Many circuits will need a voltage regulator between the power supply and the electronics. In this article, Stuart breaks down the two broad types of voltage regulators: linear and switching.
Avoltage regulator is a device that produces a constant output voltage to power an electronic circuit. This is different from a voltage reference, which produces a constant voltage as a reference for some measurement device. Functionally, the difference is that a voltage regulator sources current to whatever it is supplying, while a voltage reference may be passive (sourcing no current) or else it may source very small currents. A voltage regulator will contain some kind of voltage reference to set the output voltage. A typical application for a voltage regulator will be to accept a 9V unregulated input from a wall-mount power supply and produce a 5V or 3.3V output to power anything from a desktop clock to a printer. If you have a desktop or laptop PC, it likely has multiple voltage regulators on the main board.
Voltage regulators fall broadly into two types: linear and switching. A linear regulator is always a step-down device, regulating a higher voltage to a lower voltage of the same polarity. For example, a positive regulator may produce a fixed 5V output from an unregulated 12V input.
A switching regulator uses switching techniques to produce a regulated output voltage. A switching regulator may be a step-down regulator, or a step-up regulator. A switching regulator may also produce a negative output from a positive input.
Voltage regulators can be purchased as prebuilt modules, or as discrete components, such as integrated circuits. This article will focus on discrete regulators.
LINEAR REGULATORS
A linear regulator, as mentioned, converts a high input voltage to a lower, regulated voltage. The input and output may also be negative, but we’ll focus on positive voltages here; they are much more common.
Figure 1 shows a simple voltage regulator circuit. This consists of a transistor, resistor, and a Zener diode. The circuit takes an 8V input and produces a 3.3V output. Resistor R1 provides current into the 4V Zener, which clamps the base at 4V. The emitter will be about 0.7V lower than the base, or about 3.3V. Since the base voltage is fixed by the Zener, the output is stable. Well, kind of.
— ADVERTISMENT—
—Advertise Here—
Simple transistor voltage regulator
This circuit is a functioning voltage regulator, and it wasn’t an uncommon architecture some years ago, using a much larger transistor. It has some drawbacks and limitations that apply as well to modern IC regulators. This circuit is simple enough to make these principles easy to explain.
Dropout Voltage: To keep the emitter at 3.3V, the base has to be held at 4V. To keep the base at 4V, the input has to be high enough to provide adequate current through R1 to get the Zener diode to reach its Zener voltage. Looking at the datasheet for a typical 4V Zener, the Zener voltage tolerance is specified at about 5mA, so that’s the minimum current it needs to ensure it is operating within specification. It might work at lower currents, but operation is unspecified below 5mA.
The current through R1 must be 5mA, so the difference between the Zener voltage and the input voltage has to be 5V (4V + (1k x 5mA)), making the minimum input voltage 9V. The dropout voltage is the difference between the input and the output, so the dropout voltage is 9V – 3.3V, or 5.7V.
This particular circuit has a relatively high dropout voltage, driven by the need to pass the required minimum current into the Zener D1. If R1 were a smaller value, say 100Ω, the dropout voltage would be lower, but the Zener would also dissipate more power. At some input voltage, the maximum Zener dissipation would be exceeded.
The important thing here is to be aware that any regulator has a minimum input voltage to be able to regulate. The popular 7805 regulator, which has been around for many years, has a dropout voltage of 2V, so it needs at least a 7V input to produce a regulated 5V output. The LF33 low-dropout 3.3V regulator has a 0.7V dropout voltage, so it can produce a 3.3V regulated output with a 4V input.
Quiescent Current: Because of the need to bias the Zener, this circuit will draw 5mA at 9V input, even if nothing is connected to the output. Some IC regulators have a fixed quiescent current. The quiescent current for this example varies with the input voltage. The 7805 regulator mentioned above has a quiescent current of 5mA to 8mA. The LF33 has a quiescent current of 1mA. Quiescent current is needed even if there is no output current, so it is a constant drain on the battery or power supply. In most circuits operated from the AC line it’s not an issue, but for battery operation it’s a very important number.
