naebad wrote:

> Yes it is the magnitude response - sorry.  I have the phase
> characteristics too - they are
>
> -(pi/2)(f/fc) and it jumps at f=fc by -pi rads.
>
> regards

The filter you seek has been discussed in comp.dsp before, for example:

http://groups.google.ch/group/comp.dsp/msg/c8aecd35b58f6353

Regards,
Andor

naebad wrote:

> Yes it is the magnitude response - sorry.  I have the phase
> characteristics too - they are
>
> -(pi/2)(f/fc) and it jumps at f=fc by -pi rads.
>
> regards

The filter you seek has been discussed in comp.dsp before, for example:

http://groups.google.ch/group/comp.dsp/msg/c8aecd35b58f6353

Regards,
Andor

Yes it is the magnitude response - sorry.  I have the phase
characteristics too - they are

-(pi/2)(f/fc) and it jumps at f=fc by -pi rads.

regards

Tam

HelpmaBoab schrieb:

> I have a transfer function of a channel which is given by
>
> abs ( cos((pi/2) . (f/fc) )
>
> where fc is a constant freq and the freq f goes from dc to 2fc.

Are you sure this isn't the magnitude response of the transfer function?

I have a transfer function of a channel which is given by

abs ( cos((pi/2) . (f/fc) )

where fc is a constant freq and the freq f goes from dc to 2fc. What would
be the best way of simulating this ie passing a known signal through this
channel.?
I suppose I coudldtake an inverse FT to get the impusle response but it has
an absolute value in it.
Maybe I could multiply in the freq domain and inverse FFT?


regards

Tam