Forum Discussion

Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
16 years ago

Altera FFT - output scaling. How to do this correctly?

Hi, I do not get correct scaled results if I do everything the FFT user guide recommends. I have instantiated a 1024 points FFT, with single output and burst architecture. The data width...
Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
16 years ago

Mmmmhhh I have no Mathlab at my company. But I can check this at home this evening.

Referring to the input problem. I simulated the IP-core (this time with a quad-output architecture, for the reason of performance) with my test data, I attached it in form of two text files which are required by the simulation.

Within the simulation, I verfied again that the sine amplitude is 30146 (first screenshot) and that the exponent is also -10 (second screenshot). (I also checked this in signaltapII)

The real bin# 3 (=> 3x 2.92Hz frequency resolution = about 9Hz) has a value of 0x0D3A (with single output architecture this was 0x0D38). The imaginary bin# 3 has a value of 0xC7C7 (with single output architecture this was 0xC7C9). Look at the screenshots I attached.

So the results depend on my input OR on the FFT IP-core itself and not on my processor system or my wrapper.

What could be the reason for this behavior?

P.S.: Thank KAZ for spending time on helping me.