# how to circumvent :"Convergence problem in transient analysis"

Discussion in 'CAD' started by [email protected], Feb 9, 2006.

1. ### Guest

Hi, I'm having that famous problem. Is there any way to cirumvent it
(computing time, I have). Can anyone explain the reason for the error
? Can I go lower than the "minimum allowable step size" and what
determines the "minimum allowable step size" anyway?

Thanks,
Dan

The full error message is:
"ERROR -- Convergence problem in transient analysis at Time =
1.125E-12
Time step = 125.0E-15, minimum allowable step size =
500.0E-15"

********SPICE DECK**********
Vin 3 0 pulse (0 5 0 0.1ns 0.1ns 0.830us 4.150us)

S1=2 3 3 0 Sbreak1
S2=4 2 3 0 Sbreak2

C1=2 0 0.119244376p IC=0
C2=4 0 480nF IC=0

..model Sbreak1 VSWITCH Roff=1e12 Ron=1.0 Voff=4.9 Von=5.0
..model Sbreak2 VSWITCH Roff=1e12 Ron=1.0 Voff=0.1 Von=0.0
..OPTIONS CHGTOL=1.0e-16

..tran 0.0001ns 0.5s
..op
..probe
..end
********SPICE DECK**********

2. ### Klaus KragelundGuest

Do you need so much ratio between Roff and Ron /1E12) and do you need
so fast risetime on the pulse (0.1ns)

If you can relax on any of those parameters then you should be albe to
run without convergence problems

Regards

Klaus

3. ### Guest

I've worked out analytical equations for the circuit and can model this
thing in MATLAB but I'm trying to see if the two approaches give
identical (or nearly identical) results.

The large switch resistance I need to keep (although I have reduced it
to 1000Mohms), and I can't sacrifice much w.r.t. the pulse parameter.
Going from 0.1ns to 10ns allows the simulation to proceed, but I start
seeing significant deviations from the MATLAB at this point. Note that
my pulse wave form is high for 830ns then low for 4150ns, so a 10ns
transition is starting to be significant ...

Is there any way to have my cake and eat it too ?

4. ### Klaus KragelundGuest

Are you using Orcad PSpice? If so - it has the possibility to shedule
better resolution at specific points in time. So for example you would
set the maximum step size at your pulse generator transitions (don't

What about increasing the number of maximum iterations - perhaps it can
get over the converging point after all. (I think they are name ITL1,
ITL2, ITL4)

Regards

Klaus

5. ### qrkGuest

Looks like your running PSpice from the type of error message you got.
This deck runs fine on LTspice. Download it, it's free and works
fabulously. LTspice is also very compatible with PSpice syntax.