Hottest analysis and thermal design technology

Thermal analysis and thermal design technology

both hot and cold will have a negative impact on the circuit. Under extremely high temperature, the chip may burn out (Figure 1). More often, if your design reaches an unexpected temperature, many components may exceed the specified limits. When this happens, the circuit may show unpredictable behavior. Another situation also deserves attention, that is, the circuit temperature changes from hot to cold, and then from cold to hot. This condition can cause thermal shock and damage components. Many engineers do not care about the performance of their circuits at low temperatures, but this neglect is a mistake. The performance of semiconductor devices will change significantly at low temperature. The base emitter junction voltage of bipolar transistors will increase greatly at low temperature (Fig. 2 and reference 1). Francisco Santos, product development engineering manager of analog devices, said, "if you want to design an amplifier that can work at 1.8V at a negative temperature, you need to consider that when the room temperature drops to -40 ℃, the VBE (basic emission voltage) will increase by 130mV. This situation will force designers to adopt a completely different amplifier architecture."

many amplifiers, such as the ad8045 of analog devices, will accelerate when cooling (Figure 3), while some amplifiers (such as ad8099) will slow down when cooling. Bill gross, the retired former vice president and general manager of signal processing products of linear technology, said: "most of the troubles encountered by bipolar transistors at low temperatures are low voltage operation." He believes that higher fundamental emission voltage and smaller current gain are more difficult to meet the specification requirements. He said, "the mismatch between lower input impedance and B (current gain) will cause big problems at low temperature. Especially when they are adjusted for room temperature operation. Higher GM (transconductance) can be easily compensated by changing the working current, but in this way, the conversion rate will change."

low temperature will cause oscillation, instability, overshoot, and poor filtering performance. Parts per million measurement can change the value of your component at high and low temperatures. If you expect the IC core to work at -55 ℃ ~ +85 ℃, it only needs 60 ℃ to reach the maximum temperature limit at 25 ℃, and the temperature difference from the ambient temperature to -55 ℃ is 80 ℃. Therefore, to find out your mistakes, you should check the hot and cold conditions. James McLaughlin, a professor of electrical engineering at Kettering University (Flint, Michigan), believes that when you heat a silicon wafer for more than a few Baidu, it will be "essential". In other words, when the temperature is high enough, the dopant will migrate through the lattice, and there is no PN junction, but just an impure conductive silicon wafer. Will the connecting wire explode? Or does the silicon continue to heat until it melts and evaporates

ic damage when operating at higher temperatures is elusive. Martin delateur, a consultant and former product engineer at national semiconductor, pointed out that the molding material began to carbonize at a temperature higher than 165 ℃. At this time, the molding material will change into a hard gray material. Gas release, that is, the slow release of gases captured, frozen, absorbed or adsorbed by some materials, will cause the release of polymeric additives such as flame retardants. At low level, this kind of gas release can affect the long-term and short-term operation of an IC, because it adds ions or surface effects to the chip. The connecting wire may transmit too much current, which will also cause the carbonization of the molding material. Excessive current will harden the carbon tube, which may melt the connecting wire, thus maintaining the conductive state in the tube. Finally, higher thermal expansion will crack the passivation layer, core, or carbonization molding compound, leading to large-scale failure. (military specifications define excessive current as more than 1.2 × 105a/cm2, so the army strongly requires IC to be fully sealed.) When there is no plastic material on the core, scorching and degradation will not occur. Oil well instrument companies often test the silicon IC used in their products at 200 ℃ and determine its characteristics. The service life of these products is limited, but the working time of manual hydraulic universal testing machine is still much longer than that of plastic packaging. Even if the core temperature is lower than 150 ℃, the life cycle of IC will be shortened

in 1884, the Dutch chemist Jacobus h van't Hoff first proposed the Arrhenius equation, while the Swedish chemist Svante Arrhenius physically verified and explained it five years later. This equation is: k=ae (-ea/rt), where k is the rate coefficient, a is a constant, EA is the activation energy, R is the universal gas constant, and t is. K is the temperature in units. Arrhenius initially applied this equation to chemical reactions, describing that the reaction rate increases with temperature (references 2 and 3). Today's engineers also use it to describe the shorter life of electronic devices working at high temperatures. The equation shows that the lifetime of the device will be halved when the temperature increases by 10 ℃. Therefore, it is important to reduce the temperature of silicon wafer in the design. If you can reduce the IC temperature from 85 ℃ to 65 ℃, the service life of these components can be increased by four times

