DSPIC33EP512GP502 and FFT issue extending FFT_LENGTH

[Ttelmah]

I'd be totally scared to be using a chip with so many problems. Especially
since some of the documented errata, seem to also trigger in situations
not matching the description. CCS have got fixes for the documented
situations, but playing with this, I've seen at least two other situations
triggering problems that don't appear on the same code in other chips.

You have chosen a very large and fast CPU, but are running it very slowly.
Now there are many other PIC's that are likely to fit in the same package.
For example if you are using a small package the PIC33CK chips. Pulling
the errata sheet for these has just 5 or 6 errata, and only one of these affects
main CPU operations.

I would not trust the fixes to work in all circumstances, and with the
current errata list would not use these chips.

[Marco27293]

My problem is to find a chip with enough RAM to perform 2048 fft in an efficient and fast way.

I'll try to find another one suitable for my application...

Meanwhile where can i find free C FFT libraries ?

[Marco27293]

My device revision is:

[Ttelmah]

Look at Kiss FFT.

[Marco27293]

Can You provide me a link for a library compatible with PCD compiler ?

I've just found Kiss fft C libraries with a lot of compiling issues...

[temtronic]

Every conversion from one 'C' to another 'C' can have (will ? ) issues...
it's the nature of the beast.

One area is data types.
Depending on the compiler the 'default' can be 8 bit unsigned or 16 bit signed or ?????

Have no idea if that's your 'compiling issue' but should be investigated.....