DIY ESL build/suggestions THREAD (all questions concerning build will be in here)

[thadman]

Ok well here it goes, Im making one single thread to hopefully cover all the questions I have I dont have a massive budget, but I do want big sound. ESLs seem like the ticket (not too hard to construct, ill have to buy the parts in bulk so if I screw up I can always make more, easy to experiment on).

I'd like to cover 250/300hz --> inaudibility (I may employ a super tweeter in the future)

I have purchased Roger Sanders book and read the book from cover to cover and have simultaneously searched over here and on diyaudio.com with some success, but still have some specific questions.

What thickness mylar should I use? Roger Sanders recommends 1/2 mil (1/4 mil being ok if you get the high tensile strength stuff), but it seems like a lot of people have had great success with the 3.8 micron mylar offered by ER audio.

Are there any other polyester films that provide greater strength and durability than mylar? Has anybody heard of teonex?

What conductive coating should I use, and the adv/disadv of both? I've heard of graphite and the spray stuff.

What should I use as spacers? Roger Sanders recommends using 1/16" Lexan with epoxy...but warns of putting uneven layers (hes seen 3-17mil glue thickness on the same panel 8O 8O) which can several inhibit output performance. A few members on diyaudio.com have used urethane tape provided by 3m (foam 1/16"x1/2"x36yd #06453, urethane .045"x1/2"x20yd #06382). Has anybody on here had success with using double sided tape? Is 45 mil to small for 300hz--> reproduction?

What size D/S spacing should I use? I'd like to boost sensitivity as much as possible, to lower the reliance on transformer for output (big supplier of distortion). Im incredibly worried about acoustic coupling of the air...4 high excursion 15s operating dipole next to a 7' x 12-20" diaphragm is quite worrisome.

Whats the best way to tighten the diaphragm? Heat shrink or...use some sort of mechanical tensioning system...

What type of amplifier should I incorporate? huge pro-amp (qsc, behringer) or something along the lines of adcom?

What should I look for in high quality transformers for my intended bandwidth (250/300--->20khz)? Should I consider building my own?

Should I seperate the ESL panel and the array of RS390HFs (ie build the ESL panel and incorporate the RS390HFs in an h-frame)? Should I mount the RS390HFs opposite (ie normal/inverted/normal/inverted) to lower vibrations and non-linear distortion (specifically even order)?

[joetama]

One reply for you so CA isn't the only forum....

[thadman]

Cmon guys I know you have opinions/experience...dont be afraid to share :B

Anyways after researching a little more, I thought it would be quite impractical to use the .045" urethane tape as I need to account for acoustic coupling in my d/s spacing. 70 thou is much more practical, because it takes this into account and allows for more excursion if I ever decide to delve deeper with it.

Here are a few more questions:

Does anodizing adversely effect the thickness of the perforated aluminum? Can you use support a 7' piece of perforated aluminum, or should I use 2 pieces? How should I support the stators? Whats the thinnest stator you can use before it becomes too thin?

[JonMarsh]

I think you'll have better luck with a general literature search online. There are even ESL DIY kits available. Most of the folks here are doing dynamic driver systems- in 5 years, I don't recall anyone posting about ESL's. Now that's not to say they aren't interesting, but there are some real challenges. The most robust panel design I've ever seen is the Acoustat, but they still suffer from the typical issues with directionality. Martin Logans are a bit better, but have issues about bass performance and SPL, even/especially their hybrid designs, which are like combining a diople line source with a monopole woofer; they don't have the same SPL fall off with distance, and they don't have the same excitation of room modes, which accounts for why there's a perceived lack of integration or coherency to the sound.

Quads now are made in China, but they have some very clever engineering to get around the limitations of dispersion of conventional ESL's. Resarch reviews of the ESL63 and later models for some insight- I don't feel like writing a monograph here. Nonethess, I'd rather have an Orion system than one of the recent Quad ESL designs or other commercial ESL's I've heard lately- like at CES2006. With limited SPL goals and matched with an external dipole woofer system, I suspect the current Quads could do very well, but that's a lot of money. ESL's are NOT easy to engineer, as your current research and questions show.

As I mentioned earlier this week, Plitron makes excellent transformers for ESL's using a patented design. The key is getting a high figure of merrit, which is like primary magnetizing inductance divided by leakage inductance. The book you read should explain what those terms mean, and why they're important.

Another good resource to check out is back issues of AudioXpress/Speaker Builder. They have them available on CD, and there have been a number of articles on DIY ESL's.

The thing about DIY, which is like amateur engineering at times, is sometimes you only learn things by doing your best calculation and simulation, taking a crack at it (building), evaluating, and building again.

The physics of reproduction are not that dissimilar from dynamic drivers, so you can use SL's XL spreadsheet as a guide to required surface area and excursion, keeping in mind that while the peak excursion of the panel might be a distance "X", governed by electrode spacing and required minimum spacing for maintaining blocking voltage, that will only occur in the center of the diaphragm; it doesn't move uniformly, and you'll probably need to derate effective excusion compared with a dymaic driver by 2-3X, maybe a little more. Calculations are in order; this isn't something to do by the seat of your pants, it's engineering.

[thadman]

I'd love to build a conventional dipole line array, but sadly there are several conflicting issues with such designs such as comb filtering, distortion at extremes of spectrum, and financial cost (I dont have a huge budget).

These ESLs arent going to be cheap...im expecting a few hundred dollars worth of parts but theyll probably be cheaper than woofers + tweeters + amplification for a line source and im going to appreciate knowing that I designed/built it myself.

In Roger Sanders book he mentions that the diaphragm moves uniformly across its surface in the midrange and tweeter frequencies (didnt measure lower frequencies). He measured it with an stroboscope. The excursion is homogeneous throughout the diaphragm (no peaking), except for the edge next to the spacer (roughly 1/4" on each side) which is compromised because of the area of diaphragm under pressure by the spacer. Although the diaphragm is bowed slightly to maintain extremely high surface tension, the excursion is the same. You may perceive the slight peak in the bow of the diaphragm having greater excursion than the rest, but you wouldve been mistaken as it has been observed under a stroboscope as behaving differently.

Im not sure why you're bringing up the subject of the issues with commercially designed ESLs. Im aware of their compromises and dont plan on making the same mistakes. They will be high passed quite high to alleviate displacement requirements and mated to an additional dipole linesource which will operate into the lowest octaves (in the future I may augment the bass with a linkwitz transform involving 8 or so of those Mach5 18s )

From my understanding, there isnt much I can do wrong that will compromise sound quality, only output. After reading Roger Sanders book, the engineering behind these products is quite straightforward. The only parameters I will have to worry myself will be the diaphragm/stator spacing, spacer ratio, diaphragm, spacers, stator thickness/stator hole width/stator hole alignment, open surface area, etc. Regardless of my construction experience, I should come out with a phenomenal transducer which I can improve later on, as I anticipate buying most of my materials in bulk. I dont see why building a transducer of this simplicity would be any harder to construct than say DIY ribbons, planar magnetics, or even conventional dynamic drivers.

[cotdt]

i attempted an ESL project too, but it turned out to be really expensive to do properly as i learned more about them, so instead i turned my DIY ESL panel into headphones.

[joetama]

How is the build going?