Minimize Power Consumption in Battery-Operated Systems
A switching voltage regulator converts input DC voltage into a desired output DC voltage. This month, Stuart delves into the factors that influence the efficiency of switching regulators, and why they are crucial for embedded applications. From understanding the role of inductors and transistors to exploring practical examples, this article offers valuable knowledge for anyone interested in optimizing power supply designs.
In a lot of embedded applications, you will use a single power supply, either from the AC line or from batteries. But in some cases, you will have an application that needs more than one voltage. For example, you might have a battery-operated system that needs a 9V or 12V supply voltage for an op amp that is sensing something from a sensor. Or you might need to illuminate an LCD backlight using a voltage that is higher than the supply voltage. You might have a device that operates from a 6V battery (such as four AAA cells), but that has to power a 3.3V MCU.
In all these cases, you need a voltage regulator—a device that produces a constant output voltage to power an electronic circuit. In each instance, you can use a switching power supply to generate the needed voltage. A switching regulator, whether step-down or step-up, uses switching techniques to produce a regulated output voltage. But, especially in battery-operated equipment, you want the power supply to be as efficient as possible, so as to minimize the power it uses.
In my column on “Voltage Regulators” (Circuit Cellar #387, October 2022) [1], I looked at voltage regulators, power, and efficiency, and I touched on switching regulators. This month, I want to narrow specifically to the things that affect the efficiency of switching regulators.
SWITCHING SUPPLY TYPES
Switching supplies come in three basic configurations:
- AC-to-DC converters change AC input to a DC output. An example is the power supply in a desktop computer. I won’t be covering those here.
- Step-up or boost converters change a DC input voltage to a higher output voltage. An example would be the aforementioned switcher that converts the 3.3V MCU supply voltage to 6V to power an LCD backlight.
- Step-down or buck converters that convert a DC voltage to a lower DC voltage. An example would be a power supply that takes a 9V input from a “wall wart” power supply and produces 3.3V for an MCU.
Switching converters can also produce a negative output from a positive input or the other way around. This can be a step-up or step-down converter; the defining characteristic is that the output voltage is the opposite polarity of the input voltage. A switcher to power an RS-232 driver or the negative supply for an opamp would be examples.
Figure 1 shows the configurations of the three types of DC-DC switchers. These simplified diagrams illustrate the basic mechanism for generating a switched DC voltage. (The control and voltage regulation parts are not shown.)
DC-DC switching configurations. Top: step-up (boost) converter. Middle: step-down (buck) converter. Bottom: negative voltage generator.
All three of these use the same basic principle. An electronic switch is closed, providing current to an inductor. When the switch is opened, the energy stored in the inductor is dissipated through a rectifier to produce an output voltage.
Any inductive device, whether a coil in a switcher, the coil in a relay, a solenoid, or a motor winding, will store energy when current is passed through it. If the current is suddenly removed, energy conservation says the energy in the inductor has to go somewhere. In some applications, the energy is dissipated in a snubber circuit to protect the electronics. In a switching power supply, however, the voltage is used to drive the output voltage. The voltage generated when the inductor energy is dissipated is often called “flyback” or “kickback” voltage. The term flyback dates back to the days when televisions had CRTs instead of LEDs.
The waveforms for the inductor voltage for the three types of switchers also are shown in Figure 1. In a real switcher, the switch is a transistor that is controlled by circuitry external to the diagram.
In the top diagram, which shows a boost converter, the switch grounds one end of L1 when closed. The current builds in L1, storing energy, until the switch is opened. At that point, the driven end of L1 rises, generating the flyback voltage. The voltage is rectified through D1 and filtered with C1 to produce a DC output voltage that is higher than the original supply voltage.
The second diagram is a step-down converter, in which the switch connects the driven end of L2 to V+. When the switch opens, the energy charges C2, producing a voltage lower than V+.
The third diagram is a negative voltage generator. It functions similarly to the boost converter, except that one end of the inductor is tied to ground instead of V+. The driven end is switched to V+. When the switch opens, voltage V3 goes negative. This is exactly what happens in the boost converter, except that the polarity is reversed. The negative voltage is rectified by D3 and filtered by C3. The result is a negative DC voltage at Vout. Note that if C3 is an electrolytic capacitor, the positive terminal is grounded.
