Application of sensors in power modules

Today, operating parameter monitoring has become an integral part of power modules. Temperature sensors have become more or less standard in power modules, and even current sensors are becoming more widely adopted. In fact, integrated sensors are a more cost-effective solution compared to external sensor solutions, which brings additional protection functions to users while reducing the size of the module.

Current Sensors

If a power module is equipped with a current sensor, its signal is mainly used for output current control (e.g. in transmission applications) and can also play a role in protecting the device. The requirements of the motor control determine the characteristics of the current sensor. In many cases, the fault (including temperature drift) must be less than 1 ... 2%. The requirements for temperature (-40℃ to 125℃) and low current consumption are set by the power module itself. The device protection function sets the overcurrent capability (maximum short-circuit current is 5 times the rated current) and the upper cut-off frequency (> 100kHz).

For low to medium power devices, using a current shunt is an accurate, cost-effective solution. The current limit is about 30A to 40A. The disadvantage is that there is additional power loss, and if the shunt is used to measure the emitter current, isolation will be lost and there will be interference in the IGBT gate signal.

For high-performance and high-power semiconductor modules, electrically isolated sensors are generally used. Pure Hall effect sensors without compensation currents have poor performance in terms of error and temperature stability. The sensors can be used in user-specified modules where the requirements are well defined. Sensors with high linearity and low temperature drift operate with a compensation current. This current counteracts the magnetic field of the measuring current in the sensor core. The control signal for the compensation current amplifier is provided by a Hall effect, magnetic field or magnetoresistive probe.

For intelligent power modules (IPMs) such as the Semikron SKiiP system, the use of high-precision sensors is most appropriate due to the high performance requirements of the final application. In the final application, the sensor is directly integrated in the housing of the module, surrounding the main terminals to save space (Figure 1). The evaluation circuit for signal monitoring and conversion is part of the driver circuit. Specially designed ASIC chips guarantee high integration and high reliability, which is difficult to achieve in solutions using external sensors.

Inside the IPM, the current monitoring circuit is directly connected to the driver circuit. It can detect external short circuits in the shortest time and shut down the power semiconductor within 2 to 3µs. In the future, this feature will become more and more important because the new generation of IGBTs only allow a short circuit time of 6 µs, compared with the 10 µs allowed by previous IGBTs.

The current sensor at the AC terminal of the voltage source inverter circuit cannot detect the short circuit inside the inverter bridge. Here, by monitoring V _CE(sat) , the slope resistance of the semiconductor in the on state is used for protection purposes. This method is sufficient for short circuit protection, but is not suitable for current measurement.

Figure 1: SKiiP power module with integrated current sensor at AC terminal

Temperature Sensor

For device protection, several temperature sensors are available. These sensors have a negative temperature coefficient (NTC) or a positive temperature coefficient (PTC). NTC sensors are the most commonly used in standard industrial modules. Semikron uses its own silicon chip sensor SKCS, which has PTC characteristics, high linearity and low errors. With a suitable monitoring circuit, an IPM such as SKiiP provides an analog output signal for temperature measurement and protection functions with a failure rate of less than 5°C.

The location of the sensor within the module greatly affects its ability to provide temperature protection. In fact, the location of the sensor is more important than the sensor's error in this regard. This is especially true if the hardware trip level is set by the driver or control circuitry.

Figure 2: Case study on different temperature sensor locations within a power module; model and temperature simulation

The effect of different sensor locations was investigated. A model of a power module is shown in Figure 2. The module has no copper baseplate and is mounted on an air-cooled aluminum heat sink. The thermal coupling of the different sensors varies, from sensor A) directly connected to the power semiconductor on the same copper layer, to sensors B) and C) isolated at different locations within the module, to sensor D) placed next to the module on the heat sink. Due to the different thermal coupling, each sensor has a different junction (j) to sensor (r) thermal resistance Rth _(jr) .

