In small portable systems, marketing considerations—not engineering convenience—often drive the choice of a battery. A classic example is the battery that comprises four alkaline AA cells. AA cells are available in gift shops around the world. Though they are slim enough to fit in handheld systems, four of them can drive a 1W system all day.
Ref 1 reviews practical regulator topologies for battery-powered systems, emphasizing low-power, nonisolated regulators. It compares the circuits with each other and explains how to choose the one that best suits a given application. An important—and related—problem is deriving the regulated 5V from a 4-cell battery stack.
The voltage available from four alkaline cells in series (6.2V, gradually declining to 3.6V), is not a convenient input for a 5V regulator. The battery voltage ranges above and below 5V, so the usual designs—buck and boost regulators—won't work. Moreover, not only must the regulator boost and buck, but it also must exhibit high efficiency, low supply current, and small size. Although these requirements present a challenge, many circuits can meet them. The following are several possibilities:
Many topologies
Of the many solutions to the 4-cell problem, none stands out as the clear winner. Instead, each option offers tradeoffs in size, efficiency, input range, and other parameters. Some general circuit configurations include
Other possibilities include the Cuk and various isolated topologies, but these are either too complex or require too many energy-storage elements to be attractive for small battery-powered systems.
Why not flyback?
Flyback topologies seem an obvious first choice for the step-up/step-down problem. Flyback circuits include a transformer that electrically isolates the output winding from the input (battery) voltage, thereby solving a problem that derails the simpler boost and buck topologies. Indeed, any isolated dc-dc supply, including the forward converter, can function as a step-up/step-down converter.
The most serious contender among transformer-isolated regulators is the flyback regulator, whose simple switching circuit requires only one power transistor and a single magnetic core. Flyback circuits have poor efficiency, though, thanks to their high peak currents and consequent power losses.
Flyback vs buck or boost
To illustrate the flyback configuration's poor efficiency, compare it with the more favored buck and boost topologies. The flyback circuit's main problem is high peak current, which produces high I2R loss. Peak currents cause dissipation in small parasitic resistances: series resistance in the inductor, on-resistance in the switch, and ESR (equivalent series resistance) in the filter capacitor.
These losses are proportional to the peak current squared, so a minor change in peak current can have a substantial effect on conversion efficiency and battery life. In the 4-cell application, physics ensures that a flyback circuit's peak currents are almost double those of a buck or boost circuit.
It is intuitive that peak currents in the buck and boost topologies should be lower. Because the series connection of a boost regulator's battery and inductor aids the inductor-discharge voltage, the boost circuit needs to overcome a smaller energy "hill" in generating the output voltage (Fig 1). Peak currents in the buck regulator are lower, too, because current flows to the load during both the charge and discharge phases of the switching cycle.
IPEAK/IAVG for different topologies
The equations in Fig 2 describe the ratio of peak inductor current to average load current for pulse-width-modulated converters operating in the desirable continuous-conduction mode. In each equation, the most significant term is the first one, which represents the average dc component of inductor (primary) current. Efficiency in the flyback equation is degraded mainly by the numerator (VOUT + VIN), which represents excessive peak current.
The ac switching loss also degrades the efficiency. This parameter equals V2fC, where V is the peak voltage swing (equal to VIN for buck regulators or VOUT for boost regulators), and C is stray capacitance at the switching node. For a flyback circuit with a 1:1 transformer, V=VIN + VOUT (as a minimum).
I2R and switching losses handicap the flyback configuration. The resulting efficiency (70 to 80%) is inferior to that of buck and boost topologies (85 to 95%). The use of large and expensive power-switching components (or other drastic measures) can raise the flyback circuit's efficiency to 85% or so. Nevertheless, the flyback approach is useful if you need a wide input-voltage range or multiple outputs via extra windings, and if low cost is more important than battery life.
Inverting the battery
Another way to generate 5V from four cells is to first invert the battery voltage with a switch-mode inverter, creating -5V. By connecting this negative output to the system ground you produce +5V at the other output terminal. This approach has some disadvantages, though:
Peak currents are no lower than those in a flyback circuit (indeed, the inverting and 1:1 flyback topologies are exact electrical equivalents). The inverting circuit also joins the 5V output to the battery's negative terminal. This can be a problem if other circuit loads are referenced to ground or if other voltages are generated from the same battery stack. And, finally, the inverting circuit requires a pnp or p-channel FET high-side power switch vs a less expensive and more efficient npn or n-channel FET low-side switch.
