When current flows through a device, the power dissipated is the product of the voltage drop across the resistance of the conductive element and the current flowing through it. This power increases the temperature of the conductive element by an amount dependent on the speed with which the heat dissipates away from the element.
It has to cool off faster than it heats up …
The primary method of heat removal will be determined by the heat capacity and thermal resistance of the various materials between the junction and ambient. The junction temperature will rise until thermal equilibrium is reached.
How to cool it depends on what it is and what is used to cool it.
Thermal resistance can be considered an analogy to electrical resistance – thermal resistance opposes the flow of heat as electrical resistance opposes the flow of current. A voltage proportional to both current and resistance is developed across an electrical resistance as a thermal potential – 
T – proportional to both heat flow and thermal resistance is developed across a thermal resistance.
A resistor is a resistor is a resistor. It resists.
The value of thermal resistance is the ratio of a temperature drop across a material to the amount of heat generated under steady-state conditions. As with voltage, the thermal drop is related to some reference point – usually “ambient” which may be thought of as “ground”. Thermal resistance is usually expressed in units of degrees per watt.
The value of the thermal resistor is defined by how hot it remains between two different temperatures
The maximum power dissipation for a given application can be determined with knowledge of the thermal resistance and the maximum junction temperature rating. The maximum allowable temperature increase may be determined by subtracting the ambient temperature from the maximum allowable temperature. The resulting value is divided by the thermal resistance to obtain the maximum allowable power dissipation for that element.
If one knows how hot it can get – and how hot the place where it is is, one can tell how hot it can be before it gets too hot.
And then there’s math involved:
With an integrated circuit transistor, the heat is generated at the “junction”, then dissipated through the die to the package case, then to the ambient environment. The actual flow is much more complex than this but let’s consider a simple example. The 1N4005 is a well-established rectifier diode – I’ll use the ON Semiconductor version packaged in a plastic, axial lead, 59-10 case.
If the junction is at a temperature of 25°C, the maximum voltage drop is 1.1V with 1A current. The maximum (allowable) power dissipated at the junction is therefore:
The maximum thermal resistance between junction and ambient is specified as 
= 65°C/W when mounted flush to a PCB with ¼” leads to PCB mounting hole.
If the ambient temperature is 25°C, the junction temperature will be:
The maximum rated junction operating temperature is 175°C so there should be no thermal problem here … other than one probably wouldn’t want to touch the component (an intermediary thermal resistance – case-to-ambient – 
is not specified).
As a rough rule of thumb, if one can just momentarily hold finger on a component before flinching, the case is around 85°C. If the part glows red, it’s way too hot. If one flinches right away, it’s getting too hot. If one can hold a finger against the component, it’s probably OK even if “hot”. If it should be hot and it’s not, one may need to replace the part.
Thermal Impedance
Thermal resistance is defined under steady-state conditions. This assures a uniform temperature across the die (if applying the concept to a semiconducting element) where the power dissipation required to raise the junction temperature to a consistent and predictable temperature may be determined. For switching applications, this design criteria is excessively conservative (and possibly expensive), and the thermal capacity of the device should be considered. There are time constants involved with thermal dissipation and it may be shown that the ability to dissipate heat is frequency dependent.
A thermal equivalent circuit is shown:
Heat flow H is measured in terms of Watts (Joules/sec) and is analogous to electrical current (Coulombs/sec)
Thermal impedance 
is a complex variable due to the inclusion of thermal capacitance. The thermal resistances closest to the heat source (the “junction”) are high due to the small cross-section (low capacitance and high resistance) through which the heat flows.
The thermal capacitance varies directly with both mass and specific heat of the material. The small mass of the die has the lowest heat capacity, with capacity generally increasing as distance from the junction increases (the substrate being larger than the semiconducting junction itself). A usually acceptable assumption is that the “reference point” – the ambient temperature – has infinite heat capacity. This may be the mounting structure or surrounding air (but what if this is “space” – a vacuum?)
If a step function of power is applied to the junction, the increase in heat is related to diffusion and may be considered to have a 1st-order exponential response similar to that of a low-pass RC network (for the purposes of this discussion). Keep in mind that an actual device may have multiple and varying thermal time constants due to the different types, masses, and configurations of the various materials involved.
If the device is not mounted on a heat sink, the thermal resistance between case and ambient is usually the dominant factor between junction and ambient. A heat sink provides a means of lowering the case-to-ambient thermal resistance – effectively a parallel and lower resistance to “ground”.
The thermal capacitance of the case-to-ambient interface is used in calculating the temperature effects of pulse-mode operation. The effective time constant determines the cooling time required.
Consider a step-increase of 3V × 6A = 18 W in a package with thermal capacity of 1 J/s/°C = 1 W/°C
If the time constant of the package is 10 sec, the time required to cool the device to within 3°C of ambient (25°C) would be:
Heat Sinks
Methods to effectively remove heat from the junction are often limited to removing heat from the case – a user rarely has internal access to the packaged device. Some general guidelines for selecting a heat sink include:
– Contact surface area should be maximized
– Emissivity should approach unity (a fancy way of saying a “flat black” surface is preferred)
– Conductivity of the heat sink material should not cause excessive thermal gradients
To allow efficient heat transfer between the device and the heat sink, a thermal grease of some sort should be used – sparingly. A silicon-based grease is common but can cause problems in some applications or environments. The grease improves thermal contact by eliminating voids in the contact area, prevents detrimental oxidation of surfaces, and significantly increases heat conduction at the surface interface.
Heat sink grease will have a thermal resistance on the order of 0.05°C/W; a good heavy duty heat sink will have thermal resistance of around 2 – 5°C/W. It may be that airflow might be beneficial – a possible 10-fold improvement – although the addition of a fan to the system may not be an optimal solution.
That’s good for now.
