Tweet
Hello,
Some time ago we contacted the biomch-l
network to find poeple that are familiar with the optotrak-
system. We were having trouble with differentiating twice
using the DAP.
It turned out that the noise generated by the optotrak was
causing the problem.
Because debating the problem on the network sending the
discussions to everyone attached seemed a bit unfriendly as
not everyone migth be interested. So we now present you the
results of our research in improving the software
possibilities. We want to thank everyone who spend the time in
helping us and discussing with us.
The DAP (Data Analysis Package) is limited in its
possibilities to differentiate as it has only acces to a few
filters. With exception of the 3-point central method these
all amplified the noise; worsening the problem. Although this
method reduced the noise, it also reduced the signal amplitude
by demping it. As the signal frequency enlarges this problem
becomes bigger.
Thus the DAP is virtually useless if you want to determine the
highest acceleration acchieved by a subject who is bending its
elbow. The noise of the signal is getting the overhand and the
data you want to know disappears.
We were adviced to switch over to Matlab and use this instead
of DAP. This good advice spared us a lot of annoyance.
Before the raw data can be read in Matlab we had to convert it
into 3D-data and then into an Matlab.m-file.
Whit the knowledge that the differentiating methods provided
in DAP were not ideal we developped a method that is based on
3 steps:
First we filter out all frequencies above 50 Hz
Secondly we differentiate the remaining signal by using the
mathematical differentiating definition
Thirdly we correct for the demping of the signal caused by
this differentiating technique.
The result of this approach is that with a perfect noiseless
signal the accuracy in the range 0-50 Hz is better than
99.99%. The noise was reduced to 20% of the original.
We were very pleased with all the help we got and hope that
anyone interested or copeing with the same problem can find
anything usefull in this summary.
greetings from
Liek Voorbij and Jari Bonte
Ir. A.I.M. Voorbij (MSc)
Delft University of Technology
Fac. of Industrial Design Engineering
Dept. System and Product Ergonomics
Jaffalaan 9
2628 BX Delft
The Netherlands
Phone: +31 15 2785196
Telefax: +31 15 2787179
Some time ago we contacted the biomch-l
network to find poeple that are familiar with the optotrak-
system. We were having trouble with differentiating twice
using the DAP.
It turned out that the noise generated by the optotrak was
causing the problem.
Because debating the problem on the network sending the
discussions to everyone attached seemed a bit unfriendly as
not everyone migth be interested. So we now present you the
results of our research in improving the software
possibilities. We want to thank everyone who spend the time in
helping us and discussing with us.
The DAP (Data Analysis Package) is limited in its
possibilities to differentiate as it has only acces to a few
filters. With exception of the 3-point central method these
all amplified the noise; worsening the problem. Although this
method reduced the noise, it also reduced the signal amplitude
by demping it. As the signal frequency enlarges this problem
becomes bigger.
Thus the DAP is virtually useless if you want to determine the
highest acceleration acchieved by a subject who is bending its
elbow. The noise of the signal is getting the overhand and the
data you want to know disappears.
We were adviced to switch over to Matlab and use this instead
of DAP. This good advice spared us a lot of annoyance.
Before the raw data can be read in Matlab we had to convert it
into 3D-data and then into an Matlab.m-file.
Whit the knowledge that the differentiating methods provided
in DAP were not ideal we developped a method that is based on
3 steps:
First we filter out all frequencies above 50 Hz
Secondly we differentiate the remaining signal by using the
mathematical differentiating definition
Thirdly we correct for the demping of the signal caused by
this differentiating technique.
The result of this approach is that with a perfect noiseless
signal the accuracy in the range 0-50 Hz is better than
99.99%. The noise was reduced to 20% of the original.
We were very pleased with all the help we got and hope that
anyone interested or copeing with the same problem can find
anything usefull in this summary.
greetings from
Liek Voorbij and Jari Bonte
Ir. A.I.M. Voorbij (MSc)
Delft University of Technology
Fac. of Industrial Design Engineering
Dept. System and Product Ergonomics
Jaffalaan 9
2628 BX Delft
The Netherlands
Phone: +31 15 2785196
Telefax: +31 15 2787179