Hey Michael (reply copied to CASPER mailing list) We have had some issues with the FFT overflow output recently as well. Shifting on every stage still yielded overflows at times. We have not had a chance to get to the bottom of it yet. It could be that we are actually munging our data somewhere (although other tests don't show any distortion), or we are doing something wrong during detection. If you find anything please let us know.
Regards Andrew On Mon, Nov 7, 2016 at 11:38 PM, Michael D'Cruze < michael.dcr...@postgrad.manchester.ac.uk> wrote: > Hi Andrew, > > > > Apologies for coming to you again… > > > > I noticed something worrying while playing with wider data widths through > the FFT block, as we briefly discussed last week. Even with standard data > widths, the FFT overflow signal is high both in simulation and in practice > with a weak white noise signal. I can see some discussion of this on the > mail archive but it’s from a long time ago and it looked like the problems > were solved with subsequent git commits. I’ve just run a simulation with > “0” input into the ADC and the overflow signal is high for roughly half of > the time. > > > > I’m wondering how much I should trust the overflow signal, given this > behaviour? Is this behaviour a known issue? Again, from the mail archive it > looks like the latch should be fairly trustworthy. > > > > Given the RFI situation here, I was hoping to add a SPEAD input to > indicate in the binary data whether the spectrum overflowed the FFT. > > > > Thanks > > Michael >
