Segmented Wire Stator ESL simulator (esl_seg_ui) - Page 4 - 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 12th March 2012, 01:21 PM   #31
Calvin is offline Calvin  Germany
diyAudio Member
 
Calvin's Avatar
 
Join Date: Nov 2004
Location: close to Basel
Hi,

the 180k -in fact its 360k because of the series connection, as well as all the other doubled resistors which may be replaced by singles also- I would omit with. Instead IŽd use a low ohmic value on the primary side. Its more practical to tame the HF resonance on the primary side. Besides not to need to worry about getting shocked, the resistor on the primary always is kinda safety measurement for the amp in case something weird happens with the audio tranny.

jauu
Calvin
  Reply With Quote
Old 12th March 2012, 04:25 PM   #32
diyAudio Member
 
bolserst's Avatar
 
Join Date: Jun 2005
Location: Fort Worth, Texas
Hi,
Quote:
is there a link to the AES paper?
The above simulation is of course a simulation. Therefore it should be taken with a grain of salt.
The AES paper has been discussed a couple times on the forum.
AES E-Library Wide-Range Electrostatic Loudspeaker with a Zero-Free Polar Response
experiences with ESL directivity?

Yes, it is a simulation, but I have found it to match very closely to what I get in practice.
The reason for differences are, of course, based on what assumptions were made in the calculations.
A few practical matters I have mentioned before that are ignored by the calculations:
1) Diaphragm resonance behavior at LF is ignored
2) Transformer leakage inductance behavior at HF is ignored
3) Section to section capacitive coupling is ignored
4) HF loss due to the inertia of the diaphragm and/or the inertia of air in the holes in the stators is ignored.

Quote:
Two points:
a) The number of segments doesnŽt need to be so large. Typically 2-3 electrical segmentations, i.e 3-5 mechanical segments are sufficient.
As a rule of thumb IŽd use max. 3 segmentations with the segment area increasing by a factor of 4 (2times doubling).
b) the capacitance values of each segment is very small. This means high impedance values, hence high transformation faktors of the audio tranny. practical values should be at least 10 times higher.
c) whats the purpose of the 180k resistors? If they are intended for damping HF resonace and to tune HF-response the value will in praxis probabely be lower. 360k as in the example already attenuates at 20kHz with -0,8dB.
a) I agree, good results can be had with 3 – 5 segments. Using segments with increasing area as you recommend will enable you to achieve a wider flat bandwidth than if the same number of equal sized segments is used.(ie less total width). But, if your goal is to achieve constant directivity and wide dispersion for f > 10kHz, you have to use many equal sized segments. FWIW, I use equal sized segments for the first 4 sections, and then incrementally larger for the remaining segments.

b) Yeah, using the values for geraldfryjr’s very small panel may have not been the best example. For my 2m tall line sources capacitance is about 100pF per section.

c) The 180k resistors are there to flatten the rising response of the central segment. As you state, in practice the value will need to be adjusted somewhat to account for items 2) & 4) mentioned above. Note that these resistors are often implemented as a low value resistor on the primary side of the transformer where they can also provide some amount of resistance to core saturation. I do this, as does Audiostatic and Capaciti.

Quote:
The 180k -in fact its 360k because of the series connection, as well as all the other doubled resistors which may be replaced by singles also- I would omit with. Instead IŽd use a low ohmic value on the primary side. Its more practical to tame the HF resonance on the primary side. Besides not to need to worry about getting shocked, the resistor on the primary always is kinda safety measurement for the amp in case something weird happens with the audio tranny.
I agree concerning the 180K resistors feeding the first section. See c) above.
also, middle of following post.
experiences with ESL directivity?
and
experiences with ESL directivity?

For the remaining sections, you could combined and replace series connected resistors with singles for modeling purposes.
But, when building the panel you need to segment both the front and rear stators, so you need to keep both of the resistors…one connected between segments of the front stator, the other between segments of the rear stator.

