The most obvious approach is to connect a diode between the power source and the load, but this introduces additional power consumption due to the diode's forward voltage. While this method is simple, diodes are unsuitable for portable or backup applications because the battery must draw current when charging and supply current when not charging. Another approach is to use one of the MOSFET circuits shown in Figure 1.
Figure 1: Traditional load-side reverse protection
For load-side circuits, this approach is better than using diodes because the power supply (battery) voltage enhances the MOSFET, resulting in less voltage drop and substantially higher conductance. The NMOS version of this circuit is better than the PMOS version because discrete NMOS transistors have higher conductivity, lower cost, and better availability. In both circuits, the MOSFET turns on when the battery voltage is positive and turns off when the battery voltage reverses. The physical "drain" of the MOSFET becomes the power supply, as it is a higher potential in the PMOS version and a lower potential in the NMOS version. Because MOSFETs are electrically symmetrical in the transistor region, they conduct current well in both directions. When using this approach, the transistor must have maximum VGS and VDS ratings higher than the battery voltage.
Unfortunately, this method only works for the load-side circuitry and cannot be used with circuitry capable of charging the battery. The battery charger will generate power, reactivate the MOSFETs, and re-establish the connection to the reverse battery. Figure 2 shows an example using an NMOS version, with the battery shown in a faulty state.
Figure 2: Load-side protection circuit with a battery charger
When the battery is connected, the battery charger is idle, and the load and battery charger are safely decoupled from the reverse battery. However, if the charger goes into operation (e.g., with the input power connector connected), the charger generates a voltage between the gate and source of the NMOS, which enhances the NMOS, thus enabling current conduction. This is illustrated more clearly in Figure 3.
Figure 3: Traditional reverse battery protection schemes are ineffective for battery charger circuits.
While the load and charger are isolated from reverse voltage, a major problem with the MOSFET, which serves as the protection mechanism, is excessive power dissipation. In this case, the battery charger becomes a battery discharger. The circuit reaches equilibrium when the battery charger provides sufficient gate support to the MOSFET to absorb the current supplied by the charger. For example, if a strong MOSFET has a VTH of approximately 2V, and the charger can supply current at 2V, the battery charger output voltage will stabilize at 2V (the MOSFET's drain is at 2V + battery voltage). The power dissipation in the MOSFET is ICHARGE•(VTH + VBAT), causing the MOSFET to heat up until the generated heat dissipates off the printed circuit board. The PMOS version of this circuit works similarly.
The following section will introduce two alternatives to this method, each with its own advantages and disadvantages.
N-channel MOSFET design
The first approach uses an NMOS isolation device, as shown in Figure 4.
The algorithm for this circuit is as follows: if the battery voltage exceeds the output voltage of the battery charger, the isolation MOSFET must be disabled.
Similar to the NMOS method described above, in this circuit, MN1 is connected to the low-voltage side of the wiring between the charger/load and the battery terminals. However, transistors MP1 and Q1 now provide a detection circuit that disables MN1 in the event of a reverse battery connection. A reverse battery connection raises the source of MP1 above its gate, which is connected to the positive terminal of the charger. The drain of MP1 then supplies current to the base of Q1 through R1. Q1 then shunts the gate of MN1 to ground, preventing charging current from flowing in MN1. R1 controls the base current flowing to Q1 during reverse detection, while R2 provides discharge to the base of Q1 during normal operation. R3 gives Q1 the power to pull the gate of MN1 to ground. The R3/R4 voltage divider limits the voltage on the gate of MN1 so that the gate voltage does not drop as much during reverse battery hot-plugging. The worst-case scenario is when the battery charger is already running, generating its constant voltage level, and a reverse-connected battery is attached. In this situation, MN1 must be turned off as quickly as possible to limit the time of high power consumption. This particular version of the circuit with R3 and R4 is best suited for 12V lead-acid battery applications, but R4 can be omitted in lower voltage applications such as single-cell and two-cell lithium-ion battery products. Capacitor C1 provides an ultra-fast charge pump to pull down the gate level of MN1 during reverse battery attachment. C1 is very useful for the worst-case scenario (where the charger is already enabled when a reverse battery is attached).
The drawback of this circuit is that it requires additional components; the R3/R4 voltage divider creates a small but continuous load on the battery.
