Home > Community > Forums > Custom IC Design > zero diagonal found in Jacobian

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: *

 zero diagonal found in Jacobian 

Last post Mon, Dec 3 2012 3:27 PM by Quek. 3 replies.
Started by isazulkc 25 Oct 2012 12:17 PM. Topic has 3 replies and 2925 views
Page 1 of 1 (4 items)
Sort Posts:
  • Thu, Oct 25 2012 12:17 PM

    • isazulkc
    • Top 200 Contributor
    • Joined on Thu, Nov 27 2008
    • Quebec, Quebec
    • Posts 40
    • Points 690
    zero diagonal found in Jacobian Reply

    Hi

     I have a general question about the warning "Zero diagonal found in Jacobian"

    I use spectre(7.2) to simulate some designs and I get sometimes this warning  "Zero diagonal found in Jacobian"  (as in the example below), even if the simulation converges and the results seems correct.

    what's the effect of a " Zero diagonal found in Jacobian" on the simulation results ?

    Is it possible that Jacobian reordering lead to convergence problem or error/innacuracy in the results ?

     Thanks for your help,

    Best Regards !

    **************************************************
    Stability Analysis `stb': freq = (1 mHz -> 20 GHz)
    **************************************************
    Zero diagonal found in Jacobian at `I4.M0:int_s' and `I4.M0:int_s'.
    Reordering Jacobian.
    Zero diagonal found in Jacobian at `I4.I89.M13:int_s' and `I4.I89.M13:int_s'.
    Reordering Jacobian.
    Trying `homotopy = gmin'.

    Maximum value achieved for any signal of each quantity:
    V: V(net176) = 3.3 V
    I: I(V13:p) = 4.268 mA
        stb: freq = 2.154 mHz   (2.51 %), step = 159.2 uHz    (251 m%)
        stb: freq = 10 mHz      (7.52 %), step = 738.8 uHz    (251 m%)
        stb: freq = 46.42 mHz   (12.5 %), step = 3.429 mHz    (251 m%)
        stb: freq = 215.4 mHz   (17.5 %), step = 15.92 mHz    (251 m%)
        stb: freq = 1 Hz        (22.6 %), step = 73.88 mHz    (251 m%)
        stb: freq = 4.642 Hz    (27.6 %), step = 342.9 mHz    (251 m%)
        stb: freq = 21.54 Hz    (32.6 %), step = 1.592 Hz     (251 m%)
        stb: freq = 100 Hz      (37.6 %), step = 7.388 Hz     (251 m%)
        stb: freq = 464.2 Hz    (42.6 %), step = 34.29 Hz     (251 m%)
        stb: freq = 2.154 kHz   (47.6 %), step = 159.2 Hz     (251 m%)
        stb: freq = 10 kHz      (52.6 %), step = 738.8 Hz     (251 m%)
        stb: freq = 46.42 kHz   (57.6 %), step = 3.429 kHz    (251 m%)
        stb: freq = 215.4 kHz   (62.7 %), step = 15.92 kHz    (251 m%)
        stb: freq = 1 MHz       (67.7 %), step = 73.88 kHz    (251 m%)
        stb: freq = 4.642 MHz   (72.7 %), step = 342.9 kHz    (251 m%)
        stb: freq = 21.54 MHz   (77.7 %), step = 1.592 MHz    (251 m%)
        stb: freq = 100 MHz     (82.7 %), step = 7.388 MHz    (251 m%)
        stb: freq = 464.2 MHz   (87.7 %), step = 34.29 MHz    (251 m%)
        stb: freq = 2.154 GHz   (92.7 %), step = 159.2 MHz    (251 m%)
        stb: freq = 10 GHz      (97.7 %), step = 738.8 MHz    (251 m%)
    Gain margin = 40.6864 dB at frequency = 642.509 MHz.
    Phase margin = 86.7804 Deg at frequency = 12.8495 MHz.

    • Post Points: 20
  • Tue, Nov 20 2012 3:17 AM

    • Quek
    • Top 10 Contributor
    • Joined on Wed, Oct 14 2009
    • Singapore, 00-SG
    • Posts 1,084
    • Points 16,515
    Re: zero diagonal found in Jacobian Reply

    Hi isazulkc

    It is possible that such messages might lead to inaccurate simulation results. Would you please retry the simulation using the latest version of spectre (11.1.0.445.isr11) to see if the messages still appear?

    Best regards
    Quek

    • Post Points: 20
  • Mon, Dec 3 2012 2:39 PM

    • isazulkc
    • Top 200 Contributor
    • Joined on Thu, Nov 27 2008
    • Quebec, Quebec
    • Posts 40
    • Points 690
    Re: zero diagonal found in Jacobian Reply

    Hi Quek,

    I'm using now Spectre 11.1.0.550.isr16.  With this version of Spectre there is no more jacobian warnings with Spectre simulator (at least in the ones I have done so far), but it still appear during some AMS simulations.<

    Best regards,

    KC

    • Post Points: 20
  • Mon, Dec 3 2012 3:27 PM

    • Quek
    • Top 10 Contributor
    • Joined on Wed, Oct 14 2009
    • Singapore, 00-SG
    • Posts 1,084
    • Points 16,515
    Re: zero diagonal found in Jacobian Reply

    Hi KC

    "ncsim" executable used in AMS simulations comes from INCISIVE package and not MMSIM package. MMSIM package is only for spectre/ultrasim simulations. You can also upgrade your INCISIVE package to the latest 12.1 version to see if it can resolve your problem.

    Best regards
    Quek

    • Post Points: 5
Page 1 of 1 (4 items)
Sort Posts:
Started by isazulkc at 25 Oct 2012 12:17 PM. Topic has 3 replies.