February 16, 1998

Reap benefits while simplifying dual-battery portable power management

Mark Gurries and Timothy J Skovmand, Linear Technology Corp

Power-path-controller ICs can effectively switch power among various power sources in a minimum of board space; at the same time, they meet smart-battery standards.

Notebook-computer and other portable systems typically use dual battery packs to meet the demands for increased portable-computing power and to extend their runtime from a fully charged set of batteries. System designers have worked toward increasing the efficiency and flexibility of the main system converter to ensure that the maximum amount of energy passes through to the main processor and peripheral circuitry.

One major design challenge is grappling with the battery-interface and the power-switch-matrix functions that deliver the "raw" battery power to the input of the dc/dc converter. The energy stored in a battery pack must transfer as smoothly and efficiently as possible to the input of the dc/dc switching regulator. Arranging the appropriate battery charging while switching the correct power source to the switching regulator is also essential. Depending on the status of each battery, one of the batteries usually requires charging while the other is idle or removed.

To further complicate matters, the batteries in a portable system use different chemistries, such as lithium-ion (Li-ion) and nickel-metal-hydride (NiMH), with different charge and discharge rates. The solution to this problem is to use accurate current-, voltage-, and temperature-sensing circuits to monitor the changing slope of the battery capacity during a slow, controlled charging cycle. To achieve those goals, the battery requires an accurate "gas" gauge (which actually monitors electron flow), although this requirement adds more complexity to the system. Unless you're careful, providing such a universal system for flexible battery support can result in long initial-development cycles and considerable circuit complexity.

The convergence of power sources at the front end of the power-management system, or "power path," is where the energy from the ac wall adapter, the battery packs, the battery charger, and the standby power system all merge. This path must switch many amperes of current while protecting the various power sources and the load from fault conditions. Providing the precision isolation required for this task remains a major challenge.

Prevailing approaches to these problems are nearly as varied as the portable equipment in which they reside. Combinations of op amps, level shifters, references, comparators, regulators, switch drivers, and µP-interface glue logic can do the job. However, these combinations require the designer to create a custom power-management system for each new portable product.

Fortunately, some commonality has emerged among power-path switching schemes, and power-path controllers from several IC manufacturers now provide integrated options. These application-driven ICs act like switch masters in a railroad yard, making sure the batteries, ac adapters, battery charger, and portable computer are all properly connected or isolated as the application requires. You can now save materials and assembly costs, as well as pc-board real estate, when using these new controller ICs in portable designs.

Switching multiple batteries adds to the problem

A low-loss switching matrix at the front end of a power-management system is where you can indiscriminately connect and disconnect as many as two battery packs and a dc power source (Figure 1). The power-management µP of the host system monitors the power-path controller's voltage-sensor outputs and directs the controller to intelligently control the three main power switches. The challenge in switching among multiple sources is to seamlessly disconnect the unused power sources from the load while coupling the desired power source to the load.

Three switches--SWA/B, SWC/D, and SWE/F--direct power from either the ac adapter or one of the two battery packs to the input of the dc/dc switching regulator. Two other switches, SWG and SWH, connect the desired battery pack to the battery charger. The power-path controller, which directly interfaces with the power-management µP, intelligently controls these five switches.

The controller in Figure 2 is a multifunction system-level IC that illustrates the capabilities of power-path management. The controller's inrush-current limiting allows seamless switching between power sources and enables the use of tantalum capacitors in most applications, while battery charging can occur in the background. You can use this controller in systems with conventional battery systems or in those systems that adhere to the Intel Smart Battery Specification (SBS, see box "SBS sets standards for power-management control"). With the IC are n-channel-MOSFET gate drivers for five power pathways, a micropower switching regulator with an onboard switch to create the gate-drive supply for use by all five drivers, analog-voltage sensors that monitor the dc-input and battery voltages, and a logic interface for the host power-management processor.

Discrete n-channel MOSFETs are a good choice for switching the high currents that the system requires and are available in small surface-mount packages from several manufacturers (see box "Higher efficiencies with n-channel MOSFETs"). These n-channel MOSFETs have one main disadvantage when you use them in this circuit: an inherent body diode from drain to source that allows current to flow even when they are off. However, using two back-to-back MOSFET switches overcomes this problem and allows each switch pair to block current flow in both directions when the two halves are off (Figure 2). This back-to-back topology also allows independent control of each half of the switch pair and thus enables the use of bidirectional inrush-current limiting. Table 1 describes several examples of n-channel MOSFETs in various packages and power outputs.

You can create huge inrush currents when switching sources such as batteries and ac wall adapters into large discharge capacitors that the supply typically requires at the input to the dc/dc converter. These inrush currents may result in voltage drop out, or it may cause foldback current limiting to kick in, thereby disabling the system. Also, some tantalum capacitors are prone to failure with repeated exposure to surge currents. You can reduce inrush current to a safe value by sensing the level of current into the load and dynamically driving the switched MOSFET.

