Hello QSPICE Community,
I’m experiencing a convergence issue with a specific third-party MOSFET model and would appreciate any guidance.
The Goal: I am trying to use the STMicroelectronics SCT070HU120G3AG model (from the ST website .lib file) in a DCDC converter simulation.
The Problem: When I use this ST model in my converter, the simulation fails, always giving a Fatal error: Timestep too small....
However, if I use a standard QSPICE library part (like the UF3C120040K3S) as a direct replacement, the exact same converter schematic simulates perfectly.
Troubleshooting Already Performed:
- Model Syntax Fix: I have already modified the original
.libfile to fix thekkeyword conflict, askis a reserved constant in QSPICE.
Original:.FUNC bvd(k) {7.8+0.0026*k}
My Fix:.FUNC bvd(k_var) {7.8+0.0026*k_var}No other changes were made to the.libfile. - Standalone Test: To verify the model, I created a separate, simple characterization test bench (.qsch) to plot the switch’s characteristics. In this simple circuit, the
SCT070HU120G3AGmodel (with myk_varfix) works perfectly.
This suggests the model itself can work in QSPICE, but it is failing when interacting with my more complex DCDC converter circuit (likely during a hard-switching event or when the body diode is active).
My Question:
Since the model works in a simple test jig but fails in the converter, what should I try next?
Are there known issues with these complex, behavioral models from ST that require specific .options settings (like method=gear, reltol, or a strict max timestep) to converge in a switching application?
Attachments: I am attaching the .lib file (with my k_var fix) and the simple .qsch characterization schematic that does work. Unfortunately, due to company privacy, I cannot share the full DCDC converter schematic.
Any advice on how to make this model stable for a switching simulation would be extremely helpful.
Thank you.
