Sealed Lead-Acid Battery Charger Circuits

Manufacturers Sealed Lead-Acid (SLA) battery specifications generally prescribe a sequence of three charging stages or modes for fast charging. This usually takes the following form (4Ah Diamec batteries were used as examples):

1. Charge at a current of 0.4C amps until the terminal voltage reaches a limit typically of 2.5V per cell (7.5V for a 6 volt and 15V for a 12 volt battery, but only for SLA AGM type batteries). C is the rated battery capacity in Ampere-Hours (AH) which should be marked on the battery case.
2. Continue to charge at that constant voltage until the current falls to a value typically around 0.005C. At that point the battery is essentially fully charged.
3. Charge the battery at a reduced float charge voltage of 2.3V per cell (6.9V for a 6 volt and 13.8V for a 12 volt battery). This can be held indefinitely.

These specifications do not hold for other types of lead-acid batteries such as wet cells. SLA batteries are generally of the Absorbed Glass Mat (AGM) construction and can tolerate higher voltages and currents. Effects of temperature need to be taken into account, and the above numbers should be taken only as representative of any SLA AGM battery and only at standard room temperature.

The chargers found in consumer equipment typically require a dc voltage to be supplied and will charge the battery through a resistor for current limiting. Sometimes a zener diode controlled transistor switch will be used to prevent exceeding the specified maximum voltage limit. These very simple circuits are not optimal and do not change to the third float charge stage when the battery is fully charged. This may reduce the life of the battery if it is left on charge too long.

The best way to approach the charging of the higher capacity lead-acid batteries is to use an efficient switch-mode power source that can be controlled either by a dedicated chip or a microcontroller; the latter providing greater flexibility to adapt to the charging process and to reconfigure itself for different battery types and capacities (see e.g. Atmel Application Note
AVR450). A project of this type will eventually be described elsewhere on this site.

For the lower capacity SLA batteries, simpler analogue chargers can be built with a handful of discrete components
to provide an automatic fast charge, and can even be made to operate in each of the three modes without manual intervention. The circuits shown below are designed to charge 6V SLA batteries with a 12-14V supply and consist of three sections, one for each of the stages:

1. A constant current circuit set to give the specified maximum current for the first charging stage.

2. A constant voltage circuit set to limit the voltage to the maximum voltage for the second charging stage.

3. A circuit to drop the voltage limit when the current becomes sufficiently low.

Constant Current Limiting

The first circuit uses forward biassed diodes in the base circuit of a BJT to fix the voltage across an emitter resistor, and hence set the current. Normally a zener diode would be used to provide a stable voltage. However this circuit needs to be able to work down to quite low power supply voltages, so the forward voltage drop of a diode is used, which is nominally taken to be about 0.7V. One of the diodes shown in the circuit is intended to cancel the base-emitter voltage of the BJT and the second is to force the voltage across R2 to be constant. While the diode voltage and the base-emitter voltage of the BJT have a strong logarithmic dependency on the current, they nevertheless are not perfectly constant.

The SPICE circuit shown was tested with a dc sweep of the battery voltage (using gEDA tools and ngspice). The battery is represented by VBattery which is an ideal voltage source, but this is sufficient for comparative testing of the circuit. battery voltage is along the x-axis and current is along the y-axis.

The small decrease in the controlled current observed in the plot below is due to variations in the base current of Q1 affecting the current available to Q2. When the voltage across Q1 and R2 drops below 1.4V the BJT and diodes are starved of voltage and the current drops away sharply.

The second circuit is adapted from that given by Anantha Narayan. The controlled current is set by R2 and the base-emitter voltage of Q2, which is typically about 0.7V. R1 must be chosen to provide the base current of Q1 and also the collector current of Q5 which in turn sets its precise base-emitter voltage. If Q1 and Q5 are operating in active mode, then a momentary increase in current through R2 will result in an increase in the base-emitter voltage of Q2 which will in turn cause the base-emitter voltage of Q1 to decrease. This will pull the current back to a stable value. The advantage of using a two BJT circuit of this form is that the current can be set more precisely than the diode circuit and doesn't vary a great deal under normal operating conditions. The strong temperature variations that will occur in the power transistor Q1 will have little effect on the circuit operation. Q2 is isolated from the changing voltage and temperature environment, and is able to do the bulk of the work relatively stress free.

