Eliminating the Achilles Heel of the Linear Charger

Linear chargers are widely used for single-cell rechargeable Li-ion batteries in handheld applications due to their simplicity, low cost and small physical size. Despite all of the advantages the linear charging topology has, it does have its Achilles heel: high thermal dissipation.

The relatively low processing power, small monochrome displays and limited functionality found in early handheld devices allowed the use of low-capacity battery cells that could be rapidly charged at lower currents, thus effectively sidestepping the major thermal flaw of the linear charging topology. The industry can no longer ignore this flaw in today's handheld device designs because of the upward drive of power requirements due to high-powered processors, brightly lit color displays, increased memory, and wireless connectivity such as 3G, Wi-Fi and Bluetooth. To maintain reasonable operating times, handheld device designers are forced to increase cell capacities, which have reached 2000mAh in some applications.

Pulse charging offers the advantage of low thermal dissipation while maintaining low cost and a low component count; however, the high-amplitude pulse current it generates can degrade the life of the battery, introduce noise into the system, and cause false triggering of the protection circuit built into the battery pack. The dual-mode linear approach operates as a linear regulator, so it has no high-amplitude current pulses associated with it.

Switching chargers offer very low thermal dissipation and can accept a wider range of input voltages than a linear charger. They are not popular in handheld device designs because they require bulky filter capacitors and inductors that add cost in addition to the significant switching noise they introduce to the system. The dual-mode approach requires no inductor, uses only tiny ceramic filter capacitors, and offers thermal performance that can match or beat switcher performance.

Requirement for Rechargeable Li-ion

To charge a Li-ion battery in the shortest period of time, the charge cycle is typically broken up into two phases, a fast charge constant-current (CC) phase, identified as Phase I in Fig. 1, followed by a reduced rate constant-voltage (CV) phase, identified as Phase II in Fig. 1. The maximum charge rate in Phase I is typically limited to about 1 C-rate because the cell life decreases with higher charge rates. The C-rate is a normalized current rating defined as the nominal cell capacity/hour.

For example, if the cell capacity is 2000mAh, the corresponding 1 C-rate current equals 2000mA. In Phase II, the charger maintains a constant voltage tightly regulated to the final battery voltage (4.2V typically). This voltage is normally regulated within ±50mV because a lower final voltage will result in reduced capacity and an overvoltage will reduce cell life and can pose a safety hazard. During Phase II, the charge current gradually reduces until it reaches a minimum termination point.

Adapter and Battery Models

Two types of regulated adapters are typically used in handheld device applications: the CV type and the CV/CC type. The CV adapter's operation is straightforward, as it regulates the voltage at a fixed level for all output currents up to the over-current protection level.

The CV/CC adapter operates as a CV adapter before the load current reaches the current limit. The CV/CC adapter differs from the CV adapter in that once the current limit is reached, the output voltage will fall but the adapter will continue to source current at a constant rate instead of completely shutting down.

Fig. 2 shows the idealized V/I (voltage/current) characteristics of a CV/CC adapter output. The CV region can be modeled as a voltage source with an output resistance in series. V_NL is the no-load adapter output voltage and V_FL is the full load voltage at the current limit I_LIM. The slope r_O represents the output resistance of the CV region.

For a well-regulated CV characteristic, the output resistance is very small. The adapter can be modeled as a current source when running in the CC region. Because it is a current source, its output voltage depends on the load. In a charger application, as the battery is being charged, the adapter output rises from a lower voltage in the V/I characteristics curve, such as point A, to a higher voltage until the breaking point B. Eventually, as the charge current drops, the adapter output moves toward point C.

The rechargeable Li-ion battery is typically a pack that contains a battery cell, a protection circuit and other electronics. The battery can be approximated as an ideal cell with a lumped-sum resistance in series, also shown in Fig. 2. The resistance mainly comes from the equivalent-series resistance of the cell and the on resistance of the protection MOSFETs.

Linear Chargers

Fig. 3(a) shows a simplified block diagram of a linear charger that contains the pass element (PMOS), the reverse blocking diode, the current control loop and the voltage control loop. One external resistor is also shown to program the charge current during the CC mode. The CV type adapter is required for the linear charger.

Before the battery reaches the final voltage, the voltage loop amplifier (VA) is isolated by the diode at its output. The current loop regulates the charger to output a constant current. The I_REF pin programs the current reference I_R. The current amplifier controls the gate voltage of the PMOS so that the sensed current matches the current reference. As the battery voltage reaches the final voltage, the voltage amplifier VA outputs a current to effectively reduce the reference current to the current loop.

