Design of full-turnoff detection circuit for thyristor rectifier

This paper proposes and introduces a scheme for Ohm logic non-circulating current detection - thyristor rectifier full-off detection, and compares and analyzes it with software detection and current transformer detection, and finally draws the conclusion that the thyristor full-off detection scheme is accurate and feasible. The full-off output signal is used in combination with the above two signals, thus accurately and reliably realizing Ohm's logic non-circulating current control.

1 Introduction

The Chinese Tokamak Experimental Reactor No. 2A (HL-2A) is China's first large-scale controlled nuclear fusion research device with a divertor configuration. Its main engine is a modification of the main components of the German ASDEX device. The power supply system and other supporting systems required for its magnetic field coils are completely developed by our institute.

The role of the ohmic coil (OH) in HL-2A is to break down the gas, establish, maintain and heat the plasma current, so the ohmic power supply that powers it plays a very important role in the device experiment. The ohmic power supply is shown in Figure 1, with two positive and two negative groups, a total of four groups of power supplies.

Figure 1 Schematic diagram of Ohm power supply

Among them, No. 1 and No. 3 rectifier cabinets are positive groups, No. 2 and No. 4 rectifier cabinets are negative groups, the output voltage of the positive group is 1600V, and the negative group is 800V. The output current of both groups is 30kA.

As the experiment progresses, the experimental requirements and parameters continue to increase, which requires the realization of non-circulating operation of the positive and negative groups of the ohmic power supply. The logical non-circulating operation of the ohmic power supply can be divided into the following stages. The positive group rectification stage is the magnetization of the ohmic coil. When the discharge begins, the positive group rectifier quickly enters the inverter stage, breaks down the gas, and maintains the plasma current rising. After the positive group current passes through zero, the positive group is blocked, and then the negative group is put into operation in the rectification state, continuing to push the plasma current up and maintain a flat top. After the flat top ends, the negative group controls the plasma current to decrease in an inverter manner, and blocks the negative group after the current passes through zero, completing a discharge. For the experiment, the most critical technology to achieve logical non-circulating and ensure the safety of the device is the zero-crossing detection of the ohmic current.

In order to detect the zero-crossing of the ohm and reliably realize the logic non-circulating current control, we developed a thyristor full-turnoff detection circuit board in comparison with the actual situation.

2 Comparison of several shutdown detection methods

To achieve accurate and stable operation without logic circulation, the key is how to accurately judge the full shutdown moment of the positive group rectifier. Because if the shutdown is judged to be premature, but in fact the positive group rectifier has not been fully shut down, then the negative group rectifier will be turned on according to the set logic program, and a large circulation will be formed between the positive group rectifier and the negative group rectifier, which will pose a serious threat to the safety of the power supply equipment; if the shutdown is judged to be delayed, the dead time of switching between the positive group rectifier and the negative group rectifier is too long, which will affect the discharge of the device and even discharge failure.

Full shutdown detection has an important impact on the safety of the power system and the stability of the device discharge. Usually, the DC output current of the rectifier is detected to determine whether it is shut down. It is usually called zero-crossing detection. The following is an analysis and comparison of several detection methods.

2.1 Software Zero Crossing Detection Method

The signal of the DC sensor is sent to the computer through the acquisition board. A comparison value is set in advance and compared through the program. When the current value is detected to be less than this value, it is considered to be zero crossing. Due to the limitations of the measurement accuracy of the large current sensor and serious on-site interference, misjudgment is easy to occur. In addition, the zero crossing detection program is compiled together with the complex device discharge control program, which only detects the first zero crossing. When the current fluctuates, it cannot determine whether it has crossed zero again, as shown in Figure 2.

Figure 2 Schematic diagram of zero-crossing judgment when current fluctuates during software detection

Among them, Utk2-OH is the zero-crossing signal detected by the software, and I-OH is the ohmic power supply current. Because the measurement direction of the transformer is reversed, the ohmic current is displayed as negative (the same below). When the zero-crossing signal is reversed, there is actually a certain current, and the rectifier is not really turned off, and is in a continuous current state. If the discharge is normal, the software delays for an appropriate time, and the negative group rectifier can be controlled to be turned on when the positive group rectifier is really turned off. The length of its conversion dead time depends on the measurement accuracy of the sensor and the program speed. However, if the discharge is abnormal, just when the zero-crossing signal is reversed and the software is delayed, the plasma current breaks, and its energy is coupled to the ohmic primary side, the positive group current increases, and the continuous current time increases. The software only detects a zero-crossing point. If the negative group rectifier is turned on after the software delay (fixed value) ends, the positive group rectifier is still continuous at this time, and a circulating current will be generated.