the root cause of the problem is not only from the hot or cold static state, but also from the transition process from one temperature to another. In extreme cases, thermal shock can break circuit boards and devices into pieces. Temperature gradient (which will produce small voltage error) can also cause trouble due to the thermocouple effect of welding materials and pin materials (reference 4). In addition, the temperature gradient itself can be dynamic. The late Bob Widlar was a pioneering electronic engineer who worked for national semiconductor, Fairchild, Maxim and linear technology in the United States. He once received a prototype silicon chip that broke down at 1 kHz. Widlar believes that the heat wave comes from the radiation of the output transistor. These heat waves will spread evenly through the silicon core. The problem is that this IC has two reference nodes, and its distance from the output transistor is not equal. At the operating frequency of 1 kHz, one of the reference nodes is in a hot valley and the other is in a hot peak. This situation will lead to the imbalance of bias current and make the device unable to work normally. Due to these thermal gradients, some power supply designers prefer to use controllers rather than IC with built-in power FET. When using the controller, the heat of FET will not flow through the same core, amplifier and reference circuit

thermal analysis

thermal analysis of circuits is divided into three steps. First, estimate the heat generated in the IC. Then, estimate the heat dissipated by the circuit board or heat sink. Finally, estimate the ambient temperature at which the component will operate (Figure 4)

DC analysis is usually of little value in estimating the heat produced by the element. A resistor with a voltage of 1V and a current of 1A will generate 1W of heat. However, it is troublesome to estimate the heat generated by AC or undetermined signals. First of all, the static current from the power end to the grounding pin is always dissipating a DC power. A device with 10V power supply and 5mA quiescent current will generate 50MW heat. However, the quiescent current may change during operation. Bias current and base drive current usually increase when encountering AC signals. The biggest challenge is to calculate the heat generated by the output current of the device. This estimate may not be simple. The power provided by a device for a load is variable, but if the output transistor is normally open or closed, the power consumed inside the device is relatively small. As the traditional totem pole output stage used by most amplifiers, the heating is not the maximum when outputting a full swing square wave. The worst heating condition in IC is that the device outputs a square wave, whose amplitude is half of the power range. If the device operates at ± 12V, the square wave of ± 6vp-p will generate the maximum heat at the output stage. The internal heating of sine wave output is low. If the signal is complex or chaotic, it is difficult to estimate the real worst-case heating condition in the IC. If there are reactance loads with large capacitance and inductance components, the work of power consumption estimation will be more complex. Because the voltage and current are not in phase, the simple assumption about half amplitude square wave is not feasible

if you can determine the characteristics of IC passing signals, you can use spice to estimate power consumption. At this time, we must ensure to use appropriate SPICE models, which give reasonable results for some test signals, and power consumption calculation is of no value at this time. Figure 5 shows a spice diagram. The power consumption of the chip is different from the power reaching the load. Figure 6 is the spice curve of Figure 5. It shows the starting oscillation with a red line. Whether this oscillation will occur in the circuit is only a personal guess, but it should enable you to check this behavior after building the prototype. Remember, clicking the W key on OrCAD capture can only display the static power consumption of the chip. To obtain the working power consumption, use the power mark on the schematic diagram, and then use the RMS math function of the curve program to give the average power consumption of the device

the circuit board or radiator will dissipate the heat of the IC through convection, conduction or radiation. Conduction and heat dissipation are mainly through the metal lead frame and copper foil on the circuit board. Once the circuit board copper foil or discrete heat sink transmits heat, it provides enough surface area for convective heat dissipation to spread heat into the air. Radiation is hardly a feasible way to dissipate heat. Satellite designers use radiation because there is no other way to remove heat from the system. Since the radiation temperature in space is close to absolute zero, there is a large enough temperature difference so that a large amount of heat energy can be transmitted to space, so that the electronic equipment on the satellite will not overheat and burn out

convection heat dissipation also has some difficulties. For example, the effect of air flow on commercial fins (Figure 7). Note that at high temperatures, the thermal resistance increases fivefold. The cooling fins with forced air cooling have thinner fins with closer spacing, such as a fan type CPU cooler. If your product does not have a fan, the heat generated by the IC will be conducted and spread out, and then transmitted to the air in the machine. Next, as the temperature of the whole machine rises, heat is transferred to the surrounding air through convection. If you put the machine on your legs, part of the heat will also be transferred. The thermal resistance of the shell material becomes very important. The transfer speed of heat from inside to outside is slower than that of metal shell

engineers who make non cabin electronic equipment for jet fighters know that a jet plane has to fly up to 70000 feet. At this altitude, the air is very thin, and convective cooling is ineffective. These systems have a cold plate with glycol cooling channel to ensure that the temperature of the cold plate is not higher than 80 ℃. Each component maintains physical contact with a metal radiator, which transfers the heat of the component to the edge of the circuit board. At the edge of the circuit board, a heat transfer clamp system compresses the radiator on one side of the housing. The side of the casing transfers heat to the cold plate where the casing is located. The thermal grease can ensure the maximum heat transfer to the cold plate and the maximum conduction from the IC to the heat sink

most electronic engineers are familiar with using thermal resistance as a thermal analysis technique. Thermal resistance is expressed in degrees Celsius per watt. Simply multiply by the Watts estimated in the first step to obtain the temperature (degrees Celsius) that the component will increase. But here we need to pay attention to several issues. We should check the hidden information about the thermal resistance specification on the component data sheet. Thermal resistance from core to shell Φ JC is not a