Signal windowing: difference between FPGA and PC code

[pimacchi]

Hi all.

I noticed that there is a difference between the signal window calculated on FPGA and the same window calculated on RT: the FPGA window is not normalized and this difference is documented nowhere (even if it is possible to see that the coefficients are not normalized while configuring the FPGA express vi).

I chose hamming window but there is the same issue with all windows.

I attach a simple example for labview 13. The outcome is documented by the screenshots.

Could please someone tell me why there could be this difference?

Thank you!

P

[Raff_Gross]

Hi pimacchi

The names of the functions allude to a difference in functionality:  

FPGA » "Scaled Window" (can do several windows types)

RT » "Hamming Window" (no mention of scaling, and only does one window type)

If you open the Hamming Windows VI, you can see that calls another subVI "AllCosWindow (DBL).vi" which actually calls a DLL.  One of the unused inputs to this "AllCosWindow (DBL).vi" function is the "scale?" operator (TRUE/FALSE).  By default it passes a FALSE value, but if you add a constant to the input and make it true, you will see the same data as that output from the FPGA "Scaled Window" function.  

There is actually another function "Scaled Window VI" that functions very similarly to the FPGA function you referenced: Scaled Window VI

Best,

Raffael