You can build positive and negative boost converters using capacitors as the storage elements instead of inductors, but that’s outside the scope of this article.
WHAT IS EFFICIENCY?
The efficiency of any power supply is the output power divided by the input power. The output power will always be less than the input power, so efficiency will always be less than 100%. In the simple case, where the output load is a pure resistance, the output current is the output voltage (Vo) divided by the resistance, and the output power is Vo times the output current.
The input power is the input voltage multiplied by the input current. In a dynamic application, where the load varies, both input and output power will vary with the load current.
Any power supply efficiency less than 100% means that electrical current is being dissipated as heat — something is getting warmer than it otherwise would be. Since no power supply is 100% efficient, the goal is to get as close to 100% as practical, within the design parameters of the circuit (size, cost, power, and so on).
INDUCTOR CHARACTERISTICS
Figure 2 shows a simple switching converter. This takes a 5V input and produces a boosted voltage such as 9V or 12V that might be used to power an op amp or a backlight LED in an embedded design. It doesn’t produce much current, and is just for demonstrating some principles. By using a fixed resistive load, I can directly measure the maximum power the circuit can produce by measuring the voltage across the load. In a real application, you would probably use a dedicated switching regulator IC. However, you might use a simple circuit like this one, if you’ve already got an MCU in the design and the appropriate outputs are available. You would have to regulate the output voltage, probably using a Zener for something this simple.
Example of a simple switching converter.
IC U1 produces a PWM output on the OCR1A output. This drives transistor Q1, which in turn drives the coil. Switch S1 is a quadrature rotary encoder with a pushbutton feature on the shaft. The encoder adjusts the PWM frequency from about 5kHz to about 50kHz. The encoder also adjusts the duty cycle from 10% to 90%. The firmware toggles between frequency and duty cycle mode by pressing the shaft of the switch. The OCR1A output drives a 2N2222 transistor or a 2N7000 MOSFET transistor through resistor R2. I socketed the transistor so I can switch between the 2N2222 and the 2N7000.
This isn’t intended to be any kind of project; rather, it’s just a tool to demonstrate some basic principles. But if you want to reproduce it, the code for the MCU is available on the Circuit Cellar Article Code and Files webpage. It’s fairly simple, so it’s all in one C file.
SWITCHER EFFICIENCY FACTORS
The primary characteristics of the inductor that affect switcher efficiency are the saturation current and the resistance.
Inductor Saturation Current: The saturation current is the current at which the coil is saturated, where the inductance falls off, and the reactance goes very low. A switching supply depends on current in the coil changing. Once the current stops changing, the coil current is no longer a function of the reactance of the coil, but is instead limited by the resistance. The resistance will be significantly smaller than the reactance, so more current flows through the windings, producing more heat. Generally, you want to choose an inductor that never saturates in normal operation.
Inductor Resistance: The resistance is the ohmic resistance of the wire, itself. To demonstrate the effect of inductor resistance, I used two different inductors, one was a 330µH toroid on a ferrite drum core. The second inductor was a 330µH SMT inductor on a smaller drum core. The smaller core required smaller wire (more resistance) to achieve the same inductance. I walked through the frequencies from 12kHz to 25kHz, duty cycle 10% to 70%, and found that the maximum efficiency was 72% using the larger core, at 20kHz and 70% duty cycle. With the smaller core, the efficiency in the configuration was 62%. The larger core produced an output power of 959mW into the 500Ω load, and the smaller core produced 808mW.
I did the same experiment with two 250µH inductors, again using larger and smaller drum cores. This time, the maximum efficiency with the larger coil was 73% at 25kHz and 70% duty cycle. The smaller drum core produced 68% efficiency. Output power was 1,058mW and 1,049mW, respectively.
In a real regulator, you wouldn’t be looking for the point of maximum efficiency or maximum power, but rather, holding a constant voltage or current. In this case, the parameters that gave a specific voltage for both inductors still had less efficiency with the smaller inductor. For example, the larger 330µH inductor had efficiency of 68% at 12V, and the smaller inductor was 66% when the duty cycle was set to produce 12V (both at 20kHz).
In both the 330µH and 250µH examples, you get better efficiency with the larger (lower resistance) inductor — but that also requires more PCB real estate.
