Sampling clock subharmonics in integrated FFT spectrum
Posted: Mon Aug 26, 2019 2:57 pm
Hello guys,
I have implemented an autocorrelation FFT spectrometer. The data from the ADC is passed on to the FFT core (biplex FFT core by CASPER group), then the real and the imaginary values are squared and integrated. The data keeps streaming in, and the integration happens in such a way that the Nth spectrum is added over N-1th and so on. I then get the data out as a 32 bit number, and use Pavel's BRAM writer code to get it to the PS and to the acquisition computer.
Now, when I terminate keep the input of the ADC open or terminate using 50 ohms, I find dips at fs/4, fs/4+fs/8, fs/4-fs/8, fs/4+/-fs/16 and so on. I integrate the spectrum for ~ 140 ms. Please take a look at the attached plot.
Has anyone else using the board also seen something like this ?
Thank you,
Mugundhan
I have implemented an autocorrelation FFT spectrometer. The data from the ADC is passed on to the FFT core (biplex FFT core by CASPER group), then the real and the imaginary values are squared and integrated. The data keeps streaming in, and the integration happens in such a way that the Nth spectrum is added over N-1th and so on. I then get the data out as a 32 bit number, and use Pavel's BRAM writer code to get it to the PS and to the acquisition computer.
Now, when I terminate keep the input of the ADC open or terminate using 50 ohms, I find dips at fs/4, fs/4+fs/8, fs/4-fs/8, fs/4+/-fs/16 and so on. I integrate the spectrum for ~ 140 ms. Please take a look at the attached plot.
Has anyone else using the board also seen something like this ?
Thank you,
Mugundhan