Refining my ESL design
I'm refining my ESL speaker design...
The original beam-splitter design was drawn in DeltaCad 2D and is insanely complex and time-consuming to build. Since I've gotten a some requests for drawings on my blog page (Jazzman's DIY Electrostatic Loudspeaker Page), I've decided to modify the design to make it easier to build and also render it in 3D. So last week I downloaded Google Sketchup and started the learning process (old dogs don't learn new tricks so good).
The new version is an inch slimmer, the 5mm plywood sheathing and edge trim on the woofer box are gone, replaced with just a painted MDF woofer box, or can substitute 3/4 A-A plywood if a wood finish is desired. The frame sides and woofer baffle are 3/4 red oak A-A plywood and the front panel frame is solid red oak wood.
The transmission line is now 6 inches longer, and the angles on the beam-splitter now form a sharper blade @ 50 degrees incident to the panel (was 45), such that the rearward sound striking the beam splitter is deflected 5 degrees rearward of perpendicular. I think the slimmer version is more esthetically pleasing from a frontal view and many fewer pieces and glue ups should reduce the build time by at least 1/3.
I used an Aurum Cantus Mk II woofer originally but I would like to try out its ultra-low Q big brother: AC250/75C2C
OK... the rap on really low Q woofers is their stiffer suspension rolls off the low end so you need to EQ them and feed them more power to play really low. I get that, but their lower Q also means they have less overshoot in the midbass to better match the [near-zero-Q] electrostat. Besides, I'm using subs for the heavy lifting on the low end anyhow :-)
Since my original speakers are playing fine and still sound fantastic, and money is an issue for me (ex-wives), I can't justify building a pair of the new version for myself, but at least I can now post an easier, sexier design (in 3D) on my blogpage, for anyone interested. What can I say... designing speakers is a damnable psychosis.
Here's what I've drawn up so far:
Very cool Charlie !!
You can try modeling the difference between the two drivers using this software from a fellow DIY'er,
He did a very nice job on it and many of the experts in the field of TL's say that it is pretty accurate for the most part.
I have been playing with it alot trying to get a decent design for a more compact version using my 8" subs as they have a fairly low Q as well.
I have found the there is no substitute for the low end extension and the length of the Transmission line.
My 8's have about 50% more excursion capability than the Aurum's but they are smaller.
I would like to use at least two of my drivers in one if not three but there is no provision in the program for multiple driver's as of yet.
I will ask him on his thread sometime how to go about this.
Give it a try sometime and let me know what you find.
That sim program looks pretty user friendly. I hear the one MJ King created is great too but I never was smart enough to even figure out how to use it.
BTW, I can't wait to see those new desktop segmented wire stator stats you're building.
Yes ,I tried to use MJk's as well Horn response but there were to many terms and symbols to try and remember!!
They are nearly done Charlie and I should be having the First tests within a week or so.
Thanks for all the info on ESL setups...Bias, Bias feeds an what works best,Interfaces,Panels Mylar, coatings, best you have found....all vary needed
Beautiful work, as usual. Here's a thought on the off chance it's useful. I'm unsure of the dimensions in your drawings so I'll say at the outset that I may be off base.
That said, have you thought about the wavelengths reproduced by the woofer, especially after it's been low-pass filtered by the transmission line, compared to the dimensions of some of the contours of that line? If you're really emphasizing simplicity of construction, you might consider accepting less smooth curves in your transmission line. It could simplify construction considerably, and if the T-line output is going to be limited to a couple hundred Hz and lower, I don't think dimensions on the order of inches will matter much. In other words, you could substitute right angles, or at least 45 degree angles, where you now have (admittedly more graceful looking) segmented curves. The wavelength at 300 Hz is over 1 meter so breaking bends into curves made of pieces a few inches long may not yield audible benefits.
Just a thought; feel free to disregard it if it's not useful.
I struggled a bit deciding how much simpler to make cabinet, and I'm sure you're right that the internal angles need not approximate curved surfaces to the extent that I made them. The new version is simpler by virtue of fewer components and no glued on sheathing, yet still retains the angled woofer and pseudo-curved surfaces behind the woofer, which I find appealing (even if not needed). The angle cuts are not really a problem if done with an accurate table saw. Obviously, this isn't a speaker you can build in a weekend with just a circular saw!
I pretty much followed Sanders' Cookbook guidelines for the TL (not smart enough to model the damn thing mathematically) and I like how he did the compact line in his Eros speakers. And I became wedded to the idea of approximating curved surfaces immediately behind the woofer, so as to mitigate any standing waves that might rebound to the woofer. In any case, the configuration gives a quite seamless blend between the woofer and stat panel... actually astoundingly better than had I anticipated. Even though I can't really afford to, I'm not sure I can resist building a pair of this new version.
There is still much room to simplify and I will probably draw up yet another version, retaining the beam-splitter and TL but more focused on simplifying the build... with a straight front baffle/frame (verses the current version where the woofer is angled upward 8 degrees).
Hey, I'm still waiting to see those wire-stator ESL's you've been working on!
A thought occurred to me while pondering wire stators. I've read about segmented wire stators configured with inline resistors such that the center wires drive the treble and the outer wires drive progressively lower frequencies... in order to give wider dispersion and a wider sweet spot. I think it would be cool to make the setup switchable so you could just flip a switch turn the segmentation on and off. OFF for big-time slam when listening solo from the focal point, or ON when you have company over and want a wider sweet spot. Probably someone has already done that?? Just wondering...
It sounds like you have good reasons for your approach and since I won't be the one footing the bill, I selfishly hope you do build the new set---just so I can enjoy seeing the result.
The wire stator ESLs have been on hold---in fact LOTS of stuff has been on hold---while my wife and I finish up the new house we had built (photo below). I'm now painting the shop so I'm getting close to being back up an at 'em. In fact, I picked up an old used metal working lathe and a mill-drill to expand my capabilities.
In the meantime I've been slowly working through planar magnetic midrange-tweeter designs because I think my ESL panels may be too large when joined by the vertical array of woofers. I haven't abandoned the ESLs, though. They're carefully packed and I walk by them every day I go through the garage. They remind me that there is yet another project that I have started and haven't yet finished!
I haven't seen an implementation of the selectable segmentation of the sort you described. I had a thought that's a variation on that, but have never pursued it. Some complain that singers end up sounding larger than life when reproduced through large panels. I wondered whether it might be possible to have a selectable effective panel size in the vertical direction so that the listener could control the perceived image size. I think some sort of delay or at least low-pass filtering would be necessary. Probably something along the lines of the Quad ESL 63 but only along the vertical axis (unless you're simultaneously trying to broaden the horizontal distribution of high frequencies as well).
Since equalization is in the mix and since you will be crossing over from the panels to the woofer section high enough probably, have you considered an open baffle type of implementation for the woofers? Wouldn't it be desirable to have the dipole characteristic down to where the subwoofers take over?
|All times are GMT. The time now is 08:19 PM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Copyright ©1999-2016 diyAudio