EEVblog #1150 - Electrostatic Speaker Teardown

[EEVblog]

How do Electrostatic loadspeakers work?
Teardown of the BenQ Trevolo S Electrostatic bluetooth speaker.
https://amzn.to/2PVE9g1

[NiHaoMike]

I expect the material in the middle to be an electret providing the bias field.

[SeanB]

I expect the material in the middle to be an electret providing the bias field.

My thought exactly, looks like they are using a transformer for a CCFL tube in design, and using the DSP chip to preshape ( distort the waveform to almost look right after all is said and done) the drive to the class D chip scale amplifiers, and have a pretty steep high pass filter in tweeter side DSP to get the crossover to be less objectionable. Regular drivers probably have an incredible amount of bass boost, and the wings probably help as well to allow the audio wave to spread a bit before being cancelled, as they will look like a solid at low frequencies.

Do not think they get 30W of audio power, might be 30W of power use at full level output, mostly used by the "big" drivers, but the wings are probably only getting 2W each. Case material for the sides and the main body probably also is somewhat conductive, heavily loaded with carbon black, to reduce static charges being induced into the covers by the foil mesh, but high enough not to absorb too much power.

Going to be a very expensive stainless steel mesh, as you have to have a precision punch and careful alignment to punch out the units, though they probably also use a premoulded hot glue edging to keep loose ends in control. Life of those cutters is not long, the stainless steel wears them, and as they will probably only cut each wire in the same spot they need regular dressing to keep the edge. Just a regular PVC cutting die in about that size is around $3k, and uses precision pins and slide bearings as well to make an assembly that is the punch, then you put it in a press. Having a tab cut out makes it even more expensive, as the cutter and the backing blocks have to be even thicker not to distort. You also have to harden them to Rockwell 65C, then finish grind them. That job Robrenz has a lot of experience doing, and I doubt his experience is cheap, even if he gives it freely here.

[gamalot]

I think those tiny chips are MAX98357, not a AKM product.   

https://www.maximintegrated.com/en/products/analog/audio/MAX98357A.html

[johnlsenchak]

Sorry  Dave a real single  "electrostatic  speaker" cost  way more then  that  whole unit combined.  That's  100 percent  China  made "WANK"

[mjs]

The additional audio DSP chip is probably used for its Virtual Bass function, which enhances the bass perception by converting low frequencies to their even harmonics. We've used similar chip/integrated functionality with specialized electrostatics. Also the element construction looks very familiar to me, worked on similar stuff many moons ago.

[ssashton]

It would be interesting to see a distortion plot taken directly from the step-up TX secondary. I imagine the load on those class-D amps coupled with the small transformer will introduce enough distortion on their own, to negate any benefits of using electrostatic (or electret) panels. It's a real shame a big company like BenQ couldn't create a direct solid state high voltage driver for the panels, besides better performance it would have also saved the BOM cost on the TX.

The panel design looks like 6 separate segments which are in direct contact with the outer mesh via that silicone X shape in the middle. This is probably providing some damping for the panels, but also means the movement is extremely limited and somewhat like a bending radiator than a pistonic flat diaphragm. If the diaphragm was a single undisturbed segment suspended between the stator plates, it would likely be too delicate for a portable device like that, especially with folding wings. Speakers like the Quad ESL63 have a significant frame and ribbing to keep the stator panels stiff and maintain geometry (remember the stator panel will experience an equal opposite force as the diaphragm).



A similar design, but made using magnetic planar drivers would potentially be very much better and quite an interesting idea! Something like this which can work practically down to about 300Hz, but with a dedicated HF segment down the side would be awesome.

[darrellg]

Once upon a time, I was an R&D tech at a company that designed and sold electrostatic speakers. I don't know what those are, but calling them electrostatic speakers is intentionally misleading.

[IanMacdonald]

An electrostatic driver normally needs DC bias equal to or greater then the AC drive, otherwise it will act as a frequency doubler - Positive OR negative voltage pulls the plates together, zero volts does not.
-Rather like the way that lightbulbs flicker at twice the supply frequency.
I guess they could be using an electret film to polarise it instead of DC, but they are typically not strongly charged enough for speakers. For mikes, yes.