datasheet2spice

Traceable datasheet-to-SPICE model generation for power MOSFETs and diodes, with validation benchmarks.

View the Project on GitHub lisiqi1983/datasheet2spice

SPICE Dialects

datasheet2spice keeps one internal device project and exports simulator specific netlists through emitter dialects.

Supported dialect names:

Use all built-in dialects:

datasheet2spice emit examples/demo_sic_mosfet/device.json --out build/demo --all --dialect all

Model Family Coverage

abm-basic supports every dialect above. It emits a three-pin subcircuit with smooth channel current, datasheet capacitance tables, body diode, and starter package parasitics.

vdmos-static-fast has two export modes:

diode-basic supports every dialect above with a portable two-terminal subcircuit and native .model ... D(...) card. The same generated core model is used across dialects because the SPICE diode primitive is widely supported.

diode-abm-dynamic supports every dialect above with a two-terminal subcircuit that combines a DC diode card with behavioral current sources for nonlinear Cj(VR) and reverse-recovery charge release. It uses VALUE, CUR, or SPICE3-style B source syntax according to the selected dialect, so simulator expression compatibility matters more than it does for diode-basic.

Accuracy Notes

Dialect support means the emitter writes a netlist in the syntax family of the target simulator. It does not mean the model is vendor-qualified or fully calibrated.

For transient accuracy, prefer:

  1. abm-basic for cross-dialect behavior review.
  2. vdmos-static-fast native LTspice/ngspice for fast compact-model sweeps.
  3. diode-basic for compact diode smoke checks.
  4. diode-abm-dynamic when recovery charge and junction capacitance need to shape transient waveforms before a lab-fitted model exists.
  5. Measured double-pulse or curve fitting before design use.

The qspice dialect is intentionally marked experimental until it is backed by automated QSPICE smoke tests.