OTHER EFFICIENCY FACTORS
Transistor Drive: Another factor that influences efficiency is the transistor drive, which affects the transistor turn-on and turn-off times. R2 limits the current into the base of Q1 to about 13mA. I placed another 330Ω resistor in series with R2, and the peak efficiency with the larger 250µH inductor went down 4%. The larger resistor increases the time needed both to drive Q1 into saturation, and to bring it out of saturation.
In a high-power switcher, having the drive transistor spend any significant time in the linear region will significantly increase heating and decrease efficiency. In an IC switching regulator with an integrated transistor, all of that is beyond your control. You just pick external components that are compatible with the IC. But in situations where you are using an IC with an external transistor, make sure the transistor is driven hard enough to keep efficiency up.
Bipolar Switching Transistor: An IC may have either an internal bipolar or MOSFET transistor. A bipolar transistor has a saturation voltage that limits the voltage across the inductor and is current-dependent. For example, in a boost switcher with a 5V supply voltage, if the transistor has a 0.5V saturation voltage, then the inductor will have 5 minus 0.5 or 4.5V across it. This limits voltage across the coil, and therefore, the current through it.
MOSFET Switching Transistor: A MOSFET transistor has an on-resistance measured in ohms. I swapped the 2N2222 for a 2N7000 MOSFET. The on-resistance resistance of the 2N7000 is typically 1.8Ω at room temperature and with a 4.5V gate drive voltage. This resistance is effectively in series with the inductor and has the same effect as adding that resistance to the inductor, itself. It also causes the transistor to dissipate some heat. Staying with the 250µH inductor at maximum efficiency, the 2N7000 produces 74% efficiency, close enough to the 2N2222 to be within the measurement error.
Power MOSFET: In a higher-power switcher, you’d use a power MOSFET with lower ON-resistance, instead of the 2N7000. To demonstrate the effect of this, I replaced the transistor with a power MOSFET with ON resistance under 0.02Ω. Staying with the maximum efficiency 220µH settings, the efficiency drops to 59%. Why? Figure 3 shows what happens. The upper (yellow) trace is drive from the MCU, and the lower (blue) trace is the MOSFET drain. There is significant delay in switching the transistor, as well as a slow ramp when the transistor switches.
Effect of gate capacitance on power MOSFET response.
The reason is that the input capacitance is about 2,000pF for this transistor—40 times the capacitance of the 2N7000. So, yes, you will get better efficiency with a power MOSFET. Generally, however, the more current a MOSFET can handle, the more gate capacitance it has. So, to use a power MOSFET, you need something like 500mA of drive capability into the gate. You could add this to the example circuit, but it would require an IC driver or a few extra transistors.
External Transistor: Some switching regulator ICs have provision for an external transistor. In choosing such a transistor, and in general, a lower ON resistance (MOSFET) or lower saturation voltage (bipolar transistor) will maximize efficiency. Other components have to be compatible, of course, especially the aforementioned gate drive into a MOSFET.
EFFECTS OF DIODES AND OTHER RECTIFIERS
In the simple transistor example, I used an ordinary silicon rectifier diode. But in most switching supplies, you will use a Schottky rectifier. The 1N4001 has a forward voltage drop of about 800mV at 1A. The instantaneous current through the diode has to be higher than the switcher output current. If the diode is passing 1A during conduction, then it would use up 0.8W (800mW x 1A) while conducting. If we assume that the diode conducts 50% of the time (50% switcher duty cycle), then the average power consumption in the diode is 400mW.
Comparing the 1N4001 to a Schottky rectifier, a 1A Schottky diode might have a forward drop around 400mV, so it will waste only half the power of the 1N4001 at an output current of 1A. As with the ordinary rectifier diode, the forward drop of a Schottky diode goes up with forward current. So, to minimize losses in the rectifier, you will want a diode that has a small voltage drop at the expected output current.
The effect of the rectifier on efficiency depends on the output voltage, switcher duty cycle, and current. Say that the 1N4001 is used in a circuit with 1A conducting current and 50% duty cycle, as described. If used in a switcher supplying 12V, 500ma, the output power is 6W. The 800mV drop across a 1N4001 diode uses 800mW x 50%, or 6.6% of the output power. But if the switcher is a buck (step-down) converter with output of 3.3V at 500mA, then the power lost in the diode is 0.5 x 800mW/1.65W, or 23% of the output power—much more significant.
