hi tim, all casper fft's operate in continuous streaming mode. there is no way to start and stop them.
if you don't inject in a sync pulse, they continue working anyway. the data stream into the FFT in time domain, and after you stream in N samples (the first vector), you must continue to inject the next N samples, and then the next, etc, with no gaps, after a while, the data emerges from the output of the FFT in frequency domain - the first vector, then the next, then the next, without any gaps. the sync pulse(s) are used so that the blocks after the FFT know when the vectors start and stop. when you inject a sync pulse, that lets the FFT know the next sample is the first time domain sample in the vector. the output sync pulse tell you that the next sample emerging from the FFT block is the first frequency domain sample in the vector. you don't have to put in a sync pulse every FFT (in fact, that's not allowed). best wishes, dan On Fri, May 2, 2014 at 2:31 PM, Madden, Timothy J. <tmad...@aps.anl.gov>wrote: > Jack > > I have a problem then. Because in simulink, if I give ONE and only one > sync pulse, then the FFT stops after one compute. I am not sure what to do > about that, because the simulation seems to disagree with the documentation. > The reason I am trying to give ONE sync pulse in simulation is because > that is how I would like the hardware to run. But the simulation does not > run as advertised. I am probably doing something wrong. > > Also, the time for the sync pulse to come out of the FFT is longer than > FFTLen/N, where N is the number if inputs. > > > Tim > > > > ________________________________________ > From: Jack Hickish [jackhick...@gmail.com] > Sent: Friday, May 02, 2014 2:15 PM > To: Madden, Timothy J. > Cc: casper@lists.berkeley.edu > Subject: Re: [casper] FFT compute time > > Hi Tim, > > > > Also, if we supply ONE and only ONE sync pulse, should the fft block > compute > > indefinately? My simulink seems to require a series of sync pulses to get > > the FFt to work more than once. > > > > > > That's right, one and only one sync pulse is sufficient to keep the > FFT operating indefinitely on successive windows of data. In fact, > having multiple sync pulses without the correct period is much more > likely to cause trouble. (See > https://casper.berkeley.edu/memos/sync_memo_v1.pdf) > > Cheers, > Jack > >