The trip level for overtemperature protection can be set for each sensor in quasi-static conditions. For example, if T _j cannot exceed 140°C, the trip level for "overtemperature shutdown" for the case system studied will vary from 120°C (sensor A), 110°C (sensor B), 100°C (sensor C) to 70°C (sensor D). The better the coupling between source and sensor, the lower the impact of the cooling system. This is a great advantage of the integrated solution.

However, for other cooling conditions (heat sink material and foundation thickness, cooling medium, thermal grease thickness), the trip level has to be set to a new value. This makes it difficult for IPM manufacturers to set the overheat trip level to an appropriate value for any given application. For this reason, the sensor signal should be monitored by an external host controller and, if necessary, the thermal protection level should be matched to the cooling system.

To show the influence of the cooling system, the thickness of the thermal grease layer was increased from 50 µm to 100 µm. Sensor A, which has the best thermal coupling to the power semiconductor, shows the lowest influence on R _th(jr) , with an increase of only 3%. Sensors B and C have an increase of 7…8% in R _{th(jr) . The cooling system has the greatest influence on R th(jr)} for sensor D , with an increase of more than 25%.

Another question is whether the temperature sensors can protect the power semiconductor in the event of a short-term overload. Each sensor reacts to the increase in junction temperature with a delay that is related to the sensor's location. This behavior is described by the thermal impedance Z _th(jr) . It does not behave as expected (see Figure 3). A comparison of Z _th(jr) with the thermal impedance from junction to heat sink Z _th(js) (directly under the chip) shows that the system's junction-to-heat sink thermal impedance reaches steady-state conditions after one second, while the system's junction-to-sensor requires 100 seconds to reach steady state. The reason for this is the heat spreading inside the heat sink.

Figure 3: Thermal impedance of junction (j) to sensor (rX) and heat sink at different locations

[page] For each power semiconductor, a maximum value for its quiescent power dissipation P _tot is specified. For the example overload jump from 50% P _tot to 200% P _tot , the semiconductor will overheat after some time. Sensor A will reach its trip level of 120°C after 0.19s, providing reliable device protection and keeping the junction temperature at around 150°C. The junction temperature of the device protected by sensors B and C will be in the critical range of 160°C to 170°C; in these cases, the sensors require 0.3…0.4s to reach the trip level. Depending on the characteristics of the device, this may mean that the limits specified in the data sheet have been exceeded. Sensor D has a reaction time of more than 1 second and is therefore unable to protect the device. In the case of very high overloads and low start-up temperatures, the temperature sensors do not provide any adequate protection.

An overview of the pros and cons of the different temperature sensor locations is given in Table 1. Sensors in position B are the preferred solution today due to isolation. If in the future the driver has protection circuitry and the signal is transformed on the secondary side of the driver, this may mean that sensor position A may be a better solution.

Integrated protection

If a short-term overload occurs, there will be a gap in the device protection. The cut-out value of the current sensor is set to a high value to allow for short-term overloads, such as when the motor starts. Long-term operation at this current level will cause the device to overheat. In most cases, the reaction time of the temperature protection element is too long to detect this overheating.

One possible way to fill this gap is to use software shutdown of the current and temperature signals. The inverter controller calculates the junction temperature based on the temperature of the sensor and the electrical operating conditions. The junction temperature at time _tp can be calculated as follows:

P0 is _the power consumption at t=0s, and Pover _is the power consumption when overloaded. Here, the thermal impedance Zth _(jr) is as described in the data sheet, and the analog temperature signal Tr _is also required.

Table 1: Comparison of temperature sensors in different locations for protecting power semiconductors.

Summarize

The integrated sensor in the IPM protects power modules like the SKiiP over a wide range of operating conditions. Equipped with a suitable evaluation circuit, it can provide high-quality information for process control as a synergistic effect. This saves space, costs and development time. With an external observer, the combination of the available sensor signals can fill in the gaps for specific protection in the application.