequency is raised it transitions more in to a focused beam.

More info can be found here on this study,

A Segmented Stator Desktop ESL

I once made a bunch of frames to show this in realtime but I never did compile it into a movie clip.

I had just recently found a spreadsheet program for excel that calculates dispersion as well.
I forget that I had it and almost lost it in a system's crash (Windows, go figure) last weekend!!
I will post it if I can find it again.

Anyhow, While trying to combat this dispersion issue many have tried curved panels and curved array's but this still does nothing once you get off axis to the edge of panel or array.

The last panel in the array still is the key factor of its high frequency dispersion off axis of the edge of the speaker as a whole, due to the last panels width!

I found this out when I had 2,3 and 4 of my little panels all lined up during the testing I was doing some 4 years ago.

This was great as it did widen the overall area.

But, It did nothing once I got off to the edge and beyond as the off axis dispersion was no different then that of a single panel!

But the 3.5" wide little panels were much better for this situation than my 8.5" panels were, and is why I favored them more, even though they didn't have the displacement to produce the bass the bigger 21" x 8.5" had.

As you know I remedied this later by increasing the D/S slightly and used some Extreme voltages to boot, and, even 200Hz was enough to take you out of the room!! He,he,he,he

I found the spreadsheet so I have posted it as well.
I don't remember who wrote it or where I got it, But I do hope that it can be of use for you.

Cheers!!!

jer
Attached Files
File Type: zip Directivity Calc.zip (5.5 KB, 6 views)
  Reply With Quote
Old 24th August 2014, 08:07 PM   #3
diyAudio Member
 
Join Date: Jan 2008
Location: Toronto and Delray Beach, FL
As often, an explosion of productive ideas from Gerald. Need time to digest and respond.

But, as for the basic issue - very glad you didn't find it a simple-minded question.

But how to waft a mic on an arc in front of the elements with any kind of uniformity and repeatability (and as I tweak the knobs)??? Perhaps I'll practice in front of a mirror (until my wife sees me and asks why). Or build a swing-arm jig.

I think the peak-hold RTA display on REW using pink noise should be suitable. Length of time exposed to noise doesn't much affect the peak trace. Then just a matter of getting the arc uniform without worrying about the movement cadence.

Thanks.
Ben
__________________
Dennesen ESL tweets, Dayton-Wright ESL (110-3200Hz), mixed-bass Klipschorn (28-110), and giant OB using 1960's Stephens woofer (18-110); Behringer DSP. HiFi aspirations since 1956
  Reply With Quote
Old 24th August 2014, 08:32 PM   #4
diyAudio Member
 
Join Date: Feb 2008
Location: Jackson,michigan
Hmmm......I didn't know REW did a true RTA I will have to take a look at it a little closer.

But do get Wave Spectra I am very sure you will be very happy with that utility as well.

It does peaks as well as continuing with the RTA.

Here is a sample shot that I did of an FR sweep that I did on my Tape deck (MSR16)using it.

There wasn't any waveform displayed on the scope trace because it was after the playback sweep was done, but the noise floor was still realtime when I made this screen capture.

jer
Attached Images
File Type: jpg Track2 no DBX.jpg (88.3 KB, 161 views)
  Reply With Quote
Old 25th August 2014, 01:16 AM   #5
diyAudio Member
 
Join Date: Jan 2008
Location: Toronto and Delray Beach, FL
REW RTA is similar except it uses plain lines instead of (misusing?) colour as many software oscilloscopes do.

Thanks for link to Wave Spectra. But I think it is for Windows (as are, alas, most of these programs). Been with Apple since 1979 and have not regretted it ever longer than 30 seconds.

Ben
__________________
Dennesen ESL tweets, Dayton-Wright ESL (110-3200Hz), mixed-bass Klipschorn (28-110), and giant OB using 1960's Stephens woofer (18-110); Behringer DSP. HiFi aspirations since 1956
  Reply With Quote
Old 25th August 2014, 03:07 AM   #6
diyAudio Member
 
Join Date: Feb 2008
Location: Jackson,michigan
Oh Okay, I did not know this.

