[JakeJ]

Oh, I'd much rather be an audiophile than insane. 

[Jazzman53]

Update 2/21/15:
 
Today I finished wiring up the resistor networks, got the panels installed in my old beam splitter cabs, and I'm now ready to EQ and tune the panels.  The Soviet military rotary switches I bought off Ebay have still not arrived so I can't yet switch the resistor network in and out to make the dispersion pattern selectable, as planned.  At this time I have the network in the loop for wide dispersion mode. 
 
I will have to wait until next week to start tuning the EQ'ing (spending the weekend with GF).  However, I did play the panels for a few minutes just to make sure there was no arcing problems.   And unlike the perf-metal flat panels these replaced, there is NO pronounced sweet spot with these new segmented panels.  Right now they sound bass heavy, as the EQ is still setup for my old panels which are very bright. 
 
I will update again next week after some tuning and tweaking.
For now, here are some pics:

Screenshots from the Segmented ESL Calculator Excel spreadsheet:



The new segmented panel installed in my old beam splitter transmission line cab:


View from the back showing the resistor filter network feeding the panel segments:

[Hank]

Good work, Jazz!  Interesting build.

[Jazzman53]

Update 2/25/15: 

My initial impressions were disappointing, as the treble was a bit anemic, but this past evening's tweaks got the panels sounding pretty good and the treble can be upped a bit further if needed.     

BTW; I got those lovely resistor boards from my friend and fellow ESL builder Ken Seibert.  It was his website excellent website at Ken Seibert Audio that inspired me to build these new segmented panels.
   
In addition to the filter network I was using two 1-ohm resistors on the input side of the transformers.  I took one of those out today and then did some tuning with the Behringer DEQ which got the treble response and balance where it needs to be.  I can make the panels even brighter without extra EQ'ng if I were tp jump the first resistors feedings the center-most left/right wire groups.  This is one reason I used several lower value resistors in series adding to the desired value, as opposed to a single full value resistor.

A tougher issue is that I hear the diaphragm's drum-head resonance and it's quite bothersome.  The resonance is apparently too close to the crossover frequency and it's being excited.  Tomorrow I will try increasing the crossover filter slope from 24db to 48db/octave and, if need be, raise the x-over frequency from 218 to 250Hz to mitigate.  Also contributing, I suspect, are the higher diaphragm tension (raised the resonance frequency) and the wire stators' greater open area (about 55%) provides less damping as compared to my old 40% open perf panels.

I'm sure these new panels will sound great if I can get the drum-head resonance tamped down.  And I sure hope those Soviet military rotary switches I bought on Ebay show up soon, so that I can compare the wide/narrow modes in near real-time.

[persisting1]

Hearing things that shouldn't be there can be really frustrating. You'll sleep better once you figure it out 

[Jazzman53]

I apparently over-tensioned the diaphragms and ended up with a nasty resonance at a higher frequency than I imagined. 

I downloaded a tone generator App to my Ipad and used its tiny speaker to excite the diaphragm.  I couldn't even hear the Ipad's tiny speaker until the tone sweep rose above 80Hz where it got louder as the sweep continued upscale.  Still, I was able to hear the diaphragm respond in sympathetic resonance, like a tuning fork, reaching maximum at about 115 Hz.

Having found the resonance frequency, I used my Behringer DEQ to apply a very narrow-band -8db parametric EQ centered on that frequency; which tamped it down pretty well. 

BTW, an ESL guru on the DIY Audio forum who developed the Segmented ESL Calculator I used, ran some polar response sims for me, which I've attached below.   When you're a math dummy like me, it's nice to have smart friends 

[JakeJ]

Unfortunately, Jazzman, your images are not visible (at least to me).

[Jazzman53]

Unfortunately, Jazzman, your images are not visible (at least to me).

Sorry, I pulled them off the DIY Audio site post and I don't know how to modify them to another format.

[Hank]

I can't see them either - can someone out here help with the images?

[Jazzman53]

They are PNG files, if that helps.

[Jazzman53]

And here's the DIY Audio Forum post containing the graphs:
http://www.diyaudio.com/forums/planars-exotics/267694-thinking-about-segmented-wire-stator-esl-4.html#post4184637

[Jazzman53]

Update 3/3/15:

The segmented panels are sounding pretty sweet in the wide dispersion mode... I'm anxious to hear how they sound in the narrow mode-- should have even better imaging and slam. 
 
Hopefully this weekend I will get the switches wired in to enable selecting the [direct coupled] narrow dispersion mode.  The switches aren't shown in my schematic below, but would merely jump the resistors to couple all wire groups directly to the amp/transformers. 
 
For power handling I used at least (4) 2-watt metal film resistors in series between each segment, adding to the value specified by the Excel spread sheet calculator for my panel configuration:

[Jazzman53]

Last night I installed rotary switches to select wide or narrow dispersion-- so now my new segmented ESL's are finally completed.

I have them playing in narrow dispersion mode this morning and I've not yet decided which mode I prefer, as each has advantages.

The narrow mode gives spooky real imaging and amazing slam but can be a bit intense and piercing on loud passages. And of course, the narrow mode sweet spot is so tightly focused, it's really only good for solo listening at the focal point.

Wide mode spreads out to eliminate the "head in a vice" effect and is not as in your face as the narrow mode, yet still retains quite good imaging without giving up much slam (can crank out 105 db's at 3 meters).

Switching between wide/narrow dispersion modes takes about 90 seconds to crank the volume to zero (too much voltage to switch under load), turn the selector switches on each speaker, recall the saved EQ curve on the Behringer, and then crank the volume back up.

I'm calling this project a success !

Charlie

El Cheapo Soviet military rotary switch (mode selector):



Mode switch box, resistor network and panel connections:

[a.wayne]

Charlie how is it compared to the originals you had there before ......

[Jazzman53]

Charlie how is it compared to the originals you had there before ......

In the narrow mode they sound almost indestiguishable from my old perf metal panels, as I would expect, as they have the same dimensions and equivalent efficiency.  The highs are thrilling and imaging is superb.  These panels do have a higher resonance frequency, as the diaphragms are tensioned higher and this does make an audible difference on notes near the resonance. 

These new panels are radically different in the wide dispersion mode, however.  They still have the superb clarity and transient speed that's characteristic of ESL's but are not at all fatiguing after a long listening session, as the more intense narrow mode can be.

The only real mistake I made affecting the sonics was tensioning the diaphragm so high, which made the resonance higher and forced me to cross the panel over higher than I wanted to in order to not excite the resonance.   It's not a huge deal, since I'm using a very capable mid bass woofer that plays pretty clean at the crossover frequency but I would have preferred to cross the panel in at 175-200 Hz rather than 250 Hz.

What I really like is the ability to set a wider sweet spot when friends drop by, rather than have everyone take turns in the sweet spot.   

I didn't think I would like the wide mode for solo listening but I have to say it sounds really good. 

[Jazzman53]

What are you doing for a high voltage power supply and step up transformer?

Interesting project, I hope ends up being worth the effort.


Frank Van Alstine

Sorry I missed your post earlier Frank,
Each speaker is driven thru a pair of 50VA 230v/6v toroidal step-up transformers in tandem, giving 76:1 step up.  The bias supplies are simple half-wave Cockroft/Walton --- cheap and simple.  There is a parts list and schematic on my website here:  http://jazzman-esl-page.blogspot.com/: