Hearing Aid Design
GN ReSound uses realistic simulation from SIMULIA to optimize product performance.
Imagine you are working in an office at an average noise level of 40 decibels (dB). Now put the loudest rock band in the world (about 130 dB and a 90 dB difference) next door and try to carry on a normal conversation. Seems impossible, but that is essentially the problem that miniaturization has created for behind-the-ear (BTE) hearing aid design engineers.
Gain Without Feedback
“When a person is wearing a hearing aid, there is a distance of only 2 to 3 mm between the microphone and the receiver inside the device,” says Morten Birkmose Sondergaard, senior acoustic engineer at GN ReSound, a provider of hearing instruments and diagnostic audiological instrumentation. “We are trying to produce up to a 90 dB gain between the two without exceeding the feedback limit.”
Going beyond the feedback limit results in the output from the receiver looping back into the microphone, and the instrument will squeal at about 100 to 145 dB, depending on receiver size and applied gain — not a pleasant sound level. This fundamental performance limit must be accounted for in every BTE design.
Just a few years ago, numerous hearing aid prototypes were physically tested in the lab, and modifications in their design and composition were made according to the results. Now, the company’s engineers have deepened, yet streamlined their testing — reducing the number of prototypes they need to build and shortening the development time cycle — by adding finite element analysis (FEA) to their R&D arsenal.
GN ReSound’s current test equipment includes a laser vibrometer, a 3-D acoustic holography robot, general electro-acoustic measuring equipment, and a shaker/exciter — to analyze different velocity/sound pressure stresses. “Before simulation, we were limited to a trial-and-error approach for all our hearing aid design and testing,” Sondergaard says. “We were essentially working with a ‘black box’ we could only measure from the outside to get information. Now, with simulation, we can look inside the black box and evaluate and alter its behavior.”
GN ReSound uses Abaqus FEA software from SIMULIA, the Dassault Systèmes brand for realistic simulation, to ensure the stability of their device designs, improve hearing aid performance, and experiment with new materials and geometries.
“In some cases it is impossible to measure or visualize certain vibro-acoustic behavior without FEA,” Sondergaard says.
Hearing Aid Modeling
Abaqus software enables GN engineers to make computer models of all the critical elements of a hearing aid. They run their models through virtual vibration and sound pressure stresses that approximate real-world conditions, assess performance, and then validate the results with laboratory tests.
To model a hearing instrument, the engineers start with a simplified geometry of the device. They then use the Pro/ENGINEER Associative Interface to automatically transfer parts and assemblies into Abaqus/CAE, which enables the definition of model attributes, meshing and results visualization. The associative interface then allows for quick, automatic updates of designs in Pro/E.
Within Abaqus, models of critical connections are a particular focus for simulation. The “shrink-fit” function in Abaqus is employed to model the important pre-tension in the part of the rubber tube that stretches over the underlying receiver sound port. Most models are composed primarily of tetrahedral elements, but other shapes are used where applicable. An average model has about 200,000 to 300,000 elements and 1 million degrees of freedom.
Acoustic resonance frequencies are of importance in a sound-amplifying device: engineers study these using an FEA modal analysis, which incorporates natural vibration frequencies and the specific vibration patterns of the structure being studied.
Another important challenge when modeling the structure of a hearing aid is to account for the air, both around and inside it, that conducts the sound — and then to analyze the interaction between the air and the unit itself. This is where the multiphysics capabilities within Abaqus come to the fore: a model of the hearing aid structure can be quickly and automatically coupled to the air, using surface-based tie constraints, without the need for matching meshes between the two.
Once a model is set up, GN engineers use four Intel Xeon CPUs, averaging two runs overnight, to assess vibration velocity, sound pressure levels, and acoustic holography in both 2-D and 3-D.
“We now have a greater understanding of the causes of instability so we can eliminate them in the early design stages,” says Sondergaard. “This leads to improved performance, and also faster development times.