The value of R1 affects the range over which the current is reasonably constant and needs to be chosen carefully. It is important that R1 be connected to a constant supply voltage so that variations in the current through it are small.

The simulations showed a very similar dependency of controlled current with battery voltage.

Constant Voltage Limiting

The voltage limiting in the second phase of charging can be achieved using a common circuit in which the battery is placed in the emitter circuit of a BJT that has its base voltage limited by a zener diode.

In this circuit an extra diode has been
added to provide additional voltage drop. This will later be used as part of a strategy to reduce the charge voltage limit for the third phase of charging.

The voltage limit on the battery is to be 7.5V. While current is flowing into the battery, the diode will cancel the base-emitter voltage of the
BJT. Therefore the zener would normally be chosen as 7.5V. However the base-emitter voltage of the BJT will decrease as the current through the BJT and the battery decreases. This voltage decrease is not matched by any changes in the diode voltage, thus causing the turn off to stretch up to higher battery voltages. This would allow the battery voltage to exceed 7.5V in the latter stages of the charging. As such we would select a lower value zener diode to start turning off the BJT earlier.

Current Limit Switch

We wish to switch to a float charge when the current into the battery has fallen to a preset value in the second phase of charging. We can do this by sensing the voltage across a resistor in series with the battery, and using this to reduce the voltage limit allowed for charging. The voltage limiting circuitry described above has two devices for voltage limiting, a zener diode and a forward biassed diode. We can use the switch to short circuit the diode and so reduce the voltage limit from the nominal 7.5V to the float charge level 6.9V.

A Discrete Component Circuit

The circuit shown has the constant current circuit at the bottom. This allows the collector resistor of Q6 to be taken to the power rail and so provide a reasonably stable current to Q6.

A resistor bypassed with a diode has been added into the battery circuit to provide a measurement of the charging current. This provides base drive to Q4 which will be turned on while the battery current is flowing during the first two charging phases. The base of Q3 will be pulled low and Q3 will be off.

When the current has dropped to the preselected level to signal the end of charging, the base voltage of Q4 will drop and Q4 will start to turn off. The base voltage of Q3 will rise and it will switch on. This will short circuit the diodes in series with the zener diode, further pulling down the base of Q2 and forcing it to jump rapidly into complete cutoff. From that time on, the battery characteristics take over (these are not modelled here). Its terminal voltage will gradually decrease to the point that Q2 will start to turn back on and allow float charge current to flow. However this current will not be enough to turn Q4 back on again, and the battery voltage will stabilise at the lower float voltage. The capacitor C1 is necessary to prevent premature switching with noise and to allow the circuit to startup in the bulk charge phase.

Q5/D7/R7 are optional and provide a visual indication of charge termination.

The circuit uses a power supply of 12V. However this is insufficient to ensure that the current is maintained as the battery charges towards the 7.5V limit. Q2 is pushed into cutoff prematurely causing the current to droop after the battery voltage reaches about 6.5V. The graph below was generated with a slightly higher power supply of 14V.

The graph shows a Spice dc sweep simulation of the battery voltage showing the change in battery current (magenta) up to the point where the switch circuit cuts in Beyond that point the results are meaningless as the battery voltage never reaches to those levels. The blue curve shows how the voltage across the current sense resistor varies as the current falls, while the red curve is the collector-emitter voltage of Q3 impressed across the diode chain.

Practical Reality

The circuit as shown does work but not quite in the way that was hoped. Firstly the current sense resistor, unless it is very small, is dominated by the characteristics of the diode D5 and the turn-on characteristics of Q4, and might as well be omitted. Thus the trigger for switching is in fact somewhat higher than that desired and is controlled only by the selection of the diode (which must be able to handle the expected power dissipation). We need a better circuit for this purpose, involving a small value resistor and a high gain amplifier. This could be achieved with low cost operational amplifiers.

When the battery enters the final trickle charge phase the current is small but non-zero. This results in a voltage across the base-emitter of Q2 that can be
upwards of 0.5V. Therefore the zener diode does in fact need to be higher in value as shown, and only two additional diodes used in the chain. The variable and unknown voltage offsets in the BJTs can be eliminated also by means of operational amplifiers.