Last edited by bolserst; 12th March 2012 at 04:33 PM.
  Reply With Quote
Old 30th September 2012, 08:34 AM   #33
diyAudio Member
 
Join Date: Feb 2008
Location: Jackson,michigan
I have been playing with the segmentation program again today.
And one thing that I did not understand until now was ow the groups of wires are labeled in the program.

What I did figure out was that the very first number in the list is the center group of wires and the succeeding numbers represent each next group on either side of the center group.

I had originally thought that the first number was half of the number of wires in the center group,So when I started with 2 I thought that there was 4 wires in the center group and this was throwing off my count and spacing calculations.

I am glad to say that I have it straight now!!

Now I have come up with a new pattern with half as many sections and still looks very good.

Also it was mentioned about the 6db rise in response and this is understood.

My little panel does show that it is flat when the microphone is right on the diagphram at a few cm's away.
However, I have measured the same 6 db rise at the listening position of about .5 meter as well as 1 meter and both is shown in the first chart.
So, even this is considered nearfield and does show the rise in response in this range.

The second chart shows the ESL and the woofer in fullrange use.
These measurements were done the the Dayton Measurement microphone.

The new criteria for the simulation is the same as the earlier one except for the number of segments and the resistor values.
As well as the transformation ratio has been increased along with the bias voltage to match the test setup.

At a 5Kv bias I was getting around 89db of sensitivity and increasing the bias to 7kv in the simulation nearly match's what I have measured in realtime tests.
Although I was not able to write down any data to completely confirm this as the panel had given out before hand.

But, I did confirm a 6 db increase in SPL with a 100% increase in bias voltage and a 3 db increase of SPL with just 50% increase of the bias voltage.
I did do this using several different starting bias voltage points as well.

In the simulation I used a .5 meter distance also.

An interesting thing occurred of results when I did a simulation of the panel with no segmentation, It showed that the panels response would only have a rise of 3db per octave.

My question is, Is this the compensation for a line source?
If so I have some very nice simulations with a down sloping frequency response rather than a flat frequency response that I may have to use instead.
As it will be just a change of the resistors is all.

In the new simulation and many others there is a hump at around 200hz and this is very hard to get rid of in some of the charts I have done.
It is also the staring point of the down sloping simulations ones as well.
But, In this one it is only about a .5db to .75db boost and I doubt this would be an issue especially in that range.
I am in the process of doing this build so we shall see in the testing as to how it is going to turn out.

Cheers!!!

jer
Attached Images
File Type: jpg Small esl test setup.jpg (74.7 KB, 295 views)
File Type: jpg Small esl.jpg (248.1 KB, 286 views)
File Type: jpg Small esl-woofer.jpg (249.2 KB, 280 views)
File Type: jpg ESL SEG 12.jpg (177.3 KB, 277 views)
File Type: jpg ESL SEG 12 Bump at 200Hz.jpg (155.7 KB, 270 views)
File Type: jpg ESL SEG 12 FR 200Hz-20Khz.jpg (167.4 KB, 37 views)
File Type: jpg ESL SEG 12 HF Roll off 10Khz-20Khz.jpg (150.4 KB, 32 views)

Last edited by geraldfryjr; 30th September 2012 at 08:44 AM.
  Reply With Quote
Old 30th September 2012, 08:24 PM   #34
diyAudio Member
 
bolserst's Avatar
 
Join Date: Jun 2005
Location: Fort Worth, Texas
Hello geraldfryjr,
Good to see you are still at it, working to get simulation & measurement to agree.