Synchronous Rectifier: The losses in the diode can be reduced by using a synchronous rectifier. A synchronous rectifier is a transistor that is turned on and off by the controller. Since a saturated transistor can have a much lower voltage than even a Schottky diode, the power lost in rectification can be reduced. This requires a controller that can time the switching transistor and the rectifier transistor. The TI LM5122 [2] is an example of a controller that drives external MOSFETS to perform synchronous rectification.
Figure 4 is a simple diagram of a synchronous rectifier in a step-down configuration. The controller turns on the top MOSFET to drive current into the inductor, and then turns on the lower MOSFET to ground the driven side of the inductor for the other part of the cycle. The controller has to control the MOSFETS so they are never on at the same time, to avoid shoot-through.
Synchroxanous rectifier.a
A REAL SWITCHER
Figure 5 shows a 5V-to-12V boost converter using the TI MC34063 switching regulator [2]. The MC34063 is an older part, and it lacks features of newer parts. It’s useful for prototyping, because it is still available in a DIP package, the output voltage is programmable, and it can function as a step up-(boost) or step-down (buck) converter. It has internal switching transistors and can also drive an external transistor for higher current applications.
MC34063 example.
The circuit shown is configured to have about 100mA output current. The MC34063 datasheet [3] has the design equations for determining the various component values for the configuration you want. I won’t go through those here. Resistors R5 and R14 determine the output voltage, and the regulator keeps the voltage at the feedback pin (pin 5) at 1.25V. With the R5/R14 values shown, that occurs when the output is 12.06V. The accuracy of the output value is dependent on the tolerance of the resistors, which is why I used 1% resistors in the example.
To evaluate the effect of the inductor on the efficiency of this circuit, I calculated (using the datasheet formulas) the optimum values for the circuit. The formulas will, of course, give you ideal values out to as many decimal places you want to calculate them, but you are limited to standard values for actual construction of the circuit, which is what the schematic shows.
The formulas calculate a minimum inductance and minimum output capacitance; for this circuit, I calculated 801pF for C7 and 183µH for L5. With the values shown in the schematic, I tried different inductors to demonstrate the effect on efficiency. The circuit was loaded with 250Ω, drawing about half the maximum current, so the output power was about 580mW. Input power with different inductors is shown in Table 1.
The circuit is not all that efficient with these component values; the MC34063 can approach 90% efficiency. But this shows the effect of the inductor value. The part was able to regulate the output voltage with nearly a 4:1 span of inductor values, but the efficiency is degraded as the inductor value moves up or down from the optimum value. Of course, the optimum value depends on the size of the load. In the circuit shown, with a 350Ω load (34% of maximum), the most efficient inductor was 200µH, and efficiency with that value was 72%. The point is that the inductor providing the optimum efficiency may vary with the output current and input voltage. If you are using a different switching regulator, you’ll get different results.
A Practical Use: I used an MC34063 in the MCU switcher testing to allow measurement of input current. Any time you put a current meter in a DC circuit, there is a voltage drop across the meter. This voltage drop reduces the voltage to the circuit, so the MCU might only be getting 4.8V instead of 5V if the output power is 400mW. I constructed a 12V-to-5V buck converter to power the MCU circuit. The feedback that regulates the output voltage is connected to the MCU circuit after the DMM that is making the current measurement. A block diagram is given in Figure 6. This ensures that the voltage at the MCU is always 5V. The voltage at the DMM input might reach 6V or something, but the voltage to the MCU circuit will be constant, making the power measurements consistent.
Current measurement.
CONCLUSION
Maximizing efficiency—and thereby minimizing wasted heat and energy—involves several factors and different components in the power supply design. I’ve tried to look here at the pieces that are involved and the effects they can have on that maximum efficiency goal. Even if you never need to design a power supply, understanding it is useful.
— ADVERTISMENT—
—Advertise Here—
RESOURCES
Texas Instruments | https://www.ti.com/
REFERENCES
[1] Stuart Ball,”Voltage Regulators.” Circuit Cellar 387, October, 2022.
[2] Datasheet for the TI LM5122 wide-input synchronous boost controller: https://www.ti.com/lit/gpn/lm5122
[3] Datasheet for the TI MC34063 switching regulator: https://www.ti.com/lit/gpn/mc34063a
PUBLISHED IN CIRCUIT CELLAR MAGAZINE • FEBRUARY 2025 #415 – 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.
