Semiconductor and Device Technology Device Physics and Modeling Informational

What is the Angelov model for a GaN HEMT and when is it appropriate to use?

The Angelov model (also known as the Chalmers model, developed by Iltcho Angelov at Chalmers University) is the most widely used nonlinear transistor model for GaN and GaAs HEMT power amplifier design. It models the drain current using hyperbolic tangent functions: I_DS = I_PK × (1 + tanh(psi)) × tanh(alpha × V_DS), where psi = P1 × (V_GS - V_PK) + P2 × (V_GS - V_PK)² + P3 × (V_GS - V_PK)³ + ... The parameters: I_PK = peak transconductance current, V_PK = gate voltage at peak transconductance, P1, P2, P3 = polynomial coefficients that shape the transfer characteristic, alpha = channel modulation parameter (controls the knee region). Why the Angelov model is preferred for GaN PA design: (1) Accurate representation of the GaN I-V characteristics: the tanh function naturally captures the saturation behavior of the drain current (the current increases linearly at low V_DS, then saturates). The polynomial psi function accurately models the bell-shaped transconductance (g_m peaks at V_PK and rolls off at both higher and lower V_GS). (2) Charge-conservative model: the gate capacitances (C_gs, C_gd) are derived from a charge model (Q_gs(V_GS, V_DS), Q_gd(V_GS, V_DS)) that ensures charge conservation. This is essential for accurate harmonic and intermodulation simulation (non-charge-conservative models can produce unphysical current at harmonics). (3) Dispersion model: the Angelov model includes separate DC and RF drain current sources. The RF current source uses different parameters than the DC source, capturing the current collapse and trap effects in GaN. This is critical for GaN PA design (the RF load line and efficiency differ significantly from the DC prediction due to trap-induced current collapse). When to use: PA design (the model is optimized for large-signal behavior). Mixer and oscillator design (the harmonic distortion and nonlinear capacitance are accurately captured). Any circuit design using GaN or GaAs HEMTs where nonlinear behavior matters.

Category: Semiconductor and Device Technology

Updated: April 2026

Product Tie-In: Transistors, Simulation Tools

Angelov GaN HEMT Model

The Angelov model has evolved through multiple versions since its introduction in 1992, with each version adding capabilities for GaN-specific effects.

Parameter	Option A	Option B	Option C
Performance	High	Medium	Low
Cost	High	Low	Medium
Complexity	High	Low	Medium
Bandwidth	Narrow	Wide	Moderate
Typical Use	Lab/military	Consumer	Industrial

Technical Considerations

(1) Drain current: I_DS = I_PK × (1 + tanh(psi)) × tanh(alpha × V_DS) × (1 + lambda × V_DS). Where lambda = output conductance parameter (models the non-zero slope of I_DS vs V_DS in saturation). The psi polynomial can include up to 10+ terms for accurate fitting of complex I-V shapes. (2) Gate charges: Q_gs = C_gs0 × V_GS × (1 + tanh(phi_gs)) / 2 + C_gs_extra terms. Q_gd = C_gd0 × (1/alpha_gd) × ln(cosh(alpha_gd × V_DG)). These charge functions are integrable (they are derived from the potential, ensuring charge conservation: i_g = dQ/dt). (3) Dispersion: I_DS_RF = I_DS_DC + gm_RF × delta_V_GS + gds_RF × delta_V_DS. Where gm_RF and gds_RF are the RF transconductance and output conductance (which differ from the DC values due to trapping). The dispersion ratio: gm_RF/gm_DC and gds_RF/gds_DC capture the current collapse. Typical GaN: gm_RF/gm_DC = 0.85-0.95 (5-15% current collapse). (4) Self-heating: the model includes a thermal sub-circuit. The junction temperature: T_j = T_ambient + R_th × P_diss. P_diss = V_DS × I_DS. Where R_th is the thermal resistance (and optionally C_th for thermal transients). The temperature affects: I_DS (negative temperature coefficient: current decreases with temperature), g_m (decreases with temperature), and V_PK (shifts with temperature).

Performance Analysis

(1) DC extraction: measure the DC I-V curves. Identify I_PK (the current at peak g_m), V_PK (the V_GS at peak g_m), and alpha (from the knee region). Fit the psi polynomial coefficients to the measured g_m vs V_GS curve. Fit alpha from the V_DS dependence in the linear region. (2) S-parameter extraction: at each bias point, extract the intrinsic small-signal elements from the de-embedded S-parameters. Fit the charge model parameters to the bias-dependent C_gs and C_gd values. Fit the dispersion parameters by comparing pulsed I-V data to DC I-V data. (3) Verification: simulate a power sweep (P_in vs P_out, gain compression) and compare to measured load-pull data. Adjust the model parameters if the simulated P1dB or PAE deviates by more than 0.5 dB or 3% from measurement.

Performance verification: confirm specifications against the application requirements before finalizing the design
Environmental factors: temperature range, humidity, and vibration affect long-term reliability and parameter drift
Cost vs. performance: evaluate whether the application demands premium components or standard commercial grades

Interface compatibility: verify impedance, connector type, and mechanical form factor match the system architecture
Margin allocation: include sufficient design margin to account for manufacturing tolerances and aging effects

Design Guidelines

When evaluating the angelov model for a gan hemt and when is it appropriate to use?, engineers must account for the specific requirements of their target application. The optimal choice depends on the frequency range, power level, environmental conditions, and cost constraints of the overall system design.

Common Questions

Frequently Asked Questions

Angelov vs Curtice: which should I use?

Angelov is preferred for: GaN HEMTs (the dispersion and thermal models are essential for GaN). GaAs pHEMT PA design (the charge model and harmonic prediction are superior). Any design requiring accurate harmonic simulation (mixers, frequency multipliers, PA intermodulation). Curtice is acceptable for: simple GaAs MESFET small-signal amplifier design. Quick estimates where high accuracy is not needed. Legacy designs where Curtice models already exist and have been validated. In modern practice: virtually all new HEMT models are Angelov-based. Curtice is rarely used for new designs.

Can I get the Angelov model from the foundry?

Yes. Most III-V foundries provide Angelov models as part of their PDK (Process Design Kit). The model is extracted from test structures on the process characterization wafer. It represents the typical device performance across the wafer and across wafer lots. The model may include: corner models (fast/slow/typical) for yield estimation, temperature-dependent parameters (for simulation at different operating temperatures), and scalable models (different gate widths and number of fingers). If the foundry does not provide an Angelov model: you must extract it yourself from measured data on your specific device (following the extraction flow described above). Some foundries provide raw measured data (I-V, S-parameters) and extraction scripts.

How accurate is the Angelov model for harmonic simulation?

For PA design at the fundamental frequency: gain: ±0.5-1 dB accuracy. P_sat: ±0.5-1 dB. PAE: ±2-5% (typical). These are adequate for most practical designs. For harmonic simulation: second harmonic (2f0): ±3-5 dB accuracy (the harmonic level depends sensitively on the I-V curvature and the charge nonlinearity). Third harmonic (3f0): ±5-8 dB accuracy (even more sensitive). Intermodulation (IM3, IM5): ±3-6 dB accuracy. The harmonic accuracy improves if: the model is extracted from pulsed (not DC) I-V data (captures the RF behavior), the charge model is accurately fitted to capacitance vs bias data, and the model is verified against measured harmonic load-pull data. For critical applications (high-linearity PA for 5G base stations): measure the actual device harmonics and calibrate the model to match.

Keep Reading