In order make sound an ESL needs AC drive voltages on the stators (from the amp / step-up transformer) and a DC biasing voltage on the diaphragm (from the bias supply) for the drive voltages to react against.
If the bias supply is faulty or its wire-to-diaphragm contact is corroded, or the diaphragm's conductive coating is deteriorated/not holding charge, or there is a shorting path between the diaphragm and stator(s) draining charge off the diaphragm--- then the drive voltage has nothing to react against to move the diaphragm and make sound.
In this circumstance, and especially if the sound is merely weak and not completely gone, the tendency is to crank up the amp power way beyond what would normally be required, which can drive the step-up transformer to saturation. The amp would see this as a dead short and possibly smoke itself and/or the step-up transformer.
Replacing the diaphragms and conductive coatings is home do-able if the required diaphragm tension is known and can be reproduced. The diaphragm material is available on Ebay for about $30 and the double-sided adhesive tape is available at McMaster-Carr.
ML uses 12-micron polyester film (Mylar C) for the diaphragms, and tensions them by applying a specific amount of stretch/elongation in the lengthwise direction (only). I'm told that ML uses very high diaphragm tension.
ML tensions their diaphragms using a simple clamping jig.
The rear stator is positioned convex face up in the jig. The diaphragm film is suspended above the stator and secured to the jig with clamps at each the end. The film is then stretched to a specific elongation and hand-pressed into the adhesive tape on the stator's spars and spacers.
Because ML panels are curved, the diaphragm must be tensioned predominantly lengthwise only, as any lateral tension tends to pull the film into the rear stator. The technician hand-applies only enough lateral tension to pull out any wrinkles in the film.
This would not be difficult to DIY if one knows the amount of elongation needed to achieve design tension, and it could be done with a simple home-made jig or by even by hand alone.
Note that the stator's horizontal spacers (ML calls them "spars") are unevenly spaced on the panel, from top to bottom. This arrangement breaks up the diaphragm's fundamental ("drum-head") resonance into several softer resonance peaks at different frequencies, as opposed to equal spar spacing which would produce a much louder resonance peak at a single frequency.
Distributing the diaphragm's drum-head resonance into multiple peaks across a wider bandwidth gives the illusion of more mid-bass output, and uses the resonance energy to somewhat mitigate the dipole rolloff for a flatter response, and this also allows using a lower crossover point and/or avoid having to use a steep-sloped filter in the passive crossover. It is my understanding that ML purposely uses very high diaphragm tension (I'm guessing >200Hz resonance) for the reasons noted above.
Soundlab also uses the "distributed resonance" approach, and for their full-range designs which must play down low, there is no better option to mitigate the drum-head resonance.
I'm not a fan of this approach for hybrid ESLs, however, because the drum-head resonance is sound but not music.
A big advantage for a hybrid ESL is that the ESL panel doesn't have to play down low where the drum-head resonance occurs, and with the right setup it can be largely avoided.
I tension my diaphragms [probably] much lower than a ML does, in order to lower the drum-head resonance to around 90Hz, and then I avoid exciting it by setting the crossover frequency one octave higher, using a steep (48db/octave) digital filter.
Since I'm not then using the resonance energy to boost the mid-bass output, I offset the rolloff with additional panel area, and overlay a parametric EQ to flatten the response curve.
Back to your ML:
Since the diaphragm's tension / drum-head resonance(s) affects the frequency response, the crossover network may include one or more notch filters specifically tuned to tamp down resonant peaks, and will probably include a 6db/octave shelving filter to offset the dipole rolloff. And since the crossover filters are tailored to the diaphragm tension; matching the design tension becomes an important consideration.
In any case, it's better to tension the film high than too low-- else the diaphragm would be too-easily driven into a stator at moderate volume levels or, if very loose, pulled into a stator by the static charge alone, even with no music playing.
Matching the design diaphragm tension becomes less critical, tuning becomes much simpler, and sound quality is improved (IMHO) if using a DSP crossover with adjustable EQ's rather than fixed value passive filters