Hi Mandana On 1 March 2011 02:59, Mandana Amiri <[email protected]> wrote:
> Hi everyone, > > I download poco (tut4) on my Roach board and supply -10dbm 250MHz signals > and when I read back the value of the status register, I have x 000f0f00, > which means fft_over and eq_clip are set. I tried changing the quant0_gain, > etc values hoping that eq_clip would be cleared, but it did not not change. > > From the looks of it, the fft and equaliser are clipping, but not the ADC inputs. Have you written anything into the fft_shift register? If you have not, then overflow will occur in the FFT (and likely in the equaliser). This is especially true with a sinusoidal input, where the FFT will add a gain of logn/2 (n = FFT size) to that channel. Try writing 0x7FF to fft_shift to enable shifting on every stage, this should solve your problem. > Could someone please explain the fft_over and eq_clip functionality and > whether I need to worry about fft_over. > fft_over indicates that an overflow occurred within the FFT (either after complex multiplication when applying twiddle factors or after the addition/subtraction in the butterfly). The overflow bit is not aligned with the data output. fft_over is something to worry about as an overflow in an FFT stage affects multiple channels depending on the stage at which it occurred. > My next question is what is the dip that I see in the middle of the auto- > and cross-correlation plots, regardless of signals applied or just > terminated inputs. > > Not so sure about this, possibly include a plot if you can. > Thanks, > Mandana > > Regards Andrew
