APS-PSS error: "ERROR: ZERO DIAG reordering in Matrix factor"

 Hi Tawna,

thanks for the quick reply.

 (I'm using MMSIM 10.1)

 I think I was imprecise formulating the problem:

The above warnings arised with aps but not with spectre when trying to find the dc-solution. Both simulators find a DC-operating point and start to simulate.

 I have used tstab so that the signals can settle.

After the second iteration, the pss aborts.

...

    pss: time = 4.625 ns    (72.5 %), step = 650.1 fs      (13 m%)
    pss: time = 4.876 ns    (77.5 %), step = 1.407 ps    (28.1 m%)
    pss: time = 5.128 ns    (82.6 %), step = 3.777 ps    (75.5 m%)
    pss: time = 5.376 ns    (87.5 %), step = 2.209 ps    (44.2 m%)
    pss: time = 5.625 ns    (92.5 %), step = 3.092 ps    (61.8 m%)
    pss: time = 5.876 ns    (97.5 %), step = 964.7 fs    (19.3 m%)
Conv norm = 69e+03, max dV(I1151.hold) = -2 V, took 55.62 s.


Error found by spectre during periodic steady state analysis `pss'.
    ERROR: ZERO DIAG reordering in Matrix factor
...(plenty of repetitions here)
    ERROR: ZERO DIAG reordering in Matrix factor

Analysis `pss' was terminated prematurely due to an error.

I think that there is a problem when the simulator compares the new simulation results (2nd run) with the old ones (from the 1st run).

The pss runs perfectly fine with spectre, but it's too slow. (I have to use the pss, otherwise the results will be to imprecise.)

A wonderful day,

 Matzowicz

 Hi Tawna,

thanks for the quick reply.

 (I'm using MMSIM 10.1)

 I think I was imprecise formulating the problem:

The above warnings arised with aps but not with spectre when trying to find the dc-solution. Both simulators find a DC-operating point and start to simulate.

 I have used tstab so that the signals can settle.

After the second iteration, the pss aborts.

...

    pss: time = 4.625 ns    (72.5 %), step = 650.1 fs      (13 m%)
    pss: time = 4.876 ns    (77.5 %), step = 1.407 ps    (28.1 m%)
    pss: time = 5.128 ns    (82.6 %), step = 3.777 ps    (75.5 m%)
    pss: time = 5.376 ns    (87.5 %), step = 2.209 ps    (44.2 m%)
    pss: time = 5.625 ns    (92.5 %), step = 3.092 ps    (61.8 m%)
    pss: time = 5.876 ns    (97.5 %), step = 964.7 fs    (19.3 m%)
Conv norm = 69e+03, max dV(I1151.hold) = -2 V, took 55.62 s.


Error found by spectre during periodic steady state analysis `pss'.
    ERROR: ZERO DIAG reordering in Matrix factor
...(plenty of repetitions here)
    ERROR: ZERO DIAG reordering in Matrix factor

Analysis `pss' was terminated prematurely due to an error.

I think that there is a problem when the simulator compares the new simulation results (2nd run) with the old ones (from the 1st run).

The pss runs perfectly fine with spectre, but it's too slow. (I have to use the pss, otherwise the results will be to imprecise.)

A wonderful day,

 Matzowicz