I’ll assume you understand the basic principles of ESL operation, and are interested in specifics about the CLS.
While I’ve not done a careful study or analysis of the CLS, I can share the following general points about ML ESL’s as I’ve learned them over the years.
All membrane panel speakers are a series of design compromises (as is any speaker). The most relevant are the relationship between radiating surface and volume. The bigger the surface, the louder and lower it will play. One reason why a CLS can get away without a sub a moderate volumes.
But as we increase radiating surfaces, there is a need to manage resonances on the membranes. One reason why you never see a single large panel without any sub-divisions is that the resonances would be horrible. A CLS is sub-divided into sections that are optimized for resonant frequencies for bass, mid-bass and highs. Even smaller panel ML’s have their spars spaced such that there is a relatively even distribution of resonances.
BTW- this is one reason why SoundLab uses a matrix of cells (of varying sizes), vs single large panels that are subdivided by spars.
The next physical aspect of the speaker is the tension at which the membrane is held and the electro-motive forces the electrostatic field can apply. This impacts resonances (too loose and it ‘kazoo’s’), as well as SPL abilities (too low a field strength and it can’t play loud). As you might imagine, all of the above interact a good bit.
The hardest part to get right on basic ESL construction is the even tension of the Mylar across the stators. That’s where construction tools and techniques are so critical, and why I basically recommend getting the panels from the factory. No way you or anyone can built them better without huge (I mean six or seven figures) investment.
While it might be nice to think about optimizing Spar locations, the factory designers have been around that loop a few times, and what we have is likely pretty close to optimal.
The other physical aspect we as users have some control over is the physical stability of the radiating panel. This is why there are a variety of CLS ‘Stands’ out there. The best are ones that prevent any panel movement, which to me, would necessitate a top mounted brace to either a ceiling or wall behind for the utmost rigidity.
Finally, control of the radiating pattern of the entire speaker is the final area of physical influence, and one that gets largely ignored, yet is the one that is most in our power to do something about, and the one that has the greatest influence on final sound quality. Most specifically, management of the rear wave dispersion and frequency response.
I’ve
written much about this and will not repeat it here, other than to reinforce that it is the single, most important area to focus one’s attentions on. Forget amps, forget cables, tweaks and other such stuff until you have taken care of the speaker to room interface.
OK, on to the electrical aspects.
As noted earlier, the voltages provided to maintain the strength of the electrostatic field have a big impact, but mostly on SPL ability. Using
some Variac’s I’ve played with changing these, and results are non-conclusive, but a best, slightly lower the SPL by 2db or so.
Mods to the ESL force to raise the potential might have benefits, but you could be exceeding the calculated forces used to decide where spars go and skew the sound by engaging new resonances. But it is an area to play with.
Next is the panel audio step-up transformer. This is an area that I’ve yet to experiment with, but should have audible results. See this thread:
Replacing / Upgrading Transformers
Finally, there is the crossover section to mess with, where changes to resistors, caps and inductors all will have an effect on phase, frequency and resistance. Some will improve the speaker to amp synergy, others will affect the speakers frequency response. Much has been written about passive mods, and they are topics I’m no expert on, so I’ll not comment further on passives.
I’ve specialized in active crossovers as I believe they provide much better end results, are easier to experiment with and provide a host of options, like EQ, that are much, much harder to do in passive designs.
Finally, one of the most effective ‘mods’ is to actually do time and frequency domain room correction to the entire speaker/room interface using tools such as the Audyssey room correction. In my custom designed, acoustically treated room, with highly modified active crossover ML’s, adding Audyssey from my Denon AVP-A1HD preamp was the final icing on the cake. Huge improvement.
Coffee is finally wearing off, so signing off for now.