Home > Community > Forums > Custom IC Design > effective length variation after Monte Carlo trial

Email

* Required Fields

Recipients email * (separate multiple addresses with commas)

Your name *

Your email *

Message *

Contact Us

* Required Fields
First Name *

Last Name *

Email *

Company / Institution *

Comments: *

 effective length variation after Monte Carlo trial 

Last post Tue, Feb 21 2012 1:46 PM by Andrew Beckett. 6 replies.
Started by inessadm 18 Feb 2012 09:27 AM. Topic has 6 replies and 2093 views
Page 1 of 1 (7 items)
Sort Posts:
  • Sat, Feb 18 2012 9:27 AM

    • inessadm
    • Not Ranked
    • Joined on Wed, Sep 15 2010
    • Posts 6
    • Points 105
    effective length variation after Monte Carlo trial Reply

    Hi, when I run a Monte Carlo analysis in Cadence Spectre-v 6.1.3 (process only), I'm looking for the variation effect
    on each nominal length (L effective) of the circuit's transistors.  I find only the varied process data corresponding to
    a particular run but nothing about width or length chosen for a trial !
    Is there any way to do this please ?
    Another question : what the meaning of process data as wd108, wd015, and so on ?
    Thank you.

    Filed under:
    • Post Points: 20
  • Sun, Feb 19 2012 12:33 AM

    • Quek
    • Top 10 Contributor
    • Joined on Wed, Oct 14 2009
    • Singapore, 00-SG
    • Posts 1,070
    • Points 16,280
    Re: effective length variation after Monte Carlo trial Reply
    Hi inessadm

    You can print the actual values of the statistical parameters as follows:
    a. In ADE-XL results tab, switch to "Detail" view
    b. Right mouse click at the value column and select "Print Statistical Parameters"

    I am afraid that it is a little hard to guess what "wd108" means. Perhaps you might want to look in the PDK documentation or provide more details on where they are found (in model file, in spectre netlist, etc). If it is one of the parameters that appear in the statistical section of your model file, then it is simply a parameter that the foundry uses for varying the characteristics of a device during monte carlo. Foundry customer support should be able to provide more info on this.

    Best regards
    Quek
    • Post Points: 20
  • Sun, Feb 19 2012 12:47 AM

    Re: effective length variation after Monte Carlo trial Reply

    Quek - you beat me to it!

    In addition, it would only vary the length if the length of the devices was influenced by a parameter which had a statistical variation associated with it in a statistics block (see "spectre -h montecarlo") - and I've not seen that done very commonly in PDKs. Usually it's model parameters that are varied - although it could potentially be done with some of the model parameters that affect deltaW and deltaL - if so you'd be able to annotate the weff/leff parameters (from the same right mouse button menu that Quek pointed to) for a specific point - and then you'd be able to see on the schematic what the effective length for a specific device was. However, this assumes that the effective length was actually altered by the models. Note you'd probably also have to use the Edit->Component Display form to alter what is annotated on the schematic.

    BTW, stating what PDK you're using might mean that somebody familiar with that specific PDK might be able to give some more detailed advice - but the documentation for the PDK is a good place to start.

    Regards,

    Andrew.

    • Post Points: 20
  • Mon, Feb 20 2012 5:28 AM

    • inessadm
    • Not Ranked
    • Joined on Wed, Sep 15 2010
    • Posts 6
    • Points 105
    Re: effective length variation after Monte Carlo trial Reply

    Thanks for your quick reply !
    I don't see the detail view but only the default view ! Unfortunately, the 'print statistical Parameters' on the default view and the 'result Browser' give me anything about leff or weff. Do you know where could I possibly find documentation about the way to display parameters data as leff?
    Thanks !

    best regards

    inessadm

    • Post Points: 20
  • Mon, Feb 20 2012 9:58 AM

    Re: effective length variation after Monte Carlo trial Reply

    You must be using an old version (e.g. 6.1.3 or 6.1.2) because it was renamed from Default to Detailed some time ago.

    Anyway, as I said, you won't see leff in the statistical parameters if it is not being statistically varied. 

    I was mistaken about annotating it on the schematic. The leff should be in the "output" parameters in the result browser - you can enter this expression in the calculator and then plot it (assuming  you've saved family data on the Monte Carlo options form):

    pv("/I7/M3" "leff" ?result 'output)

    Note that /I7/M3 is the hierarchical path to the device in the schematic. If I do this, I see the effective length of that transistor versus the iteration number. In my case it doesn't vary because it's not a statistical parameter...

    Andrew.

    • Post Points: 20
  • Tue, Feb 21 2012 9:31 AM

    • inessadm
    • Not Ranked
    • Joined on Wed, Sep 15 2010
    • Posts 6
    • Points 105
    Re: effective length variation after Monte Carlo trial Reply

    I've got only "dc operating points" or "design parameters" folders but anything about "outputs parameters" in my result browser ! I have to add specific option to view it ?

    inessadm 

    • Post Points: 20
  • Tue, Feb 21 2012 1:46 PM

    Re: effective length variation after Monte Carlo trial Reply

    Double click on the test name in ADE XL to bring up the Test Editor, and then do Outputs->Save All. I suspect that "Save output parameters info" has been turned off.

    That said, this is extremely unlikely to help you, because the chances are the length is not being altered by the statistical models.

    Andrew.

    • Post Points: 5
Page 1 of 1 (7 items)
Sort Posts:
Started by inessadm at 18 Feb 2012 09:27 AM. Topic has 6 replies.