Abstract: This paper introduces the composition of an unattended DC power backup system for power plants and the charging process of the battery banks. The functional requirements of the key equipment in this system—the intelligent charger—are discussed in detail, along with the hardware and software design methods for implementing this equipment using multiple microprocessors. This system has been successfully applied in several power companies. Keywords: DC power backup system, multi-microcontroller charger. A DC power backup system is essential for various power plants, substations, and power stations. Its main function is to serve as a substitute for the main power supply, providing power to key control equipment, fault monitoring systems, and fault protection systems when the main power supply is suddenly interrupted. It can include multiple battery banks, each with up to one hundred lead-acid or other types of batteries, with a capacity of over 2000 amp-hours. A complete unattended, maintenance-free DC power system consists of battery banks, a floating charging device, battery monitoring (capacity and voltage) devices, insulation monitoring devices, AC monitoring devices, silicon chain voltage regulators, a series of remote control switches, protection subsystems, and an industrial communication network and central controller connecting these devices into a unified system. Multiple sets of the above devices (except for the central controller) can be used. The communication network uses an RS485 interface. To improve the anti-interference capability on site, the RS485 interface should be optically isolated. Industrial fieldbuses such as CAN bus and LonWorks can also be used. The central controller has a human-machine interface with a Chinese character LCD display. On the one hand, it can communicate bidirectionally with each subsystem (device) through the communication network to obtain its real-time operating data and remotely control and adjust these devices. On the other hand, it is also connected to the power station integrated automation system to transmit information of the entire DC system to these higher-level systems. The system structure diagram is shown in Figure 1. Among these devices, the charging float charging device (i.e., the intelligent charger) undoubtedly occupies an extremely important position. Its role is to provide intelligent charging and discharging process control, automatically replenish the energy lost by the battery due to accidental discharge and closing operations, so that the battery pack is always in the optimal energy storage state, ensuring the reliability of the DC backup power system. 1 Battery Charging and Discharging Control Process and Functional Design of Intelligent Charger The batteries in the DC power system may encounter various operating states during system operation, such as discharge caused by AC interruption, and self-discharge loss caused by internal chemical reactions during long-term operation. To ensure battery capacity, the battery must be charged using a specific control process. Figure 2 is an oscilloscope diagram of the operating procedure of the microcomputer-controlled DC power supply system, which was organized and formulated by the former Ministry of Electric Power. As can be seen from the figure, the charging process of lead-acid batteries consists of the following parts: (1) Start-up stage In order to avoid voltage surges that could impact the battery, the charging voltage must rise smoothly upon power-up and reach the given value after a few seconds. This stage is called the soft-start stage, and the charger should implement constant current control. (2) 0.1C10A constant current charging After the soft start is completed, the charger charges the battery with a constant current of 0.1C10A. At this time, the battery terminal voltage will gradually rise. When the terminal voltage rises to 2.35×n (n is the number of batteries), the constant current charging ends and the next stage begins. (3) Equalization stage This stage is constant voltage control, with a given value of 2.35×n. At this time, the charging current will gradually decrease. When the current decreases to 0.01Cl0A, the timing system starts timing. After the "equalization to float setting time" is completed (this time is adjustable, ranging from 0 to 72 hours), the system enters the next stage - the float charging stage. (4) Float charging stage This stage is also constant voltage control, but its voltage setpoint is 2.25×n. After a settable "activation time" (1 to 3 months) in the float charging stage, the system returns to the above stage (2). (5) AC interruption and power restoration When the AC power grid is interrupted, the battery discharges to provide backup power to the system. When the AC system restores power, the charger automatically performs constant current charging on the battery and returns to the above stage (2). In summary, the functions of the intelligent charger are as follows: (1) Complete charging and discharging control of the battery. (2) All control parameters can be set by the user, including equalization charging, float charging voltage setpoint, constant current charging current setpoint, equalization to float charging time, activation time, etc. (3) Comprehensive alarm protection functions, including AC interruption, phase loss, overvoltage, overcurrent, short circuit, etc. (4) RS485 communication interface and corresponding communication protocol, including providing host computer remote control function. (5) Constant voltage and constant current accuracy can reach ±1%, ripple coefficient is below 1%, and power factor and efficiency can also meet the user-given indicators. (6) Temperature has a certain impact on the charging characteristics of the battery, so temperature compensation function is required. 