Abstract – In this master thesis the design
and implementation of an interpolation filter
for hearing-aid applications will be discussed.
The aim of the design will be to minimize the
current consumption, hardware demand and
area needed for the implementation of the
design.
Keywords – Hearing aids, interpolation
filter, sigma-delta modulator, D/A converter
Hearing aids are devices with very strict
specifications. They are attached in its owner’s ear,
therefore to make them comfortable they must be
small and light. The biggest and heaviest part of a
hearing aid is the battery. This means that if a
significant reduction of the size and weight of the
device wants to be achieved, the battery’s
dimensions and in consequence, its capacity, must
be reduced. This solution conflicts with the fact that
since the user will probably be wearing it all day,
the working time per charge should be maximized.
The design of the hearing aid will aim at the
reduction of hardware demands that will lead to a
reduction of current consumption that will allow
the battery to be smaller.
The D/A converter, which is the back-end stage of
the audio processing chain of a hearing aid can be
seen in Figure 1. It consists of an interpolation
filter, a sigma-delta modulator, a digital pulse
width modulation, a class-D output stage, a
feedback chain and an output filter. The sigma-delta
modulator is an oversampled data converter, so a
previous oversampling is needed for its correct
operation. This oversampling will be performed by
the interpolation filter.
This oversampling process will increase the
sampling frequency by the oversampling ratio
needed for the sigma delta modulator, which will
lead to an increase of the band of interest. Since the
input of the interpolation filter is discrete, its
frequency spectrum will be repeated at every
multiple of fs. Those frequency spectrum
repetitions or images, will appear in the output
band of interest of the filter, therefore they will
need to be suppressed by the interpolation filter.
The interpolation filter of this design will be
separated into four stages as it can be seen in
Figure 2. The first stages of the filter will be the
sharpest ones, and in consequence the most
hardware demanding ones since the will have to
attenuate the closest images. The last stages of the
filter will attenuate the furthest images, so filters
with less hardware requirements will be suitable.
This multistage approach will also allows the first
stages of the filter to work at lower frequency, since
the sampling frequency will be increased step by
step.
This paper will deal with the design of the first
stage of the interpolation filter since it is the most
hardware demanding one. An optimized design will
be critical regarding the overall hardware savings
of the interpolation filter.
