The problem
SiC MOSFETs are replacing silicon IGBTs in many traction inverters, especially at 800 V, because they switch faster and have no knee voltage, giving lower losses at part load. The real energy benefit depends on the drive cycle, the voltage class and the switching frequency chosen, and claims are often made without a transparent model. A careful, reproducible study helps decide when wide-bandgap devices are worth it.
Objectives
- Build datasheet-based loss models for an Si IGBT and a SiC MOSFET module of similar rating.
- Generate inverter efficiency maps over torque and speed for each device and voltage class.
- Combine them with a motor map and vehicle model to compute drive-cycle energy.
- Study the effect of switching frequency and junction temperature.
- Quantify range or battery-size implications under stated assumptions.
What you need
Tools and toolboxes
- MATLAB — Loss models, efficiency-map generation, scripted studies and plots.
- Simulink — Vehicle and drive-cycle model with lookup tables for motor and inverter efficiency.
- Simscape Electrical — Detailed phase-leg switching models to validate the analytical loss model at selected points.
- Powertrain Blockset — Optional: reference vehicle and Mapped Motor blocks. A hand-built longitudinal model also works.
Software
- MATLAB R2023b or later
- Simulink
- Simscape Electrical
- WLTP, MIDC and FTP-75 speed–time data
Extra parts and data
- Datasheets for a comparable Si IGBT module and SiC MOSFET module — Fair comparison requires real conduction, switching energy and thermal data at the same voltage and current class.
Key concepts
- Knee voltage versus resistive conduction
- An IGBT has an on-state voltage with a fixed knee plus a resistive part; a MOSFET behaves as a resistance, so it has much lower loss at light load where EVs spend most time.
- Switching energy
- SiC MOSFETs have much lower turn-on and turn-off energy and no diode reverse-recovery tail of the same size, so they can switch faster or at higher frequency.
- Third-quadrant conduction
- A SiC MOSFET can conduct reverse current through its channel when on, reducing body-diode losses during dead-time-free intervals.
- System-level benefit
- Inverter energy saving reduces battery energy per kilometre; for a fixed range it allows a slightly smaller battery, which can be stated only with clearly stated assumptions.
Step-by-step guide
- 1
Define the comparison
Choose a labelled example vehicle (a mid-size car with stated mass, Cd·A and Crr), a motor efficiency map from an earlier project, and two voltage classes: 400 V and 800 V. For each class pick comparable Si IGBT and SiC MOSFET modules with similar current rating. Define cycles (WLTP Class 3, MIDC, FTP-75), switching frequencies (10, 20, 40 kHz) and coolant temperature.
Checkpoint: A study matrix listing vehicle, devices, voltages, frequencies and cycles.
- 2
Build device loss models
Digitise on-state characteristics and switching energies at two junction temperatures for each device and its diode or body diode. Implement analytical sinusoidal-PWM loss expressions for conduction and switching, including third-quadrant conduction for SiC with a chosen dead time. Scale switching energy with DC-link voltage and current as the datasheet conditions allow. Record the gate resistance and test conditions of every datasheet curve you use.
Checkpoint: Loss model outputs reproduce datasheet conditions and are plotted against current for both devices.
- 3
Validate with switching simulation
Build a Simscape Electrical phase-leg model for each device with tabulated losses and run it at several operating points (low, medium and high current; two modulation indices). Compare losses with the analytical model. Note where the analytical model is less accurate, for example very low current or high power factor angle. Use a fine fixed step, for example 10 ns, and confirm convergence by halving it.
Checkpoint: Analytical and simulated losses agree within a stated margin at all validation points.
- 4
Generate inverter efficiency maps
For each device, voltage and switching frequency, map inverter loss over the motor torque–speed plane using the motor's current, voltage and power factor at each point. Include a thermal estimate so junction temperature is consistent with losses at each point. Plot efficiency maps and difference maps between SiC and Si. Keep the same motor map and current limits for every device so that differences come only from the inverter.
Checkpoint: Difference maps show the largest SiC benefit at light load, consistent with the device physics.
- 5
Build the vehicle energy model
Build a backward-facing longitudinal model in Simulink using F = m·g·Crr·cosθ + ½·ρ·Cd·A·v² + m·g·sinθ + m·a, a fixed gear, the motor map and an inverter map as 2-D Lookup Tables, and battery power with regeneration. Check the baseline energy per kilometre against a sensible range for the example vehicle. Also check regeneration energy recovered during braking, since inverter efficiency affects it too.
