On Jun 29, 6:35=A0am, "bwake" <pietrof...@[EMAIL PROTECTED]
> wrote:
> Hi all.
> Let me introduce my problem to you, hoping you can give me some useful
> directions to solve it. I'm developing my bachelor thesis on a smart THD
> sensor for a power distribution line but I'm just new to digital signal
> processing and I don't know much about implementing FFTs (or, better,
DFTs=
)
> in c language. I've search extensively through the net but all the
> algorithms I've found failed to compile (using MPLAB C30) because of
error=
s
> I really don't know anything of! This is not the real problem, anyway.
> My main concern is I don't know exactly what to look for... Here in
Europe=
> the power distribution system works at a fixed frequency of 50 Hz and my
> application should be able to detect harmonics of the 50=B0 order,
accordi=
ng
> to the specifications of power quality guidelines.
> According to Microchip's datasheet, using a DSPIC33 clocked at 120 MHz,
> the device is able of 1.1 Msps ADC acquisition, so a 256 points FFT
could
> detect frequencies up to 19 kHZ if applied to this kind of signal and
this=
> is dramatically low for my application!
>
> Hoping this brief explanation is enough for you to understand my needs,
> I'll be looking forward to having directions: articles, books, internet
> resources... Help me, I want to get into programming this sensor and I
nee=
d
> knowledge!
>
> Filippo
the relation****p between sampling rate, bin resolution and FFT size
(N) is
Bin_resoution =3D Sample_Rate/N
So you pick two and get the other, if the sample rate is 10Hz and N is
10, the bin resolution is 1 Hz, ( each bin is seperated by 1 Hz, so
the bins would center on the following frequecies...0Hz, 1Hz, 2Hz, 3
Hz, 4 Hz, 5 Hz, 6Hz, 7Hz, 8Hz, 9Hz)
in the above case you need to filter out (in hardware) above 4Hz or so
to prevent aliasing (depends on how steep your filter is).
|
