The variable PFBsize is the same as your FFT size. The number of taps determines the number of filter coefficients in the PFB, which is TotalTaps*2^PFBsize. See the wiki page for reference: https://casper.berkeley.edu/wiki/The_Polyphase_Filter_Bank_Technique#The_Math_Behind_the_PFB
On Mon, Dec 14, 2015 at 6:16 AM, Mugundhan vijayaraghavan < [email protected]> wrote: > Hello again, > > One other question: > Is there a relation between the taps of the PFB FIR and the number of > input streams to the FFT_Biplex module ? > > Size of PFB: (2?)PFBSizeThe number of channels in the PFB (this should > also be the size of the FFT which follows).Total Number of Taps:TotalTapsThe > number of taps in the PFB FIR filter. Each tap uses 2 real multiplier cores > and requires buffering the real and imaginary streams for 2*P**F**B**S**i* > *z**e* samples.Windowing FunctionWindowTypeWhich windowing function to > use (this allows trading passband ripple for steepness of rolloff, etc).Number > of Simultaneous Inputs: (2?)n_inputsThe number of parallel time samples > which are presented to the FFT core each clock. The number of output ports > are set to this same value. > > From the above tab in the blockumentation I see that *number of > simultaneous inputs* is the parallel time samples presented to the fft > core. > > Does this mean say, I have a 16384 PFB size, setting the taps as two, this > would give me 2 8192 sample streams, so now, will these 2 8192 sample > streams be the input of the subsequent FFT core ? I.e the number of inputs > must be 1 and the number of input streams must be 2 ? > > Kindly clarify ! > > Mugundhan Vijayaraghavan > Junior Research Fellow > Indian Institute of Astrophysics > -- > the giver of moksha >
