Real-time online prediction method for dynamic junction temperature of semiconductor power device

Abstract

The present invention discloses a real-time online prediction method for a dynamic junction temperature of a semiconductor power device. The present invention has advantages as follows: the sampling value of electrical parameters required for system closed-loop control is multiplexed as inputs, and no additional system hardware circuits and costs are needed; the processor resources can be saved to the utmost extent by using the idea of discrete iterative calculation, online calculation can be realized, and real-time performance of dynamic junction temperature calculation can be ensured; an optimal fitting dynamic thermal resistance discretization model is creatively proposed, and is used to perform iterative calculation, so that while real-time performance of dynamic junction temperature calculation of the power device is ensured, calculation accuracy is also ensured, meeting the requirements of protection, life prediction, and reliability design of the power device, and this method is very suitable for actual engineering application.

Claims

1. A method of making a semiconductor power device having an improved lifetime, comprising implementing a real-time online prediction method for a dynamic junction temperature of a semiconductor power device, comprising the following steps: step 1: sensing working state parameters of the semiconductor power device: an output current of the power device, a bus voltage of the power device, and a temperature of a radiator just under the power device; step 2: calculating loss parameters of the semiconductor power device by fitting curve method: a conduction voltage drop U.sub.CE, a turn-on energy E.sub.on, and a turn-off energy E.sub.off corresponding to a current I.sub.T flowing through the semiconductor power device at a given temperature; step 3: calculating a conduction loss P.sub.DC of the semiconductor power device in a calculation period T.sub.S according to the obtained conduction voltage drop U.sub.CE corresponding to the current I.sub.T flowing through the semiconductor power device:
P.sub.DC=U.sub.CE*I.sub.T*D wherein D is a conduction duty cycle of the semiconductor power device in the calculation period T.sub.S a time interval from step 1 sensing working state parameters to step 7 performing real-time online prediction to obtain the junction temperature of each semiconductor power device; step 4: calculating a switching loss P.sub.sw of the semiconductor power device according to the obtained turn-on energy E.sub.on and turn-off energy E.sub.off corresponding to the current I.sub.T flowing through the semiconductor power device:
P.sub.sw=(E.sub.on+E.sub.off)*f wherein f is a switching frequency of the semiconductor power device; step 5: acquiring a total loss P of the semiconductor power device according to the conduction loss and the switching loss acquired in the step 3 and the step 4:
P=P.sub.DCP.sub.sw step 6: calculating a temperature difference ΔT.sub.tjh1 between the junction of the semiconductor power device and the radiator surface just under the power device:
ΔT.sub.tjh1=ΔT.sub.tjh0+P*R.sub.thjc[@T.sub.S.sub.]−ΔT.sub.tjh0*λ wherein ΔT.sub.tjh0 is the temperature difference between the junction of the power device and the radiator surface just under the power device at the end of the previous calculation period, λ is a discrete coefficient that is obtained after optimal fitting is performed on a dynamic thermal resistance curve of the power device and that is related to the calculation period T.sub.S and the dynamic thermal resistance curve, and R.sub.thjc[@T.sub.S.sub.] is a transient thermal resistance of the power device corresponding to the calculation period T.sub.S; step 7: iteratively calculating the junction temperature of the power device at each moment: repeating the step 1 to the step 6, then adding the real time temperature of the radiator surface just under the power device to the calculation result ΔT.sub.tjh1 of the step 6, and performing real-time online prediction to obtain the junction temperature of each semiconductor power device.

