Fourier interpolation labview FPGA

[mikeki]

Hi,

I'm trying to perform a Fourier interpolation on Labview FPGA. 

The steps are: 

- Acquire Data from a NI 1483 adapter to the PXI 7975r FPGA

- Perform a FFT of 2048 point using the FFT Express VI of  the FPGA & Math Analysis palette( 

- Perform a IFFT of 8192 point (by puting zeros in the middle)

The FFT is in Continuous input indexes / M-interval output indexes so the it output [x0, x1024] for the first ouput [x1, x1025] for the second and so on.

The IFFT is in M-interval input indexes /  Continuous output indexes so the first input is [x0, x1024, x2048, x3072, x4096, x5120, x6144, x7168] and the first output [x0, x1, x2, x3, x4, x5, x6, x7] and so on. For the input of the IFFT, the x0 corresponds to the x0 of the FFT and x7168 corresponds to the x1024 of the FFT. The rest are replace by zero.

I'm supposed to get at the end of IFFT a nice gaussian with 8192 point because my input is a gaussian with 2048 points. But I'm getting instead 8 gaussian of 1024 point each. I really don't understand why it's happening. 

I attached a pictured of the portion of FPGA VI that does this part.

Does someone has an idea ?

[Sekine108]

Can we see what your current output looks like?

[mikeki]

Here is the result. I'm supposed to have a nice gaussian of 8192-points but I'm receiving a 8 x 1024 points (8 gaussian au 1024 points each). 
Is there anything to do with applying the scaled window function in the FPGA before the FFT and the IFFT?

[Sekine108]

I am uncertain if that is the cause or not. But have you tried running it without the scaled window function to see the results? 

[mikeki]

Yes I've tried. It gives me the same think. I added the scaled windows because I thought it was a problem of spectral leakage.  Using the hanning windows it seems to resolved it but it's hard to say because the gaussian is distorted. I tried different window, the same problem is happening but this time the Gaussian is distorted whatever the window I apply

[Sekine108]

Can we see what the your output using the different types of windows? I'm interested to see how the Gaussian is distorted.