Characteristics Simulation for a VSI system UoD 04/12/2008

Aims of study A simulation study based on semiconductor devices physical models is carried out for a voltage-source-inverter (VSI) to examine the effects of device degradation on the converter thermal and electrical characteristics. It’s hoped that this work can lead to the development of a condition monitoring (CM) technique for power electronic systems. ESR Network

Failure Mode, Mechanism and Effect Analysis (FMMEA) of IGBT Cases studied Failure modes Failure causes Failure mechanisms Failure effects represented in study Case 1: Solder fatigue Temperature swings Different CTEs Thermal gradients Thermal resistance (Rjc) increases 20% Case 2: Gate degradation High temperature High electric field Over voltage High current density Time dependant dielectric breakdown Hot carries Gate threshold voltage (Vt) decreases 20% Case 3: Bond-wire lifting Number of chips in parallel decreases (e.g. 8=>7) Failure Mode, Mechanism and Effect Analysis (FMMEA) of IGBT ESR Network

Saber physic-based IGBT and diode models Advantages: Detailed physic model Electrothermal simulation available Experimentally verified Disadvantages: Difficult to parameterize Encryption of MAST template Hefner IGBT model [1] H. A. Mantooth Diode model [2] [1] A. R. Hefner and D. M. Diebolt, “An Experimentally Verified IGBT Model Implemented in the Saber Circuit Simulator,” Proc. of IEEE Power Electronics Specialists Conference, Cambridge, MA, 1991. [2] H. A. Mantooth, R. G. Perry, J. L. Duliere, “A unified diode model for circuit simulation,” IEEE Trans. On Power Electronics, vol. 12, No. 5 pp. 816-823, September, 1997. . ESR Network

Diode Electrical Parameters Diode thermal Dependent Coefficients Diode parameters [3] Diode Electrical Parameters Description ISL = 1.25 mA Saturation current for low level recombination NL = 3.89 Emission coefficient for low level injection BV = 600 V Breakdown voltage CJO = 3 nF Zero-bias junction capacitance TSW = 130 ns Charge Sweep out time TT = 165 ns Carrier Lifetime Tm=20ns Charge diffusion time Diode thermal Dependent Coefficients XTI = -2.36 ISL temperature exponent TNL1 = -6.35 E-4 Linear NL temperature coefficient TNL2 = -4.14 E-6 Quadratic NL temperature coefficient BETA = 1.71 Temperature exp. of carrier lifetime TT BETASW = 1E6 Temperature exp. of carrier lifetime TSW TRS1=3.12E-3 Linear RS temperature coefficient TRS2=1.39E-5 Quadratic RS temperature coefficient SKM100GB063D Diode parameters (H. A. Mantooth model) [3] John Vincent Reichl, “Inverter Dynamic Electro-Thermal Simulation with Experimental Verification”, M.sc Thesis, Virginia Polytechnic Institute and State University, 2005 ESR Network

IGBT Electrical Parameters IGBT thermal Dependent Coefficients IGBT parameters [3] IGBT Electrical Parameters Description A = 1cm2 IGBT chip Area Isneo = 1.5 x 10-12 A Emitter electron Saturation Current Wb = .015 cm Metallurgical drift region width VTO = 6.4 V MOSFET channel threshold voltage Nb = 1.5 x 1014cm-3 Base dopant density Kplino = 23 A/V2 Linear region transconductance Kpsato = 11.5 A/V2 Saturation region transconductance τHLO = .651 μs High level minority carrier lifetime Cgs = 3.98 nF Gate-source capacitance Coxd = 15.1 nF Gate-drain overlap oxide capacitance IGBT thermal Dependent Coefficients τHL1 = -2.36 High Level minority carrier lifetime temp. coeff. Isne1 = 1.189 Emitter electron saturation current temp. coeff VT1 = -7.2mV/K MOSFET channel threshold voltage temp. coeff. Kplin1 = 1.0286 Linear region transconductance temp. coeff Kpsat1 = 2.1032 Saturation region transconductance temp. coeff. SKM100GB063D IGBT parameters (Hefner model) ESR Network

Thermal model and parameters [4] IGBT Zth(j-c) Values Diode Zth(j-c) R1 160m k/W 400m k/W R2 88m k/W 165m k/W R3 18m k/W 30.5m k/W R4 4m k/W 4.5m k/W τ1 0.0447 s 0.0613 s τ2 0.0087 s 0.0085 s τ3 0.0015 s 0.0045 s τ4 0.0002 s 0.0003 s 4-order Forster RC thermal model for SKM100GB063D [4] [4] Datasheet of SKM100GB063D, SEMIKRON, 2006 ESR Network

Electrothermal calculation for comparison Thermal calculation results are in close agreement with our simulation results, as shown next. ESR Network

Solder fatigue simulation (case 1) (a) Junction temperatures Tj_t ↑15C , Tj_d ↑1C Tc ↑ Change of case temperature detectable! (b) Current harmonics Ia_5th ↓5.66 mA Ua_5th ↓9.4 mV (↓0.8%) Small! ESR Network

Solder fatigue simulation (case 1) (c) On-state resistances Ron_t ↑0.91 mΩ (↑2.42%) Ron_d ↑0.01 mΩ (↑0.09%) Small! (d) dv/dt td_off ↑ 13 ns td_on little change Small! ESR Network

Gate degradation simulation (case 2) (a) Junction temperatures Tj_t and Tj_d: little changes (b) Current harmonics Ia_5th ↓38.1mA Ua_5th ↓63.3 mV (↓5.3%) Perhaps Detectable ESR Network

Gate degradation simulation (case 2) (c) On-state resistances Ron_t ↓1.17 mΩ (↓3.12%) Ron_d ↑0.006 mΩ (↑0.05%) (d) dv/dt td_off ↑ 80 ns td_on ↓ 13 ns Notable! ESR Network

Bond-wire lift simulation (case 3) (a) Junction temperatures Tj_t ↑5C , Tj_d: little change (b) Current harmonics Ia_5th ↑29.0mA Ua_5th ↑48.2 mV (↑4.1%) Perhaps Detectable ESR Network

Bond-wire lift simulation (case 3) (c) On-state resistances Ron_t ↑3.526 mΩ (↑ 9.12%) Ron_d ↑0.007 mΩ (↑0.05%) Detectable (d) dv/dt td_off ↓54.0 ns td_on: little change Perhaps detectable ESR Network

Results and discussions Simulation results show that there are changes of converter electrical and thermal characteristics due to the device degradation. With respect to CM, some points can be made as below: Solder fatigue mainly influences the thermal behaviour of converter, while it has very small effects on the converter terminal electrical characteristics. It’s suggested that thermal characteristics based CM method might be suitable for detection of solder fatigue. Gate degradation mainly influences the electrical characteristics, especially the harmonics and dv/dt. Bond-wire lifting mainly influences the electrical characteristics, especially the harmonics and On-state resistances. Electrical characteristics based CM method might be more suitable for bond-wire lifting and gate degradation failures. However, high sensitive signal detection method is essential for the techniques. ESR Network

Targeted CM methods Thermal resistance observation based CM The accuracy of Rth observation under various operating conditions is important. Characteristics harmonics identification based CM About 0.1% precision could be achieved according to our preliminary research. On-state resistances identification based CM The change of converter terminal resistance can be identified according to the machine’s electromagnetic transient characteristics, when it is comparable to the machine winding resistance. EMI characteristics based CM The dv/dt of output PWM voltage will influence the converter EMI characteristics. It is expected that the change of high-frequency characteristics could be used for failure detection. A structured combination of different techniques might be appropriate in practice! ESR Network

Thanks for you attention!