The signal of the DC sensor is sent to the computer through the acquisition board, and a comparison value is set in advance. When the current value is detected to be less than this value, it is considered to be zero-crossing. Due to the limitations of the measurement accuracy of the large current sensor and the serious interference on site, it is easy to cause misjudgment. In addition, the zero-crossing detection program is compiled together with the complex device discharge control program, which only detects the first zero-crossing. When the current fluctuates, it cannot judge the zero-crossing again, as shown in Figure 2. Among them, Utk2-OH is the software detection zero-crossing signal, and I-OH is the ohm power supply current. Because the measurement direction of the transformer is reversed, the ohm current is displayed as negative (the same below).

When the zero-crossing signal is reversed, there is actually a certain current, and the rectifier is not really turned off, and is in a continuous current state. If the discharge is normal, the software delay can be used to control the negative group rectifier to be turned on when the positive group rectifier is really turned off. The length of the conversion dead time depends on the measurement accuracy of the sensor and the program speed. However, if the discharge is abnormal, the plasma current breaks when the zero-crossing signal is reversed and the software delays, and its energy is coupled to the ohmic primary side, the positive group current increases, and the continuous current time increases. The software only detects a zero-crossing point. If the negative group rectifier is turned on after the software delay (fixed value) ends, the positive group rectifier is still continuous at this time, and a circulating current will be generated.

2.2 Hardware Zero Crossing Detection Method

This method is implemented by hardware, using Hall elements to measure the total current signal of the positive group of the ohmic rectifier as the input signal. It can also accurately detect the shutdown time during normal operation. However, when the power supply has some abnormal conditions, such as inverter failure or early blocking when the current is not zero, and the current happens to be near the zero point, the detection method will display the zero crossing condition many times. When this method detects the first zero crossing signal, the rectifier is not necessarily completely turned off. It is just because the voltage of two phases of the power supply is added to the OH coil through a pair of thyristors. On the basis of a DC excitation current, it is continuously excited and demagnetized. The energy of the OH coil is consumed through the loop resistance until the DC current of the OH coil decays to zero and the rectifier is truly turned off. As shown in Figure 3.

Figure 3 Waveform when hardware zero-crossing detection is blocked in advance

Where V-OH is the ohmic power supply voltage, and Utk1-OH is the hardware detection zero-crossing signal.

2.3 Tube voltage drop zero-crossing detection method

In view of the defects of the above two methods, a new detection method is now used to determine whether it is completely turned off by detecting the tube voltage drop at both ends of the thyristor. This is a direct and effective method. If all tubes are turned off, there is no current in the load, and the tube voltage drop is several hundred volts. If there are still tubes turned on, there is still current in the load, and the tube voltage drop is several volts. By detecting the tube voltage drop to determine whether it is fully turned off, it can be determined whether there is load current. In view of this feature, a thyristor full-off detection circuit is designed.

In order to reliably judge whether the thyristor is turned off, the moment of judgment is taken as 15 degrees of the phase voltage = 700×0.25=175V, that is, when the voltage across the tube is higher than 175V, the tube is judged to be in the off state; when the voltage across the tube is less than 175V, the tube is judged to be in the on state, as shown in Figure 4.

Figure 4 Analysis of the full shutdown detection principle

(a) are the three-phase voltages of A, B, and C, (b), (c), and (d) are the status signals of the three thyristors in the common cathode group, which include forward voltage and reverse voltage, with a 30-degree low level in the middle. The three signals b, c, and d are ANDed to obtain a pulse series signal e, indicating that all the tubes are turned off. This signal is shaped into a level signal by a monostable to indicate the turn-off signal.

3 Thyristor full turn-off detection circuit

FIG5 is a schematic diagram of a circuit in which all thyristors are cut off.