Despite the drawbacks, the inverting regulator's simplicity and wide range of input voltage make it attractive for many portable-equipment designs. The wide input range lets the system accept alternate power sources such as ac/dc adapters and 12V lead-acid batteries. As another advantage, the inverter output moves to zero in shutdown mode, a condition not always guaranteed for other regulating topologies.
Low-dropout linear regulators
A step-down, low-dropout linear regulator would seem a poor choice for the 4-cell application. It converts only so much of the battery's energy; spent batteries still have considerable energy left in them. Even so, the linear regulator offers better battery life than some switching regulators. In the 4-cell application, the theoretical efficiency is lowest when the battery is fresh (5/6V×100%=83%) and rises toward 100% as the battery voltage approaches 5V.
What's more, the heat and I2R losses associated with pulsed current are absent in a linear regulator, and the continuous supply current has a gentler effect on the battery chemistry. Though its battery life is generally lower than that of switching regulators, the linear regulator's cost, size, and low noise make it more attractive in some applications.
Switching regulators usually provide tightly regulated outputs even at low battery voltages; when the output finally collapses, it does so in milliseconds. Linear regulators, on the other hand, drop out slowly and gracefully as the battery voltage decays. This behavior complicates the comparison of switchers and linears. When do you consider the battery to be discharged? Simply defining a dead battery as one that produces an output of 4.5V instead of 4.75V increases a linear regulator's battery life by more than 50%. For the 4-cell application, good linear regulators should have low dropout voltage (100 mV) and low quiescent current (10 µA) (Fig 3).
Boost with linear postregulation
The best 4-cell regulators use boost or buck topologies in a way that overcomes input-voltage limitations. Boost circuits, for example, feature low peak current and a simple schematic. They just keep boosting the battery voltage (to 5V) until the battery's energy completely dies.
Adding a linear regulator to a boost regulator prevents the series connection of the inductor and rectifier from pulling the output above 5V when the battery is fresh (Fig 4). In this case, the linear regulator is implemented with an active (pnp) internal rectifier instead of the usual Schottky diode.
The switching-regulator IC in Fig 4 is also unusual. Instead of a standard CMOS or "junk" bipolar process, this chip is fabricated with an advanced, complementary-bipolar RF process. The result is a combination of high switching frequency (normally the strength of CMOS) and operation below 1V (normally a strength of bipolar processes). Synchronous rectification overcomes many of the limitations inherent in the simple boost topology. In addition to the step-up/step-down function, synchronous rectification allows the output to be shorted to ground, and it automatically (and completely) disconnects battery from the load when the IC is placed in shutdown mode.
Boost with linear preregulation
A second boost-plus-linear approach is to preregulate the input to a boost switcher (Fig 5). The switching regulator is disabled when the battery is fresh, so an external silicon rectifier (D1) drops the worst-case high input voltage from 6.3 to 5.4V. This cheap-and-dirty equivalent to a linear regulator requires a minimum load of 100 µA or so to prevent output overvoltage due to the diode leakage current. The circuit continues to operate even with battery voltages of 3V and below. Typical efficiency is 80% when the battery is fresh, rising to 90% as the battery voltage declines to 4V.
To accommodate higher input voltages, you can easily substitute a linear regulator for D1. And if low cost is your goal, omit the p-channel FET switch-over circuit. Performance with the diode alone is still comparable to that of a flyback circuit.
Step-up/step-down topology
The step-up/step-down regulator achieves high performance at the cost of complexity by switching from buck mode to boost mode as the declining battery voltage passes through 5V (Fig 6). And it does all this with a single inductor. Switch-mode operation over the entire range of battery voltage yields higher efficiency than does the boost-plus-linear approach, yet the step-up/step-down regulator does not experience the high peak current and consequent I2R losses of inverting and flyback approaches.
Efficiency exceeds 90% over most of the battery's range, and the step-up/step-down circuit extracts nearly all of the battery's energy. The penalty for this high performance is complexity. The circuit requires three MOSFETs (or four—as shown—if you parallel two p-channel devices for a lower rDS(ON)). Also, the switch from buck to boost causes an ~±2% change in the output voltage as the battery voltage reaches 5V.
Built into IC1 is a comparator that decides when to switch from step-down to step-up operation. The comparator monitors the battery or output voltage, whichever is higher, via a diode-OR connection. As the buck regulator begins to lose control (drop out), the output begins to fall. When the input reaches 4.95V, the circuit switches from buck to boost, causing the output-regulation point to shift from 4.92 to 4.98V. If for some reason the battery voltage rises above 5.15V, the circuit switches back to buck mode.
1. Moore, Bruce D, "Regulator Topologies Standardize Battery-Powered systems," EDN, January 20, 1994, pg 59.
For Bruce Moore's biography, see Ref 1.