2. A semiconductor power device system, comprising a semiconductor power device integrated thereon, having a dynamic junction temperature; a computer readable storage medium, having stored thereon a computer program and integrated in the semiconductor power device system, said program arranged to: operate a processor of the semiconductor power device to perform a method of predicting a real-time dynamic junction temperature of the semiconductor power device, comprising the following steps: step 1: sensing working state parameters of the semiconductor power device by original sampling circuits of the semiconductor power device working state parameters of the semiconductor power device: an output current of the power device, a bus voltage of the power device, and a temperature of a radiator just under the power device; step 2: sensing working state parameters of the semiconductor power device by original sampling circuits of the semiconductor power device loss parameters of the semiconductor power device: a conduction voltage drop U.sub.CE, a turn-on energy E.sub.on, and a turn-off energy E.sub.off corresponding to a current I.sub.T flowing through the semiconductor power device at a given temperature; step 3: calculating a conduction loss P.sub.DC of the semiconductor power device in a calculation period T.sub.S according to the obtained conduction voltage drop U.sub.CE corresponding to the current I.sub.T flowing through the semiconductor power device:
P.sub.DC=U.sub.CE*I.sub.T*D wherein D is a conduction duty cycle of the semiconductor power device in the calculation period T.sub.S, a time interval from step 1 sensing working state parameters to step 7 performing real-time online prediction to obtain the junction temperature of each semiconductor power device; step 4: calculating a switching loss P.sub.sw of the semiconductor power device according to the obtained turn-on energy E.sub.on and turn-off energy E.sub.off corresponding to the current I.sub.T flowing through the semiconductor power device:
P.sub.sw=(E.sub.on+E.sub.off)*f wherein f is a switching frequency of the semiconductor power device; step 5: acquiring a total loss P of the semiconductor power device according to the conduction loss and the switching loss acquired in the step 3 and the step 4:
P=P.sub.DC+P.sub.sw step 6: calculating a temperature difference ΔT.sub.tjh1 between the junction of the semiconductor power device and the radiator surface just under the power device:
ΔT.sub.tjh1=ΔT.sub.tjh0+P*R.sub.thjc[@T.sub.S.sub.]−ΔT.sub.tjh0*λ wherein ΔT.sub.tjh0 is the temperature difference between the junction of the power device and the radiator surface just under the power device at the end of the previous calculation period, λ is a discrete coefficient that is obtained after optimal fitting is performed on a dynamic thermal resistance curve of the power device and that is related to the calculation period T.sub.S and the dynamic thermal resistance curve, and R.sub.thjc[@T.sub.S.sub.] is a transient thermal resistance of the power device corresponding to the calculation period T.sub.S; step 7: iteratively calculating the junction temperature of the power device at each moment: repeating the step 1 to the step 6, then adding the real time temperature of the radiator surface just under the power device to the calculation result ΔT.sub.tjh1 of the step 6, and performing real-time online prediction to obtain the junction temperature of each semiconductor power device.

3. An on-line real time dynamic junction temperature prediction tool integrated on a semiconductor power device system, for use to predict a junction temperature for a semiconductor power device, included in the semiconductor power device system, comprising a computer readable storage medium of the semiconductor power device system, having stored thereon a computer program, said program arranged to: operate a processor to perform a method of predicting a real-time dynamic junction temperature of the semiconductor power device, comprising the following steps: step 1: sensing working state parameters of the semiconductor power device by original sampling circuits of the semiconductor power device, working state parameters of the semiconductor power device: an output current of the power device, a bus voltage of the power device, and a temperature of a radiator just under the power device; step 2: sensing working state parameters of the semiconductor power device by original sampling circuits of the semiconductor power device, loss parameters of the semiconductor power device: a conduction voltage drop U.sub.CE, a turn-on energy E.sub.on, and a turn-off energy E.sub.off corresponding to a current I.sub.T flowing through the semiconductor power device at a given temperature; step 3: calculating a conduction loss P.sub.DC of the semiconductor power device in a calculation period T.sub.S according to the obtained conduction voltage drop U.sub.CE corresponding to the current I.sub.T flowing through the semiconductor power device:
P.sub.DC=U.sub.CE*I.sub.T*D wherein D is a conduction duty cycle of the semiconductor power device in the calculation period T.sub.S, a time interval from step 1 sensing working state parameters to step 7 performing real-time online prediction to obtain the junction temperature of each semiconductor power device; step 4: calculating a switching loss P.sub.sw of the semiconductor power device according to the obtained turn-on energy E.sub.on and turn-off energy E.sub.off corresponding to the current I.sub.T flowing through the semiconductor power device:
P.sub.sw=(E.sub.on+E.sub.off)*f wherein f is a switching frequency of the semiconductor power device; step 5: acquiring a total loss P of the semiconductor power device according to the conduction loss and the switching loss acquired in the step 3 and the step 4:
P=P.sub.DC+P.sub.sw step 6: calculating a temperature difference ΔT.sub.tjh1 between the junction of the semiconductor power device and the radiator surface just under the power device:
ΔT.sub.tjh1=ΔT.sub.tjh0+P*R.sub.thjc[@T.sub.S.sub.]−ΔT.sub.tjh0*λ wherein ΔT.sub.tjh0 is the temperature difference between the junction of the power device and the radiator surface just under the power device at the end of the previous calculation period, λ is a discrete coefficient that is obtained after optimal fitting is performed on a dynamic thermal resistance curve of the power device and that is related to the calculation period T.sub.S and the dynamic thermal resistance curve, and R.sub.thjc[@T.sub.S.sub.] is a transient thermal resistance of the power device corresponding to the calculation period T.sub.S; step 7: iteratively calculating the junction temperature of the power device at each moment: repeating the step 1 to the step 6, then adding the real time temperature of the radiator surface just under the power device to the calculation result ΔT.sub.tjh1 of the step 6, and performing real-time online prediction to obtain the junction temperature of each semiconductor power device.

