Implement a readout temperature measuring mechanism

Article By : Ralf Ohmberger

A direct voltage method allows engineers to implement a readout temperature measuring mechanism using the MOSFET body diode of a PG pin.

Designers often want to measure the junction temperature of a DC switching power supply. Knowing a system’s junction temperature is an important and basic requirement from safety and failure mode and effect analysis (FMEA) standpoints. In many applications, it is important to measure the junction temperature under the maximum specified load and ambient temperature.

However, it’s often difficult in temperature chambers because a thermal camera is inaccurate and can be damaged under high environmental temperatures. Moreover, an external temperature sensor is difficult to fix onto small packages. This article describes a practical method to measure the junction temperature of an IC by demonstrating how to implement a direct readout temperature measuring method using the MOSFET body diode on a power good (PG) pin. It’s a direct voltage readout method that uses the diode voltage and temperature relationship.

Figure 1 Here’s the block diagram of a step-down converter that uses the PG N-channel MOSFET body. Source: Monolithic Power Systems

A fully integrated, fixed-frequency, synchronous step-down converter— the MPQ4572 in this case study—can achieve up to 2 A of continuous output current with peak current control. The device offers a 4.5 to 60V input voltage range to accommodate a variety of step-down applications. Apply a 1 mA current source in the forward direction through the body diode—a part of the MOSFET—on the PG pin (Figure 1).

The diode voltage vs. temperature curve can be measured using the EVQ4572-QB-00A evaluation board (Figure 2). It can also be measured directly from a custom board. Here, it’s important to note that the diode curve characteristic depends on temperature, rather than on the PCB dimensions.

Figure 2 The 4-layer evaluation board measuring 8.9×8.9 cm is used to measure diode curve characteristics. Source: Monolithic Power Systems

Measuring junction temperature with PG body diode

The PG pin has an internal N-channel MOSFET with a body diode. To accurately measure the junction temperature, the forward diode voltage and temperature must be calibrated. Follow the steps below for calibration:

1. Disconnect any resistor, microcontroller, or other part from the PG pin.

2. Glue a temperature sensor—like a 4-wire PT1000—on top of the device package that is being tested. Another method is to solder a floating thermocouple close to the device being tested; it’s recommended to solder this thermocouple to GND. Fixing temperature sensors to the package is a demanding task, so use the smallest possible sensors. The temperature sensor should not act as heat sink for the small package. Use thermal conductive glue to fix a PT1000 temperature sensor to the package, or use a thermocouple soldered directly to a part of the board that has EMC quiet potential (Figure 3).

Figure 3 The thermocouple is soldered to the PCB. Source: Monolithic Power Systems.

1. Connect a precision multimeter with a built-in diode test function and a 1 mA current source to the PG pin, as shown in Figure 1 and Figure 5. Smaller currents can be used, but the system must have the same current while it is calibrated and while measurements are taken.

2. Measure the forward diode voltage vs. junction temperature relationship in a climate chamber.

3. Measure the diode voltage when the device is powered by a supply voltage under the desired input voltage (V_IN). Determine which V_IN values have valid calibration because V_IN can influence the efficiency, and can therefore influence the device temperature as well. Do not connect a load on the DC/DC converter output.

4. Take measurements with the evaluation board or on a custom PCB.

5. Turn the device off.

6. Start the climate chamber, for example at 25°C, and ensure that the external temperature sensor shows a stable reading.

7. Turn the device on for a short period, and read the voltage on the multimeter. Without a load, the junction temperature should not rise significantly, because the power loss in the junction is low (only a few milliwatts). If possible, use advanced asynchronous mode (AAM) for its ability to ensure low quiescent current under small loads.

8. Turn the device off.

9. Set the climate chamber to the next selected temperature and let the PCB temperature settle for approximately 20 to 30 minutes, depending on the PCB’s specific heat capacity and size.

