Temperature Problems Solved for You (III) High-Performance Processor Mold Temperature Monitoring

In the previous article , we have introduced how to monitor circuit board temperature. However, power management in high-performance processors such as central processing units (CPUs), graphics processing units (GPUs), application-specific integrated circuits (ASICs), and field-programmable gate arrays (FPGAs) is often more complex. With temperature monitoring, these systems can not only initiate safe system shutdown procedures, but also use temperature data to dynamically adjust performance.

Monitoring process temperature can improve system reliability and maximize performance. As shown in the figure below, high-performance processors often use heat sinks to absorb excess heat from the die. Higher temperatures may activate cooling fans, modify system clocks, or quickly shut down the system if the processor exceeds its temperature threshold.

Motherboards equipped with high-performance processors usually require a heat sink

Design Considerations for Die Temperature Monitoring

To achieve efficient temperature monitoring, high-performance processors have two design considerations: temperature accuracy and sensor placement. The temperature accuracy of the processor is directly related to the sensor location.

Improve system performance with high-precision temperature monitoring

As shown in the figure above, processor performance can be maximized through high-precision temperature monitoring, pushing the system to its temperature design limit. Although most integrated circuits have built-in temperature sensors, the accuracy of these sensors is not consistent due to differences between wafers and other batches. In addition, the processor must be conditioned based on a benchmark to adjust the coefficient relative to the die temperature. High-performance processors have complex circuits and cause self-heating, which will produce temperature errors that increase with temperature. If the system is designed with lower accuracy and temperature errors, the performance of the system will not be maximized within its temperature design limit.

Sensor placement and accuracy

An integrated temperature sensor or temperature diode or an external temperature sensor can monitor the thermal performance of the processor. In some cases, using both internal and external sensors can maximize system performance and improve reliability.

Bipolar Junction Transistor Integrated Temperature Sensor

Some high-performance processors contain bipolar junction transistors (BJTs) for temperature sensing. BJTs have a transfer function that is temperature dependent and very predictable. Remote temperature sensors use this principle to measure the die temperature. The most common BJT in complementary metal oxide semiconductor processes is the P-channel N-channel P-channel (PNP). The following figure shows a remote temperature monitoring circuit for measuring a PNP transistor connection configuration.

Measuring the base-emitter voltage change (ΔVBE) using two currents

The process of designing a remote temperature monitoring system can be challenging due to noise and errors caused by wafer-to-wafer and batch-to-batch variations. Temperature diode errors can be caused by:

• Ideality factor variation. The characteristics of a BJT temperature diode depend on process geometry and other process variables. If the ideality factor, n, is known, the n-factor register can be used to correct for n-factor errors. Alternatively, a software calibration method can be used to correct for ideality factor variations over the desired temperature range.

• Series resistance. Any resistance in the signal path will cause a voltage offset due to the current source. Modern remote temperature sensors use a series resistance algorithm to eliminate the temperature error caused by resistances up to 1-2kΩ. This algorithm produces robust, accurate measurements even when combined with a resistor-capacitor filter.

• Noise injection. When the diode traces are run in parallel with high-frequency signal lines carrying high currents, electromagnetic interference or inductance coupled into the remote pc board traces can cause errors. This is one of the most important board design considerations for remote temperature sensors.

• Beta compensation. Temperature transistors integrated into FPGAs or processors may have a Beta value less than 1. Remote temperature sensors with Beta compensation are designed to work with these transistors and correct for the temperature measurement errors associated with them. The Beta compensation feature does not provide any benefit when used with discrete transistors.

Device Recommendations

The TMP421 provides a single channel to monitor a BJT; there are also multi-channel remote temperature sensors supporting up to eight channels for measuring temperature locally and remotely.

The TMP451 provides high accuracy (0.0625°C) temperature measurement both locally and remotely. Server, notebook, and automotive sensor fusion applications can benefit from multi-channel remote sensors.

External temperature sensor

Although the built-in temperature sensor is optimally located, its accuracy is as low as ±5°C. Adding an external local temperature sensor can improve the die temperature accuracy and improve system performance. A local temperature sensor can also be used when an integrated die temperature sensor is not available. However, with a local temperature sensor, sensor location is an important design consideration. The following figure shows some options for placing a local temperature sensor: positions a, b, and c.

High-performance processor temperature monitoring via sensor placement

• Position a. The sensor located in the center drilled hole of the microprocessor heat sink is very close to the die. The heat sink can be clamped to the processor or epoxied to the top of the processor. Temperature sensors in this location generally require long leads, and the sensor data will become inaccurate as the thermal conductivity from the heat sink to the microprocessor decreases.

• Position b. Another potential location for the sensor is in the cavity below the processor socket, where assembly is very straightforward. Since the sensor is isolated from the airflow, the ambient temperature has minimal effect on the sensor reading. Additionally, if the heat sink is separated from the processor, the sensor will show an increase in processor temperature. Nonetheless, with this sensor placement, the temperature difference between the sensor and the processor could be between 5°C and 10°C.

• Position c. The sensor can be mounted on a circuit board next to the microprocessor unit (MPU). While this mounting arrangement is easier to implement, the correlation between the sensor temperature and the MPU temperature is much weaker.

Device Recommendations

Footprint is a factor to consider when selecting a local temperature sensor. The TMP112 is available in a 1.6mm x 1.6mm package, allowing it to be used close to the processor. The 0.5°C accuracy of the TMP112 device maximizes performance compared to temperature sensors integrated into the processor, which typically have an accuracy of only 5°C to 20°C.

Click here to quickly locate the TI analog column and view the latest and most comprehensive information on TI sensor products. At the same time, in the next few articles, we will focus on the design considerations of various applications, evaluate the trade-off between temperature accuracy and application size, and discuss sensor placement methods.

Temperature problems solved for you (I) Basic principles of temperature sensing

Temperature Problems Solved for You (II) System Temperature Monitoring