CMSIS-DSP DC offset problem [Solved]

[Rashad Shubita]

Hi everyone,

I am developing a sound acquisition node with MEMS-Microphone and everything works fine except that the FFT output not as it's supposed to be! even it's work and frequencies in the input data is detected good but imag[0] != 0 why? and it seems that real[0] doesn't represent the DC offset!

Any help?

Thanks in advance!

[Rashad Shubita]

up

[Rashad Shubita]

I already use HPF with cut-off frequency ~10Hz and also try to average the PCM data before windowing&filtering but still approximately the same!

Any suggestion?

[Ozone]

A magnitude/phase representation of your data instead if re/im would be more clear.

If you use ADC input immediately, you will surely get a large DC component.

If you do any level-shifting to get your signal matching to the ADC input range, you need to correct that for the FFT.

[Rashad Shubita]

I already use magnitude representation of the PCM microphone data.

[Rashad Shubita]

up

[gregstm]

I don't use these packages (I wrote my own) , but I found the following web site useful when I was testing/validating my routines ( http://scistatcalc.blogspot.com/2013/12/fft-calculator.html ). Perhaps capture the data you put through your FFT and also plug it into the calculator (or Matlab etc. ) and compare the results - or feed both routines pseudorandom test data.

[Rashad Shubita]

no good result after all of this:( also why the other frequency pins have a value?

Any help?

[Rashad Shubita]

Solved

[DTomi.1]

What was the solution?