The circuit exhibits hysteresis. As such the battery needs to be discharged below a certain level in order for the circuit to begin operation reliably in the high current bulk charging phase (a capacitor at the base of Q3 can help provided the battery is connected before power is applied). A push button switch at the base of Q3 can also be used to manually reset the circuit. This will raise the voltage cutout limit temporarily and allow high current to flow. This current will in turn disable the switch.

An Operational Amplifier Circuit

A quad operational amplifier chip (14 pin DIP) will cost barely more than two small signal BJTs. In the above circuit three or four BJTs can be replaced and a lower cost, conceptually simpler and more precise circuit will result.

Choose a current sensing resistor small enough that the voltage across it at the highest current flow is negligible in the circuit. The value 0.7V is a suitable value but could be smaller. This will eliminate any need for diode limiting. The circuit is to switch to float charge when the current falls to the selected trigger point: for the example Diamec batteries the maximum current is to be 0.4C and the termination current is to be 0.005C.

An operational amplifier can be used, boosted with a power BJT, to set a constant voltage between the BJT emitter and a more negative point. This simple device can be used to establish both a constant current circuit and a voltage limiting circuit to replace those used in the original circuit above. If a comparator were to be used rather than an operational amplifier then all four of the small signal BJTs may be eliminated with the four comparators in a quad package. Note that it is worthwhile including diode protection for the LM324 positive power in case the supply is reversed.

The operational amplifier chosen is the LM324 which is a low cost quad opamp device that can be used with a single supply. In the circuit shown the BJT of the constant current circuit is replaced with the operational amplifier X3 to ensure that the voltage across R2 is precisely that of the diode D4. The diode voltage is unaffected by changes in the heavy battery current as it is isolated by X3. The resistor R1 is selected to set the voltage across the diode D4 to about 0.7V.

Similarly X2 isolates the diode chain D1 and D2 from the base current of Q2 and allows a more precise setting of the limiting battery voltage. Resistor R3 is chosen to set the voltage across D2 to about 0.7V.

The resistor R5 provides a measurement of the charging current. This is amplified by X1 used as a comparator which turns Q3 off during bulk charging and turns it on when the current has dropped to the termination point.

Diode D3 and the associated resistor chain provide a reference voltage to define the voltage across R5 at which bulk charge termination occurs. The value of R10 can be adapted to set this point.

It is tempting to remove R5 and use R2 as a measure of current. This does not work because Q1 moves into deep saturation as the battery current drops and as X3 responds by trying to force Q1 to pull more current. The result is that the base current of Q1 increases in R2 and the voltage across it does not fall low enough.

There is still some uncertainty in the circuit as the voltages across the diodes will depend on the diode choice and the current through them. However we have effectively removed the variability of these voltages. The circuit can be tuned by adjusting the diode currents, although it is not necessary to have very great precision for this application.

As with the previous discrete component circuit, when the current reaches the termination point the diode D2 is shorted and the circuit switches off all current to the battery until the latter settles to below 6.8V, after which it enters the float charging phase.

The graph shows a Spice dc sweep simulation of the battery voltage showing the change in battery current (red) and the voltage across the diode D2 (blue). It can be seen to hold the desired voltage limit very closely as the current reduces to the termination point. The slight rounding in the battery current curve is due to Q2 moving from saturation to cutoff as the battery voltage increases. This can be eliminated by increasing the power supply voltage beyond 12V.

Prototype Circuit Measurements

The final graph for the operational amplifier based charger shows measurements taken over time of a small 6V SLA battery being charged from an initial terminal voltage of 6.3V. The curves clearly show the voltage rising to about 7.3V while the current is essentially constant at around 850mA. The voltage remains at 7.3V while the current drops toward the termination point at which the circuit will switch to a float charge mode. The actual termination current was set somewhat high at 94mA for illustrative purposes. Note that the terminal voltage doesn't drop instantaneously when the current is switched off. Ultimately the system settles to a float charge of about 5mA at a terminal voltage of 6.6V. The vertical scale represents battery terminal voltage in volts, while the current in amps is scaled up by a factor of five for better visibility.

In practice the circuit didn't achieve the desired 0.4C (1.6A) bulk charging current, but that should only require a reduction in the current setting resistor R2.

Discrete Circuit Prototype

Operational Amplifier Circuit Prototype

First created 19 May 2010

Last Modified 31 August 2010
© Ken Sarkies 2010