Regarding GaN, based on the latest QSpice updates, Mike seems working on GaN models with EPC. Here is the latest entry from the revision history 04/15/2026 Corrected a number of errors in the EPC level 2026 models.
It appears that Mike has included a new NMOS model, level 2026, specifically to describe GaN devices. If you add a NMOS, right-click to open the Selection Guide, and navigate to EPC devices, you will see their .model is listed as nmos level=2026, which matches the revision history.
If you download the SPICE models from the EPC GaN website: EPC Device Models, they model GaN using subcircuits with parameter names like agn, agd, etc. The nmos level=2026 model also uses these parameters, suggesting it is a new dedicated NMOS model for GaN in related to how EPC modeling GaN.
The model generator in QSpice is designed to create VDMOS models. Based on you input, you feel that VDMOS cannot describe GaN accurately enough for your specific usage, right?
What challenges are you currently encountering?
Thank you for the update. I reviewed the EPC models available in the QSPICE library, and I noticed that the model representation is quite different.
I would like to evaluate the GaN subcircuit model provided by the manufacturer, then generate a corresponding model using the QSPICE model generator, and finally compare it with the Level 3 model developed by the research team.
Since this model also includes parameters such as channel length and channel width, I was wondering if these can be utilized to further improve the accuracy of the model and enable a better comparison. Is there a way to do this?
QSpice describes Cgso and Cgdo in its Help file. However, for in-depth information, you generally have to search the internet or papers. HSPICE has a manual of nearly 1,000 pages dedicated solely to MOSFET models. In general, HSPICE and NGspice provide more detailed documentation, whereas QSpice and LTspice only provide rough information.
Sometimes asking AI can provide some insight as well. I am not sure if this forum has experts in modeling, as it is such a specialized topic that even advanced users may only scratch the surface.
Cgso and Cgdo are the capacitances across Cgs and Cgd, but they scale with the gate width (W).
As for Cjo, it belongs to the diode model (.model MD D) and represents the junction capacitance.
I just shared the image of the GaN Level 3 model for reference. My intention was not to specifically highlight the red box parameters (Cgso, Cgdo, and CJO).
I am able to run the Level 3 model, but it is not providing reasonable characteristics when compared with the datasheet.
If we are able to fine-tune this Level 3 model and then compare the same model with the subcircuit model and the generator model, I believe it would be very valuable work and will contribute a great knowledge.
