Is there an easy way to find the SNR of the output of a pipelined A/D with cadence?  I'm trying to avoid piping the data into matlab.

rg

Thanks for the reply Sheldon.

I gave up and went to matlab, which turned out to be less painful than I thought since I found out that you can write data directly to a file... but I would like to know how to do a calculator function for something like SINAD if it doesn't require much of your time.  I expect it will teach me something about the calculator.

I'm only interested in the noise from quantization at this time.  Right now I'm taking the "digital" output and converting it back into analog with an ideal D/A converter so I don't have to work with binary numbers.  It should be a simple matter of summing the fft of the ideal dac output minus the fundamental... but my calculations are never correct (unless I use matlab).

Just curious... when you say transient simulation for noise I assume you mean some sort of periodic steady state? or can you easily simulated thermal and 1/f noise in the transient domain.

I'm also curious how you would pull out the distortion tones for SNR.  Do you just throw away the bins that would correspond to the aliased components?

The FFT functions in the Cadence calculator suffer in two ways trying to get the SINAD.
1st it doesn't deal directly with the sampled values (which might be effected by jitter on the clock)
2nd It doesn't deal directly with the digital values.

MOST sinad programs out there handle both of these, so using the external program (MATLAB) is really probably a better choice here.
The IFFT functions needed are quite simple, and fast, and you don't need the confusion of resampling the data, nor of using continous valued data.
what might work, would be to pass the Integer data vector by skill to an external command and get the answer you want.. but you almost need a verilogA model to sample the bits after the clock and write the integer data to a file... unless you have "save ahdl variables" on.  Then, if you write your model to sample exactly the right number of values, you can take the whole wave from the ahdlvar, and reformat it for the external call. -- of course if youve done THAT you might be able to write the model to calculate the SINAD too.. (In AMS I was thinking about learning PLI So that I could just call IFFT once I had the vector. why reinvent the wheel)

(or write the function to use a starting sample number, and to pass on only the right number of samples.)
IE you don't have to pipe the data into matlab, just into IFFT.. (or SPW as I did a few years ago)  
Once the calculator provides a foolproof way to get the vector of integers, then the function can be easily created to process the data and give you SNR and SINAD.