I have gotten it to work in Linux under WINE as well but I haven't fully tested it yet.
I do believe the colors can be changed but I don't remember exactly as they work for me for now.

jer
  Reply With Quote
Old 25th August 2014, 02:10 PM   #7
diyAudio Member
 
bolserst's Avatar
 
Join Date: Jun 2005
Location: Fort Worth, Texas
Quote:
Originally Posted by bentoronto View Post
How do you measure the upper treble response of curved panels or arrays?
(I use REW and trusty old lab meters too.)
...measurement of a single ESL in isolation is as swell as I'd want and with dispersion a lot wider than the panel angle), the listening position measurement of the whole array in action gets horrible north of maybe 7 kHz.
When measuring at the listening position, are use using the Impulse Graph in REW to adjust the measurement window to remove reflections? Typical window length would be 3mS to 5mS. If you haven't adjusted the window, you are most likely seeing reflections contaminating your measurement. It is also possible that the size/spacing/angles of the 4 panels are causing some interference effects at HF.

A few questions:
1) Any chance you can post a pic of the backside of the panel array?
2) How big is the radiating area of the individual ESL panels? (ie don't include surrounding frames.)
3) What is the width of the gaps between the radiating areas?
With 2) & 3) I can analyze/estimate what intereference effects you might expect.

4) When you measured one in isolation, was it a near-field or far-field measurement?
If far-field, was it suspended in space? or placed in front of your DW as shown.
5) Same question as 4) but for the array...wondering if you are getting reflections off your DW.

Is REW still being supported for the Apple platform?
I thought I read it had been dropped....
  Reply With Quote
Old 26th August 2014, 08:45 PM   #8
diyAudio Member
 
Join Date: Jan 2008
Location: Toronto and Delray Beach, FL
Wonderful to have bolserst sophisticated thoughts here too.

Hard to photograph rear. Back of 3-inch ESL disks has some damping paper. Not much influence from the rear the way I have them sitting draped over the Dayton-Wright boxes - as judged by the absence of change when I stuff hi-tech absorbent behind them.

Likewise for room reflections because I favour truly damped surroundings*, walls at some distance, and impulse display shows little rumbling even within 3 milliseconds.

So that leaves the curved geometry, I'd guess. the tweeter speaker is a 14 x 4.5 inch, segmented curve. The peak distance from a chord between the ends makes the middle two drivers about .5 inch closer to my ears than the outer two. Which means some kind of interference beyond maybe 3 kHz, assuming pure on-axis mic response.

In practice, the total experience of treble music playing (and through two speakers and including some mid-range driver output) may not be diminishing quite like the on-axis geometry/acoustics may suggest. But I certainly don't fancy doing the most authentic measurement: psycho-acoustic response.

So back to the question. Is there some way to re-think the on-axis mic response as the final word?

Gerald - worked on RTA-noise measurements using swing the mic in an arc, as you suggested. Funny, not getting much different treble response than a stationary mic. Maybe I need more practice. Holding the mic opposite one of the outer elements, I do get a response closer to a single cell, as might be expected.

bolserst - measurement of a single ESL tweeter element was made just inches from the driver, just hanging over my workbench. Output seemed to sail well over the limits of my existing test gear (a Dayton EMM-6 on-order).

Ben
*Many of the pictures I see at DIYaudio of music rooms are furnished in tasteful modern (read: sparse, unupholstered) ways. Frankly, my ears hurt just looking at these noisy, reverberant rooms. How can music be reproduced in such echo chambers? Some people fuss endlessly about different crossover curve models but play sound in rooms that have been given no acoustic fore-thought.
__________________
Dennesen ESL tweets, Dayton-Wright ESL (110-3200Hz), mixed-bass Klipschorn (28-110), and giant OB using 1960's Stephens woofer (18-110); Behringer DSP. HiFi aspirations since 1956

Last edited by bentoronto; 26th August 2014 at 08:55 PM.
  Reply With Quote
Old 28th August 2014, 03:50 AM   #9
diyAudio Member
 
bolserst's Avatar
 
Join Date: Jun 2005
Location: Fort Worth, Texas
Quote:
Originally Posted by bentoronto View Post
… Back of 3-inch ESL disks has some damping paper. Not much influence from the rear the way I have them sitting draped over the Dayton-Wright boxes - as judged by the absence of change when I stuff hi-tech absorbent behind them.
I have always found placing ESLs just in front of other objects whether designed to be acoustically absorbent or not tend to result in reflections that propagate back thru the ESL and combined with the forward firing wavefront resulting in periodic peaks and valleys in the response. Does this look at all like what you saw? Can you post the data file or response graph of what you are measuring at the listening position?

