Fatal error: Timestep too small (1.25e-19) at t=0

QSPICE
1

Hi everyone,
I’m trying to run a transient simulation in QSpice, but getting “timestep too small” convergence error. I tried same netlist in LTspice and it runs perfectly fine without any errors.

I have also tried adding:
.option fastmath=0
shunts: gshunt and cshunt
.option maxstep

Even with these, QSpice either fails with the timestep error or gives unstable or incorrect results. Are there QSpice-specific solver/convergence options or netlist tweaks that typically help? Netlist content is attached below:

** NETLIST **
V1 1 2 800.0
VG1 3 0 PULSE(0 15 7.5e-6 16e-9 16e-9 35e-6)
VG2 5 8 0
RG1 4 3 2.0
RG2 G2 5 2.0
L1 1 7 {0.5 * 15.0n}
L2 2 0 {0.5 * 15.0n}
X$Q1 6 4 0 AIMBG120R010M1_L1
X$Q2 7 G2 8 AIMBG120R010M1_L1
I1 7 8 93.0
C1 7 8 1.0p
VIPROBE 8 6 0

** SIMULATOR **
.TRAN 0 50u 0 10n

** PARAMETER **
.TEMP 150

** PATH **
.inc model.lib

.OPTIONS

  • method=Gear
  • abstol=1n
  • vntol=10u
  • reltol=1m
  • itl1=200
  • itl2=200
  • itl4=25
  • trtol=7
2

Please try to increase your user level to let you upload your schematic.

3

@awais.muhammadi Well, this is not the first time I have seen Qspice have a hard time resolving an Infineon model. Infineon describes a MOSFET with many equations, making it hard to figure out what is causing the challenge. Maybe you can copy the netlist and model file to Mike for review, provided this model can run in LTspice. I am just not sure how Mike feels about this kind of model. Even if LTspice can run this netlist, it may encounter issues if the circuit is set up another way.

I modified the netlist and it can run, but it is very slow.

netlist.cir (497 Bytes)
SiC-MOSFET-AUTOMOTIVE_1200V_AIMBG120R010M1_L3.lib (18.4 KB)

image
image1361×684 52.9 KB