Segmented Wire Stator ESL simulator (esl_seg_ui)
I thought I had posted this information before, but when glorocks asked about it in the ESL woofer thread, I couldn't locate it.
Sooooo, I figured it deserved it's own thread.
“esl_seg_ui” is a program written by Edo Hulsenbos for modeling segmented line source ESLs.
I think you can still download a copy at the following link:
- See Post#9 for info on additional archive file that is required.
- See Post#39 and Post#75 for English Translation version of esl_seg_ui posted by Bazukaz.
Used with knowledge of its assumptions, I have found the results to be accurate.
The most important assumptions that you need to keep in mind are:
1) ESL panel is assumed to be infinitely tall, or fully filling the floor to ceiling space. If this isn’t the case, keep the listening distance parameter no greater than the height of your ESL panel and the plotted LF results will be reasonably accurate. In practice moving further away than this will result in rolling off of the LF response, but the program does not model this.
- See Post#20 and Post#27 for more details on panel height assumption and LF response.
2) ESL panel width is assumed to be small compared to listening distance. In practice moving closer than about 10x the width of the panel, and the HF response will start to roll off or flatten, but the program does not model this.
3) Audio signal amplitude is assumed to be uniform for all frequencies. So, transform leakage inductance which affects HF bandwidth and peaking/roll-off in the response is not considered.
4) Diaphragm resonance at low frequency is not modeled.
5) It is assumed that there is no capacitive coupling between segments. In practice there is enough coupling that appreciable HF signals are coupled from segment to segment which results in a slightly rising response and poorer off-axis dispersion than the program predicts.
6) No checks are made on the entered voltages to see if they would produce ionization of air in the gap.
7) No checks are made to see if diaphragm would hit stators for entered parameters.
8) The wire spacing and number of wires are used to calculate the width of each segment. This determines the directivity of the segment, and the capacitance which directly influences the acoustic output. It is assumed that the electrostatic field is uniform between the stator wires and diaphragm no matter what the spacing. In general this will be the case if the diaphragm-stator spacing is > openings between wires.
- Some details for estimating efficiency loss with large stator openings can be found here: First time ESL Builder
The simulator was written to model symmetrically segmented ESLs driven either by ladder resistor or parallel resistor networks. Input parameters are entered along the left of the screen. The total number of wires and total width of ESL panel is shown at the bottom left corner. The upper plot shows the NF response of each of the segments. The lower plot shows the summed response at the listening positions for the various off axis angles that have been entered.
Attachment 1: Sample 3 segment ESL output with English labels pasted over the Dutch ones in the GUI screen to help you figure out what the input parameters are.
Attachment 2: Figure showing schematic for the Ladder resistor arrangement.
Attachment 3: Figure showing schematic for the Parallel resistor arrangement.
From the figures you can see that there will actually be 5 segments of wires when building a symmetrical 3 segment ESL.
- See Post#28 for more details on translating the input parameters to electrical layout for building the segmented panel .
Wonderful contribution. Thanks!
Wonderful contribution. Thanks!
Thanks,Bolserst, For this thread!
I have been searching for that program as well.
I have been just now contemplating my next panel build and hashing through the power requirements and ratio's of equalizing a panel using a 6db slope filter for five octaves.
So far I have found out that by using this method greatly reduces the power requirements and stresses on the amplifier at the higher frequency's.
Although I have not been able to put this into real world tests until I get some new panels made.
My next little panel will be using some 1/16" TIG rod so that I can try the segmentation method of the stator.
This new panel will be about 50% wider and should result in a little better low end response compared to my last panel.
The horizontal dispersion of my last little panel was at a very respectable -/+ 25 to 30 degrees before any noticeable significant loss of the higher frequency's off of the center axis.
So this will be a very interesting build.
The overall dimensions will be about 11" X 6" for the TIG wire build, and, 12 7/16" X 6" for a new wire mesh panel to compare it to.
The reason for the TIG design to be shorter is due to the size 8" X 10" X .050" thick material that is readily available (home depot).
Also to meet the requirements of fitting on my desktop while increasing the surface area to about 60% more than my original panels.
I still have one panel that should still work to compare the new ones to as well.
All of the panels ave about a 1/2" to 5/8" border.
The diagphram sizes are as follows,
Original 3 1/4" X 9 3/4"
Tig 5" X 10"
New wire mesh 4 5/8 " X 11 1/8"
The D/S has not been determined as of yet but it will be adjustable as with all of my last builds.
The one question that I do have is about the widths of each successive segment.
I have seen two methods have been used in past documentation.
One is with equal widths and another one were each next segment is double the width of the one before it.
I plan on a starting center segment of around 3/8" or 1/2" but I will have the opportunity to change these ratios at will and may even try using every rod as a separate segment too.
I would like too know your opinions on this matter if you would.
Meanwhile, I will start messing with the new program.
Thank you B!
I have been considering for quite a while building wire stator panels. As I see it, the primary obstacle for most builders is the fabrication of the frames, and more specifically, the wire separators.
I have been in contact with a dude that has a CNC facility, and if we can agree on a few parameters, we could do a group buy.
Once we know the wire type we wish to use and it's diameter, the spacers can be produced on a linear basis. (By the foot)
In this way the construction of the frames may be more easily accomplished by those who don't have access to heavy shop equipment.
Just thinking here---
As I recall, there were two parts to this file. Your link only hits one. Ideas?
Yes, it is looking for a part 2 file.
But meanwhile in my search for it, I found what looks like a very interesting paper but I believe that it is in german or something and it can be found here,
I found it !!!!
P.S. If you are using Google Chrome you just Highlight the link of the missing file part2 and right click search google and the file will down load . :) :)
I have the program working !!!! Yeeeeah !!
The other part you need is:
Place both the *.exe file I linked to in the first post and this *.rar archive in the same folder before running the *.exe self-extracting archive.
Glad to hear you got the simulator working...I thought I had posted it back in 2010 when I started the thread on the dutch full range ESL projects.
Oh well, better late than never I guess ;)
If your panels are closer to square than a line source then the response slope will be 6dB like you mention. In this case, all you need is one resistor in series with the whole panel to flatten the 6dB slope. If you desire to segment to reduce the capacitive load, then you would need to use an inductor ladder network(like Quad ESL63) between segments rather than resistors. Multiple segments driven with ladder resistor networks is mainly useful for flatten the 3dB slope of line sources.
More information on LC transmission line segmentation here:
Using the simulator you will see that off-axis response changes more smoothly if you keep segment size the same rather than increasing size. You will have to decide how sonically important this is to you.
|All times are GMT. The time now is 10:25 AM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Copyright ©1999-2016 diyAudio