Can I do PSS analysis by the AMS simulator in ADE?

No, this is  not possible. It sounds as if it doesn't even meet the suggestion I made that the digital portion would need to be static during the PSS run (it's not possible even if it was static), but it's even less likely if the digital signals are moving as then there's a hidden state.

Andrew

Dear Paulinho,

> I`m simulating a DDS which has a digital/verilog portion
> followed by a DAC. So it would have been good to check
> the freq spectrum of the analog output signal using PSS
> analysis.

To be honest, my first thought is that a PSS analysis in this case is not warranted and may make a difficult problem even more challenging, A PSS analysis is designed for a simulation that requires a significant number of periods of some fundamental frequency to reach a periodic solution. In this case, the instantaneous frequency will be constantly varying to create the desired average frequency.

Depending on the length of the digital sequence required to generate the desired synthesized frequency, I think the accuracy required to distinguish between the desired average frequency and the instantaneous frequency as it varies with time would pose a significant PSS convergence issue.

In my experiences with designing and simulating a DAC based DDS, I have opted for analyzing the digital output steam with an ideal DAC (mathematical or veriloga based) and analyzing the output of a circuit based DAC in a circuit based simulation. In the case of the latter, I perform a transient simulation for a time period of at least one period of the digital sequence used to generate the specific frequency of interest, The simulation accuracy settings and number and interval of the saved data points is set by the frequency resolution and bandwidth of the desired frequency analysis. I then compare the spectrum obtained using the ideal DAC and the circuit based DAC to understand how the non-linearities of the circuit based DAC impact the frequency spectrum.

Perhaps this is all obvious to you paulinho, for which I apologize! However, I at least wanted to add my thoughts based on my experiences!

Let me know if I can provide any more information or clarify any of my thoughts.

Shawn

Dear paulinho,

I should have also added that it is best if the output transitions of the "ideal" DAC are exponentially shaped y, for example, a first order low-pass filter. Output transitions that have a linear slope (i.e., a fixed "slew rate") are by their nature non-linear and will induce distortion in the output waveform. In practice, a filter often follows a DAC to provide this function and is termed a "glitch filter".

Shawn