IGBT Interview Questions

Q1. What is an IGBT and how does it combine BJT and MOSFET properties?

An IGBT (Insulated Gate Bipolar Transistor) is a four-layer power switch that uses a MOSFET input structure for voltage-controlled, high-impedance gate drive while its output is a bipolar PNP transistor that provides high current density at low on-state voltage — combining the best of both devices. An Infineon IKW40N120T2 (40 A, 1200 V) has a gate drive requirement of ±15 V at a few milliamps, like a MOSFET, but a saturation voltage VCEsat of 2.1 V at 40 A — far lower than an equivalent MOSFET RDSon × I product at that voltage class. IGBTs dominate applications above 600 V and a few hundred watts where MOSFET RDSon becomes impractically high.

Follow-up: At what voltage and current level does a MOSFET become preferable to an IGBT and why?

Q2. Explain the internal structure of an IGBT.

An IGBT's structure from gate to collector is: gate oxide on a P-body diffusion in an N-drift region, with an N+ cathode (called emitter) and a P+ collector substrate injecting minority carriers into the N-drift for conductivity modulation. The P+ substrate causes hole injection into the N-drift on turn-on, which increases conductivity by 10–1000× compared to the background N-doping — this is the 'bipolar' part enabling low VCEsat. The N-buffer layer in punch-through (PT) IGBTs controls minority carrier lifetime and determines the trade-off between VCEsat and turn-off speed (tail current).

Follow-up: What is the difference between punch-through (PT) and non-punch-through (NPT) IGBT structures?

Q3. What is the tail current in an IGBT and why does it cause turn-off losses?

Tail current is the slowly decaying collector current that persists after the IGBT is turned off because stored minority carriers (holes) in the N-drift must recombine before conduction fully ceases, unlike a MOSFET where current stops almost instantly when gate turns off. A Semikron SKM75GB12T4 at 75 A shows a tail current lasting 1–3 µs after gate voltage falls, during which full DC bus voltage is reapplied across the device — the product of tail current × DC bus voltage × tail time is the turn-off energy Eoff. In a 20 kHz drive application, this turn-off energy of perhaps 2 mJ per switching event dissipates 40 W per IGBT, which sets the thermal limit on maximum switching frequency.

Follow-up: How does IGBT generation affect tail current magnitude and turn-off loss?

Q4. How do you design a gate drive circuit for an IGBT?

An IGBT gate drive circuit must provide +15 V for turn-on (to ensure full saturation and low VCEsat) and -8 to -15 V for turn-off (to prevent spurious turn-on from dv/dt-induced displacement current through gate-collector Miller capacitance), with sufficient peak current to charge and discharge gate capacitance within the required switching time. For an Infineon IKW75N60T IGBT with gate charge Qg = 200 nC, achieving a 200 ns switching time requires a peak gate current of 1 A; gate drive ICs like the Infineon 1ED020I12-F2 or TI ISO5852S provide 4–8 A peak and isolated ±15 V supply in a single package. The gate resistor Rg balances switching speed against dv/dt and di/dt levels, which directly affect EMI and overvoltage spikes.

Follow-up: What is the effect of gate resistance on IGBT switching speed, switching losses, and turn-off voltage spike?

Q5. What is the latch-up failure mechanism in an IGBT?

Latch-up occurs when the parasitic NPN transistor (formed by N+ emitter, P-body, N-drift) and the PNP output transistor create a four-layer PNPN structure that turns on and cannot be turned off by the gate — analogous to an SCR turning on inside the IGBT. In a 600 V IGBT, if collector current density exceeds the critical latch-up current (which decreases with temperature, so latch-up is most likely during overcurrent at high junction temperature), the device locks into conduction and destroys itself. Modern IGBT cell designs use heavily doped P-body regions and optimized emitter cell geometry to push the latch-up current density well above the rated current to prevent this failure mode.

Follow-up: How does temperature affect the latch-up current threshold in an IGBT?

Q6. How do you measure and compare switching losses (Eon, Eoff) for an IGBT?

