|
How TSTEP Affects a Transient Simulation
How does the value of TSTEP affect my transient simulation?
8 y- ]4 ?* }: oTSTEP is specified in the transient analysis netlist command.9 |( P/ w" w9 J8 u
.TRAN TSTEP1 TSTOP1 TSTEP2 TSTOP2 ... TSTEPN TSTOPN
! ?2 M" y, @2 R7 W5 Y ]) BThe most common usage is a single TSTEP/TSTOP pair. For example:
. t) h4 r( D9 o; l* m.TRAN 0.1ps 100ns. l2 R+ g2 i4 `0 i0 I l7 t
TSTEP has a variety of effects on the operation of HSPICE.$ F% C: G( K2 }- ~$ E# e T
1. When the RUNLVL algorithm is in use:
4 O2 R! Q* @$ s- K6 L8 W - RUNLVL "in use" is defined as HSPICE version >= 2004.09 and OPTION RUNLVL > 0.
7 S1 o/ j/ C# P For HSPICE version >= 2006.09, RUNLVL=3 by default.
! x' Q2 h, D3 [+ ?2 z$ x" q! { - The minimum timestep is determined by the RUNLVL algorithm. The RMIN option is ignored.
, {2 \0 J2 L: E' y5 `. E5 i4 Z& s/ ? - The maximum timestep has a "soft" limit of RMAX*TSTEP. The RUNLVL algorithm; ~; B0 b- B3 ~$ ]% I
allows the timestep to exceed RMAX*TSTEP as long as tighter Newton-Raphson
; k' V; `& l1 I% s tolerances are met (voltage, current, charge).The default value of RMAX is 5.5 n7 K7 m( P, O. n
- The DELMAX option can be used to override the RUNLVL algorithm's choice of3 Z" |" K. i0 O5 o
maximum timestep. When DELMAX is used, TSTEP has no impact on simulation accuracy (except2 l7 c. ^0 B1 w0 J8 I; b- {3 S! z
when used in conjunction with .OPTION INTERP, see #3 below)., w8 t1 U+ U0 [8 {
2. When RUNLVL=0 (discouraged):4 d" E; w' z. y) s
- The minimum timestep is defined as TSTEP*RMIN. The default value of RMIN is 1e-9.
' b$ [+ Q* S ? - The maximum timestep is defined as TSTEP*RMAX. The default value of RMAX is 5.3 [" }5 s6 G) V9 H+ l: F: v6 `4 m
- These limits are "hard" limits.
% F2 J( M; z8 J4 t0 ^% E - If the simulation engine attempts to choose a timestep < TSTEP*RMIN, a
7 J. p4 i- S! C7 f Timestep Too Small Error will result.
+ x1 P* c8 Q! {# Y8 y8 W! y - The simulation engine is never allowed to take a larger timestep than/ }0 W/ y5 g; g7 O. I0 ^' s
TSTEP*RMAX." h8 B3 f* j2 ?2 O. C1 u% d
Summary1 p! a+ G1 y: f+ `9 E
With the introduction of the RUNLVL algorithm, the impact of TSTEP has been
: E5 H5 x! O' o5 y! igreatly reduced to the point that it can be ignored except for niche applications
' l; _6 v/ o: W# ]( @" ?like INTERP. One should still be careful not to set TSTEP too small for long,
3 s3 r- e' H: r7 y+ f5 }low frequency simulations in which the soft timestep upper limit of RMAX*TSTEP- c j5 p$ C! N) `. I$ Q7 H9 ]
may still force more timestep solutions than is necessary which will cause0 v/ m+ V$ v ?& X0 }
excessive simulation time. |
|