The Designer's Guide Community Forum
Analog Verification >> Analog Performance Verification >> Monte Carlo analysis interpretation in cadence.

Message started by VINAY RAO on Jan 31st, 2015, 12:22am

Title: Monte Carlo analysis interpretation in cadence.
Post by VINAY RAO on Jan 31st, 2015, 12:22am

Hi all,
I am new to monte carlo analysis features and I am unable to interpret results from that simulation in cadence. I have designed band-gap current reference and I want to know how the resultant current varies with process and mismatch. Can any one tell me why I am not getting Gaussian distribution curve for the resulted current (figure is attached in pdf for 2 different cases). These are the procedure I followed.

Set-up: I opened Montecarlo → selected, sampling method to Random - >  Number of points to 500 → save process and mismatch data → run nominal simulation → all instances are selected in specify instances - > ok.

Case 1: Temperature is not varied, only DC operating save option is selected in analysis for MC. It is shown in bottom figure in pdf, where 497 samples are concentrated at only one point. Its mean is showing to be 1.18uA where 'nom' is designed for the current of 1.050uA at 27 Deg C. Why Gaussian plot isn't appearing?

Case 2: Temperature range is selected from -10 to 100 Deg C in analysis option in ADE for the MC run. Its mean is showing as 1.1uA (top figure in pdf). Here 499 samples are concentrated at one point. Here again why Gaussian response isn't appearing and different samples are taken at which temperature (as I varied temp from -10 to 100Deg C)?

Title: Re: Monte Carlo analysis interpretation in cadence.
Post by sheldon on Jan 31st, 2015, 5:52pm


  Have you checked the models to verify that they include mismatch


Title: Re: Monte Carlo analysis interpretation in cadence.
Post by VINAY RAO on Jan 31st, 2015, 9:13pm

Hi Sheldon,
Thanks for your reply. MC model has got separate mismatch and process file and that are coded. So I checked in detail → print statistical parameters. In that I checked for few samples in which I could see it is taking different values for Vth and mobility under process and mismatch variations.  So model point of view I think it's fine. Moreover, for some runs it generates Gaussian form but for some it doesn't. Yet I am asking for few clarification regarding MC analysis.

1> How do we go about MC analysis? Whether we have to start from some definite seed, where we have to do some trial and error and find out which seed gives the best? I think in my case, it may be generating all those samples which are very close, so all the samples' response is concentrated in one point.

2> What is child analysis? Whether every general analysis such as DC, AC, Tran, SP etc. are called child analysis?  

3> How samples are taken when I have selected a child analysis. Ex: When I am varying temperature from -10 to 100 Deg C, each other samples are considered at which temperature? It is important because response is a scalar quantity and don't know those responses are w.r.t which temperature value.

4>How to plot or record responses for AC, Tran, SP etc for different samples ? Ex: I want my resulted current to be plotted w.r.t temperature along the x-axis but for different MC samples.

Thanks & regards,
Vinay Rao.

Title: Re: Monte Carlo analysis interpretation in cadence.
Post by sheldon on Feb 25th, 2015, 5:58am


  Can't replicate your results, my Bandgap shows the effect process
variation and mismatch.

1) The choice of seed shouldn't matter. Since a pseudo-random
    sequence is used, the seed is used to make sure that you
    replicate your results. That is, same seed same results.
    Choosing different seeds will result in different sequences
    and slightly different results which is why confidence intervals
    are supported. They show the range of values for your
    results: yield, mean, std, based on the number of runs, ...

2) Not sure where the term, child analysis comes from.
   All the analysis defined are run a for each seed so the results
   are consistent.

3) Again all the analysis are run as a set. If the seeds are not
   same, then the results would be inconsistent rendering them

4) This requirement is well understood and the simulation      
    environment can even manage, re-use the same seed, when
    simulating with trimming/calibration.  

5) BTW, the recommended sampling method is LDS.


Title: Re: Monte Carlo analysis interpretation in cadence.
Post by weber8722 on Apr 2nd, 2015, 3:33am


a good bg should deliver almost Gaussian histograms! Best create some statistical corners in ADEXL and debug them in detail to understand what your circuit is doing.

Also run e.g. a simple MC for a very simple circuit like pure Resistor or current mirror.

Bye Stephan

The Designer's Guide Community Forum » Powered by YaBB 2.2.2!
YaBB © 2000-2008. All Rights Reserved.