Quote:
And one thing that I did not understand until now was ow the groups of wires are labeled in the program. What I did figure out was that the very first number in the list is the center group of wires and the succeeding numbers represent each next group on either side of the center group.
This was the key point of possible confusion I tried to address in post #13.
Glad you got it figured out now.
Segmented Wire Stator ESL simulator (esl_seg_ui)

Quote:
An interesting thing occurred of results when I did a simulation of the panel with no segmentation, It showed that the panels response would only have a rise of 3db per octave. My question is, Is this the compensation for a line source?
Yes, an uncompensated line source has a response that rises +3dB/octave. Reducing the driven area with increasing frequency with segmentation is what flattens the response. Remember, the simulator assumes the dimensions you enter for your panel will describe an ESL that behaves as a line source, so it always uses the +3dB/oct slope for the starting(non-segmented) response. It is up to you to make sure that is the case.
You might reread post#20.
Segmented Wire Stator ESL simulator (esl_seg_ui)

When measured at 1m, your small panel will behave more like a point source than a line source over most of the frequency range with a +6dB/oct slope rather than +3dB/oct.

Quote:
In the new simulation and many others there is a hump at around 200hz and this is very hard to get rid of in some of the charts I have done.
The hump is from impedance miss-match at the end of the R-C ladder network. The AES paper had some suggestions on how to deal with it.
More information in Question 3) at bottom of post#27
Segmented Wire Stator ESL simulator (esl_seg_ui)
  Reply With Quote
Old 30th September 2012, 10:00 PM   #35
diyAudio Member
 
Join Date: Feb 2008
Location: Jackson,michigan
Thanks Bolserst it is becoming more clear to me now.

In further studying after my last post I had modeled the results from the your excel file and had found this same compensation.

The results of my earlier simulations that had this compensation and had a much flatter and extended roll off for all of the angles from 10Khz and above.
This when I started to wonder what was happening and which is the proper way to do this.

Second,
Without doing some spice simulations yet, I am finding that the very first resistor is of a much lower value and this resistor needs to be a much higher wattage than I anticipated for the higher voltages.

Realizing that this controls the overall slope compensation my idea is to eliminate this resistor and use active equalization instead and leave the rest of the resistors as normal for the panels off center equalization.

Even by using a resistor in the primary brings the value into a more workable range.
But, It will still need to be in a rather large power resistor and is a waste of precious amplifier power as well.
Not to mention the cost and size of such resistors.

I will know more of the results once I get it running.

On the last testing of the little panel the dispersion wasn't too bad and I did have to cut back the highs a bit as a form of this compensation, and, this greatly reduced the ear fatigue.

But even at the very close range the drop of highs in the 15 degree to 30 degree off center range was noticeable with some movement.

Bettering the dispersion to the 30 degree range will be a vast improvement even (especially) for nearfield listening.
Even if this means a 1 db or 1.5 db drop of SPL's in the off center range.

I believe that this will be much less of noticeable difference as the frequency response will still be flat rather than suddenly dropping off.
I am expecting that this will help for when I get up and walk around as well.

When I got up to walk around the room it still wasn't too bad.
It just lacked the crispness and sparkle of the high end is all, as they played plenty loud enough for me to hear plain as day when I went downstairs as well.

Also they definitely didn't have the beaming issues that my 8" wide panels had as I now understand why this is.
Even at a much greater distance of 12' I still felt like my head was in a vice compared to the little panels and is what made me fancy them more back in 2003 when I made them and knew nothing about hows and why's about them.

Thanks for your help !!!

jer

Last edited by geraldfryjr; 30th September 2012 at 10:03 PM.
  Reply With Quote
Old 5th October 2012, 02:39 PM   #36
Bazukaz is offline Bazukaz  Lithuania
diyAudio Member
 
Bazukaz's Avatar
 
Join Date: Nov 2005
Location: Vilnius
Send a message via Skype™ to Bazukaz
Its interesting where does this program come from.
Anybody knows the developer? If I could get the source code I could translate to English + Make save/load function.
  Reply With Quote
Old 5th October 2012, 03:40 PM   #37
diyAudio Member
 
bolserst's Avatar
 
Join Date: Jun 2005
Location: Fort Worth, Texas
Quote:
Originally Posted by Bazukaz View Post
Its interesting where does this program come from.
Anybody knows the developer? If I could get the source code I could translate to English + Make save/load function.
Refer to post#1 and #9 in this thread for downloading and extracting the program files.

