It seems like what you want is to determine the -10dB frequency, or
maybe -3dB would be a better choice depending on filter and number of
poles, etc..., and then set your filter for that frequency. Look at the
fft of your input (impulse) signal. You find that the magnitude is
fairly flat until it starts to roll off that first time. Kind of like:
_________________
\ /------\ ___
\/ \/ \__
| | | \/--\/-\_________
(*I hope that looks the way I meant it!)
Then you can search the force input (impulse) spectrum for the point
where the magnitude starts to roll off. I'm thinking sort of cross
platform here, in that I am trying to convey what I would do with
a hardware signal analysis, in te
rms of doing it in software. Then
you can apply a multiplier function (vs frequency) to the (each) FRF.
This is something that, at least to me, seems like it should be done
to the results in the frequency domain rather than to the raw data
in the time domain. You may want to ask about this in a signal
analysis or modal testing newsgroup. I'd like to know if there is one.
Dave Karon
Nigel Randell wrote:
>
> I am using LabView to acquire and process data from impact tests. I need to
> filter the raw data to the standard laid down in SAE J211 which specifies
> tolerances on the cut-off frequency and roll-off rate. This has been
> implemented using a Butterworth filter of the appropriate cut-off frequency
> and order.
>
> What I need is to be able to produce an attenuation vs. frequency plot of a
> given filter to compare with the standard. Does anyone know of a reasonably
> straightforward way of doing this?
>
> Thanks
>
> Nigel RANDELL
> Project Engineer
>
> Cranfield
Impact Centre
> U.K.
>
> e-mail n.k.randell@cranfield.ac.uk
> URL www.cicl.co.uk
