Infinity Sensor: Temperature Sensing in GaN Power Devices using Peak di/dt

Jianjing Wang, Mohammad Hedayati, Dawei Liu, Salah-Eddine Adami, Harry Dymond, Jeremy Dalton, Bernard Stark

Research output: Chapter in Book/Report/Conference proceedingConference Contribution (Conference Proceeding)

32 Citations (Scopus)
1096 Downloads (Pure)


Di/dt has previously been proposed as a temperature indicator for Si and SiC devices, however, the evaluation of its viability for GaN devices is challenging as known current sensors introduce significant unwanted parasitic inductance. This work presents a figure-of-eight magnetic field sensor ( ∞ -sensor) that permits, for the first time, high-bandwidth floating current sensing, with negligible insertion impedance and influence on switching performance, in high-speed GaN and SiC switching circuits. The pair of coils are connected in a way that the measurement is immune to currents outside of the sensing region. The simulated bandwidth of the sensor, taking into account the loading by the probe connected to its output, is 225 MHz. The insertion inductance is 0.2 nH, and the insertion resistance is 4.2 mΩ at 100 MHz. This sensor is used to investigate the temperature dependency of turn-on di/dt in a 650 V, 52 mΩ GaN device. It is found that both average and peak turn-on di/dt decrease with temperature. Peak di/dt appears to be the preferred temperature indicator due to its high sensitivity and linearity.
Original languageEnglish
Title of host publication2018 IEEE Energy Conversion Congress and Exposition (ECCE 2018)
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
Number of pages7
ISBN (Electronic)9781479973125
ISBN (Print)9781479973132
Publication statusE-pub ahead of print - 6 Dec 2018

Publication series

ISSN (Print)2329-3721


  • Current sensing
  • Gallium nitride (GaN)
  • Silicon carbide (SiC)
  • Low insertion impedance
  • High bandwidth
  • Di/dt
  • Temperature


Dive into the research topics of 'Infinity Sensor: Temperature Sensing in GaN Power Devices using Peak di/dt'. Together they form a unique fingerprint.

Cite this