PSS sweep in ADE-XL

Dear ZoltanT,

ZoltanT said:

Is there any way to use this PSS sweep or I have to sweep the parameter globally?

I'm not sure what you are asking to be honest. Are you looking to re-use the tstab portion of a PSS analysis from ADE-L in ADE-XL corner simulations? If so, the recover command that is saved in, for example, ADE-L is only valid for that specific corner and cannot be used if you change the corner in any way. If you saved the recover file for each of a set of ADE-XL simulations at the end of a set of PSS simulations, then you can uniquely set the recover files for each corner simulation in the future if you happen to repeat them (or if they crashed for some reason).

I'm thinking I am probably not addressing your question as my guess is I don't fully understand - sorry!

Shawn

I have to sweep a parameter for every corner (for example: tuning voltage of an oscillator). If the sweep is done inside the PSS analysis (=Set up in the Choosing Analyses form), the initial transient is run only at the beginning to get a good starting point for the algorithm. This takes most of the time. If I need to sweep a variable globally to reach the same effect the simulation time increases 15x. I want to avoid that. Since this feature was built already into ADE-L, it has surprised me, that it does not work.

I am not sure how to use this recover file. If it is similar to the writefinal option of transient or DC, then I would need to run manually one simulation, save the result for every corner and parameter setting and load it in again. I hope I am wrong in this, because this does not seam a pleasant way.

Dear ZoltanT,

Thank you for your explanation - I now understand exactly what you want to accomplish!

ZoltanT said:

If the sweep is done inside the PSS analysis (=Set up in the Choosing Analyses form), the initial transient is run only at the beginning to get a good starting point for the algorithm.

I did not check ADE-XL, but I did check Assembler (essentially the new version of ADE-XL as the former is no longer actively supported) and the PSS sweep capability is supported. As a check, I created a netlist for a PSS analysis of an oscillator with a value of tstab and where I sweep the control voltage from 500 mV to 900 mV in 100 mV increments. I created a netlist for the defined test in Assmbler and following appears in the netlist.

pss  (  iclkoutp  iclkoutn  )  pss  fund=16e9  harms=9
+    errpreset=conservative  tstab=100n  saveinit=yes  oscic=lin
+    tunerange=[0.5  0.9]  tunestep=0.1  tuneparam=pvdda  annotate=status

A portion of the panel I used to create the netlist follows:

ZoltanT said:

I am not sure how to use this recover file. I

Basically, and I do this all the time for PSS/pnoise analyses, you can instruct spectre to save the circuit state (not just the circuit node values) at either some time or periodically in a simulation. The resulting file is stored in the netlist directory. The filename you select for the "recover" file and its options are in the "reuse" tab of the pss options GUI panel. I've indicated the entries that might go into the various dialog boxes both to save a restart file periodically at some list of times and, following the completion of the pss portion, where you might want to insert one of the recover file locations to indicate that pss shoud start its analysis from that file time (as opposed to starting from 0 sec).

Hence, in summary, I might suggest you try Assembler and use your ADE-L state file to generate a Maestro view (essentially the new "state" file that replaces former state files in Assembler/Explorer). Assembler works in a very similar fashion to ADE-XL. Hence, I think you will have very little trouble getting started.

Hope this helps a bit...

Shawn

Thanks! It worked with Assembler. Also thanks for the recover file idea!