These components are mostly compact. MP1 and Q1 are not power devices and are typically available in SOT23-3, SC70-3, or smaller packages. MN1 should have very good conductivity as it is a transmission device, but its size does not need to be large. Because it operates in the deep transistor region and receives significant gate enhancement, its power consumption is low even for devices with moderate conductivity. For example, transistors below 100mΩ are often packaged in SOT23-3 packages.
Figure 4: A feasible reverse battery circuit
However, a drawback of using a small transmission transistor is that the additional impedance in series with the battery charger prolongs the charging time during the constant voltage charging phase. For example, if the battery and its wiring have an equivalent series resistance of 100mΩ and a 100mΩ isolation transistor is used, the charging time during the constant voltage charging phase will double.
The detection and shutdown circuit consisting of MP1 and Q1 doesn't shut down MN1 particularly quickly, and it doesn't need to. Although MN1 generates high power consumption during reverse battery connection, the shutdown circuit only needs to disconnect MN1 "last." It must disconnect MN1 before its temperature rises to the point of causing damage. A disconnection time of tens of microseconds might be suitable. On the other hand, it is crucial to shut down MN1 before the reverse battery has a chance to pull the charger and load voltages to negative values, hence the need for C1. Essentially, the circuit has one AC and one DC shutdown path.
This circuit was tested using a lead-acid battery and an LTC4015 battery charger. As shown in Figure 5, the battery charger is in the OFF state when the battery is hot-plugged in the reverse direction. Reverse voltage is not transmitted to the charger or the load.
Figure 5: NMOS protection circuit when the charger is off
It is worth noting that MN1 requires a VDS rating equal to the battery voltage and a VGS rating equal to half the battery voltage. MP1 requires both VDS and VGS ratings equal to the battery voltage.
Figure 6 illustrates a more severe scenario: the battery charger is in normal operation when the battery is being hot-swapped in reverse. Reverse battery connection pulls down the charger's voltage until detection and protection circuitry pulls it out of operation, allowing the charger to safely return to its constant voltage level. The dynamic characteristics will vary depending on the application, and the capacitance on the battery charger will significantly influence the final result. In this test, the battery charger combined a high-Q ceramic capacitor with a lower-Q polymer capacitor.
Figure 6: NMOS protection circuit when the charger is in operation
In summary, it is recommended to use aluminum polymer capacitors and aluminum electrolytic capacitors in battery chargers to improve performance during normal forward battery hot-plugging. Pure ceramic capacitors can generate excessive overshoot during hot-plugging due to their extreme nonlinearity. The reason is that their capacitance can drop by a staggering 80% as the voltage rises from 0V to the rated voltage. This nonlinearity induces high current flow at low voltage conditions, while causing the capacitance to decrease rapidly as the voltage rises; this is a fatal combination leading to very high voltage overshoot. Empirically, a combination of a ceramic capacitor with a lower Q-value, voltage-stable aluminum capacitor or even a tantalum capacitor appears to be the most robust configuration.
P-channel MOSFET design
Figure 7 illustrates the second method, which uses a PMOS transistor as a protection device.
Figure 7: PMOS transistor transport element version
In this circuit, MP1 is a reverse battery detection device, and MP2 is a reverse isolation device. The source-to-gate voltage of MP1 is used to compare the positive terminal of the battery with the output of the battery charger. If the battery charger terminal voltage is higher than the battery voltage, MP1 will disable the main transmission device MP2. Therefore, if the battery voltage is driven below ground potential, the detection device MP1 will obviously drive the transmission device MP2 to the off state (interfering with its gate to its source). This operation will be performed regardless of whether the battery charger is enabled and generating a charging voltage or disabled (0V).
The biggest advantage of this circuit is that the PMOS isolation transistor MP2 has no capability to deliver negative voltage to the charger circuit and the load. Figure 8 illustrates this more clearly.
Figure 8: Schematic diagram of the cascode effect
The lowest voltage achievable on the gate of MP2 via R1 is 0V. Even if the drain of MP2 is pulled far below ground potential, its source will not experience significant downward voltage pressure. Once the source voltage drops to VTH, which is above ground potential, the transistor will debias itself, and its conductivity will gradually disappear. The closer the source voltage is to ground potential, the greater the degree of debiasing of the transistor. This characteristic, coupled with a simple topology, makes this method more favored than the NMOS method described earlier. Compared to the NMOS method, it does have the disadvantages of lower conductivity and higher cost of PMOS transistors.
