When you watch a movie on your laptop, the fan almost immediately starts to sound like a jet engine, and feels warm enough to start a fire. If you leave your smartphone in the glove box in the summer, it won’t work until it cools down. Keeping electronics a reasonable temperature is not magic, although it may feel like it at times as the rules and situations are not very familiar to electrical engineers. The effects of heat on electronics are also somewhat difficult to predict and often can be better understood using statistics, such as mean time between failures to show the lifetime averages.
When current flows through a resistor, power is dissipated in the form of heat. Resistors are rated for the amount of power they can dissipate before they burn out or burn up. Electronics are filled with transistors that turn on and off at literally unimaginable speeds but in some ways act as resistors. Transistors are typically understood to be in two states, open, wherein no current flows, or closed, wherein the resistance is nominally zero and current flows unimpeded. However, sometimes reality does not exactly match the concept. For example, when a transistor is open, there is a small leakage of current through what is considered a very high resistance burning a small amount of power. When the transistor is closed, the resistance is not truly zero, and that minor resistance also adds to the total used power. However, the greatest amount of power used is during the transition between the two states.
Although assumed to be a perfect square wave for most considerations, clock signals have finite edge widths where a transistor is partially on and partially off. During this time, a current is flowing yet the resistance is much higher than when the transistor is in the closed state. In the large view, the amount of power in one clock cycle for that transistor is minuscule. However, when you multiply that one transistor by the hundreds of millions, or even billions of transistors in a processor, it jumps phenomenally. Multiply that by the billions of cycles per second and suddenly the trivial amount of power consumed has become a serious concern. Designers have tried many different methods to reduce the power consumed by making the clock edges sharper, at the expense of increased cost and electromagnetic emissions, or by decreasing the voltage level, which increases the susceptibility to spurious signals or digital errors. Many processors now are able to shut down portions of the chip when not in use to decrease power usage. This is very helpful, yet does not decrease the peak power usage.
In addition to avoid burning holes in desks, there are many reasons why we care about how much heat is being generated by electronics. The effects of heat on electronics vary greatly, and they are not always negative. For example, the chemical reactions within batteries tend to work better at room temperature. Yet batteries store better at cooler temperatures and becoming too hot while running can lead to catastrophic and explosive results.
Most electronics have a minimum and maximum temperature rating, typically 0C to 70C for commercial applications. The minimum temperature rating of electronics is usually not an issue except during the winter when electronics are left in the car or attic. Most of the concern related to cold is either the condensation that could be formed when bringing the electronics into the warmth, much like glasses fogging when coming inside, or the rapid expansion of traces and connections when electricity flows reheat them. Fortunately, these problems are easily avoided by either keeping electronics at a reasonable temperature or letting them come to room temperature before using them.
The damage caused by heat has some similarities with those with cold. Much as a cold part coming to room temperature causes unequal expansion of the different materials, changing from room temperature to levels better suited for baking cakes also causes different expansions. As different materials have different coefficients of thermal expansion, some materials will expand much more than others and break connections. Another issue created by excessive heat is the increased resistance of copper, which creates a spiral effect and can contribute to runaway conditions. Even if the circuit remains functional and the temperature levels off at acceptable levels, the increased temperature and resistance means that more power is being used by the circuit to accomplish the same things, thus decreasing the overall efficiency. While copper increases its resistance with increasing temperatures, the silicon substrate of most integrated circuits is also susceptible to heat, causing it to lose its semiconductor properties over a certain threshold. This leads to completely unpredictable behavior changes in the circuit. Finally, excessive heat can also encourage whiskering, a concern that has already increased due to the transition back to lead-free solder.
There are a myriad of ways to approach heat management. The simplest could be to reduce the amount of heat created by your product. Microprocessor manufacturers over the last two decades have struggled with keeping the temperature of their products at a reasonable level as they produce a large amount of heat in a very small area. As mentioned, decreasing the voltage, and therefore power consumption, has been very successful in the past. There are other ways to decrease the power output at more of a board versus IC level. Any power conversion is a great opportunity to find savings in heat and efficiency, such as replacing a low drop-out voltage regulator with a switching regulator. This has its own trade offs, efficiency for cost and electromagnetic noise, however, it is a viable option. Other ways to increase efficiency are to reduce any bit-banging that may be required on any embedded microcontrollers. By choosing systems that are able to handle complex or difficult data management in hardware, the amount of required instruction cycles is reduced, allowing more time to go into sleep mode, wherein the chip has time to cool.
When testing for temperature reliability, try to emulate real-world conditions as much as possible. Temperature testing is not purely about surviving certain temperatures, it includes other variables such as air flow, humidity levels, and concurrent physical strains. A product that has been tested up to one hundred degrees Celsius with forced air does not indicate whether or not that product will, in an enclosed environment without forced air, heat itself up to even higher temperatures. Actual tests, like computer simulations, are subject to the same rule of garbage in, garbage out. If the parameters are not correctly setup, then the test results will not be helpful.
Yet another aspect of engineering that must be taken into consideration when designing is that temperature control and testing are not the be all, end all of any product design. A well-designed product will balance all of the needs of the project in an efficient, cost-effective solution to best provide a solution for your customer. Creating a device that always runs well below the temperature threshold but is too bulky to be conveniently used is a failure compared to a device that runs barely within the temperature threshold and has a convenient form factor. However, creating a device that has a great interface but survives less than three days under normal operating conditions is an absolute failure. Keep heat management in your personal toolbox and pull it out every once in a while during the design to make sure that you are on the right track, not just designing another fancy looking brick.