Arioso Systems is a Dresden-based mems loudspeaker spin-out from Fraunhofer-IPMS, which recently closed a venture capital seed round.

It is aiming its silicon loudspeakers at ear buds.

They are made by a combination of machining and stacking silicon-on-insulator wafers using a CMOS-compatible process.

How do they work?

Electronics Weekly got in touch with Arioso to find out.

And the answer is multi-layered, as there is a lot going on.

To the outside world, the proof-of-concept speaker is a silicon die with a 6 x 1.5mm rectangular grid of rows of slots in its upper surface, from which sound emerges.

These lead to a flat rectangular chamber ~77μm high inside the chip.

Stretching from end to end of this chamber are pairs of long tall wavy multi-layered ribbons of silicon standing up on edge (shown green in the diagram above). These are stiff springy (rather than floppy) ribbons, suspended between ceiling and floor at a few points along their lengths.

The above diagram is an odd view – an up-side-down view along the cavity looking under its roof though a transparent floor.

Just in case it helps, the same diagram is turned the right way up on the left – you are looking up, through the floor, past the ribbons and through the output slots (black) towards an ear somewhere in the distance.

In a pair, the ribbons face each other symmetrically, with a row of output slots between them.

To output a positive pressure wave, the pair of ribbons is brought together, pushing air out of the slot.

When allowed to relax again (the drive is single-ended) the ribbons spring back sucking air in through the slots and emitting a negative pressure wave towards the ear.

To push air effectively, the ribbons have to be a close fit within the cavity – which they are, being 75μm tall in the 77μm gap between floor and ceiling.

To prevent a partial vacuum forming behind the ribbons when that are actuated, which would slow movement dramatically, there are rows of slots in the floor/rear of the device to vent these back spaces – shown as white rectangles in zig-zag lines in the diagram.

Effectively, in-phase sound comes out of the front of the speaker, while out-of-phase sound exits from the back.

Drive mechanism

Arioso-MEMS-speaker-ribbon-detailWhat drives the ribbons is a C-shaped basic unit, repeated many times along each ribbon (diagram right).

Each shallow C-shaped curved section consisting of three layers of machined silicon with their ends bonded firmly together with aluminium oxide.

Being an insulator, the oxide does not electrically short the layers, allowing them to be used at three separate electrodes.

Unseen in the diagram, each layer continues to run lengthways through the oxide bond, electrically daisy-chaining each layer along the ribbon.

In operation, the centre layer is grounded to the substrate at one end of the ribbon and, at the other end of the ribbon both outer layers are connected to the same input signal.

Any potential difference (regardless of polarity) between input (outer layers/electrodes) and substrate (inner layer/electrode) will cause the outer layers to be attracted to the inner layer, with the force dependant on the potential applied.

Through the geometry of the C-shape, this attractive force curl the C slightly tighter, pulling the end oxide bonds closer together (they are not attached to the substrate) – which has the effect of pulling the whole ribbon tighter.

The same scheme would also work with a two-layer C-shape, but three layers increases the force.

Why not more than three layers?

“A configuration with more electrodes is possible, but would give only a very slight improvement in terms of maximum deflection of the actuator, which is one of the key parameters we optimised the geometry of the actuator for,” Arioso head of marketing Lutz Ehrig told Electronics Weekly.

Why does shortening the ribbon push air out?

Arioso-MEMS-speaker-quad-STo turn the ribbons into air movers, each active element consists of two ribbons, each with four C-shapes (diagram right, individual layers not visible). Only the ends are attached to the substrate, with the six intermediate oxide joints (just visible) free to move with the ribbons.

Together these form four shallow S-shapes, and it is the way these are mechanically interconnected that causes the pair of ribbons to move together (pushing air out of the black slits) when a potential is applied, and to spring back away from each other (sucking air back through the slots) when the potential is removed.

In total, 14 of these ribbon pairs are arranged in a 2 x 7 grid (diagram left) with adjacent pairs offset lengthwise to prevent the backs from colliding when relaxed.

As this leaves spare space as the end of each row, seven half-length pairs fill in the gaps – so each of the seven rows has two full length and one half-length ribbon pair.

And driving them? 

In operation, the ribbons are biased away from their rest position with a dc potential of 40V, then audio is superimposed on this as 5V ac signal. Arioso is working on suitable amplifiers.

“This technology can only unfold its true potential in combination with a suitable electronic driving circuit,” said Ehrig. “Since it is an electrostatic actuator, the electrical load seen by the amplifier is purely capacitive, which means it is essentially a short circuit for higher frequencies. Therefore all existing amplifiers for inductive loads are very inefficient. Because of that we developed a new amplifier concept which is way more efficient and will be the subject of future publications.”

Arioso mems speaker photoBeyond the proof-of-concept (p-o-c pictured right)

As well as creating amplifiers, the company is developing a bi-directional actuator

“The actuators we are discussing here only use attractive forces and can only be moved actively in one direction, therefore it is necessary to chose a working point by applying a dc voltage. The movement in the other direction is due to the elastic behaviour of the structure,” said Ehrig. “This mixing of forces results in non-linearities. To get rid of these non-linearities, a push-pull actuator that can actively be moved in both directions is necessary. We have tested this concept successfully.”

Information on this bi-directional actuator will be released in a scientific paper.

Will they be suitable for ear bud use?

“For meeting the specifications of ear buds, hearables or even hearing aids our main focus is currently the increase of sound pressure level and the development of electronics to drive such a mems speaker,” said Ehrig. “For applications including active noise control, sound pressure levels of well over 100dB are necessary. To achieve this we need to increase the volume of air the actuators are moving.”

This can be done by increasing deflection of each actuator, by increasing the number of actuators on the chip or by increasing the height of the actuators.

Increasing the deflection is only possible to certain limits as the ribbons movement is increasingly non-linear as the ribbons in a pair get close, and then they stick if they touch – known as ‘electrostatic pull-in’.

Increasing the number of actuators is limited by the die cost as the active area increases.

“Thus, increasing the height of the actuators is a very attractive solution, since the area of the chip remains the same,” said Ehrig.