After the system was built, it is important to calibrate the measurements to boost its ac- curacy. This section introduces the calibration methods and the preparations needed before the power cycling test.
5.2.1
Calibration of Junction Temperature Measurement
There are many different methods to measure Tj. They can be sorted into three categories:
electrical, optical and physical contact [88]. Both optical and physical contact measurements can sense rapid variations in temperature and can easily provide temperature distribution of the chip surface. However, they require optical access to the chip which is not possible for packaged device.
Many electrical properties of the IGBT are highly temperature dependent. PN junction voltage drop, gate threshold voltage and leakage current are examples of temperature sensi- tive parameters (TSPs). A measurement of any above parameters can be used to estimate
Tj. Although TSPs can only provide average temperature of the chip, it is widely used in
real operation measurement since optical access is not available in most applications. In this work, PN junction voltage drop is selected as the TSP.Tj is estimated by measuringVCE at
a small calibration current (100 mA). According to Held [17], VCE decreases linearly with a
increasing Tj from 0 to 150 .
The calibration circuit is shown in Fig. 5.9. Three randomly selected test modules are connected in series and placed in the environmental chamber. The temperature of the chamber is controlled to increase from 25to 150 whileVCE of each IGBT are recorded
every 25. The test results are listed in Table5.3, whereVCE1.1andVCE1.2indicate the upper
Ic 15V 15V 15V 100 mA current source 15V 15V 15V
Envir
onm
enta
l c
ham
ber
Figure 5.9: Test configuration of junction temperature measurement calibration. between Tj and VCE can be observed and is described by Eqn. 5.2. The fitting result is
shown in Fig. 5.10. To obtain a more accurateTj measurement, further calibration for each
individual IGBT is designed before the power cycling test. Tj for each IGBT are measured
at room temperature when the rig is off. Due to the individual differences, slight adjustment of Eqn. 5.2 is normally necessary to ensure all the measured Tj are equal to the room
temperature.
Tj =−489VCE + 307.2 (5.2)
In order to build confidence on the TSP measurement method, infrared camera (IC) is used to measure Tj simultaneously. An IGBT module is opened and its silicone coat
which covers the chip is removed, as shown in Fig. 5.11. This module is then mounted on the heatsink and connected to the test circuit used for the thermal network extraction, as shown in Fig. 4.13(a). Run the test program and measure Tj throughout the test by IC.
Table 5.3: Relationship between Tj () and VCE (V)
Tj VCE1.1 VCE1.2 VCE2.1 VCE2.2 VCE3.1 VCE3.2 Average
25 0.576 0.576 0.576 0.575 0.577 0.576 0.576 50 0.527 0.526 0.526 0.525 0.527 0.527 0.526 75 0.476 0.475 0.475 0.476 0.475 0.475 0.475 100 0.427 0.427 0.425 0.426 0.425 0.424 0.425 125 0.374 0.374 0.373 0.373 0.373 0.373 0.373 150 0.321 0.320 0.321 0.320 0.321 0.319 0.320 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0 20 40 60 80 100 120 140 160 180 Tj ( o C ) VCE(V) TjVS... .VCE Linear fitting
IGBT on state voltage drop at different temperatures
Figure 5.11: IGBT module under test.
The measured temperature curves are shown in Fig. 5.12. This TSP measurement method only sense the temperature at time point 2 as shown in Fig. 4.13(b) which is considered to be the maximum junction temperature of one load cycle. While the IC could measure the chip temperature in real-time, hence it recorded the temperature drop due to the cut-off of the main current as well. However, the TSP method can be considered as a temperature measurement method with low sampling frequency (same as the frequency of the cyclic current) since the temperatures measured by TSP and IC at the same time point are very close as shown in the figure. In this work, only the maximum and minimum Tj needs to be
recorded hence real time measurement of Tj is not required. Therefore, the proposed TSP
measurement method can be used and the accuracy meets the design requirement.
5.2.2
Calibration of Thermocouples
Any conductor subjects to a thermal gradient will generate a voltage, known as ther- moeelctric effect. The main principle for thermocouple is using two different metals to complete a circuit in which two legs generate different voltages. The temperature difference between hot and cold junctions can be calculated by measuring the voltage difference at the cold junction, as shown in Fig. 5.13. To measure a single temperature, one junction, normally the cold junction, needs to maintain at a known constant reference temperature.
0 5 10 15 20 25 30 35 40 45 30 40 50 60 70 80 90 100 110 Temperature ( o C) Time (s) Tc Tj - TSP Tj - IC
Figure 5.12: Comparing the Tj measured by TSP and IC.
V Cold junction Hot junction Metal A Metal B Metal C
Thermocouples
Figure 5.14: Thermocouple placement.
In this work, K type thermocouples are selected to measureTc. This type of thermocouple
is the most common general purpose thermocouple with a sensitivity of approximately 41
µV /K. It is able to sense temperature between -200 and +1350 , which covers the
designed Tc range. Four slots are trenched on the heatsink to bury the thermocouples. It
is important to place the thermocouples in the same position right under each IGBT chip, as shown in Fig. 5.14. The misplacement of the thermocouple normally leads to a greater measured Rth, since the measured Tc will be smaller than the actual Tc. The measured Tc
can be adjusted in Labview by changing the CJC (cold junction compensation) value. Tc
measurement is calibrated by adjusting all measured Tc to the room temperature when the
rig is off.