Description

BRIEF DESCRIPTION OF THE DRAWING

(1) FIG. 1 is a program block diagram of a real-time online prediction method for a dynamic junction temperature of a semiconductor power device according to the present invention.

DETAILED DESCRIPTION

(2) The following further describes the method of the present invention in detail with reference to embodiments.

(3) Embodiment: An IGBT module (FF600R12ME4) of a power device is used as an example to describe specific implementations of the method of the present invention (the present invention is effective to all power devices, and is not limited to the IGBT module). As shown in FIG. 1, a specific implementation of the method of the present invention mainly includes the following steps:

(4) Step 1: Obtain working state parameters of the IGBT module (FF600R12ME4): an output current of the power device, a bus voltage of the power device, and a temperature of a radiator just under the power device.

(5) For example, an average current of the IGBT module (FF600R12ME4) in a time period T.sub.S=10 ms is I.sub.T=300A, the bus voltage of the power device is 600V, and the temperature of the radiator is 100° C.

(6) Step 2: Obtain loss parameters of the IGBT module (FF600R12ME4) of the semiconductor power device, including a conduction voltage drop U.sub.CE, a turn-on energy E.sub.on(IGBT), and a turn-off energy E.sub.off(IGBT) corresponding to a current I.sub.T flowing through the IGBT of the power device at a given temperature.

(7) For example: at a given temperature, corresponding to the current I.sub.T=300A flowing through the IGBT module (FF600R12ME4), an IGBT conduction voltage drop U.sub.CE (FF600R12ME4)=1.4V, a turn-on energy E.sub.on(FF600R12ME4)=0.033J, and a turn-off energy E.sub.off(FF600R12ME4)=0.034J.

(8) Step 3: Calculate a conduction loss of the IGBT module (FF600R12ME4) of the semiconductor power device: calculating an IGBT conduction loss P.sub.DC(IGBT) corresponding to a calculation period T.sub.S according to the obtained conduction voltage drop U.sub.CE(IGBT) corresponding to the current I.sub.T(IGBT) flowing through the IGBT of the semiconductor power device:
P.sub.DC(IGBT)=U.sub.CE(IGBT)*I.sub.T*D

(9) where D is a conduction duty cycle of the semiconductor power device in the calculation period T.sub.S;

(10) For example, the conduction loss of the IGBT module (FF600R12ME4) is calculated as follows:
P.sub.DC(FF600R12ME4)=U.sub.CE(FF600R12ME4)*I.sub.T*D=1.4*300*0.5=210W

(11) The conduction duty cycle of the IGBT module (FF600R12ME4) in the calculation period T.sub.S=10 ms is 0.5.