Fig. 3(b) illustrates the operation of the linear charger. The power dissipation in the linear charger is, P_CHGR = (V_IN - V_BAT)*I_CHG (Eq. 1) where P_CHGR is the charger power dissipation, V_IN is the input voltage, V_BAT is the battery voltage, and I_CHG is the charge current.

Since the input and the output voltages have a large difference at the beginning, the power dissipation is at its maximum. For example, if the battery voltage is 3V, the adapter voltage is 5V, and the charge current is 750mA, the power dissipation is 1.5W (difficult to dissipate in a small handheld device). As the battery voltage rises during the charge, the power dissipation drops. After entering the CV mode, the power dissipation drops further because the charge current starts to fall.

Linear Chargers with Thermal Fold-Back

As illustrated in Fig. 3(b), linear chargers can have high thermal dissipation during the initial CC fast charge phase. However, more advanced chargers such as ISL6291 and ISL6292 have a charge current thermal foldback function to prevent over-dissipation. Thermal foldback greatly simplifies thermal design and provides protection against thermal damage to the charger or to the application.

The concept of thermal foldback is illustrated in Fig. 4. The difference between the block diagram in Fig. 4(a) and Fig. 3(a) is the addition of the thermal monitoring block shown in blue. Before the IC internal temperature reaches a threshold, typically 100°C in the ISL6291 [2], the thermal monitoring block has no effect on the charger. Once the temperature reaches the threshold, any further increase of the IC temperature results in reduction of the charge current, the cause of the temperature rise. Consequently, the temperature rise is limited.

Fig. 4(b) shows the typical operating curves for such a charger. The curves are almost the same as those for a traditional linear charger, with the exception that the effects caused by thermal foldback are shown in black. The maximum power dissipation causing the IC die to reach the temperature threshold depends on the layout, PCB size, and the ambient temperature. Because the battery voltage is relatively low at the beginning of a charge cycle, the power dissipation is large if the charge current is constant. With the thermal foldback, the charge current is reduced and power dissipation and temperature are controlled.

The linear charger that has thermal foldback capability is an improvement over traditional linear chargers, but the power dissipation is still large and the charge process can be retarded if the power dissipation capability of the printed circuit board is small.

Dual-Mode Chargers

What characteristics would an ideal charger for one-cell Li-ion application have? Low thermal dissipation, low cost, compact size and low noise are all important. The dual-mode charger offers a solution that provides all of those characteristics.

Fig. 5 on page 47 shows a block diagram of the dual-mode charger modeled after ISL6292. Notice that the block diagram is identical to the one in Fig. 4 except the adapter is a CV/CC type instead of a CV type. One other difference, which is not shown in Fig. 5, is that the dual-mode charger requires that the IC be capable of operating at a much lower input voltage (to be explained soon). From the block diagram, one can find that the dual-mode charger is a linear charger. The difference happens when a CV/CC adapter is used to power the charger and, in addition, the reference current is programmed to be higher than the current limit of the CV/CC adapter.

Let's consider what happens when a CV/CC adapter is used. Before the battery voltage reaches the final voltage, the voltage loop is isolated. The current control loop tries to regulate the charge current by controlling the gate voltage of the PMOS. Since the current is limited by the adapter and never reaches the reference current, the controller will continue to enhance the PMOS by reducing the gate voltage until the current-loop amplifier CA saturates. Under this condition, the gate-source voltage of the PMOS is already higher than the gate threshold voltage; therefore, the PMOS is actually turned on as a switch. Since the PMOS if fully turned on, the adapter voltage drops to a voltage slightly greater than the battery voltage. The difference is:

V_IN - V_BAT = R_DS(ON) * I_LIM (Eq. 2) where the R_DS(ON) is the PMOS on resistance and I_LIM is the limited current of the CV/CC adapter. Thus the power dissipation is:

P_CHGR = R_DS(ON) * I_LIM² (Eq. 3)

The actual power dissipation is dependent on the R_DS(ON). In general, using an external MOSFET is easier to obtain lower on resistance but that will also require a reverse blocking diode and result in higher total loss.

Fig. 5(b) shows the operating waveforms. Notice that the input voltage is very low when the battery voltage is low. That is the reason the IC needs to be capable of operating at a much lower voltage than a traditional linear charger. The charge current is shown in the green color with the programmed reference charge current I_REF shown with a light green dotted line. The power dissipation is shown in the orange color. The yellow dotted line shows the power dissipation of a traditional linear charger for comparison.

