P/Z, root locus & Spice Simulaiton

Hi All,

I am looking for a methodology or a way to do effective pole/zero and root locus analysis for analog blocks(both open & closed loop) effectively (no 1st order or 2nd order MOS equations).


The only way i am aware is to do operating point analysis in SPICE, extract the approx. values for all my elements in small signal model, write down the TF, then use matlab to plot root locus.

This method is cumbersome when the no. of elements grow...

Is there a viable solution with better accuracy...?

i gone to literature and came to know abt symobilc P/Z analysis tools, does any used them before? will it take spice netlist as input? where i can get it?

Let me know if i m not clear..


Please Help me...

Thanks for your time..

Since you speak about "analog blocks" I assume that you are looking for a program on block level rather than on component level. Right ?
In this case, I recommend to use one of the programs available to simulate control systems which are capable for p/z and root locus simulations.

Hi LvW,

sorry for misleading... i am looking for program which can take in my spice netlist(component level) to the P/Z analysis (either using muller algo. or symbolic P/z) plot them, also option to do plot root locus.

hope this time im clear..

please let me if i m not.

thanks,

I am trying to do something similar; I thought it would be very useful to have a root-locus tool for certain designs. My thoughts so far are to use the spectre .pz analysis with a sweep variable; you could very easily set up an ocean script that defined a pz analysis that swept one parameter, then the next, then the next, etc. The next step would be to read the pz data into MATLAB, separated by certain sweeps, and then you could look at how the poles and zeros moved as different circuit parameters were varied.

This is extremely cumbersome due to the number of simulations required. Also, I have to think of a way to "track" which pole is which. (you can imagine a design with pole A at lower frequency than pole B; A will always be listed first by .pz analysis. However, if you change a circuit parameter (like changing a pole-splitting resistor) that now makes A a higher frequency than B; .pz will list B first now. MATLAB has no way of knowing which pole is which so you couldnt plot the path with interpolated line segments any more.

If anyone has an idea of how to do this, let me know.

I will definitely read up on symbolic PZ analysis. Post if you get anywhere!