10. Turn the device on for a short period, and read the voltage on the multimeter.

11. Turn the device off again. Continue with the next selected temperature for the climate chamber.

12. Measure the forward diode voltage under the maximum desired load and maximum ambient temperature.

When measuring the PG forward diode voltage, keep the following in mind:

• The slope of this calibrated voltage vs. junction temperature is almost linear. For the highest accuracy, use more points and a polynomial fit function. Check the calibration for repeatability.
• Devices of the same type have similar slopes, but often different offsets.
• Similar devices will typically have slightly different slopes.
• Side effects such as a small change to V_OUT are possible. This should not be considered a failure as the coupled currents within the junction can cause such effects.
• The main advantage of this measuring method is that the forward diode voltage can be used to calculate the junction temperature under any load.
• A temperature sensor is not necessary.
• Note that not every part can use the PG pin to measure the current; contact the part manufacturer for product guidance.

Measured calibration curve

Figure 4 shows a first-order PG forward diode voltage vs. junction temperature graph, which has a linear fit function. The PG diode is driven by the external 1 mA current source shown in Figure 1.

Figure 4 The measured calibration curve on the EVQ4572-QB-00A has a linear fit function. Source: Monolithic Power Systems

By measuring the diode voltage, the junction temperature can be calculated using equation 1:

Diode and thermal camera readouts

Table 1 shows a direct comparison between the readout of the junction temperature and a visual thermal camera. The ambient temperature is measured with a platinum resistance type PT1000 (1,109Ω for 28°C).

Table 1 PG forward diode temperature readouts are compared with thermal camera readouts. Source: Monolithic Power Systems

Table 1 shows that the measured junction temperature is comparable to the thermal camera on the PG diode section of the package. The camera method shows lower temperatures, caused by thermal resistance of the mold compound between the junction and the package top surface.

The camera is adjusted to a 0.95 emissivity, which is a good fit for the mold compound of the package. The junction temperature is not consistent between components; for instance, the PG section inside the die is colder than the MOSFET section. Figure 5 shows the PG diode section and MOSFET section.

Figure 5 The block diagram of the MPQ4572 package highlights the MOSFET and PG sections. Source: Monolithic Power Systems

As shown in Figure 5 and Figure 10, the small signal section and the power MOSFET section are at different positions. The PG forward voltage diode measures the junction temperature at the PG position, so the diode temperature must be compared to the camera temperature at that position. Because the MOSFET has a higher temperature by a few degrees, that small offset must be added to the maximum junction temperature.

Figures 6-13 show the camera measurements corresponding to Table 1. These measurements were all taken using the EVQ4572-QB-00A evaluation board.

Figure 6 This photo shows the measurements when I_LOAD = 0 mA. Source: Monolithic Power Systems

Figure 7 This photo shows the measurements when I_LOAD = 10 mA. Source: Monolithic Power Systems

Figure 8 This photo shows the measurements when I_LOAD = 100 mA. Source: Monolithic Power Systems

Figure 9 This photo shows the measurements when I_LOAD = 500 mA. Source: Monolithic Power Systems

Figure 10 This photo shows the measurements when I_LOAD = 1,000 mA. Source: Monolithic Power Systems

Figure 11 This photo shows the measurements when I_LOAD = 1,500 mA. Source: Monolithic Power Systems

Figure 12 This photo shows the measurements when I_LOAD = 2,000 mA. Source: Monolithic Power Systems

Figure 13 This photo shows the measurements when I_LOAD = 2,500 mA. Source: Monolithic Power Systems

Regarding measurements shown in Figure 13, it’s worth noting that a continuous 2.5 A current is not recommended.

This direct temperature readout method simplifies the process for design engineers when testing a custom PCB in a temperature chamber, where a thermal camera cannot be utilized. It achieves fast, reliable, and accurate junction temperature data without the complicated and often intensive processes such as fixing temperature sensors on a device package.

This article was originally published on EDN.

Ralf Ohmberger is applications engineer at Monolithic Power Systems.

Related articles:

• Accurate temperature sensing with an external P-N junction
• Temperature Measurement, Part 1
• Transistor junction monitors remote temperature
• Make accurate temperature measurements using semiconductor junctions
• Thermocouples: Basic principles and design essentials
• Temperature-Monitoring Systems Optimize Cooling in Power Designs
• PCB Design Hints for Improving IGBT Thermal Performance

Subscribe to Newsletter