Figure 5 Schematic diagram of full shutdown detection circuit

The two ends of each thyristor are divided into two voltages, and two optocouplers are connected in the positive and negative directions. When the voltage at both ends of the tube is positive, one of the optocouplers is turned on. If the voltage at both ends of the tube is negative, the other optocoupler is turned on. When the optocoupler is turned on, the output signal is 1, otherwise it is 0. These two signals are processed by "OR" to obtain a signal. If the signal after "OR" (that is, the signal of a thyristor) is still 1, it can be judged that this thyristor is in the off state, otherwise it is in the on state.

When the signal after the "AND" of the six thyristor signals is 1, it is judged that the thyristors are all turned off at this time. By adjusting the resistance and capacitance of the monostable trigger RC and C, the output high level is maintained for 60 , which can ensure accurate reflection of the pulse state. When all tubes are turned off, it is a high level. When there is no pulse after 60 , there is no high level output. In other words, as long as there is a thyristor turned on, the output state signal is 0. Therefore, when the plasma breaks, the positive group freewheeling time increases, and there will be a low level output.

In this circuit, the optocoupler works in the linear region, at least in the lower part of the sine wave, that is, when the voltage at the bottom of the sine wave is very low, the optocoupler can also be turned on, which truly reflects the voltage at both ends of the thyristor. The part with a higher sine wave voltage can make the optocoupler turn on more easily. Now take the output of an optocoupler as an example to introduce the working principle of the board.

First assume that an optocoupler is turned on, and its output signal is amplified by a transistor, and then divided into three paths and passed through three comparators, one of which is a common-phase comparator and two are inverting comparators. The common-phase comparator is used to obtain the leading edge of the pulse. When the sine wave voltage reaches the comparison level, the comparator flips to produce a rising edge, and its output is connected to the clock end of a D flip-flop, and the Q end starts to be "1"; when the sine wave voltage reaches the inverting comparator comparison level; a rising edge is also generated, and this rising edge is sent to the reset end of the D flip-flop after a monostable to make the Q end become "0", which means that the trailing edge of the pulse is determined. After this link, the sine wave has been shaped into a square wave.

The function of another inverting comparator is to ensure that a signal that can reset the D flip-flop can be found. This is because if the trailing edge comparison level is higher than the leading edge comparison level, it is possible that the thyristor will be turned on after the leading edge is detected, and the voltage across the thyristor is only a few volts, so the trailing edge cannot be found, and the output of the D flip-flop will always be "1", which incorrectly reflects the state of the tube. Therefore, the comparison level of this comparator must be set below the comparison level of the leading edge comparator, and its output is "OR" with the output of another inverting comparator and connected to the input of the monostable; the final output signals of the corresponding forward and reverse optocouplers are "OR" and then "AND" with the other six signals to obtain a full shutdown signal.

4 Debugging Results

First, use a standard sinusoidal wave signal source to adjust the operating point of the high isolation voltage optocoupler, try to make its delay time short within the voltage range to be detected, and make the performance of the optocoupler basically consistent, and then use the comparator to adjust the consistency of each output.

Since the ohmic power supply works in pulse operation mode, on-site debugging is difficult. Through the data recorded in each discharge, the signal of each channel is optimized and adjusted to make it work normally and accurately detect the zero-crossing moment of the positive group rectifier current, as shown in Figure 6, where Utk3-OH is the zero-crossing signal detected when the thyristor is fully turned off.

Figure 6 Schematic diagram of full-shutdown zero-crossing detection

The thyristor full-off detection circuit can not only truly reflect the full-off moment of the tube in theory, but also has been well applied in experimental operation. In the logic non-circulating current experiment under the condition of OH power supply dummy load, the full-off signal and other zero-crossing signals (software zero-crossing detection signal, hardware zero-crossing detection signal and plasma existence signal) are combined as the criterion for whether the rectifier is turned off, which has a good effect. In the experiment, after the full-off signal appears, there is a delay of 6ms. If the above three signals exist at the same time, the conditions for the negative group to open are met, which provides a guarantee for the safety of the experimental device and greatly improves the experimental parameters.

References

[1] Yao Lieying, Xuan Weimin, Li Huajua, et al, Design and Development of the Power Supply for HL-2A Tokamak, 23rd Symposium on Fusion Engineering20-24 Septemher 2004-fondazione Chni, Venice, Italy.