(12) Step 4: Calculate a switching loss of the semiconductor power device: calculating a switching loss P.sub.sw(IGBT) of the IGBT of the semiconductor power device according to the obtained turn-on energy E.sub.on(IGBT) and turn-off energy E.sub.off(IGBT) corresponding to the current I.sub.T flowing through the IGBT of the semiconductor power device:
P.sub.sw(IGBT)=[E.sub.on(IGBT)E.sub.off(IGBT)]*f

(13) where f is a switching frequency of the semiconductor power device.

(14) For example, the switching loss of the IGBT module (FF600R12ME4) of the semiconductor power device is calculated as follows:
P.sub.sw(FF600R12ME4)=[E.sub.on(FF600R12ME4)d+E.sub.off(FF600R12ME4)]*f=(0.034+0.033)*5000=335W

(15) A working switching frequency of the IGBT module (FF600R12ME4) is 5000 Hz.

(16) Step 5: Calculate a total loss of the semiconductor power device. A total loss P.sub.(IGBT) of the semiconductor power device may be acquired according to the conduction loss and the switching loss acquired in the step 3 and the step 4:
P.sub.(IGBT)=P.sub.DC(IGBT)+P.sub.sw(IGBT)

(17) For example, the total loss of the IGBT module (FF600R12ME4) of the semiconductor power device is calculated as follows:
P.sub.(FF600R12ME4)=P.sub.DC(FF600R12ME4)+P.sub.sw(FF600R12ME4)=210+335=545W

(18) Step 6: Calculate a temperature difference ΔT.sub.tjh1 between the junction of the semiconductor power device and the radiator surface just under the power device. According to the obtained total loss P.sub.(IGBT) of the IGBT corresponding to the calculation period T.sub.S, a temperature difference ΔT.sub.tjh1(IGBT) as mentioned above after a calculation period T.sub.S can be accurately calculated by using a formula as below, using an optimal fitting dynamic thermal resistance discretization model:
ΔT.sub.tjh1(IGBT)=ΔT.sub.tjh0(IGBT)+P.sub.(IGBT)*R.sub.thjc[@T.sub.S.sub.]−ΔT.sub.tjh0(IGBT)*λ

(19) where ΔT.sub.tjh0(IGBT) is the temperature difference between the junction of the power device and the radiator surface just under the power device at the end of the previous calculation period, λ is a discrete coefficient that is obtained after optimal fitting is performed on a dynamic thermal resistance curve of the power device and that is related to the calculation period T.sub.S and the dynamic thermal resistance curve, and R.sub.thjc[@T.sub.S.sub.] is a transient thermal resistance of the power device corresponding to the time T.sub.S.

(20) For example, the temperature difference ΔT.sub.tjh1 between the junction of the IGBT module (FF600R12ME4) and the radiator surface just under the power device at the end of the calculation period T.sub.S=10 ms is calculated:
ΔT.sub.tjh1(FF600R12ME4)=ΔT.sub.tjh0(FF600R12ME4)+P.sub.(FF600R12ME4)*R.sub.thjc[@10 ms]−ΔT.sub.tjh0(FF600R12ME4)*λ
=25+545*0.015−25*0.366=24.025

(21) where ΔT.sub.tjh1(FF600R12ME4) is the temperature difference 25° C. between the junction of the IGBT module (FF600R12ME4) and the radiator surface just under the power device at the end of the previous calculation period, is a discrete coefficient that is obtained after optimal fitting is performed on a dynamic thermal resistance curve of the power device and that is related to the calculation period T.sub.S and a dynamic thermal resistance curve of the IGBT module FF600R12ME4, where 0.366 is selected herein, and R.sub.thjc[@10 ms] is a transient thermal resistance 0.015 of the IGBT module (FF600R12ME4) corresponding to the time 10 ms.

(22) Step 7: Iteratively calculate the junction temperature of the power device at each moment. Step 1 to step 6 are repeated, then the real-time temperature of the radiator surface just under the power device is added to the calculation result ΔT.sub.tjh1 of the step 6, and real-time online prediction may be performed to obtain the junction temperature of each power device.