Output Current: In the example, the 2N3904 transistor has a maximum current rating of 200mA, so that would be the maximum output current that could be drawn from this circuit. The 7805 maximum current is 1 amp, and the LF33 maximum current is 500mA.
Power: Regulators have a maximum power rating, the maximum power they can dissipate. For the example circuit, the 2N3904 has a maximum power rating of 625mW. Ignoring the quiescent current, the power is the current through the regulator x the voltage across the regulator. So if the transistor circuit had a load that uses 100mA, it will dissipate (9 – 3.3) x 0.1, or 570mW. This is close to the maximum power of the transistor. At 150mA, the transistor would probably melt.
The regulator output current is limited by the specified current limit, or by the total power dissipation—whichever is less. For the example circuit, the regulator can dissipate 625mW or carry 200mA—again, whichever is less. The power dissipation in turn depends on the input voltage. The 7805 and LF33 datasheets don’t specify the maximum power dissipation. They only state that the regulator internally limits the dissipation to prevent damage.
— ADVERTISMENT—
—Advertise Here—
Temperature: The regulator also has a maximum operating temperature. The 2N3904 in Figure 1 has a maximum operating temperature of 150ºC. The 7805 and LF33 have a maximum operating temperature of 125ºC. However, the power dissipated in the regulator will raise the temperature. Many IC regulators such as the 7805 have a power tab that can be bolted to a heatsink to reduce the effects of self-heating. Surface mount regulators such as the LF33 have a tab that can be soldered to a copper land on the PCB to improve heat transfer to, for example, the PCB ground plane. Figure 2 shows a 7805 next to an LF33.
LM7805 and LF33 linear regulators
The maximum temperature rating provides a third limit on output current. In addition to the specified current limit and the total power dissipation, the temperature increase generated by dissipating power will also limit the output current. All else being equal, an LF33 regulator soldered to a large heatsink area can provide more output current than an LF33 soldered to a pad that is just large enough for the tab. This is because a good heatsink reduces the temperature rise caused by the power dissipation in the part.
For a surface-mount regulator, I typically use a sizeable power pad in my designs, even if I don’t expect to draw much current, and I use multiple vias (without thermal relief) to provide a thermal connection to the ground plane. If I’m hand-wiring a prototype, I’ll use adhesive-backed copper tape for a heatsink. Any time you do this, of course, the heat must have a way to dissipate. In a very hot environment, such as an automotive or military application, the part is already operating at an elevated temperature so the heatsink will dissipate heat less efficiently than it will in a cooler environment. In those situations, if you are operating the regulator anywhere near its limits, you may need a thermal analysis to ensure it doesn’t overheat.
Bolt-on heatsinks from companies such as Aavid specify what kind of package they fit (such as TO-220) and how much the temperature will rise for a specific wattage dissipation in specific amounts of airflow. All those factors affect the size and type of heatsink needed. For an SMT regulator, the amount of copper pad needed to manage the temperature is dependent on the thickness (weight) of the copper, the thermal resistance of the device, and the amount of temperature rise allowable. Rohm has an application note showing how you can do the calculations for this; it’s listed in the references section below [1]. You can also get heatsinks that solder to the PCB over an SMT power-tab regulator to minimize the amount of PCB copper area needed to cool the device.
I said the 7805 has a power tab for a heatsink and the LF33 has an SMT tab. The 78xx parts are also available in an SMT version with the solderable tab, and the LF33 is also available in a power tab with a heatsink mounting hole. It’s common for linear regulators to be available in multiple packages.
Accuracy: The example shown in figure 1 has a Zener with some tolerance, the deviation from the nominal 4V value. The Zener voltage varies slightly with current, so as the input voltage goes up, the output voltage will go up as well, if only slightly. The 2N3904 transistor doesn’t always have a 0.7V base-emitter voltage—this also varies with the emitter current and can go to 0.95V. So overall, the simple transistor regulator doesn’t have high accuracy, especially for low voltages where all these variations stack up to be a significant percentage of the intended output voltage.
