Digital Signal Processing Reference
In-Depth Information
The variety of design alternatives offered by DASY
Lab
make one want to design a
convenient development tool for digital filters where the filter range can be set on the
screen and the filter coefficients of the Si-shaped pulse response appear at the touch of a
button.
Solution:
•
An Si-function generator is created by means of the formula module in
which the Si function can be selected “at will” at the inputs of the
module using a slider module.
•
The possibility of multiplying the Si function by the mid-frequency of
the bandpass should be provided for. The module “list” is connected
with the output of the formula module. It indicates the filter
coefficients set. These can be copied from the list on to the clipboard.
•
As it is difficult to allocate the relevant frequency range to each
Si-function at the output of the list an FFT is carried out with the
subsequent indication of the frequency domain. It can now be clearly
recognised in the frequency domain how a change set by the slider
makes itself felt (see Illustration 213).
The string of numbers of the filter coefficients must be represented in a particular form as
a “vector file”. For this purpose an
Editor
is used which is to be found under
Accessories
under
Programs
in the
Start
menu:
(1)
The list is placed in the
Clipboard
via the menu (by clicking the left mouse key on to
the list option)
Edit
and
List to Clipboard
.
(2)
Start the Editor - under
Accessories
in the Windows
Programs
overview.
(3)
Activate the menu of the
Convolution
module and then
Help
. The design of the
vector file is described here. Examine the example and note the structure of the
vector file.
(4)
Write the header of the file (see Illustration 214), copy the filter coefficients from
the clipboard, delete everything apart from the “string of numbers” and add EOF
(end of file) at the end.
(5)
Store the vector file first as a *.txt-file in a filter folder. Then change the file ending
from “txt” to “vec” in
Explorer
.
(6)
Now load this
vec-file
into the
Convolution
module and the digital filter is complete.
In designing a filter the following facts which result from the Uncertainty Principle
(
UP
) and the Sampling Principle (
SP
) are of fundamental importance:
• If you select for example a block length and sampling frequency of n = 128 for the
pulse response of the planned filter (n = 128 filter coefficients) you can at first only
select a maximum filter bandwidth of 64 Hz (more precisely, from -64 to +64) on