(23) For example, the junction temperature of the IGBT module (FF600R12ME4) after the calculation period 10 ms is calculated as follows:
T.sub.tjh1(FF600R12ME4)=24.025+100° C.=124.025° C.

(24) Based on this method, the junction temperature of the IGBT module (FF600R12ME4) at any moment can be calculated.

(25) In the present invention, the sampling value of electrical parameters required for system closed-loop control is multiplexed as inputs, and it is realized by only adding a software algorithm to the system original control platform. On the one hand, the electrical parameter sampling circuits are multiplexed, and no additional hardware costs are needed; on the other hand, calculation parameters of software algorithms can be adjusted according to different measurement objects, so that junction temperature calculation for different series and different model numbers of power devices can be realized without cost addition.

(26) In this method, processor resources can be saved to the utmost extent by using the idea of discrete iterative calculation, online calculation can be realized, and real-time performance of dynamic junction temperature calculation can be ensured.

(27) In this method, an optimal fitting dynamic thermal resistance discretization model is creatively proposed, and is used to perform iterative calculation, so that while real-time performance of dynamic junction temperature calculation of the power device is ensured, calculation accuracy is also ensured, meeting the requirements of protection, life prediction, and reliability design of the power device.

(28) In the present invention, a direct calculation method of the junction temperature is used, no special processing needs to be performed on a power device, and the performance of power device is not affected at all, so that this method is very suitable for actual engineering application.

(29) The most important advantage of our junction temperature prediction method is that it is a real-time online junction temperature achieved by an optimal fitting dynamic thermal resistance discretization model with minimum CPU resources (storage space, computational complexity). In our motor drive system, 2 us is used for DSP 28377S. Completion of junction temperature calculation). In the method disclosed herein, it reuses the control CPU in the system, and sampling value of electrical parameters required for system closed-loop control is multiplexed as inputs from original sampling circuits without adding any hardware cost.

(30) The algorithm in the present invention is practical and has been used in motor drive systems. It can provide reliable real-time protection and life prediction for power semiconductor devices in motor drive systems.

(31) The present invention is not limited to the practical application background of the motor drive system, because in any device with power semiconductor devices, such algorithms and its possible improvement, can be integrated. For example, not being limited to, wind power drive, photovoltaic inverter, electric car and electrical drive and so on.

(32) One example of the power semiconductor device of the present invention is the power semiconductor device in a motor drive system. The algorithm disclosed herein is not an offline/online tool that performs operations on an external computer, but an algorithm that is integrated in the power electronic device, multiplexes the CPU of the power electronic device, and its hardware sampling circuit, and calculates the data in real time with minimal resources, a tool is disclosed herein, comprising an algorithm is integrated in the electronically controlled CPU of the semiconductor power device system.

(33) In one embodiment of the present invention, a semiconductor power device system is described. Power electronics systems includes semiconductor power devices. The semiconductor power device system comprises a semiconductor power device having a dynamic junction temperature; a computer readable storage medium, having stored thereon a computer program and included in the semiconductor power device system, said program arranged to: operate a processor of the semiconductor power device to perform a method of predicting a real-time dynamic junction temperature of the semiconductor power device, comprising the steps 1-7 of the detail description section.

(34) In another embodiment of the present invention, an on-line real time dynamic junction temperature prediction tool integrated on power electronics systems which including semiconductor power devices for use to predict a semiconductor power device junction temperature is described. The tool comprises a computer readable storage medium of the semiconductor power device system, having stored thereon a computer program, said program arranged to operate a processor to perform a method of predicting a real-time dynamic junction temperature of the semiconductor power device, comprising the following steps comprising the steps 1-7 of the detail description section.

(35) In the above aforementioned embodiments, the power electronics systems are considered as semiconductor power device systems, which including semiconductor power devices and the on-line real time dynamic junction temperature prediction tool is integrated on power electronics systems.