The 7805 regulator, depending on which manufacturer’s part you use, can be obtained with accuracy from under 2% up to 4%. The LF33 is available in 1% to 2% tolerance. IC regulators achieve this by using feedback. Figure 3 shows a simple feedback regulator where the output transistor is in the feedback loop of an op-amp. The op-amp will adjust the drive to the transistor base to compensate for the varying base-emitter voltage. In this case, the accuracy of the regulator would be the same as the accuracy of the Zener. IC voltage regulators use this type of feedback to provide better accuracy, and can have more accurate references than a simple Zener.
Simplified linear regulator with feedback
Because of the feedback circuitry and the possibility for oscillation, many linear regulators require a large value capacitor at the input and output. LF33, for example, requires an output capacitor of at least 2.2µF for stability. Many years ago another engineer came to me and said he didn’t understand why his circuit wasn’t working. I looked at his schematic and said, “The voltage regulator is oscillating. Add an output capacitor.” Voltage regulator instability is a real thing in some parts.
Output protection: Because of the internal feedback, IC regulators typically have protection against some kinds of failure. For example, the 7805 and LF33 regulators have thermal overload protection that will shut down the regulator if the part gets too hot. The transistor regulator has no such protection. The IC regulators are also internally protected against a shorted output or any overcurrent condition. The transistor circuit will just get very hot or even melt if the output is shorted.
SWITCHING REGULATORS
Figure 4 shows a simplified diagram of buck and boost converters. The top figure is a buck converter that would be used to step down a higher voltage to a lower voltage. When S1 is closed, the junction of C1, D1, and L1 is pulled to the supply voltage V1. Current flows through the inductor to the load, R1. However, because the inductor functions as a low-pass filter, the load voltage slowly rises toward the supply voltage V1 (“slowly” is relative, it may be less than a microsecond in a real circuit). If S1 is then opened, the stored energy in the magnetic field of the inductor is dissipated into the load with the circuit completed through D1. In a real circuit, the switch would be a transistor and the duty cycle is controlled to maintain the desired voltage at the load. The output voltage is fed back to the switch control logic, and if the load changes, the duty cycle is adjusted to make the output correct again.
Buck (a) and Booster (b) regulator configurations
The lower diagram in Figure 4 is a boost converter that performs a step-up function, converting a lower voltage to a higher voltage. When S2 closes, current flows through L2, building up a magnetic field. When S2 opens, there is a flyback effect where the voltage at the junction of L2 and D2 rises above the source voltage. Current flows through D2 into the load. Again, by adjusting the duty cycle of the switch, a constant output voltage that is higher than the input voltage can be maintained.
Efficiency: The efficiency of a regulator is measured as:
Output power is the power dissipated in the load, voltage x current. Input power is the power into the regulator, again, voltage x current. The input power equals the output power plus whatever is dissipated in the regulator itself.
The efficiency of a linear regulator is dependent on the input voltage. The efficiency of a linear regulator that has a 5V input and a 3.3V output at 500mA is about 66%. If the same regulator has a 9V input, the efficiency is about 36%, with about 64% of the power lost in the regulator as heat. So two thirds of the power is dissipated in the regulator. For a linear regulator, the power dissipation goes up with the input voltage, even if the output current remains the same.
In both of the switching regulator circuits, the transistor switch dissipates very low power. It is either passing zero current or zero voltage (almost zero—there are no perfect switches). A switching regulator can have efficiency over 90%, which results in less wasted energy and cooler operation. The switching regulator efficiency is less affected by variations in the input voltage as long as the input stays within the range specified by the manufacturer. Better efficiency is the primary reason for using a switching regulator instead of a linear for step-down applications.
— ADVERTISMENT—
—Advertise Here—
Noise: There is a catch to the high efficiency though. The switching regulator current isn’t constant. Current passes through the switch and into the inductor when the switch is closed, and no current flows through the switch when the switch is open. If you measure the current into the switching regulator, it will vary at the switching frequency. So switching regulators are more electrically noisy than linear regulators. The current into a linear regulator changes only when the input voltage changes or the output current changes. Switching regulators, especially those with high-frequency switching, require care in bypassing and grounding the input and output.
