High accurate SPICE model for low voltage MOSFET (12V-300V)

Structure and features

As shown in Fig. 1, our new generation low voltage MOSFET (12V-300V) has a Trench Field Plate (FP) under a gate electrode, and low drain-source on-resistance is achieved by increasing the concentration of the drift layer. Fig. 2 shows the schematic of SPICE model for FP-MOSFET which is developed by Toshiba. As a core MOSFET model, it is excellent in convergence and computation speed, and it adopts BSIM3v3 model which is implemented in most of circuit simulators of every EDA vendor. And the Field Plate shown in Fig. 1 producing additional capacitance components while relaxing electrolytic concentration. In order to express the non-linearity of capacitance characteristics due to this Field Plate structure, a non-linear capacitance Cgs/Cgd/Cds using optional function are constructed between each terminal.

Fig. 1. Cross section of FP-MOSFET (outline drawing)
Fig. 1. Cross section of FP-MOSFET (outline drawing)
Fig. 2.  Schematic of FP-MOSFET SPICE model (outline drawing)
Fig. 2. Schematic of FP-MOSFET SPICE model (outline drawing)

MOSFET fitting accuracy

Fig. 3 shows the reproduction level (called "fitting accuracy") by comparing the actual measurement of ID-VGS characteristics and ID-VDS characteristics of our FP-MOSFET product TPH1R306PL with the simulation using the SPICE model. Our SPICE model for FP-MOSFET does not match the saturation characturistics in the high current region of ID-VGS characteristics commonly found in power MOSFET. On the other hand, it can be seen on Fig. 3 (a) and (b) that our SPICE model have higher reproducibility mainly on the reproducibility near the threshold voltage (Vth) and the reproducibility of the drain-source on-resistance (RDS (ON)) and the slope of the drain current ID in the ID-VDS characteristic.

Fig. 3. Comparison of the measured (symbols) and the simulated (solid line) I-V characteristics using the SPICE model (G2 model) of TPH1R306PL. (a)ID-VGS Characteristics at VDS = 10V, (b) ID-VDS Characteristics
Fig. 3. Comparison of the measured (symbols) and the simulated (solid line) I-V characteristics using the SPICE model (G2 model) of TPH1R306PL. (a)ID-VGS Characteristics at VDS=10V, (b)ID-VDS Characteristics

Figure 4 shows a comparison between actual measurement and simulation of the VDS dependencies on the Input capacitance (Ciss), Output capacitance (Coss) and Reverse transfer capacitance (Crss). The Trench Field Plate structure enables lower on-resistance by reducing the electric field concentration around the gate along with the enhancement of non-linearlity on the characteristics between gate capacitances and VDS. Toshiba SPICE model for FP-MOSFET reproduces the non-linear gate capacity property of the actual measurement with high accuracy by utilizing the non-linear capacitance combining several arbitrary functions as Cgs, Cgd, Cds.

Fig. 4. Comparison of the measured (symbols) and the simulated (solid line) C<sub>iss</sub>/C<sub>oss</sub>/C<sub>rss</sub>-V<sub>DS</sub> characteristics at V<sub>GS</sub> = 0V using the SPICE model (G2 model) of TPH1R306PL.
Fig. 4. Comparison of the measured (symbols) and the simulated (solid line) Ciss/Coss/Crss-VDS characteristics at VGS = 0V using the SPICE model (G2 model) of TPH1R306PL.

Transient characteristics (resistive load switching operation) fitting accuracy

Fig. 5 shows a resistive load switching circuit diagram using FP-MOSFET product, TPH1R306PL. And Fig.  6 (a) and (b) show the comparison between the actual measurement and the simulation for turn-on and turn-off characteristics using the resistive load switching circuit, respectively. As a result, it can be seen that the simulation waveform can reproduce the measured waveform with high accuracy for both turn-on and turn-off characteristic. 

Fig. 5. Schematic for resistive load switching characteristics
Fig. 5. Schematic for resistive load switching characteristics
Fig. 6. Comparison of the measured (dashed line) and the simulated (solid line) switching waveforms using a resistive load switching circuit (a)turn-on waveform, (b) turn-off waveform
Fig. 6. Comparison of the measured (dashed line) and the simulated (solid line) switching waveforms using a resistive load switching circuit (a)turn-on waveform, (b) turn-off waveform

High accurate SPICE model (G2 model) to evaluate systems (300W isolated DC-DC converter)

The power conversion efficiency and the switching waveform of the MOSFET are compared by using the actual 300W isolated DC-DC converter reference design and the simulation circuit that applies the high accurate SPICE model (G2 model). Fig. 7 shows the appearance of the actual 300W isolated DC-DC converter and a schematic diagram of the simulation circuit.

Fig. 7. 300W isolated DC-DC converter actual board and simulation circuit diagram
Fig. 7. 300W isolated DC-DC converter actual board and simulation circuit diagram

Fig. 8 shows the comparison of the load current dependency on the power conversion efficiency between actual measurement and simulation when TPN1200APL is mounted on the primary side of the DC-DC converter circuit and TPH2R408QM is mounted on the secondary side. It can be seen that the efficiency difference between measurment and simulation is 2% or less. This difference is mainly due to the factors not taken into account in this simulation circuit, such as loss of the controller and the temperature dependency on each component constituting the circuit.

Fig. 9 shows a simulation result of the loss calculated for the main components of a 300W isolated DC-DC converter. The total MOSFET loss on the primary and secondary sides at 25A output current accounts for approximately 43% of the total loss (the subtotal loss ratio of Pre_Side and Sec_Side in Fig. 9), and this ratio increases with the decrease of output current.

Fig. 8. Comparison of the measured and the simulated power conversion efficieccy depending on output current.
Fig. 8. Comparison of the measured and the simulated power conversion efficieccy depending on output current.
Fig. 9. Power losses for main components on the 300W isolated DC-DC converter using a simulation circuit
Fig. 9. Power losses for main components on the 300W isolated DC-DC converter using a simulation circuit

This result indicates that it is very important to reduce the loss on MOSFETs in order to improve the power conversion efficiency of the DC-DC converter. The detailed loss analysis during the switcing of the MOSFETs on the primary and/or secondary side is required for reduction of MOSFET loss, and the reproducibility of the switching waveform of the SPICE model is important.
Fig. 10 shows a switching waveform comparison of TPH2R408QM used in the secondary side synchronous rectifier circuit of the DC-DC converter between actual measurement and simulation. According to this, a switching waveform simulated by G2 model that reflects the non-linearity of the capacitance characteristics with high accuracy is closely reproduced with the actual measurement can be obtained by simulation, therefore the quantitive MOSFET loss analysis is expected to be realized.

(Note) The high accuracy simulation circuit is developed uniquely for loss analysis.

Fig.10. DC-DC converter circuit drain-source voltage (VDS) waveform of the secondary side synchronous rectifier MOSFET
Fig.10. DC-DC converter circuit drain-source voltage (VDS) waveform of the secondary side synchronous rectifier MOSFET

Our high accurate SPICE model (G2 model) can reproduce the actual switching waveforms with high accuracy. The circuit verification by using high-accuracy simulation in advance of the actual production may reduce the double work for the circuit modification and help the development time shorten. Toshiba's high accurate SPICE model (G2 model) supports your circuit development and analysis effectively.

開啟新視窗