Design and implementation of intelligent high-power DC power supply

In high-power DC power supplies, the main circuit generally uses a thyristor three-phase fully controlled bridge rectifier circuit. The key lies in how to accurately, reliably and stably control the conduction angle of the thyristor.

At present, the most common control method in the field application of high-power DC power supplies mostly uses KC or KJ series small-scale integrated circuits, that is, a phase-shifted signal obtained by comparing the three-phase sawtooth wave signal with the DC control signal. However, the slope, duty cycle, amplitude, etc. of the three-phase sawtooth wave signal are closely related to the device parameters of each phase, and small interference in the comparison signal may cause a large phase shift error, so the reliability and automatic balancing ability of the circuit are poor.

Using a single-chip microcomputer as a control circuit, according to the logical relationship between the three-phase fully controlled bridge trigger pulses, a six-phase highly balanced trigger pulse is directly generated, which can overcome the shortcomings of poor balance of the KC and KJ series circuits. However, since the field system works in the place where strong electric interference is serious, in order to reduce the interference that may cause the program to run disorderly, causing the system to lose control and cause damage to the main circuit components; in addition, in order to enhance the function of the system, strengthen the human-computer dialogue ability, and realize the functions of display, printing, command input, cycle detection, overvoltage and overcurrent protection, and software PI regulator, it is necessary to use dual CPUs to work in parallel. However, the parallel operation of dual CPUs not only increases the complexity of the system, but also reduces the reliability and practicality of the system.

In order to overcome the above limitations, the 8098 single-chip microcomputer is used as the main control unit, and the anti-interference performance of WATCHDOG is fully utilized. The general trigger board with phase-locked loop (PLL) as the basic control principle is used as the intermediate interface to form an intelligent high-power DC backup power supply for power plants. Figure 1 shows the control system block diagram.

Figure 1 Control system block diagram

1 System working principle

Taking the example of the power system for forced charging and floating charging of batteries, the working principle of the system is explained. According to the site requirements, the system has 7 working modes, see Figure 1.

1) Manual mode (M)
The system works in an open-loop state, using the PWM port of 8098 to output a 0~5V control voltage signal to the trigger board after filtering, so that the corresponding output voltage of the rectifier bridge is 0~300V. This mode is mainly used for the inspection and maintenance of the main circuit of the system.

2) Voltage stabilization mode (V)
The voltage stabilization mode (V) is also called the floating charge mode. The system operates in a closed-loop as a voltage stabilization source.

In order to enhance the flexibility and versatility of the system, PI adjustment is implemented by software.
(1) Standard digital PI algorithm

FIG. 2 shows a block diagram of a computer control system with a digital PI regulator.

Figure 2 Block diagram of a typical computer control system

The Z transfer function of the digital PI regulator is:

Where: Ki -- integral coefficient, Ki = KoT/Ti; T -- sampling period; Ti -- integral time constant; Kp -- proportional factor; U (Z) -- z transfer function of the control quantity output; E (z) -- z transfer function of the deviation.

Expanding formula (1), the following positional algorithm can be obtained:

Where Uo -- initial value; Uk -- control quantity obtained at the kth sampling point; Ek -- deviation obtained at the kth sampling point; Ej -- deviation value of the jth sampling point; k -- kth sampling point.

Arranged into a recursive formula form:

According to the above recursive formula, the PI regulator can be easily implemented by software.

(2) Improved digital PI algorithm
The standard PI algorithm generally cannot meet the requirements of the site. For example, when starting, stopping or significantly changing the set value, the system deviation changes dramatically in a short period of time, which is easy to cause a large integral accumulation ∑ (Ek), causing the control output to change sharply, the system overshoot is serious, and the dynamic performance deteriorates.

In order to prevent this phenomenon from happening, the integral separation method, the over-limit weakening integral method and the effective deviation method are often used to improve the standard PI algorithm, which is more common in the design of the servo system.

Since this system belongs to a constant value control system, it requires soft start and soft stop functions, and the above improved algorithm can no longer meet the requirements. For this reason, a new constant deviation algorithm is adopted.

The constant deviation method is similar to the effective deviation method. The effective deviation method is also called the inverse algorithm. That is, when the control quantity Uk exceeds the limit, Uk takes the boundary value Umax or Umin. The deviation value Ek'' is calculated from the boundary value to replace the original deviation value Ek. However, in the constant deviation method, the attenuated Ek'' is used to replace Ek. When the system responds to the step, it actually works in an over-damped state, thereby reducing the impact on the main circuit components during starting and stopping. Figure 3 shows the step input response curves of the two algorithms. Among them, curve (a) is the standard PI algorithm response curve, and curve (b) is the constant deviation method response curve.

Figure 3 System step response curve

(3) PI parameter setting
① Sampling period T
Since the output filter network of the main circuit determines the maximum cutoff frequency f of the system output ripple, the upper limit of the sampling frequency f1 = 2f can be determined according to Shannon's theorem. In engineering, f1 = 10f is generally taken.

Since the main circuit parameters are known, it can be obtained that:

its lower limit T2 is determined by the execution time of the 8098 software. If a 12M crystal oscillator is used, the average execution time of each statement is 2us, and the program requires about 500 statements to run, then T2 = 1ms. Therefore:
1ms≤ T≤6ms .