The sense resistor, a low-value resistor in series with the load, provides a sense voltage proportional to the inrush current (Figure 2). This resistor measures the instantaneous current flowing through SWA/B, SWC/D, SWE/F. The gate drivers then control the inrush current during the transition from one power source to another. This control dramatically reduces source-current loading and capacitor inrush current. A micropower step-up regulator that continuously generates 36V supplies the gate drive for all five low-loss n-channel switches. The drivers, in turn, supply a regulated 5.7V gate-to-source voltage, V_GS, when activated.

The micropower boost regulator uses power that it takes from three internal diodes connected to the DCIN, BAT1, and BAT2 main power sources (Figure 2). The diodes direct the highest voltage potential to the top of an inexpensive 1-mH surface-mount inductor, L1. A fourth internal diode directs the current from the inductor to the V_GG gate-supply output capacitor. In the circuit using the LTC1479, the V_GG regulator requires only L1, C1, and C2.

Dual Li-ion system illustrates operation

When your system uses dual Li-ion batteries and if power is available from the ac adapter, both MOSFETs in switch pair SWA/B are on (Figure 3). This configuration provides a low-loss path for current flow to the input of the dc/dc converter. Switch pairs SWC/D and SWE/F are off.

This circuit example uses a constant-voltage/constant-current (CV/CC) source, such as the LT1510 battery-charger IC, to alternately charge the two Li-ion battery packs. The power-management µP decides which battery needs recharging by directly querying the smart-battery pack. After the processor determines which one, the power-path controller turns on switch-pair SWG or SWH to pass charger current to the battery. The charging-battery voltage returns to the voltage-feedback input of the CV/CC battery charger via a multiplexer in the power-path controller. After the first battery is charged, the controller disconnects it from the charger circuit and charges the second battery. When you remove the ac adapter, the controller instantly informs the power-management µP that the dc input is bad and then connects the desired battery pack to the input of the dc/dc converter.

A portable power system also needs a backup source if the main power and the two batteries are unavailable. A standby switching regulator powered from a small, rechargeable "bridge" battery provides the backup power for the circuit in Figure 3. This circuit automatically goes into operation when there are no other available sources of power and the V_OUT bus starts to collapse. The only energy source that remains to hold up the main V_OUT bus voltage is the bulk capacitance on the bus itself.

The input voltage to the main switching regulator must not fall below a predetermined voltage, usually 6V. While the system applies power in a backup situation, power-sequencing problems can arise that can cause lockup of the power-management system.

To deal with this situation, devices such as the LTC1479 offer a "three-diode-mode" operating state. Under normal operating conditions, the controller simultaneously turns on and off both halves of each switch pair. For example, when the input power source switches from a good ac adapter to battery 1, both gates of switch-pair SWA/B are off and both gates of switch-pair SWC/D are on. The back-to-back body diodes in switch-pair SWA/B block current flow in or out of the ac adapter.

On the other hand, in three-diode mode, only the first half of each power-path switch pair--SWA, SWC, and SWE, for example--is on. The second half--SWB, SWD, and SWF--is off. These three switch pairs now act as three "diodes" connected to the three main input-power sources (Figure 4). The power-path diode with the highest input voltage passes current to the input of the dc/dc converter to ensure that the system does not lock up, regardless of the order in which the power is applied. After you reconnect a good power source to one of the three main inputs, the power-path controller fully turns on the appropriate switch pair. The other two switches are off, thus restoring normal operation.

Any power-path controller needs intelligent communications to provide effective control and monitoring of the various parts of a power system. A simple µC normally accomplishes this function, along with other system tasks, and many inexpensive µPs are available that can fulfill these requirements. You can easily program these µPs to accommodate the custom requirements of each system and to allow performance updates without resorting to costly hardware changes.

The power-path controller should take logic-level commands directly from the dedicated power-management µP and make changes at high-current and high-voltage levels in the power path. The controller also provides information directly to the µP on the status of the ac adapter, the batteries, and the charging system.

Some delays are acceptable in certain portions of the power-management system, but it is imperative that the power-path switching control be made through a direct connection to the power-management µP. Excessive latency during a critical switch-over period can result in main µP reset or loss of memory because of supply-voltage dropout at the dc/dc-converter input.

SBS sets standards for power-management control

The Smart Battery Specification (SBS) targets portable computers almost exclusively. It uses state machines, such as dedicated logic or microcontrollers, along with other ICs, to control the charging cycle and monitor the discharge cycle of rechargeable batteries. At a minimum, two physical devices --a battery and a charger--constitute the SBS. If the system supports more than one battery, the system designer adds a selector (or a controller, such as the LTC1479) to control the multiple power paths. You can combine the charger and selector functions into one physical device. Another key device you need for an SBS implementation is the host.

