[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
## AW: Bessel-filter problems

**From**: |
Bernd Heinze |

**Subject**: |
AW: Bessel-filter problems |

**Date**: |
Tue, 26 Apr 2011 21:10:36 +0100 (BST) |

Hello Christian,
thanks for your reply. Unfortunately I'm not an filter expert, I played around
with Diadem and decided that a Bessel filter with said parameters gives me the
best results. I assumed if it works in Diadem it will also work in Octave.
>* But are you sure that you really want to construct such a filter? Your ratio*
>* between corner frequencies and the Nyquist frequency (2.5 GHz ???) is quite*
>* high. Throw in a filter order of 10 and you will get a filter with*
>* coefficients that are numerically ill-behaved.*
The data is taken from an oscilloscope which has a sampling rate of 5 GHz. As
my signal contains many high frequencies this sampling rate is necessary.
(Although it might be possible to reduce it by a factor of 2 or 4)
>* Constructing a butterworth filter from your data shows what I mean: The*
>* resulting coefficients range from 1 to 1e5, i.e. they differ in scale by*
>* about 17 bits.*
I assume this means that by using the filter I might get larger "errors" or
imprecissions in my signal?
Regards,
Bernd