Once the charger enters the CV charge mode, the charge current starts to drop, so does the adapter current. For the CV/CC type adapter to reduce the current, it needs to enter the CV region. Usually a voltage discontinuity occurs during the transition. The adapter voltage jumps to V_FL, as shown in Fig. 5(b). If the charge current reduces gradually, the power dissipation curve will also show a discontinuity. The worst-case power dissipation for the dual-mode charger usually occurs during this transition. Even so, taking 750mA charge current and 5V adapter voltage as the example, the peak power dissipation is (5V - 4.2V)*0.75A = 0.6W, much less than the 1.5W worst case in a linear charger. The 4.2V is the typical final charge voltage for rechargeable Li-ion batteries.

The charger in Fig. 5(a) still has the thermal foldback function, thus, if the IC internal temperature exceeds the threshold, the charge current will also be reduced, so will the power dissipation. The gray dotted lines show such a case for the charge current and the power dissipation respectively.

Further Cuts In Power Dissipation

Proper design of the CV/CC adapter can further reduce the peak power dissipation in the dual-mode charger. The following demonstrates that the thermal dissipation can be minimized if the adapter output reaches the full-load output voltage (point B in Fig. 2) before the battery voltage reaches the final voltage (4.2V).

The entire charging system can be modeled as the circuit shown in Fig. 6(a) when charging in the CC region of the adapter output. The pass element in the charger is fully turned on, and the charger is equivalent to a resistor. The charge current is the current limit I_LIM and the charger input voltage V_IN can be found in Equation 2.

A critical condition for the adapter design occurs when the adapter output reaches the full load voltage (point B in Fig. 2) at the same time as the battery pack voltage reaches the final charge voltage V_CH (4.2V typically), that is:

V_CFL = R_DS(ON) * I_LIM + V_CH (Eq. 4) where V_CFL is the critical full load voltage of the adapter. For example, if the final voltage is 4.2V, the R_DS(ON) is 350mΩ, and the current limit I_LIM is 750mA, the critical adapter full-load voltage is 4.4625V. When the critical condition is true, the charger enters the CV mode simultaneously as the adapter exiting the CC region.

The charging system can be modeled as the circuit in Fig. 6(c) when in the CV mode. The charger outputs a constant 4.2V. Because the charge current drops at a higher rate than the rising rate of the adapter voltage in the CV mode, the power dissipation decreases as the charge current decreases. Therefore, the worst-case thermal dissipation occurs in the constant-current charge mode.

Fig. 7(a) shows the V/I curves of the adapter output, the battery pack voltage and the cell voltage during the charge. The 5.9V no-load voltage is just an example value higher than the full-load voltage. The cell voltage 4.05V is under the assumption that the pack resistance is 200mΩ. Fig. 7(b) illustrates the adapter voltage, the battery pack voltage, the battery cell voltage, the charge current and the power dissipation in the charger respectively in the time domain. The power curve illustrates that the worst-case power dissipation occurs during the CC charge mode.

If the adapter voltage reaches the full-load voltage before the battery voltage reaches 4.2V, the charger will enter a resistance-limit mode. When the adapter voltage reaches the full load voltage, it cannot rise further without exiting the CC region. At the same time, the battery voltage will not have yet reached the final voltage, thus the charger will attempt to regulate the charge current to the reference. Fig. 6(b) shows the equivalent circuit for this mode. Eventually the battery pack voltage will reach 4.2V because the adapter no-load voltage is higher than 4.2V. At this point the equivalent circuit becomes the one shown in Fig. 6(c) again.

As with the traditional linear charger, the worst-case thermal dissipation also occurs in the CC charge mode. Fig. 8(a) shows the V/I curves of the adapter output, the battery pack voltage, and the cell voltage for the case V_FL = 4V. The rest of the assumptions are the same as the critical condition case. In this case the full-load voltage is lower than the 4.2V final charge voltage, but the charger can still fully charge the battery as long as the no-load voltage is above 4.2V. Fig. 8(b) illustrates the adapter voltage, the battery pack voltage, the cell voltage, the charge current, and the power dissipation in the charger respectively in the time domain.

Based on the above discussion, the worst-case power dissipation occurs during the CC charge mode if the adapter full-load voltage is lower than the critical voltage given in Equation 4.