While simpler than the NMOS method, this circuit has a significant drawback. Although it always provides protection against reverse voltage, it may not always connect the circuit to the battery. When the gate is cross-coupled as shown, the circuit forms a latch-up storage element that can potentially select an incorrect state. Although difficult to implement, there is a scenario where the charger is generating a voltage (e.g., 12V), and the battery is connected at a lower voltage (e.g., 8V), causing the circuit to disconnect.
In this scenario, the source-to-gate voltage of MP1 is +4V, thus boosting MP1 and disabling MP2. This situation is illustrated in Figure 9, with the stable voltages listed at the nodes.
Figure 9: Diagram of possible blocking states when using a PMOS protection circuit
For this condition to be met, the charger must already be running when the battery is connected. If the battery is connected before the charger is enabled, the gate voltage of MP1 is fully pulled up by the battery, thus disabling MP1. When the charger is turned on, it generates a controlled current (instead of a high current surge), which reduces the likelihood of MP1 being turned on and MP2 being turned off.
On the other hand, if the charger is activated before the battery is attached, the gate of MP1 simply follows the output of the battery charger, as it is pulled up by the bleed resistor R2. When the battery is not connected, MP1 has no tendency to turn on or deactivate MP2.
A problem arises when the charger is already powered on and running, and the battery is attached. In this situation, a momentary difference exists between the charger output and the battery terminals, which causes MP1 to disengage MP2 as the battery voltage forces the charger capacitor to absorb charge. This creates a competition between MP2's ability to draw charge from the charger capacitor and MP1's ability to disengage MP2.
The circuit was also tested with a lead-acid battery and an LTC4015 battery charger. Connecting a heavy-duty 6V power supply as a battery simulator to an enabled battery charger never triggered a "disconnect" state. The tests performed were not comprehensive and should be more thorough in critical applications. Even if the circuit is locked, disabling and re-enabling the battery charger will always result in a reconnection.
The fault condition can be demonstrated by manually manipulating the circuit (establishing a temporary connection between the top of R1 and the battery charger output). However, this circuit is generally considered more suitable for connection failures. If connection failure does become a problem, a circuit that disables the battery charger using multiple components can be designed. Figure 12 shows a more complete circuit example.
Figure 10 shows the effect of the PMOS protection circuit when the charger is disabled.
Please note that under no circumstances will negative voltage be transmitted between the battery charger and the load voltage.
Figure 11 shows the circuit in the unfavorable condition that the charger is already in operation when the reverse battery is hot-plugged.
The effect is almost identical to that of an NMOS circuit. Before disconnecting the circuit and taking the transmission transistor MP2 out of operation, the reverse battery slightly pulls down the charger and load voltage.
In this version of the circuit, transistor MP2 must be able to withstand VDS twice the battery voltage (one for the charger, one for reverse battery connection) and VGS equal to the battery voltage. On the other hand, MP1 must be able to withstand VDS equal to the battery voltage and VGS twice the battery voltage. This requirement is unfortunate because for MOSFET transistors, the rated VDS always exceeds the rated VGS. Transistors with 30V VGS tolerance and 40V VDS tolerance can be found, suitable for lead-acid battery applications. To support higher voltage batteries, Zener diodes and current-limiting resistors must be added to modify the circuit.
Figure 12 shows a circuit example capable of handling two series-stacked lead-acid batteries.
Figure 10: PMOS protection circuit when the charger is off
Figure 11: PMOS protection circuit when the charger is in operation
ADI believes the information it provides is accurate and reliable. However, ADI is not responsible for its use or any potential infringement of third-party patents or other rights as a result of such use. Specifications are subject to change without notice. No license to use any of ADI's patents or patent rights may be implied or otherwise granted.
Figure 12: High voltage reverse battery protection.
D1, D3, and R3 protect the gates of MP2 and MP3 from damage caused by high voltage. D2 prevents the gate of MP3 and the battery charger output from rapidly dropping below ground when a reverse-connected battery is hot-swapped. MP1 and R1 detect reverse-connection or fault disconnection latch-up and utilize the missing RT feature of the LTC4015 to disable the battery charger.
in conclusion
A reverse voltage protection circuit for battery charger-based applications can be developed. Several circuits have been developed and briefly tested, with encouraging results. There is no magic bullet for the reverse battery problem; however, it is hoped that the method presented in this article will provide sufficient insight into the existence of a simple, low-cost solution.
