Active control of boundary conditions for dynamic thermal characterization of electronic components

Abstract

This paper presents a practical realization of the system for the active control of boundary conditions during the dynamic thermal characterization of electronic components. The control of boundary conditions is exercised by the dual cold plate cooling assembly equipped with Peltier thermo-electric modules and an appropriate control circuit. Additionally, a tensometer bridge is used to assure the parallel alignment of surfaces and to adjust the contact thermal resistance between particular layers. The operation of the entire control system is illustrated based on a practical example where Peltier module currents are adjusted in real time so as to impose isothermal or constant heat flux boundary conditions on a power diode package surface during measurements performed with time varying power dissipation in the device.

Introduction

Thermal characterization of electronic system components requires reproducible transient temperature measurements taken in precisely defined cooling conditions. Such measurements usually are performed using the so-called Dual Cold Plate (DCP) cooling assembly which is supposed to stabilize temperature on both faces of a package during the entire measurement time. For example, the DCP cooling is used in the standard methodology elaborated within the DELPHI and PROFIT projects. According to this methodology, boundary independent dynamic compact models of electronic systems are generated based on the experimental data obtained from four transient temperature measurements performed in a DCP system; each time forcing different cooling conditions of a package, i.e. when package faces are either cooled or thermally insulated (see Refs. [1], [2] for details). However, the main drawback of this procedure is that the exact cooling conditions are not known and in addition they are not constant even during a single measurement because the use of cold plates alone is not sufficient to stabilize package temperature.

This problem is illustrated by a following simple example presenting the measurement results of a silicon carbide power diode, rated for 6 A and 600 V and packaged in the standard TO-220 package. This diode, applying some thermal grease, was squeezed between the cold plates with the copper spacers placed on the top and the bottom of the package, thus assuring good thermal contact with the cold plates during the measurement. The measurements were taken for three different values of the forward current flowing through the diode (4, 6 and 8 A) and three different values of cooling liquid temperature (30, 60 and 90 °C). The temperature values were measured inside the copper spacers with thermocouples located 1 mm from the surfaces contacting with the diode.

The measurement results, presented in Fig. 1, show the temperature rise of both spacers over the cold plate temperature due to the power dissipation in the diode. The solid lines correspond to the bottom spacer contacting the heat slug whereas the dashed ones indicate the top spacer contacting directly the package. Different markers are used to distinguish the cold plate temperature values.

Theoretically, this measurement configuration should assure boundary conditions on both sides of the package close to the isothermal ones, but as can be seen such conditions cannot be attained in practice. Moreover, they differ significantly on both sides of the package. Namely, on the top side the temperature is fairly constant and increases slightly only at higher power dissipation values, especially when the cold plate is cooler because of poor radiation cooling. Quite the opposite, on the side of the heat slug, where the most of the generated heat is evacuated, there is an important temperature rise, exceeding even 16 K at the power dissipation of 20 W. Consequently, the boundary conditions on this side of the package resemble rather the constant heat transfer coefficient or the constant thermal resistance case.

The above-described problem could be solved by inserting on both sides of the package the Peltier Thermo-Electric Modules (TEMs) which should be placed between the cold plates and the spacers. When, based on the indications of temperature and heat flow sensors, the current flowing through the TEMs is controlled using an appropriate algorithm, the temperature of the package outer surfaces could be kept a desired level virtually for any power dissipation in the diode, thus allowing the practical realization of the variable thermal resistance. This concept, adopted also by the authors in this paper, is not new and was already presented in literature [3], [4], [5].

Another problem occurring in the thermal measurement procedures discussed here is how to assure reproducible contact resistance values in various experiments, especially when the measurement setup is disassembled in the meantime, e.g. in order to change the spacers. This problem could be solved introducing a special force balancing system and a thin membrane tensometer bridge allowing the proper alignment of surfaces and the precise adjustment of the squeezing force.

Concluding, the authors propose in this paper an enhanced version of a DCP system, whose cross-sectional view is presented in Fig. 2. This system realizes in practice not only the concept of variable thermal resistance but also it allows dynamic thermal measurements with adjustable contact thermal resistance. Although the main idea of the proposed system was adopted from the previously published literature, the practical realization differs substantially, particularly in the design of the electronic control system. Moreover, the current version possesses some novel features which, at least to the authors' knowledge, have never been considered before. As a result, the new system allows the registration of thermal transients with sensor sampling rates reaching 1 Msps not only in the precisely controlled cooling conditions but also with known contact resistance values.

The remainder of this paper is organized as follows. First, the technical details of the cooling condition control system practical realization are presented. Then, the system test and calibration results are discussed. Next, the cooling condition control algorithm is described. Finally, the operation of the entire system is verified based on practical examples where the temperature and the heat flux are stabilized on a surface of a power diode package during the measurements carried out with variable power dissipation.