FFT?

[GTBecker]

I've done it in X86 VB - and it's really not too bad - and I am tempted to try on a -24u with its 30MHz, 128k of program space and 8k of RAM. Has anyone ventured into spectrum analysis in ZBasic?

[mikep]

Save yourself a lot of trouble and use one of the C function libraries that already exists and wrapper the native code. You can do it all from scratch but look out for all of the shortcuts people have used such as DFTs, fixed point arithmetic, dividing by powers of 2 etc.

I have only done FFTs on a SuperComputer - not quite the same until AVRs have vector math instructions - which will be never.

[twesthoff]

It would depend on what you were trying to do. If you didn't need too much resolution it would probably work.
Many years ago (hard to remember the details) I think used a DFT algorithm to decode DTMF tones on a 6502 and on a 68xx micro. Both of those were way lower power than the -24u.
Tom W

On 2/20/2014 8:04 PM, ZBasic wrote:

I've done it in X86 VB - and it's not really too bad - and I am tempted to try on a -24u with its 30MHz, 128k of program space and 8k of RAM. Has anyone ventured into spectrum analysis in ZBasic?



Tom
http://67.207.143.181/amb9 Cape Coral Ambient Stereo
http://67.207.143.181/vhf9 Cape Coral All-channel Marine VHF
http://67.207.143.181/vlf9 Cape Coral VLF <16kHz, Lightning, spherics

[GTBecker]

It appears that 512 bins of 7200Hz-sampled time domain data will fit, using Murphy McCauley's Deeth code as a template (http://web.archive.org/web/200503110956 ... VBFFT2.bas, largely expired from the net but frequently plaigerized under other credits).

I should have a progress report in a few days.