FFT IP not working

[Isnan]

Dear All,

I did some projects using FFT IP to process 20Khz signal from two-channel 5MSPS simultaneously ADC. however, the FFT IP from my design is not working properly. The source_imag of FFT Ip is always zero. and source_real can't detect magnitude and freq. the detail of my problem is in the figure attached. Any help would be greatly appreciated
Thank you very much.

[CheePin_C_Intel]

Hi,

As I understand it, you have some inquiries related to the FFT IP. Just to check with you if you have had a chance to run a Modelsim simulation with the FFT IP. This would be helpful to narrow down any functional related problem prior to hardware testing.

Please let me know if there is any concern. Thank you.

Best regards,

Chee Pin

[Isnan]

Dear Chee Pin,

Thank you for your reply,, 

after your suggestion, I try to simulate using ModelSim, and the result is similar to real hardware, please see the figure attached.
I think the sop and eop signals are correct. I still can't understand why.
Also, I attach the full Quartus project of FFT simulation (.rar), hopefully you (or anyone) can help me to find the problem. Thank you

Sincerely,
Isnan

[CheePin_C_Intel]

Hi,

Thanks for your update. Since the hardware observation is matching the FFT, it means that the behavior of the FFT is something expected. I would suggest for you to look into your input data source to check if it only consist of real data or it is consisting of both real and imaginary.

Please let me know if there is any concern. Thank you.

[Isnan]

I am sorry I don't understand about matching the FFT, what should I do with input signal? do I need pre processing before FFT?. my input data for sink_real is directly from ADC sampling, whereas for sink_imaginer is zero.  basically I expect output that contains spike represent frequency and magnitude. however, the FFT output is still in sinusoid type.
Anyway,, I will try again with several input variations.
Thank you very much.

[CheePin_C_Intel]

Hi,

Sorry for the typo. I am actually referring to "hardware observation is matching the FFT simulation". Sorry for missing out the "simulation" which cause confusion.

To ease the debugging, you can try to use Matlab's simulation result as golden reference. Then, feed the same set of data to the FFT simulation. Generally the FFT would have almost similar behavior as the Matlab golden FFT.

Please let me know if there is any concern. Thank you

[CheePin_C_Intel]

Hi,

Just to follow up with you on this. Thank you.

[Isnan]

Hi CheePin,

yeah, still stuck to the last problem, I can't get a signal that represents real and imaginer. I think the sampling data needs some pre-processing or signal conditioning before wiring to FFT IP. btw I did a simulation using Matlab Simulink and the output is good as expected. well, I need to learn more about this FFT IP.