FIR filter with integer coefficients

Jun 30, 2008 1 Replies
R

Hai,



Can i implement FIR filter in FPGA using fixed point number..



I referred various FIR filter implementation in FPGA all have used integer coefficients for their implementation and i am interested to know the difficulties of using fixed point numbers.I knew fixed point numbers utilize more hardware.



I understand fixed point number is scaled and rounded to get integer values from various tools(FDAtool) hence some information will lost..Ultimately this will alter the filter response(cutoff,passband,stopband)



am i correct?



if it is possible to implement in fixed point can anyone compare the accuracy of integer FIR implementation and fixed point implementation?



pls clarify ..



regards, raj



Vote
Twitter Facebook LinkedIn
A

Hi Raj, This is correct. There is well know result that relates the number of bits you use for quantization with the level of noise added to your signal (about 6dB per bit). Note that quantizing your filter coefficients will result in a completely different filter. There are implementations that are less sensitive to coefficient quantizations than others. Particularly, FIRs are better than IIRs in that sense. Check a DSP book on filters theory. You will find your answers there. Alonzo.

. I th> Hai,

Vote
Twitter Facebook LinkedIn

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required