Finally, the size of 1 can be selected through on-site debugging.

② Proportional factor Kp and integral time constant Ti The critical proportionality method is often used in engineering to adjust the constants Kp and Ti. That is, under closed-loop conditions, temporarily remove the integral effect and gradually increase the proportional gain until the closed-loop system reaches a critical stable state and continuous oscillation occurs. Record the critical gain Ku and oscillation period Tu at this time, and obtain the approximate values ​​of Kp= and Ti by looking up the table, and then correct them by debugging the whole machine.

The results of the voltage stabilization closed loop experiment of this system are: Ku = 6, Tu = 12.0ms.

From the table, we can get: Kp = 0.45 x Ku = 2.7, Ti = Tu ÷ 1.2 = 10ms.

Through the whole machine debugging, we can get the values ​​of parameters A and B in the recursive formula: A = 3, B = 2.

(4) Algorithm
In order to simplify the program design, the unsigned number algorithm is used when programming the 8098 software.

3) Steady current mode (I)
Steady current mode is also called strong charging mode. The system operates as a current source closed loop. The principle is the same as above, but the difference is that the integral time constant is smaller and the adjustment speed is faster.

The experimental data are as follows: Tu = 6.0ms, Ti = 5ms, A = 4, B = 2.

4) Automatic conversion of voltage and current stabilization (V/I)
When the battery is low, the system works in strong charging mode and the battery voltage gradually increases. When it exceeds the set value, it changes to floating charging mode.

That is, the system can automatically select the charging mode according to the load conditions. Figure 4 shows a typical two-stage charging curve.

Figure 4 Battery two-stage charging curve

5) Working mode memory (M1.M2)
The system can store two sets of commonly used data, namely working mode, voltage and current given values, voltage and current conversion values, and overvoltage and overcurrent values.

The current working record can also be protected when power is off.

6) Soft start and soft stop mode (SS)
Press this key and the system works in soft stop mode. The trigger board pulls the control pulse phase to the maximum and then blocks the pulse output. The key pops up, which is the soft start mode. The control output slowly rises from the minimum to the given value.

7) Battery detection mode (TEST)
The system can cyclically detect the battery voltage, which is displayed by the digital tube and can also be printed out through the serial port, and there is an alarm prompt.

2 System workflow diagram

After the system is powered on, the peripheral interface of the microcontroller is first checked to ensure that the human-computer dialogue channel is unobstructed. Then a power-on signal is issued to reset the LED, given register, etc., and then a square wave signal with a frequency of 2Hz is generated as the given input of the system.

To enter the above 7 working modes, the system has three ways.

Way I is the power-off restart mode. That is, after the system loses power, it can record the current working mode and directly return to the original state after the power is on.

This path is the entry channel when the system is working normally.

Paths II and III are generally used for debugging. Path II is to enter the corresponding working mode through M and M, while the user needs to use Path III when setting the given value, conversion value and protection value.

After the system enters the corresponding working mode, it can accept user commands through interruption mode and change the current working state (interrupt program flowchart omitted). Figure 5 shows the system workflow diagram.

Figure 5 System workflow diagram

3 Introduction to the general trigger board

The thyristor general trigger board is a thyristor trigger system designed based on the logical relationship between the three-phase synchronous signals locked by the voltage-controlled oscillator (VCO) using a 40-core CMOS large-scale integrated circuit as the core and using the phase-locked loop control technology (PLL). Given a 0~5V DC control signal, it can generate a three-phase, six-phase or twelve-phase strong trigger pulse with a phase shift range of 0°~180°. Due to the use of the above new technology, many shortcomings of similar products in the KC and KJ series are overcome, making the control pulse output by the trigger board highly symmetrical and balanced. In addition, the anti-interference ability and various additional functions of the board also greatly enhance the practicality of the board, so it has a very high performance-price ratio and is suitable for various high-power thyristor control circuits such as rectification, inversion, and AC side primary control.

The trigger board does not require a synchronous transformer, has the ability to automatically measure and control the phase sequence, and has a phase loss protection function and a pulse prohibition interface; dual 30° and 120° wide high-frequency modulation trigger pulses can be provided through the DIP switch. Experiments have shown that this pulse can directly drive thyristors above 1000A, and is an ideal product for thyristor triggering systems in field applications.

4 Conclusion

The feature of this system is the flexible application of the software PI regulator. On the one hand, a comprehensive comparison can be made on various PI algorithms, and some new algorithms can be verified. On the other hand, based on the versatility of the system, different control modes and parameters can be given for different controlled objects. For example, batteries with 100A and 500A require different PI adjustment parameters; even if the capacity is the same but the manufacturer is different, the required parameters are inconsistent, which is determined by the chemical reaction speed inside the battery.

Another feature of the system is the use of a universal trigger board. It is between the computer PI regulator and the thyristor main circuit. As a good buffer interface, it can ensure the safety and reliability of the system when the regulation is out of control. On the other hand, the WATCHDOG of the 8098 microcontroller also enhances the anti-interference ability of the system, making the system very suitable for field control.