Inductor: The value and type of inductor is crucial to proper operation of the regulator, and the manufacturer’s datasheet will specify how to choose an inductor. If the inductor saturates in operation, meaning the magnetic field stops increasing, then the regulator won’t be able to regulate the output voltage, and the efficiency will also go down. If the inductor resistance is too high, then the inductor will be dissipating power and efficiency will again go down. So selection of the inductor is critical to proper operation of the circuit. Most switching regulator IC datasheets provide guidance on inductor selection, and sometimes recommendations for specific inductors from specific manufacturers.
In a linear regulator, the regulator will work the same at any current up to the limits of the device. For a switching regulator, the inductor value is often dependent on the input voltage and output current. IC manufacturers will include this information in their datasheets, but the point here is that designing with switching regulator ICs is a little more complex than using a linear regulator.
Capacitor: The output filtering capacitor selection may also be important, depending on the regulator used and the application. The output will have some ripple at the switching frequency, which may be in the megahertz range. The proper capacitor minimizes the ripple. Again, the regulator datasheet may have information about this, but usually lower equivalent series resistance (ESR) capacitors are better. In some cases, where current is low, a ceramic capacitor may be adequate. Like the inductor, different ICs have different output capacitor requirements. See the datasheet for details.
IC enhancements: You can build a switching regulator with just transistors or transistors plus op-amps. The earliest switching regulators were built exactly that way. But using an integrated circuit provides many advantages. These include:
- Overcurrent protection for the switching device (transistor)
- Automatic handling of very low current loads (some discrete switchers could not regulate at low currents because the duty cycle could not go below some fixed percentage such as 10%)
- Wider input voltage range
- Higher efficiency due to better control of the drive transistor, faster switch time, and ability to incorporate internal functions such as constant-current sources
- Output short-circuit protection
The problem with an IC regulator, of course, is that you are limited to available parts. An intermediate approach is use of a switching controller, where the controller IC has all the pulse-width modulation (PWM) and other functionality inside, but it drives an external transistor. This allows the regulator to handle much higher currents than could be handled by a transistor inside the IC package.
Linear Technology (now part of Analog Devices) has an old but good application note on switching regulators that is worth reading. It is listed in the resources below. Texas Instruments has a more recent application note that is also good, also listed in the resources.
Buck-Boost: Regulators that combine both buck and boost functionality are available, and are capable of operating with an input voltage either above or below the output voltage. These are useful in cases where the input voltage may vary, such as in battery-operated applications where the circuit needs to compensate for declining input voltage as the battery discharges.
Example circuit: Figure 5 is a schematic of a simple switching regulator using the TI LM2675 part. The LM2675 has an input range from 8V to 40V, up to 96% efficiency, and is available with 3.3V, 5V, 12V, or adjustable outputs. Only five external components are needed. The inductor value is chosen based on the maximum output current needed and the input voltage. The datasheet for the LM2675 is included in the references, below. The LM2675 is an older part, but still available, and is a good example of how switching regulators using ICs are implemented.
LM2675 example schematic
CONCLUSION
Many if not most circuits will need a voltage regulator of some kind between the power supply and the actual electronics. Hopefully this has given you an overview of the two basic types of regulators and the tradeoffs in using them.
RESOURCES
Rohm application note on copper pad heatsinks:
https://fscdn.rohm.com/en/products/databook/applinote/ic/power/linear_regulator/linearreg_heat_calculation_appli-e.pdf
Linear Technology switching regulator application note:
https://www.analog.com/media/en/technical-documentation/application-notes/an25fa.pdf
TI switching regulator application note:
https://www.ti.com/lit/an/snva559c/snva559c.pdf?ts=1642458903581
TI LM2675 datasheet:
https://www.ti.com/lit/ds/symlink/lm2675.pdf
PUBLISHED IN CIRCUIT CELLAR MAGAZINE • OCTOBER 2022 #387 – Get a PDF of the issue
Stuart Ball recently retired from a 40+ year career as an electrical engineer and engineering manager. His most recent position was as a Principal Engineer at Seagate Technologies.