2 Hardware Design of Intelligent Charger This charger adopts the phase-controlled rectifier main circuit of thyristor three-phase fully controlled bridge. The characteristics of this system are high control accuracy requirements and many tasks. In order to improve hardware reliability, the system adopts the design of multiple microcontrollers (single-chip microcomputers). The entire operation task of the system can be roughly divided into the following four major sub-tasks according to the principle of functional independence and load uniformity: synchronous pulse generation, phase-shift trigger pulse generation, process control and constant voltage and constant current algorithm implementation, interface and serial communication, etc. These tasks cannot be completed by one or two microcontrollers, so four microcontrollers are used, each completing one of the above sub-tasks. The microcontrollers communicate with each other via hardware handshake signals and shared RAM. Figure 3 shows the logic diagram of these microcontrollers. The synchronization microcontroller generates synchronization pulses, including phase sequence determination and phase loss determination, thus enabling the system to meet the requirement of not needing to adjust phase sequence. The triggering microcontroller generates and distributes six trigger pulses (determined based on the phase sequence signal and the single-phase conduction signal from the control microcontroller). The human-machine interface microcontroller has many extended functional circuit units, and its principle block diagram is shown in Figure 4. First is temperature sampling. Since the battery cabinet and controller can be hundreds of meters apart, the temperature and current sensing chip AD590 is used. The AD590 outputs a current signal proportional to the absolute temperature value. Due to its high-impedance constant current source output characteristics, it is insensitive to long-line impedance and can also resist field interference. The temperature sampling circuit is shown in Figure 5. The two potentiometers in the figure can be used for zero-point and full-scale calibration. Due to the large number of batteries and their large footprint, three evenly placed AD590s are connected in parallel, and the resulting measurement is the average temperature at three points. The display interface uses two serial 7219 LED display driver chips, expanding to a 4x4-bit LED digital tube display and several LED indicator lights. The 7219 can automatically control the dynamic scanning of the LED digital tube display. The 12C887 calendar clock chip provides the date and time scale required for charging process control, and its nearly 128 bytes of on-chip non-volatile memory can also be used to store control parameters. The DS1609 is a dual-port RAM with 256 bytes. In this system, it serves as shared memory between the HMI microcontroller and the control microcontroller, transmitting common information. Preventing conflicts caused by two microcontrollers simultaneously reading and writing to the same memory cell on the DS1609 is crucial; the system uses a signal mailbox to solve this problem, as described later. The wiring of the control microcontroller is relatively simple, mainly involving the output of a single synchronous phase-shift trigger signal from the trigger microcontroller. 3. Software Design Due to space limitations, only the software design of the interface microcontroller and the control microcontroller will be discussed. The interface microcontroller software is more complex, with the entire software operation driven by two interrupt sources: one is the external interrupt of the main synchronization signal, and the other is the serial port interrupt. The overall software is a multi-task background switching structure. Tasks include voltage and current sampling, keyboard scanning, communication with the 1609, scaling transformation of sampled values, display of sampled values, alarm handling, temperature sampling and display, determining whether complete serial communication data and command frames have been received, and interpreting serial communication command frames. After each external interrupt of the main synchronization signal occurs, the interface microcontroller must complete the first four tasks. Other less urgent tasks are selected and activated by the main control program in turn, with the only criterion for activation being the order. The left microcontroller refers to the interface microcontroller, and the right microcontroller refers to the control microcontroller. The so-called "left set, right clear" means that the flag is set by the interface microcontroller and reset by the control microcontroller. Figure 6 shows the flowchart of the left microcontroller reading the shared information of the 1609. It can be proven that by using the above four flags, read and write conflicts of the shared memory unit can be completely avoided. The main task of the microcontroller is to complete the control algorithm and output the control quantity. Its software operation is driven by the external interrupt brought by the synchronization pulse every 20ms. The constant voltage and constant current algorithm adopts the anti-differential saturation PID algorithm. Since the control quantity output is a phase shift angle, this angle is determined by the timing time of the on-chip timer D (t0). According to the relationship between the timing time constant of the 89C52 timer and the main frequency (11.059MHz), it can be deduced that: timing constant = 216 - 921.6 × 1/18 × α (degrees) = 216 - 51.2 × α (degrees). The main flowchart of the microcontroller software is shown in Figure 7. The external interrupt service routine of the synchronization pulse only sets one occurrence flag, while the timing interrupt service routine of t0 needs to output a phase shift pulse of about 1ms. Therefore, it is necessary to set the timing time constant of 1ms and restart the timing. The relevant software block diagram is omitted.