In a typical portable-computer implementation, the host monitors the available life of the battery and reports the information to the user. Optionally, the host may do more. Armed with additional information that the battery, charger, and optional selector provide, the host may reallocate the power resources and modify the processes involved in charging the battery itself. Regardless of the number of physical devices, all of them must communicate over a common two-wire serial bus, the Smart Management Bus (SMBus).

In one implementation of a combination charge-selector device, you use a µC to implement the SMBus interface. The reprogrammable µC has the advantage of supporting specification upgrades and, optionally, implementing automatic power functions without the need for a host. The µC translates the SMBus commands and, along with a battery-charger IC and power-path-controller IC, implements the SBS charger and selector functions.

Nickel-metal-hydride (NiMH) and the newer lithium-ion (Li-ion) batteries do not charge and discharge in a linear fashion. The problem for portable-computer users is that the slope of the discharge curve often changes abruptly. At one instant, the computer's battery-monitoring software indicates that the battery has 20% capacity left, and, in the next instant, the computer flashes for an emergency shutdown. To solve this problem, the industry has defined the SBS.

An SBS system is not a good choice for every application. Its flexibility results in high complexity and design costs. This flexibility may be acceptable in a portable-computer system, but it may be more than you need in an embedded system. To determine if SBS is appropriate, ask these questions:

Do you need to support multiple battery chemistries (Li-ion, NiCd, NiMH, and sealed lead-acid) at the same time?
Do you need to support multiple cell configurations?
Do you need a superaccurate battery gas gauge, and do you need a gas gauge at all?
Will you frequently change or replace the batteries?
Do you need to support more than one battery at a time?

Match your answers with the capabilities that the SBS offers:

100% chemistry independence,
support of as many as four batteries,
standard software interfaces and protocols,
standard serial communication bus--the SMBus,
standard functions for each component in the SBS system,
three levels of system complexity allowing scalable cost vs flexible configurations, and
overall lower system cost through industry standardization.

The SBS system is divided into three levels of sophistication. You can distinguish the levels by determining what is implementing the battery-charging algorithm or communication. The implementor is the active or master device. At Level 1, the simplest level, the battery is passive; the charger is master and controls the charge. At Level 2, the charger is passive; the battery is master and controls the charge. And at Level 3, the most complex level, the battery and charger are passive; the host is master and controls the charge.

Please note: Per version 1.0 of the released SBS, Level 1 chargers no longer comply with the SBS. Level 1 is not chemistry-independent. It targets the lowest cost systems in which fixed battery chemistry and configurations are acceptable. Level 2 and higher levels target more mainstream portable applications.

The solution to this problem is to use accurate current-, voltage-, and temperature-sensing circuits to monitor the changing slope of the battery capacity during a slow, controlled charging cycle. An ac/dc converter gives this information to the µP, which compares the data with certain known characteristics of the battery and calculates its probable degradation curve as well as the optimum charging characteristics. Smart-battery advocates believe that this technique will not only enable more accurate battery-fuel gauges, but also increase battery life by reducing charging-cycle stress.

Higher efficiencies with n-channel MOSFETs

Portable power systems have traditionally relied on logic-level p-channel MOSFETs to switch loads and supplies. P-channel MOSFET-device shortcomings include their high on-resistance, R_DS(ON), and marginally higher cost than their n-channel counterparts.

Discrete n-channel MOSFETs, which have lower on-resistance, are preferable. These devices' lower on-resistance results in reduced I²R losses, greater efficiency, and cooler operation than p-channel MOSFETs can yield in power-management-system design. N-channel MOSFETs are available in small, surface-mountable power packages.

One drawback of n-channel MOSFETs is that they require a separate gate voltage that must be higher than the supply rail being switched. This voltage is normally unavailable in a typical portable system, but you can supply it with special gate-drive circuitry. A micropower boost converter inside a modern controller generates a regulated 36V and provides sufficient head room above the maximum 28V operating voltage of the three main power sources to ensure that the logic-level MOSFET switches are fully enhanced. The voltage at each gate driver is regulated to be a minimum of 5.7V above the MOSFET source.

04M289MG

04M289T5

Authors' biographies

Mark Gurries is an applications engineer at Linear Technology Corp (Milpitas, CA), where he designs dumb, smart, and intelligent battery-charger systems. He has done similar work for Apple Powerbooks and has a BSEE from the University of Santa Clara.

Tim Skovmand is a senior IC-design engineer at Linear Technology Corp, where he designs high-side switches, power-management devices, and dc/dc converters. He has spent 17 years in IC design--eight of them at Linear--and holds six patents.

| EDN Access | Feedback | Table of Contents |