Forums

Re: window function-dsp

Started by Jerry Avins January 19, 2011
On Wednesday, January 19, 2011 2:25:07 AM UTC-5, rashmi venugopal wrote:
> >Bharat: > >Total harmonic distortion can be measured in many ways, but Rashmi asked > ab= > >out filter design. (A filter to suppress aliasing is needed even for your > F= > >FT approach.) > > > >Rashmi: > >Windowed-sinc design is historically interesting, but rarely the best > desig= > >n method when appropriate software is available. Windows suppress unwanted > = > >responses in the stopband, but they increase the length of the filter > neede= > >d to meet the cut-off spec. Different windows make the tradeoff to > differen= > >t extents. Nowadays, one usually supplies the specification to a program > an= > >d an complete design is produced. I don't know what tools you have. Matlab > = > >and its clones can do this, also ScopeFIR (free trial version at > http://www= > >.iowegian.com/scopefir), and several Java applets on the web, such as > http:= > >//www.dsptutor.freeuk.com/remez/RemezFIRFilterDesign.html. > > > >Jerry > > sir,, > as iam doing project in a company,, FFT has already bean implemented, but > FFT not showhing accuracy,, so i am thinking to use DFT,
Bharat has already addressed this point. The Discrete Fourier Transform requires many computations if it is calculated in a straightforward way. There is a way to calculate precisely the same results much more quickly by reordering and combining some operations. That way is called the Fast Fourier Transform. Multiply six and nine; first by adding nine sixes, and second, by consulting a multiplication table. If the results differ, at least result is wrong.
> i am thinking to design a FIR filter in window method and apply DFT to the > coeffcients Obtained applying the window,, > is this my method is correct,,
Given the modern tools now available, the window method of filter design is cumbersome and leads to filters that require more than the optimum number of taps. What do you expect the Fourier Transform of a filter's coefficients to tell you?
> and if correct, then i am confuse which window i can use to design the fir > filter so that its coefficients are passed through DFT and applied . > > whole process i need to do in LAB VIEW.
LabView has a filter-design utility: http://www.ni.com/pdf/products/us/200549868221101dlr.pdf
> KINDLY REPLY > WAITING FOR U R REPLY
Your understanding of text-message idiom is on a par with your understanding of filters and FTs. English isn't your first language. Avoid embarrassment by not getting cute with it. Jerry
Jerry,

    I agree with you, an anti-alias filter is needed even in my approach.
    I was under the assumtion that the OP has already put in AAAF + ADC
    and that the digital samples are now available for analysis.

Regards
Bharat