Switching losses are measured using a double pulse test: two pulses of precise width are applied to the IGBT under test into a clamped inductive load, and a differential voltage probe and Rogowski current probe on an oscilloscope measure VCE and IC simultaneously during turn-on of the second pulse and turn-off of the first, integrating the instantaneous power VCE × IC to get Eon and Eoff in millijoules. An Infineon FF450R12KT4 1200 V, 450 A IGBT module shows Eon = 40 mJ and Eoff = 25 mJ at rated current and 125°C junction temperature at 600 V DC bus. Switching loss data from datasheets is given at specific test conditions and must be scaled to actual operating conditions using the graphs of Eon and Eoff versus collector current and junction temperature.

Follow-up: How does DC bus voltage affect IGBT switching losses and what scaling factor is typically used?

Q7. What is desaturation protection in an IGBT gate driver?

Desaturation (DESAT) protection monitors VCE of the conducting IGBT via a high-voltage diode; if VCE rises above a threshold (typically 7–9 V) during on-state, it indicates overcurrent or short circuit — the device has come out of saturation — and the gate driver immediately applies a controlled soft turn-off to prevent di/dt from causing destructive overvoltage. An Infineon 1EDC60I12AH gate driver detects DESAT within 2 µs and initiates a soft turn-off by charging the gate to a lower intermediate voltage before fully turning off, limiting di/dt to a safe level even for bolted bus faults. DESAT protection is mandatory in IGBT motor drives and converters because short circuit current rises to 5–10× rated current in microseconds and would destroy the device in < 10 µs without intervention.

Follow-up: What is the short circuit withstand time (SCWT) of a modern IGBT and why is it important?

Q8. What is the safe operating area (SOA) of an IGBT?

The SOA defines the region of VCE-IC space within which the IGBT can operate without damage; it is bounded by maximum VCE (breakdown voltage), maximum IC (metallization limit), maximum power (thermal limit), and for hard switching, the boundary set by latch-up current and avalanche energy. A Semikron IGBT module's datasheet SOA curve at 150°C junction temperature shows a lower IC limit at high VCE, meaning the device must be derated in both current and switching frequency as junction temperature rises toward maximum. Operating outside the SOA — even for microseconds during fault conditions — is the most common cause of IGBT failure in converter field returns, making SOA analysis a mandatory step in converter design review.

Follow-up: How does the SOA of an IGBT compare to that of a power MOSFET?

Q9. What is the difference between 4th generation and 7th generation IGBTs?

Fourth-generation IGBTs used planar cell technology with turn-off tail times of 3–5 µs and switching frequencies limited to 5–15 kHz; 7th-generation Infineon TRENCHSTOP 7 IGBTs use fine-pitch trench gate topology with N-field-stop buffer, cutting tail time to below 200 ns and enabling reliable operation at 30–60 kHz. An IKW40N120T2 (7th gen, 1200 V, 40 A) achieves Eoff = 0.35 mJ at 25°C and 600 V, compared to ~2 mJ for a comparable 4th-gen device — a 5× improvement reducing switching losses and enabling compact, high-frequency inverter designs. Each IGBT generation has brought roughly 30–50% reduction in both VCEsat and switching losses, driving down inverter size, weight, and cooling requirements in industrial and EV applications.

Follow-up: What technology changes in TRENCHSTOP vs planar IGBT cell design account for the improved switching performance?

Q10. How are IGBTs used in electric vehicle traction inverters?

In an EV traction inverter, six IGBTs in a three-phase bridge switch at 8–16 kHz to convert 400 V or 800 V DC battery voltage into three-phase AC for the drive motor — a Tesla Model 3 uses silicon carbide MOSFETs but earlier EVs like the Nissan Leaf used 650 V IGBT modules from Renesas in a water-cooled aluminum housing. The inverter manages up to 200 kW peak power, requiring IGBT modules with gate drive PCBs, current sensors, NTC thermistors, and DSP-based vector control running field-oriented control at 20 kHz. The shift from IGBT to SiC MOSFET in newer EVs reduces inverter switching losses by 50–70%, extending EV range by 5–8% for the same battery size.

Follow-up: Why are SiC MOSFETs preferred over IGBTs in EV traction inverters despite higher unit cost?

Q11. What is the gate threshold voltage of an IGBT and why does it matter?

