Thermal measuring and testing – Determination of inherent thermal property
Reexamination Certificate
1998-10-28
2001-03-20
Bennett, G. Bradley (Department: 2859)
Thermal measuring and testing
Determination of inherent thermal property
C374S029000, C374S020000, C374S039000
Reexamination Certificate
active
06203191
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to electrical current carrying devices. More specifically, the invention relates to semiconductor components such as diodes, transistors and microprocessors. More specifically, the invention relates to a temperature sensing method for remotely monitoring the junction temperature of a semiconductor component.
DEFINITIONS
Junction—A junction is an interface in a semiconductor device between regions with differing electrical characteristics. These characteristics determine the logic of the semiconductor device. The number of junctions per device can vary greatly. A diode may have a single junction while a microprocessor may have many millions of junctions. In the description and claims to follow, the use of the expression junction is intended to embrace both single and multiple junctions in semiconductor devices.
T
j
or junction temperature—The temperature of the substrate at the junction is often referred to in literature as T
j
.
IHCP or Inverse Heat Conduction Problem—This is a method that has been employed during the last 30-40 years in thermal analysis to determine the surface temperature or surface heat flux from transient temperature measurements at one or more points inside the body of the part. It can also be used to determine the heat flux and temperature at the source of a self heating body such as a semiconductor device. This type of problem is one in which the solution does not depend directly on the measured data. Methods for solving this type of problem are described in the Publications section above.
BACKGROUND OF THE INVENTION
As is well known in the art, semiconductor devices are widely used in various electronic components and devices such as transistors, integrated circuits, lasers, and the like. It is also well known that the passage of current through a junction results in a certain amount of power loss and heat generation therein. Continuous operation or frequent activation with minimal off periods may result in elevating the junction temperature. This elevated temperature can cause two problems. First, some integrated circuit (IC) devices are susceptible to drift (e.g., lasers). A temperature that is a function of loading can be a source of drift that is difficult to predict. Second, many devices such as microprocessors have a high number of junctions per volume. This results in devices that have very high power densities and are susceptible to overheating. This overheating may result in the failure of the semiconductor to perform its assigned circuit function and may sometimes involve the destruction of the semiconductor device itself. It is therefore important to monitor the junction temperature and perform a control or alert function based upon the results.
Prior art has taught several different methods of determining the junction temperature of IC devices.
1. Temperature sensing directly on the die of the IC.
2. Recreation of the IC into a more thermally predictable device.
3. Pure computational methods (no sensing).
4. Single point temperature sensing remote to the die of the IC (e.g., on the package of the IC) followed by an extrapolation of the junction temperature.
5. Single point temperature sensing remote to the die of the IC (e.g., on the package of the IC) and a measurement of the ambient temperature prior to extrapolation of the junction temperature.
Sensing on the die itself is accomplished in a variety of different ways. Some proposals (e.g., U.S. Pat. No. 5,639,163-Davidson et al., U.S. Pat. No. 5,555,152-Brauchle et al., U.S. Pat. No. 5,422,832-Moyal, U.S. Pat. No. 5,291,607-Ristic et al., U.S. Pat. No. 3,383,614-Emmons et al., U.S. Pat. No. 4,896,199-Tsuzuki et al., U.S. Pat. No. 5,406,212-Hashinaga et al., and U.S. Pat. No. 5,206,778-Flynn et al.) include the use of a monolithically integrated environmental sensor. One typical implementation of this environmental sensor is a pair of on-chip thermally responsive diodes coupled to a remote current source. The diode pair generates differential voltage output proportional to temp. Other proposals (e.g., U.S. Pat. No. 4,896,245-Qualich, U.S. Pat. No. 3,521,167-Umermori et al., U.S. Pat. No. 4,970,497-Broadwater et al., and U.S. Pat. No. 4,039,928-Noftsker et al.) rely on the fact that the impedance of internal circuitry varies as a function of temperature. Similarly, these circuits are driven by an external source and the resulting voltage drop is correlated to junction temperature. Despite the theory, in practice, the output of these circuits varies from one manufactured on-chip circuit to another to an extent that calibration particular to each on-chip circuit is required. Differences in construction and operation between sensors and semiconductor devices such as microprocessors have led the semiconductor industry to shun their integration into the same substrate. In operation, most sensors generate analog signals that have been difficult to process in digital microprocessors. Interface circuits used to couple the analog sensor signal to a microprocessor require additional semiconductor devices and further discourage monolithic integration of sensors and microprocessors. In addition, the inclusion of a sensing circuit on an IC die naturally results in a larger die to be fabricated. The manufacturing yield of devices such as microprocessors is inversely proportional to die size. Thus, the inclusion of sensing circuits into an IC result in a circuit that is more difficult to manufacture. Another difficulty with this type of technique is that the solution must be designed into a particular device. Devices already in existence cannot be sensed with this technique since they do not have the sensor on the substrate. Even if sensors are monolithically integrated onto the substrate, some environments that those components go into may be thermally challenging while others may not. Even if the environment is not thermally challenging and no sensing or control is required, the purchaser of this device to be used in this environment is still burdened with the extra cost and size of these devices.
Squires (U.S. Pat. No. 3,502,944), Demarest et al. (U.S. Pat. No. 4,117,527), and Barker et al. (U.S. Pat. No. 4,669,025) teach methods for recreating or simulating the thermal condition of the IC in question into a different form. The goal with these techniques is to overcome the shortcomings of on-die measuring mentioned above. This simulated IC is monitored and the control circuit drives the actual IC in response to the thermal state of the simulated IC. In practice, this simulation is very difficult to achieve. The actual IC and the simulated IC cannot occupy the same space. Therefore, the simulated IC and the actual IC are operating in different environments. Often in electronic devices, the environment can vary greatly between chips that are even right next to each other. In addition, if a thermal solution (heat sink, heat pipe, heat spreader, peltier junction, etc.) is imposed upon the IC of interest, the same thermal solution must be imposed on the simulated IC. Besides generating additional cost and complexity, an identical thermal solution is difficult to achieve primarily because of thermal impedances across interfaces of different materials. In other words, an IC and a corresponding simulated IC can be attached to identical heat sinks. The surface roughness of the components and heat sinks at the interfaces can vary. In addition, the forces clamping the heat sink to the IC may not be identical to the forces clamping the heat sink to the simulated IC. These and other factors can contribute to thermal impedances that are not identical between the component in question and its simulation. These differences can be dramatically reflected in the output making the simulation inaccurate. Even if the simulation and the actual component are reasonably similar, the solution is still problematic as the simulation generates additional heat. Thus, the overall heat generated by the system is greater than the heat generated by the component itself. The total h
Bennett G. Bradley
Fish & Neave
Pisano Nicola A.
Speculative Incorporated
Verbitsky Gail
LandOfFree
Method of junction temperature determination and control... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of junction temperature determination and control..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of junction temperature determination and control... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2457868