The portability of portable devices is closely related to the development of batteries, from the initial lead-acid batteries, nickel-cadmium (Ni-Cd) batteries to nickel-metal hydride (Ni-H), lithium-ion (Li-ion) batteries until recently. Lithium polymer (Li-polymer) batteries, the energy density is gradually improved, the mobile performance is getting stronger and stronger, and the shortcomings of the battery are constantly being overcome. This article will introduce the management system design of a portable lithium polymer battery.
The overall structure of the system
The application entity of this design is a portable device used in industry, which adopts Altera's FPGA and NIOS II embedded processor on it, and uses the USB interface to connect with the computer, which is oriented to the application of large data volume. This device requires 30V DC voltage, so it is planned to use a battery pack with 4 1000mAh lithium polymer batteries connected in series; in addition, for waterproof and dustproof considerations, only a square USB interface (USB B Type Socket) is used externally. This USB port At the same time, it has the functions of data transmission and charging.
Figure 1 Block diagram of the overall structure of the system
The actuator needs a DC voltage of 30V and a current of about 80mA. It uses a boost DC/DC circuit. This circuit is controlled by the control core. It usually does not work and only turns on before it needs to act.
The charging uses an external 20V power supply, which is connected through the USB interface. The consideration for using this power supply is for 1C or 0.5C high-current high-speed charging. Since it shares a port with the ordinary USB, in order to avoid entering the charging procedure when the ordinary USB is connected, a voltage judgment circuit is required for judgment.
Since it is difficult to find a suitable chip solution in the market, it is decided to use the remaining logic resources of the FPGA to implement the control function of the charger and add a small amount of analog circuits to assist. This requires that the power supply to the control circuit cannot be interrupted, the battery pack must be online all the time, and the negative electrode of the battery needs to be connected to GND all the time.
1 Voltage sampling
The most important part is the design of the voltage sampling circuit, which requires high precision and is less affected by temperature. The difficulty with this design is that the battery voltage is floating with respect to GND. Many schemes adopt the scheme that the differential op amp is converted to the ground voltage and then input to a dedicated ADC for AD conversion. But this scheme has many problems due to the introduction of differential op amps. First, the voltage is relatively high, and the op amp is difficult to find; second, the power supply of the op amp and the input voltage use the same power supply, which requires the op amp to have the function of rail-to-rail input; third, a negative power supply may be required, Using DC/DC introduces noise; in addition, op amps and the use of matched resistors reduce accuracy.
Figure 2 RC charging circuit
In order to simplify the circuit as much as possible, an integral ADC is constructed here, which converts the high precision of FPGA timing into the high precision of voltage measurement.
This is a simple RC charging circuit (see Figure 2). The work flow is: J1 is closed first to release the charge on C1; then J1 is opened, and C1 is charged by R1; the voltage comparator U1 compares the voltage on C1 with the reference voltage V2, and outputs a high voltage when the voltage of C1 exceeds V2. flat. Counting the time from when J1 is turned on to when U1 outputs a high level, the voltage of V1 can be determined. It can be intuitively seen that the higher the V1, the shorter the period of time.
The actual circuit is shown in Figure 3. Note that this picture only shows the measurement circuit for the first battery. Among them, R1 and C1 are the resistors and capacitors used for integration, Q1 is a commonly used P-MOSFET, which is used here to realize the function of J1 discharging the capacitor, and U5 realizes the dual functions of voltage reference and voltage comparator at the same time. X1 is the discharge control, which comes from the FPGA, and X2 is the switch output, which goes to the FPGA. The voltage comparator is MAX921 from Maxim.
Figure 3 Actual sampling circuit diagram
This circuit only consumes 4μA of MAX921 current and leakage current of C1, Q1, Q2 in static state, which is basically negligible and very power-saving.
Another feature of this circuit is that the often-used optocoupler is omitted and capacitor C2 is used instead. In static state, both ends of C2 reach voltage balance and do not consume power. At this time, the voltage of X2 is 0. When U5 outputs a high level, because the voltage across C2 cannot be transient, the voltage of X2 is boosted. The two Schottky diodes D1 and D2 play a limiting role. By carefully adjusting the values of C2 and R4, the switching information can be transmitted smoothly.
2 Balance charge
Balanced charging is the charging method required for all lithium battery packs, but there is no balanced charging in many low-power applications, such as most laptop battery packs, which actually has a considerable impact on battery life.
Existing equalization technologies are mainly divided into energy transfer equalization between batteries and external energy input equalization. Energy balance between batteries is to charge the energy of the high-battery battery to the low-battery battery. The biggest problem with this method is that it is very complicated to control.
Many special-purpose chips or single-chip microcomputer solutions now use an external equalization method, which is achieved through controllable energy consumption. In this method, an energy-consuming element is generally used to consume energy, thereby waiting for other battery cells to be fully charged or reducing the voltage of some cells. The disadvantage of this scheme is that the energy consumption on the Zener diode is too large, and the resulting heat is unbearable.
Figure 4 Actual charging method
The actual charging method is shown in Figure 4. Of course, this is just a schematic diagram, excluding the current detection circuit (between the input and the transformer) and the voltage detection circuit (transformer secondary winding). Among them, the switch array is realized with power MOSFET.
In this way, the tubes work in the on-off state and consume very little energy. In addition, the battery does not have a diode in series, so the maximum output can be obtained. The disadvantage is that the circuit is more complicated. Since the voltage of each battery needs to be matched, the input charging circuit is required to be isolated. The T1 transformer is used here as isolation, because the switching frequency can be made very high, and the size of the T1 transformer is very small.
The whole charging circuit works in the switching state, no control module is added, the FET is directly controlled by the FPGA, and the output of the current detection and voltage detection circuit is also converted into a switching value and directly transmitted to the FPGA.
Charging is divided into four steps:
a) Detect whether there is a battery cell lower than 2.5V, if so, use a 5% duty cycle to charge the battery lower than 2.5V in turn to boost the voltage to 2.5V;
b) Turn on J1 and J8, charge the whole with high current, and measure the voltage of the battery cells at the same time, if any battery cells reach 4.2V, go to the next step;
c) Gradually reduce the duty cycle to maintain the highest voltage of the single battery at 4.2V until the duty cycle is < 5%;
d) Charge the batteries under 4.2V in turn, when the duty cycle drops to 5%, the charging ends.
It should be noted here that the alternate charging in steps a) and d) is realized by the switch matrix, and the alternate charging will not prolong the charging time, because the duty cycle at this time is much less than 25%, which can be used in a Charge each of the four batteries during a charge cycle.
3 Overcurrent and low voltage protection
In order to ensure the absolute safety of the battery pack, the over-current and low-voltage protection of the battery pack are independently set, and the output of the battery pack can be directly cut off when a problem occurs. This type of circuit is also very common, and will not be repeated here.
In addition, it should be noted that the control system also contains non-volatile memory and battery output detection circuit. When the action of the protection circuit is detected, the current information will be saved in the non-volatile memory for future analysis.
The battery pack composed of multi-cell lithium batteries is an inevitable choice for portable high-power devices. How to manage and maintain this battery pack to make it work with high efficiency and long life is also a task for electronic designers.
This paper provides a new idea, which is to use a simple and accurate circuit to convert complex analog quantities into digital quantities, thereby simplifying the design of external circuits, and handing over complex charging sequence control to programmable logic for processing. Doing so is not only very flexible and accurate, but also reduces cost.
Responsible editor: gt