transient noise

hello,

1) I am running a transient simulation for a deisgn (which is not an oscillator), only cells ramping one after the other. How much do I need to choose the Noise Fmax in cadence virtuoso? any rule of thumb, such that to have a good estimation about what to expect in reality? 

2) If I run 3 different times the same simulation, will the effect of noise be the same in these 3 simulations? because in reality the effect of noise, if I will run 3 times will not be the same. In this case is multiple runs effective or noise seed? can someone please elaborate more on that?

3) can someone elaborate more on the difference between conservative, moderate and liberal in transient simulation?

Please read the Guidelines for the Custom IC Design Forum (the pinned post at the top of each forum). This asks you not to ask the same question in more than one forum - as a result, I merged your posts into one. Because the forum is based on community responses (including those of us from Cadence who answer in your spare time), you might not get an answer straightaway (or even an answer at all).

Anyway, a very quick answer to your questions:

1. The noisefmax should be set based on the bandwidth of your circuit - if you have sharp transitions, that might help set the maximum noise - it rather depends on what your circuit is doing. Of course you can do some experiments and observe the difference. There are a few resources on the Cadence support site and also on this forum which can help make some suggestions.
2. The default behaviour would be that the simulations are repeatable (i.e. give the same result). If you use the "multiple runs" feature in ADE, it runs by sweeping the noiseseed - each noise seed would give a different random result.
3. This is covered in the Spectre User Guide <SPECTREinstDir>/doc/spectreuser/spectreuser.pdf in the transient section of the analyses chapter.

Andrew