How PWMSim is validated
Validation is part of every release. Each published result tells you which circuit and LTspice version were used, how the value was measured, and what tolerance it had to meet.
Public evidence tiers
| Tier | Meaning |
|---|---|
| Promoted | The portable test passes an independent formula or physics check and the stated LTspice tolerances. Its sources can be reproduced. |
| Matched-derived | PWMSim and LTspice agree on the listed result, but the source or controller is not equivalent enough for promotion. |
| Deferred | Some files or measurements exist, but a required check is missing. The reason is shown. |
| Source-only | The source file and checksum are recorded, but the test has not run in the required software. |
A result never earns a higher tier just because its waveform looks reasonable.
Reproduction discipline
- Each portable benchmark includes its netlist and measurement steps. When the syntax is supported, LTspice opens the same file.
- Formula checks and LTspice comparisons are shown separately. They answer different questions.
- Buck CCM, buck DCM and boost CCM use the same 4001-point grid over the last 20 switching periods for output-voltage and inductor-current mean, RMS, minimum, maximum and peak-to-peak.
- Every row says what it does not test. A passing operating-point result does not validate startup, loss, stress, controls, heat, or hardware.
- Wrong answers follow the public correctness policy. Failed evidence is not quietly removed.
Current boundary
The scorecard includes all 17 examples. Eight can be exported faithfully and pass their stated LTspice checks: three standard buck, boost, and flyback examples; four related variants; and the diode reverse-recovery example. The other nine use PWMSim-only control or device models. Each row names the part that prevents an accurate LTspice export. Approximate replacements do not count as a match.
Rows for parameter-sweep examples test the default circuit, not every suggested value. The wider comparison includes 38 same-file LTspice measurements. Of these, 30 cover the operating points for buck CCM, buck DCM, and boost CCM. Other converter types remain deferred when a required test is missing.
Use the benchmark scorecard to download the files and repeat a test. If you find a difference, report it through the correctness policy.