i am simulating hspice code for 16 bit adder but i am getting error that internal time step too small and convergence issues. how to fix this problem? i have used following commands but its not working.
.options POST=1
.options AUTOSTOP
.options INGOLD=2 DCON=1
.options GSHUNT=1e-12 RMIN=1e-15 GMIN=1e-5
.options ABSTOL=1e-9 ABSVDC=1e-4 VNTOL=1M
.options RELTOL=1e-2 RELVDC=1e-2
.options NUMDGT=8 PIVOT=13
.OPTION NOMOD
.OPTION METHOD=GEAR ITL4=100 LVLTIM=2
.OPTION RAMPTIME=10ns
Seeking for help.
It's related to your inputs and .tran setting. Try with large time pulses and large step times while you have to set .tran step time very small compared to your input period.
Hello Shikha Singh
If the recommendation of Sadegh Biabanifard doesn't help there also is a brute force method:
Add a minimum capacity (some fF) to every node. This reduces the speed of the switching slopes. The timings you get are wrong then but at least you have a chance to make it converge. (In spectre there even is an officially supported parameter called cmin that does exactly this. May be the parameter cmin is common for HSPICE too.).
I can provide a pearl script that may be useful (I use it if I have convergence trouble with ngspice).
But as I said, the result after this manipulation is deviating from the perfect simulation.
thanks ricardo for the help. i have connected load capacitance at output nodes only. with this, simulation happens but a variation in on and off period of the pulse is seen .
- Badges
- Science topic
- Similar topics
- Biological Science
- Histology
- Fixatives