Design of single chip computer system based on fault reproduction

Experimental method of electromagnetic

pulse irradiation effect The experimental method of electromagnetic pulse irradiation effect on electronic system is to place the electronic system under test in the electromagnetic pulse radiation field, receive the electromagnetic pulse irradiation, and study the interference and damage of the system under test under the electromagnetic pulse irradiation. The

experimental configuration is shown in Figure 1. It mainly consists of a gigahertz transverse electromagnetic wave transmission chamber (GTEM Cell), a Marx generator, a control console and a test system. The Marx generator is used to generate high voltage and cooperate with the GTEM chamber to generate a uniform electromagnetic field in the GTEM chamber. The control console is mainly composed of an oscilloscope, an optical receiver and a Marx control panel . The optical receiver and the electric field sensor form an analog optical fiber field measurement system, which is mainly used to convert the radiated electromagnetic field into a voltage signal; the oscilloscope is used to display the electric field waveform; the Marx control panel is used to control the charge and discharge operation of the Marx generator and the adjustment of the steepening gap.

Fault reproduction Principle

Concept of fault reproduction

Computer Under the action of electromagnetic pulse, the system may produce hardware damage, increased data acquisition error, memory data change, program jump, restart and freeze. These fault phenomena are the collection of fault phenomena caused by a large number of different computers under interference in different environments and at different times. If you take out any computer for experiment, only a few fault phenomena can be produced. Because these computers do not have automatic detection functions, some faults cannot be observed even if they occur.

If the fault phenomenon is not fully observed, it is impossible to find out the law and cause of the fault, let alone conduct research on protection technology. Therefore, it is necessary to design a computer system specifically for electromagnetic pulse effect experiments, which has the following functions:

· Automatically detect and display faults that occur in the system itself;

· Faults are most likely to occur during interference;

· The most types of faults occur;

· Fault reproduction function.

Fault reproduction refers to actively taking certain technical means to make the fault recur. You can see any kind of fault you want to see, and you can make it appear as many times as you want. This is completely different from the design idea of ​​taking effective measures to resist interference in general circuits. Conditions

for fault reproduction Fault reproduction is not to simulate

the fault with computer software , but to reproduce the fault in reality. In order to reproduce the fault, in addition to the amplitude of the radiation field being strong enough, the system under test must also have the necessary hardware circuit and software environment. The software environment refers to the program that controls the working of the functional circuit when the interference occurs, that is, time alignment. For example, if you want to examine the impact of electromagnetic pulses on the conversion accuracy of the A/D conversion circuit, you must first have an ADC , and secondly, ensure that the ADC is working during irradiation. Technical means to achieve fault reproduction The requirements for radiation field intensity and hardware circuits are relatively easy to achieve, but the difficulty lies in ensuring time alignment. Of course, the reproduction of some faults has very loose requirements for time alignment, such as freezing and restarting. These two fault phenomena may occur when the computer runs almost any program. The solution to this difficulty is to adopt program modularization and loop waiting technology. Program modularization makes each fault (effect) correspond to a program module. If you want to see which kind of fault, run the corresponding program module; if you want the fault to occur multiple times, repeat the effect experiment. The loop waiting technology is to let the computer always run a certain section or a certain sentence of the program, which can greatly increase the probability of successful interference and make the fault most likely to occur. System composition and working principle Hardware composition This system uses 51 series single-chip microcomputers . In order to facilitate the study of the effect of the program memory, the 8031 ​​without EPROM inside is selected as the central processor , and the program is solidified in the external program memory. Here, the convenient E2PROM (2864 or 28C64) is used. Since 8031 ​​contains CTC and SIO, no external CTC and SIO are required. In order to enable the system to reproduce as many fault phenomena as possible, the peripheral chips used include: external data memory (6264) and ADC (AD0809). In addition, a 4-bit digital tube is added for information display, and the display data is stored by 4 latches (74LS373). The above hardware circuit not only completes certain functions but also is the test object. The system composition is shown in Figure 2. [page]

Program execution flow

The system software consists of 8 program modules: a program module for instructing the microcontroller to restart; a program module for checking the operation of CTC; a program module for checking the serial communication function; a program module for judging whether the content of the external RAM has changed and whether there is an error in reading and writing; a program module for judging whether the content of the internal RAM has changed; a program module for checking whether the conversion error of the A/D conversion circuit has increased; a program module for judging whether the external interrupt foot is triggered by mistake; and a program module for displaying whether the content of the E2PROM has been rewritten.

The working process of the system is the running process of the above 8 program modules, which are executed sequentially under the control of the execution switch K. Figure 3 shows the flow chart. Almost every effect experiment corresponds to a program module. Since the hardware damage failure has little to do with the software operation and the failure phenomenon is obvious, there is no need to set up a special program module. The re-shoulder effect experiment can work in any program module except the program module for instructing the microcontroller to restart. The deadlock effect experiment can work in any program module. Since the failure phenomenon is obvious, there is no need to detect the program. The specific implementation of fault reproduction and detection Different fault phenomena have different reproduction and detection methods. Due to limited space, only three methods for fault reproduction and detection are given.

