Many electronic designs require more than one voltage rail for different portions of a system. For example, one might have a higher voltage motor which is controlled by lower voltage electronics. In such a system, it is most convenient to supply the overall system with the highest voltage required, and then step that voltage down to a lower voltage rail for the control electronics.
The go-to device for many people designing such systems is a linear voltage regulator. Linear regulators are conceptually simple devices, and they typically do the job without any problems. However, reality is not always so kind. What makes sense in concept, and what might work on the prototyping bench, doesn’t always work in the real world.
A common complaint we get here at DigiKey comes from people who buy a linear regulator from us and then call in, telling us that they “got a bad regulator”. The symptoms are either that the part shuts down (no output voltage), or in some cases, it actually overheats and dies.
Most problems people have with linear regulators are caused by a lack of understanding of how they work. They attempt to drop too much power across the regulator, which creates more heat than it can dissipate. Even with a good sized heat sink, the device may overheat if the output current is toward the upper end of its specified range and the differential between the input and output voltage is too high.
For example, if one required an output voltage and current of 5V and 1A, respectively, and had an available supply voltage of 24V, one might assume a regulator from the venerable 78xx family, such as the MC7805BTG would suffice. After all, according to its specifications, it is rated to handle an input voltage of up to 35V, and it can output a regulated 5V at up to 1A, steady-state.
The problem is that the device can only meet those individual specifications so long as the maximum internal die temperature (often referred to as the “maximum operating junction temperature”) is not exceeded. Unless heroic levels of cooling are applied to the body of the regulator (think refrigeration and liquid cooling), it is generally not possible to reach all maximum specifications of the part simultaneously. To make a functional design, one must either:
- Reduce the load current
- Reduce the voltage differential between the input and the output
- Increase heat sinking
- Do some combination of the above
Typically, the load current and voltage requirements are fixed, so one can’t do much there. Reducing the input voltage is sometimes possible, and will have the largest impact on internal power dissipation if it can be lowered. Finally, a larger heat sink can be added, or forced air can be applied with a fan, to more effectively remove heat from the package.
To figure out how much heat sinking is required for a particular application, one must consider the maximum internal die temperature of the part, the maximum expected air temperature around the circuit board when the end-device is running (referred to as the “maximum ambient air temperature”), and the thermal resistance of the interface between the internal die and the ambient air.
Thermal resistance is specified in °C/W, and both the device package and any applied heat sink will have a thermal resistance characteristic. For instance, the common TO-220 case of the MC7805BTG has a thermal resistance of about 5°C/W and a maximum junction temperature of 150°C. A typical TO-220 heat sink might have a thermal resistance of around 15°C/W. This combination gives a combined thermal resistance of 20°C/W between the internal die and the ambient air.
From MC7805 datasheet:
Using the previously described example, the approximate level of power dissipated by the linear regulator can be calculated by finding the difference between the input power and the output power of the system. Note that with a linear voltage regulator, the input current is nearly the same as the output current (the input current is actually slightly higher, but insignificantly so, for the purposes of this discussion).
- Input power: P(in) = V(in) x I(in) = 24V x 1A = 24W
- Output power: P(out) = V(out) x I(out) = 5V x 1A = 5W
- Regulator power dissipated: P(reg) = P(in) - P(out) = 24W - 5W = 19W
So, how high will the die temperature rise given the above set of conditions?
- Die Junction Temperature = (Case Thermal Resistance + Heatsink Thermal Resistance) x Internal Power Dissipated = (5°C/W + 15°C/W) x 19W = 380°C above ambient air temperature!
That’s pretty darn hot! Even with a really efficient heat sink with a thermal resistance of 5°C/W, the die junction temperature would rise 190°C above ambient air temperature. If the ambient air temperature was 25°C, the die temperature would be 215°C, which is well above the maximum 150°C limit of the regulator.
The point here is that a linear voltage regulator is not a magic component. One has to consider not only the voltage and current requirements of the system, but also the power handling limits of the regulator. As power dissipation requirements of the regulator increase, one may need to improve heat sinking and/or lower the input-to-output voltage differential to prevent the die junction temperature from exceeding the specified maximum temperature under all expected load and ambient air temperature conditions.
As has been described above, there may be cases in which a linear regulator is simply not capable of handling the requirements of a system, and alternatives will need to be considered. One of the most useful alternatives to try is a special form-factor of switching DC-DC converter designed to fit into the same footprint as the typical TO-220 style linear voltage regulator. Digi-Key refers to these as “Linear Regulator Replacement” DC-DC converters.
These devices can drop into the same footprint as the typical 78xx type linear regulator and generally do not require any heat sinking at all. They are much more efficient than a linear regulator (usually 80% or higher), so not only do they generate far less heat, but they can extend battery life significantly in battery powered applications.