LTspice As a circuit simulation software, LTspice is probably familiar to every hardware developer. The brain behind LTspice was Mike Engelhardt, which many probably know, as he participated in the LTspice World Tour for many years and held seminars on LTspice.
At “The Amp Hour Electronics Podcast” There's an episode from 2014 that I can recommend to every hardware developer. In the episode, Mike Engelhardt talks about Spice simulators and offers, among other things, the following tips for LTspice:
- The GMIN parameter should ideally be set to 0. This parameter is especially needed when simulating diodes in series. If you experience problems, you can enable the GMIN hack in LTspice.
- Always start with the Normal Solver. It's about 10 times faster than the Alternate Solver, but much less accurate. If you experience problems with convergence or artifacts, you can switch to the Alternate Solver.
- The TRTOL parameter is normally set to 7 in Spice programs. In LTspice, the parameter can be set to 1, and only if problems occur is it set to 7. Lower values result in more accurate solutions (fewer artifacts); higher values make the simulation run faster, but may contain artifacts that experienced circuit engineers can recognize.
The parameters can be set in LTspice via simulate-> Control Panel -> SPICE.
If anyone wants to delve deeper into the topic of circuit simulation, I can recommend the book “Inside Spice” by Ron Kielkowski.
The book describes how Spice works, the differences between the various simulation types (AC, Operating Point, Transient, etc.), and how to handle non-convergent circuits. His clearly defined approach is particularly helpful. For convergence problems in transient analysis, for example, he suggests varying the RELTOL, VNTOL, and ABSTOL parameters (starting on page 53).
Its algorithm for selecting parameters is:
- RELTOL usually fits with 0.001 (0.1%).
- VNTOL = RELTOL * V_small, where V_small is the smallest voltage in the circuit (e.g. an opamp offset voltage).
- ABSTOL = RELTOL * I_small, where I_small is the smallest current that is relevant, e.g. the leakage current of a diode.
- The values can then be added to the simulation via options, for example: .OPTIONS RELTOL=.0001 VNTOL=.0001V ABSTOL=10NA
Following this pattern, there are tips for various problems related to SPICE simulations.
Maybe these tips will help with convergence problems occasionally.
Good luck with the simulation
Martin Bosch