Hi friends,
I want some suggestion about a problem:We know that ac response can be got from transient response at different frequencies.Now i take DFT of my transient response for a particular input signal and definitely DFT will contain the signal peak alongwith several harmonics due to nonlinearity of the system.So for a filter i want to check passband and stopband rejection.Now i pass input sinusoids at different frequencies ,the magnitude of the o/p at signal frequency for a particular amplitude input signal can give me gain.So at different frequencies the signal gain can be calculated.Suppose the system has some offset ,still can we apply the same method (i.e DFT shows some offset alongwith  common mode,that means for a 0.6V common mode i/p and o/p offset shows more than 0.6V)??
What is the implication of other frequencies(harmonics) in DFT at a particular i/p frequency??

Hi Tosei,
my doubt is that in case of a filter upto certain frequency the signal will be transmitted to o/p with gain and passed after cutoff frequency with very low gain.Now the amplitude can be large as we want to see how much signal is passed at those frequencies.Now AC analysis does it with small signal at different frequencies..

If this is a continuous time filter, why would you want to get the gain vs. frequency from a transient analysis?

It might be a good idea to clarify what you are trying to do, and then perhaps the group can help you work out an appropriate method.

Jerry

manodipan wrote on Jun 24^th, 2008, 12:51am:

You can do AC analysis at specific time point.
First run transient analysis until time you want to observe AC characteristics and save final state.
Next run AC analysis using final state of transient analysis.

Hi manodipan,

I am not familiar with spectre (I suppose you are using it), however, independent on the program used, in general you can perform an ac analysis for S/C circuits when you have transformed it into a time continuous equivalent before (using delay elements). This method works exact and reveals all properties specific to S/C circuits (periodic functions, zero´s, poles,..)

manodipan wrote on Jun 24^th, 2008, 12:51am:

As far as clock and input signal are periodic and circuit is stable(this means circuit is never oscillative), circuit always results in steady state.
Settling transient is repeated periodically in steady state.

If you want to consider input signal amplitude, apply two large signal drives(one is clock, the other is signal).

There are four possibilities in simulation.

(1) PSS(two large tones, signal frequency is swept as large.
(2) PSS(two large tones/PAC, signal frequency is fixed as large but you can observe signal frequency characteristics as small.
(3) QPSS(two large tones), signal frequency is swept as large.
(4) QPSS(two large tones/QPAC, signal frequency is fixed as large but you can observe signal frequency characteristics as small.

If you want to insist on DFT from transient analysis, explain situation assuming actual instruments.
What instruments do you use ?
Have you measured actual DUT using actual instruments ?

Do you assume real time spectrum analyzer or FFT analyzer where short time interval overlapped FFT is invoked ?