12-02-2009 11:19 AM
Mike,
You can get 1 Hz resolution with a 5 Hz update. What I described earlier was a way to do the FFT once per second.
If you want to update the FFT 5 times per second, just do the FFT after each acquisition. You still need to accumulate 1 second of data in the circular buffer.
Suppose you have 0.2 second acquisitions numbered 1 through 5. An FFT on {1,2,3,4,5} will give the 1 Hz resolution. 0.2 seconds later you have acquisition 6. Then do FFT on {2,3,4,5,6}. Some of the data overlaps so rapid changes will not show up in the transformed result immediately, but that is a consequence of the resolution you have selected. This is called a "sliding window" calculation.
Lynn
12-02-2009 01:18 PM
Hi again,
Many thanks for your time and effort you have shown in helping me out here.
One final question as i now understand what you have said so far, but how would i create a circular buffer for the smaller samples.
Any chance you could put a small pic up showing how its done?
Again , thanks for your help
Mike
12-02-2009 03:54 PM
Mike,
I have been pretty busy today. I will try to post something tomorrow.
Lynn
12-03-2009 03:17 AM
Thanks Lynn, look forward to getting this working 🙂
Cheers
Mike
12-03-2009 04:24 AM

12-03-2009 04:56 AM
Hi, i've tried that, but most probably incorrectly as it still did not work, could you provide a quick diagram if you have time
Many thanks
Mike
12-03-2009 05:31 AM
Hi all (again!) 🙂
I think i may have got it working using the chart as suggested, below is what i have done , all seems to work fine - is this what you guys meant?
This was for 10hz and a history of 10 samples
Cheers
Mike
12-03-2009 05:44 AM - edited 12-03-2009 05:45 AM

12-03-2009 05:49 AM
hi, not sure if there is supposed to be an image attachment?
Just looking at my example too..i think i probably have to add from earliest to latest time wise not from latest to earliest as i have done?, or does it not really matter?
12-03-2009 05:51 AM