External RAM Effect

This part of the experiment includes three parts: first, check whether the external RAM content is rewritten when no read or write operation is performed; second, check whether the read operation is wrong; third, check whether the write operation is wrong.

In the first part of the experiment, the RAM content is rewritten due to interference with the RAM chip, and only a detection program needs to be compiled. First, write the same data ("AA") to the 0000H~1FFFH unit of the RAM, and then wait for the execution switch K to be pressed, and perform the impact experiment during the waiting period. After the impact is completed, read the RAM content and determine whether it has changed.

The second and third parts of the experiment check whether the RAM read and write operations are wrong due to interference. Let the interference pulse with a duration of only microseconds interfere with the read and write instructions with an execution time of only a few microseconds. The probability of this event occurring is almost zero. For the case where the interference source can work in a repetitive working mode, it can be made to work in a repetitive working mode, which is undoubtedly a good idea. However, since the repetition frequency of the repetitive working mode cannot be made very high, the maximum can only reach about 1kHz, so its effect is not obvious. The most effective method is to make the program repeatedly execute a read or write instruction. Although there are several instructions between the two reads or writes to determine whether the read or written data is correct, the time interval between the two reads or writes is only in the order of tens of microseconds. This is equivalent to letting the read and write instructions wait for electromagnetic pulses to interfere, which greatly increases the probability of interference.

At the beginning of the compilation of the second and third parts of the program, in order to make it more representative, all units of the RAM are read or written, that is, the 0000H~1FFFH units of the RAM are cleared to 0 first, and then the program is made to read these units in a loop, or to write data "AA" to these units in a loop, and check in real time whether the read or written data is correct. In the experiment, it was found that the number of errors in the second and third parts of the experiment was more than the number of errors in the first part of the experiment. Despite the adoption of the above-mentioned loop waiting technology, a certain instruction

The possibility of interference is still very small. After many experiments, the answer was not clear. Later, when the reading experiment showed an error message, the contents of each RAM unit were checked. It was found that each time there was an error in the contents of a part of the RAM unit, and the error in the read operation could not cause the RAM content to change. Therefore, it was not or not all errors in the read and write operations, but because the RAM content was rewritten and misjudged as a read or write error. The solution to this problem is to make the read and write operations only for a certain fixed RAM unit. Since the probability of a unit being rewritten is 1/2 of the probability of all units being rewritten, this greatly reduces the probability of false alarms. [page]

Serial port SIO effect

Serial port SIO effect experiment mainly checks whether the serial port communication is wrong. To observe this fault phenomenon, the microcontroller must run the serial port communication program. Since the 51-type microcontroller has only one serial port, at least two serial ports are required to make it communicate, which requires at least two sets of microcontroller systems, which will complicate the equipment and experiments. After carefully studying the working principle of the serial port, I finally found a solution to simulate serial port communication using only one microcontroller : short-circuit the TXD and RXD of the CPU , and directly send the data sent by the TXD end to the RXD for reception, so that the microcontroller works in a self-transmitting and self-receiving state, and judge whether the communication is normal by checking whether the received and sent data are equal. Of course, it is still necessary to use the loop waiting technology to make the communication cycle. When it runs normally, a pulse signal is generated at the P1.1 port to light up the red LED. If the communication is abnormal, the digital tube will display an error message and turn off the LED.

Timer CTC effect

In order to reproduce the CTC working error, an instruction to allow CTC interrupt can be added to the main program, so that when the program is running, CTC is always working and waiting for electromagnetic pulses to interfere. The timer uses the CPU internal timer 0, and the working mode is mode 1. Write the interrupt subroutine of CTC0, cooperate with the software counter R0, generate a square wave signal at the P1.1 port, and drive the LED to flash. The main program waits for K to press the instruction, and conducts interference experiments during the waiting period. If the LED flashes abnormally, it means that the CTC is not working properly. The following is the interrupt subroutine of timer 0:

Experimental results

Before designing the single-chip system, a single-chip minimum application system was used for effect experiments, and only the crash phenomenon was observed. After using the system for effect experiments, various fault phenomena were observed, such as hardware damage, increased A/D conversion error, memory data change, program jump, crash, CTC work error, serial communication error, and program memory E2PROM content being rewritten. Through a large number of repeated experiments, the thresholds of various faults were measured and the causes of the faults were analyzed.

Figure 4 is the normal signal on the serial port RXD pin and the interference waveform when the communication is wrong, recorded by the oscilloscope. Figure 4 shows that there is a strong interference signal on RXD, and the low level is widened by 3 to 4 times. According to the experimental data and the working principle of the serial port, there are two reasons for the communication error: 1. Interference causes the SIO circuit to work incorrectly, such as the change of the content of the serial port control register SCON, the change of the content of the send or receive SBUF, etc., which may cause the received data to be inconsistent with the sent data, thus causing communication errors; 2. The interference signal on the RXD line causes confusion in the serial data, thus causing errors in the received data.