Checkpoint: Baseline Wh/km on WLTP is in a plausible range for the example vehicle.
- 6
Run the study matrix
Script all combinations with sim() or parsim and record inverter loss energy, battery energy per kilometre and operating-point histograms. Present the results as tables and bar charts. Separate the effect of device technology from the effect of voltage class. Keep the vehicle model, motor map and cycle files identical across runs, and save every result with the settings that produced it, so that tables can be regenerated and checked by a reviewer.
Checkpoint: A complete results table for every combination of device, voltage, frequency and cycle.
- 7
Study switching frequency trade-offs
Higher switching frequency reduces motor current ripple and harmonic losses but increases inverter loss. If possible, add a simple estimate of motor harmonic loss against switching frequency, and find the frequency that minimises total drivetrain loss for each device. Explain why SiC moves the optimum higher. If you have no data for motor harmonic loss, say so and present inverter loss alone, noting that the true optimum is likely to sit at a higher frequency than inverter loss suggests.
Checkpoint: A plot of total loss against switching frequency for each device with optima marked.
- 8
Translate to range or battery size
Under clearly stated assumptions (fixed battery usable energy, or fixed range target), compute the range change or battery-size change. Make clear these are model results for the example vehicle, not general claims. Include an uncertainty range from the sensitivity of key inputs. Avoid presenting a single headline percentage; give the range across cycles and voltage classes, together with the assumptions that produce the largest and smallest values in that range.
Checkpoint: Range or battery-size implications with an uncertainty band and stated assumptions.
- 9
Sensitivity analysis
Vary coolant temperature, dead time, motor map, vehicle mass and cycle choice, and recompute the energy benefit. Report which assumptions change the conclusion most. This is what makes the study credible. Vary each input over a plausible range, one at a time from the baseline, and record the change in the SiC benefit. Then check a few combined worst and best cases. If any single assumption can reverse the conclusion, state this clearly in the results.
Checkpoint: Tornado-style sensitivity chart of the SiC benefit.
- 10
Research write-up
Document devices, models, validation, maps, cycle results, switching-frequency study, range implications and sensitivity. Release scripts and digitised data where allowed. State limitations such as no EMI, no gate-driver losses unless modelled and no cost analysis. Compare your results with published comparisons of Si and SiC traction inverters, explain any differences in assumptions, and state clearly what the study adds, such as Indian drive cycles or the frequency optimisation.
Checkpoint: A thesis chapter draft with reproducible results and transparent assumptions.
Testing & validation
- Device loss models reproduce datasheet values.
- Analytical losses validated against switching simulation.
- Baseline vehicle energy in a plausible range.
- Complete study matrix over devices, voltages, frequencies and cycles.
- Sensitivity of the conclusion to key assumptions.
Expected results
A transparent, reproducible comparison of Si IGBT and SiC MOSFET traction inverters at 400 V and 800 V, with efficiency maps, drive-cycle energy, optimum switching frequency and sensitivity-bounded range implications for an example vehicle.
Research angle for your PhD
Research questions: for Indian urban cycles such as MIDC compared with WLTP, how large is the drive-cycle energy benefit of SiC over Si in 400 V and 800 V systems, what switching frequency minimises total drivetrain loss for each, and how sensitive are conclusions to assumptions? Metrics: inverter loss energy, Wh/km, optimum frequency, sensitivity. Baselines: Si IGBT at standard switching frequency. Validated loss models and transparent assumptions are essential. A reproducible cycle-level study including Indian drive cycles and a sensitivity analysis is the kind of work suitable for a journal or conference paper.
Take it further
- Add GaN devices for a lower-voltage two-wheeler drive.
- Include a cost model with clearly sourced assumptions.
- Add gate-driver and EMI filter losses.
- Couple a thermal lifetime study to the device comparison.
Viva and review questions
Why does SiC give the biggest benefit at light load?
An IGBT has a fixed knee voltage, so its conduction loss does not fall proportionally at low current, whereas a MOSFET's resistive conduction loss falls with the square of current.
Why might the benefit be larger at 800 V?
Silicon IGBT switching losses rise strongly with voltage, while SiC switching energy stays much lower, so the gap widens at higher voltage.
How do you keep the comparison fair?
By choosing modules of similar voltage and current rating, using the same thermal conditions, validating the loss model, and showing sensitivity to assumptions instead of a single number.
Learn it at DIYguru
These DIYguru courses and programmes teach what this project needs.