It is a MATLAB based program.
All of the window layout and labels are contained in the file "esl_seg_ui.fig".
I have tried modifying it with MATLAB's GUI editor, changing all the labels to English.
But, when I save the layout file and try to running the program again with this file it crashes.
I am not familiar with creating and linking GUI interfaces with MATLAB.
If you are, I'm sure you can get it working in short order.
  Reply With Quote
Old 5th October 2012, 07:17 PM   #38
diyAudio Member
 
AcoustatAnswerMan's Avatar
 
Join Date: Feb 2011
Location: Chandler Arizona
Quote:
Originally Posted by geraldfryjr View Post
....The the only thing I am wondering is how do you determine the wattage of the resistors.
I do plan on doing sweep tests on the panels after they are built and I don't need any failures.
I will try to plot this up in circuit maker and as well as LTSpice as I am just now learning to use that program.

If this works as planned then my planned 6' (2 X 3 foot peices stacked) line source will be perfect for my current system.

Any Suggestions?

jer
As you probably know, the Acoustat Spectra series is a segmented ESL. The various sectors are fed with resistors acting against the capacitance of the panel to effect the appropriate roll-off frequencies. Those resistors are axial-lead models, available from several manufacturers (Dale/Vishay is one) rated for 5-watts at 5-kV. This rating seems to be quite adequate for the task, as I am not aware of any field-failures.
  Reply With Quote
Old 7th October 2012, 01:55 AM   #39
Bazukaz is offline Bazukaz  Lithuania
diyAudio Member
 
Bazukaz's Avatar
 
Join Date: Nov 2005
Location: Vilnius
Send a message via Skype™ to Bazukaz
Quote:
Originally Posted by bolserst View Post
Refer to post#1 and #9 in this thread for downloading and extracting the program files.

It is a MATLAB based program.
All of the window layout and labels are contained in the file "esl_seg_ui.fig".
I have tried modifying it with MATLAB's GUI editor, changing all the labels to English.
But, when I save the layout file and try to running the program again with this file it crashes.
I am not familiar with creating and linking GUI interfaces with MATLAB.
If you are, I'm sure you can get it working in short order.
Looks like its a compiled MATLAB program, so simply editing .fig file is not possible. It had to be translated with a hex editor.
Maybe I will implement external project saver/loader but that's a bigger job so no promises .
First extract all .zip files and then run self-extracting archive. I could not post here otherwise because of file size and type restriction.
Attached Files
File Type: zip esl_seg_ui.EN.part01.zip (904.7 KB, 64 views)
File Type: zip esl_seg_ui.EN.part02.zip (956.6 KB, 33 views)
File Type: zip esl_seg_ui.EN.part03.zip (957.3 KB, 36 views)
File Type: zip esl_seg_ui.EN.part04.zip (957.3 KB, 34 views)
File Type: zip esl_seg_ui.EN.part05.zip (956.4 KB, 33 views)
File Type: zip esl_seg_ui.EN.part06.zip (955.3 KB, 31 views)
File Type: zip esl_seg_ui.EN.part07.zip (751.1 KB, 32 views)
  Reply With Quote
Old 7th October 2012, 03:59 AM   #40
diyAudio Member
 
bolserst's Avatar
 
Join Date: Jun 2005
Location: Fort Worth, Texas
Your English translated version works great. Thanks!
  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
ESL stator rho Planars & Exotics 13 4th January 2008 11:37 PM
ESL stator fixing 405man Planars & Exotics 9 29th August 2007 11:19 PM
Wire Insulation for Wire stator Audio_idiot Planars & Exotics 6 25th May 2007 12:42 PM
single stator esl. tade Planars & Exotics 2 4th July 2005 06:00 PM
angles for segmented ESL? Few Planars & Exotics 21 25th February 2005 04:30 PM


New To Site? Need Help?

All times are GMT. The time now is 01:56 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