Gate threshold voltage VGE(th) is the minimum gate-emitter voltage at which the IGBT begins to conduct collector current, typically 5–7 V for standard IGBTs; it decreases with increasing temperature at about -10 mV/°C. For a device with VGE(th) = 5.5 V at 25°C, the threshold at 150°C drops to about 3.8 V — meaning a negative gate bias of -8 V at turn-off is necessary to prevent spurious turn-on due to Miller current and reduced threshold. In automotive-grade IGBT drivers, the gate voltage levels are chosen with this temperature coefficient in mind so that the margin between off-state gate voltage and VGE(th) is maintained across the full operating temperature range.

Follow-up: What is the Miller plateau in IGBT gate charge and what happens during it?

Q12. How is IGBT module thermal management designed?

IGBT thermal design follows the thermal resistance chain: Tj = Tamb + P × (Rth_jc + Rth_ch + Rth_ha), where Rth_jc is silicon-to-case (~0.1°C/W for a 1200 V half-bridge module), Rth_ch is case-to-heatsink (0.01–0.03°C/W for thermal paste), and Rth_ha is heatsink-to-ambient. A Semikron SKiiP 1242GB120-4DW module dissipating 400 W total (conduction + switching) with Rth_jc = 0.075°C/W on a forced-air heatsink at 50°C ambient reaches junction temperature of 50 + 400 × (0.075 + 0.02 + 0.15) = 50 + 98 = 148°C — within the 150°C maximum. Water cooling with Rth_ha = 0.02°C/W replaces the 0.15°C/W air heatsink, halving the temperature rise and allowing either higher power or higher ambient temperature operation.

Follow-up: What is the transient thermal impedance Zth and when is it more relevant than steady-state Rth?

Q13. What is IGBT paralleling and what are the challenges?

Paralleling IGBTs increases current capacity by connecting two or more in parallel on the same DC bus and load terminals, but requires careful attention to current sharing since small differences in VCEsat or gate threshold cause the device with lower VCEsat to take more current — a self-aggravating condition because higher current raises temperature and lowers VCEsat further. Infineon recommends using devices from the same manufacturing batch (same VGE(th) bin) and placing individual gate resistors of 10–15% of the total Rg for each parallel device to equalize dynamic current sharing during switching transients. Press-pack IGBT designs used in HVDC valves have inherently matched VCEsat because multiple dice are pressed into the same package and thermally equalized.

Follow-up: What is the difference between static and dynamic current sharing in paralleled IGBTs?

Q14. What is the collector-emitter saturation voltage (VCEsat) and how does it affect efficiency?

VCEsat is the voltage across the conducting IGBT at rated current, typically 1.7–2.5 V for 600–1200 V devices; the conduction power loss is IC × VCEsat per device, and for a 3-phase bridge carrying 100 A load, two IGBTs conduct at any time giving 100 A × 2.2 V × 2 = 440 W conduction loss. An Infineon IKD10N60R at 10 A and 25°C shows VCEsat = 1.65 V; replacing a standard device with a low VCEsat variant in a 10 kHz drive reduces conduction loss by 15–20% but the tradeoff is higher switching losses since low VCEsat devices have higher stored charge. The VCEsat–switching loss trade-off is the central design decision in selecting an IGBT for a given operating point.

Follow-up: Why do lower VCEsat IGBTs typically have higher switching losses and what fundamental physics causes this trade-off?

Q15. What are the differences between IGBT and power MOSFET in terms of switching behavior?

Power MOSFETs switch faster (tens of nanoseconds) with no tail current because conduction is by majority carriers only, making them preferred below 200–300 V; IGBTs switch more slowly (hundreds of nanoseconds to microseconds) due to minority carrier tail but have lower on-state voltage above 600 V where MOSFET RDSon becomes excessive. An IRFP260N MOSFET at 200 V, 50 A has RDSon = 40 mΩ giving 100 W conduction loss at 50 A; a comparable 1200 V MOSFET would need RDSon > 1 Ω giving 2500 W — impossible, whereas a 1200 V IGBT at the same current has VCEsat = 2.5 V and only 125 W. SiC MOSFETs bridge this gap at 650–1700 V, combining low on-state resistance with fast switching, but at 3–5× the cost of silicon IGBTs.

Follow-up: At what voltage level does silicon MOSFET on-resistance exceed the practical limit, and how does SiC change this boundary?