John,
My experience has been that you don't need to test all the codes to
estimate the dynamic performance, e.g., SFDR. However, there is
another phenomena you might want to consider. The noise floor of
the FFT is a function of the number of points in the FFT,
FFT Noise Floor is 10 log10 (M/2). See the link for more detailed
discussions,
http://www.analog.com/en/analog-to-digital-converters/ad-converters/products/tut...I have actually found that for low-order data converters, 4-bit~6-bit,
sometimes you need to use more points for the FFT to get a clean result,
128~256 points.
Note:
This observation is just to accelerate simulation time, on the bench use as
many points as you like.
Best Regards,
Sheldon