Hi Marcus, hi Adrian, @Adian Thank you for your suggestions, I think it's the right way to go, I will try it soon! @Marcus I am trying to do Optical Frequency Domain Reflectometry with Gnu Radio. Basically sending a periodic Frequency sweeping signal to the fibre, and receiving reflected signals from the other side of the fibre to compare the frequency difference. Due to the fibre delay, part of the measurements in each sweeping period are useless, so I drop them, and do FFT with the rest of the measurements. I think you are right, if I set start and stop for the measurement window in each sweeping period, the length might vary. I will do as you suggest, just set a "start" time, and record the following N samples in each period for FFT. Thank you very much for pointing it out, I didn't consider it : )
Best regards, Wei On Wed, 13 Nov 2024 at 10:52, Marcus Müller <mmuel...@gnuradio.org> wrote: > Hi Wei, > > as Adrian said, sounds like a good job for a block that you write > yourself! The GNU Radio > tutorials on https://tutorials.gnuradio.org explain how you can do that :) > > I come from a bit of a different perspective on this: > > What is the thing you're computing here? I ask because you have "start" > AND "stop" tags, > which suggests the amount of samples you are trying to subject to the FFT > might be > varying? (Otherwise, you'd just have a "start" tag, and would take the > next N samples > after that, N const., I'd assume?). > > Note that DFTs of vectors of different length are very hard to compare, > and you might want > to be careful with statements you make based on these. > > Can you explain what the purpose of this block would be, from a higher > level? > > Best regards, > Marcus > > On 11.11.24 15:34, Huang Wei wrote: > > Hello everyone, > > > > For my program, in every 20 ms, I need to choose 60% of data in the > stream (I marked them > > with stream tags "start" and "stop"), then do FFT and plot, and discard > the following 40%. > > Is it possible to do all of this in gnuradio with OOT blocks? > > > > I was thinking of keep M in N, but when I look at the block > introduction, I am worried > > that it may cause inaccuracy due to the changed data interval? > > I am also looking at ZMQ sink, to send truncated data to an external > Python code for FFT, > > but my sample rate is quite high, I am afraid the python code can't > handle it. > > > > Please could you guide me in the right direction to do it properly? > > > > Thank you and best regards, > > Wei > >