I posted an example measurement showing the effect of placing a section of fiberglass damping behind an ESL here:
ESL with acoustic foam

Quote:
I'd guess. the tweeter speaker is a 14 x 4.5 inch, segmented curve. The peak distance from a chord between the ends makes the middle two drivers about .5 inch closer to my ears than the outer two.
Does the attached figure approximate your ESL array?
If so, I can simulate the polar response to see what interference effects you might expect from the array spacing and angles.

Quote:
So back to the question. Is there some way to re-think the on-axis mic response as the final word?
I don’t think there is anything different to consider when measuring curved ESL panels or arrays than measuring any other speaker. You want the anechoic response to be flat-ish and trend smoothly with horizontal off-axis angle. The final sonic result does depend, as you mentioned, to a large part on the acoustics of the room . But I have found that starting with a speaker that has a flat-ish smoothly trending anechoic response vastly increases the odds of getting a nice sounding setup.

Quote:
…measurement of a single ESL tweeter element was made just inches from the driver, just hanging over my workbench. Output seemed to sail well over the limits of my existing test gear
If you would like to understand what you ESLs tweeters are doing, I would recommend a set of 4 measurements taken at 1m:
1) one of the ESL tweeter elements suspended as much as practical away from any objects.
2) the array of 4 ESL tweeters suspended as much as practical away from any objects.
3) the ESL tweeters mounted on your DW speaker, but only one ESL tweeter element driven
4) same as 3) but with all 4 ESL tweeters driven.

Since these are tweeters(how low do you play them?), I'd say use a measurement window of 3mS for all measurements.
Attached Images
File Type: jpg ESL4_curved.jpg (21.5 KB, 106 views)
  Reply With Quote
Old 28th August 2014, 09:43 PM   #10
diyAudio Member
 
Join Date: Jan 2008
Location: Toronto and Delray Beach, FL
bolserst - your interest and efforts much appreciated. Not to mention your good advice.

Yes, that sketch is accurate. It would be good to see the geometric/acoustic modeling, when you have the spare time.

Below is a freq response at 36 inches of the tweeter array, on axis. The orange trace is with a bundle of commercial absorbent behind the array. The set-up, angles, distances, etc. may be a bit discrepant between the traces. And for sure, my ancient Radio Shack SPL meter with a large electret is not good for higher frequencies, even with the web-endorsed calibration correction plugged into REW.

Never the less, there seems to be some impact of the absorbent, albeit not with any interpretable regularity that I can discern.

All I can now declare about my free-air up-close tests of a single tweeter element (and using a gen-U-ine Dayton-Wright cosmos-class transformer and different HV bias) is that, at the time and mental state I was in, the result for these little Dennesen guys seemed amazingly flat and extended. I'll try to test some again.


About measured sound versus heard sound, I believe there are differences, but based on psycho-acoustics not mythical powers. For equivalent mic response in a certain part of the music compass, a room with more reverb in that band will sound different - including possibly the perception of loudness. For sure, the dipole adherents like their sound and probably could prove they can tell it apart in an ABX blind test, even if the freq response was matched by mic. Such an experiment is inconceivable, of course. And not sure what kind of a blind test it would be if you couldn't "match" reverb.... which is the point of the test.

Ben
Attached Images
File Type: png w and wo absorb 36 in.png (61.7 KB, 95 views)
__________________
Dennesen ESL tweets, Dayton-Wright ESL (110-3200Hz), mixed-bass Klipschorn (28-110), and giant OB using 1960's Stephens woofer (18-110); Behringer DSP. HiFi aspirations since 1956

Last edited by bentoronto; 28th August 2014 at 09:46 PM.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Curved Vs. Flat ESL panels? markusA Planars & Exotics 31 9th November 2011 04:38 AM
FS: Martin Logan curved panels from SL3 $275 homemade Swap Meet 9 8th November 2010 03:37 PM
Curved ELS panels Johan_Ch Planars & Exotics 4 2nd November 2007 08:20 AM
how to measure freq response skyraider Solid State 23 20th January 2004 07:08 AM
How to measure open loop freq response yves Solid State 6 30th July 2002 10:51 PM


New To Site? Need Help?

All times are GMT. The time now is 12:21 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2