A Brief Discussion on the Software Reset Method of Single Chip Microcomputer

Some single-chip microcomputers (such as 8098) have special reset instructions. Although some enhanced MCS-51 system single-chip microcomputers do not have reset instructions, they have integrated WATCHDOG circuits in the chip, so anti-interference is not a problem. Since the popular MCS-51 series single-chip microcomputers (such as 8031 ​​and 8032) have no reset instructions and do not have hardware WATCHDOG circuits, if there is no external hardware WATCHDOG circuit, software anti-interference technology must be used. Commonly used software anti-interference technologies include: software traps, instruction redundancy, software WATCHDOG, etc. Their role is to detect when the system is interfered with in time, and then use software methods to reset the system. The so-called software reset is to use a series of instructions to imitate the reset operation. This is the software reset technology unique to the MCS-51 series single-chip microcomputers.

Now let's use a simple experiment to illustrate. The experimental circuit is shown in the attached figure. The light-emitting diode LED0 connected to the simulation socket P1.0 is used to indicate the working status of the main program, the light-emitting diode LED1 connected to P1.1 is used to indicate the working status of the low-level interrupt subroutine, the light-emitting diode LED2 connected to P1.2 is used to indicate the working status of the high-level interrupt subroutine, and the button connected to the P3.2 port is used to set up an interference flag. After the program detects the interference flag, it deliberately enters an infinite loop or falls into a trap to simulate the interference situation, thereby verifying the actual effects of various reset methods. The test initialization program is as follows:

ORG ００００Ｈ

STAT: LJMP MAIN; reset entry address

LJMP PX0; button interrupt vector (low-level interrupt)

ORG ０００ＢＨ

LJMP ＰＴ０ ; t0 interrupt vector (low-level interrupt)

ORG ００１ＢＨ

LJMP PT1; T1 interrupt vector (high-level interrupt)

ORG ００３０Ｈ

MAIN：

CLR EA

MOV SP, #7

MOV P1，#0FFH

MOV P3，#0FFH

ＭＯＶ ＴＭＯＤ，#11Ｈ

CLR ００Ｈ ；Interference flag initialization

SET Ｂ ＥＴ０

SETB ET1

SET EB OX0

SET PB PT 1

SET TB TR ０

SET TB TR 1

SETBB EA

LOOP: CPL P1.0; main program LED flashes

ＭＯＶＲ６，#８Ｈ

MOV R7, #0

ＴＴ１：

DJNZ R7, TT1

DJNZ R6, TT1

SJMP LOOP

PX0:

SET Ｂ ００Ｈ ; Set up interference flag to simulate interference

ＰＴ０： ＣＰＬ Ｐ１.１ ；Low-level interrupt program LED LED1 flashes

RETI

PT1: CPL P1.2; Advanced interrupt program LED LED2 flashes

RETI

EN

The experimental steps are as follows:

1. Start the execution according to the above program, and all three LEDs should flash (otherwise, the fault should be eliminated first), indicating that the main program and each interrupt subroutine are normal. Because the analog interference mark is not detected, it is not affected by the button.

2. Modify the main program as follows. When the button is pressed, the main program falls into an infinite loop.

LOOP: CPL P1.0

ＭＯＶＲ６，#８Ｈ

MOV R7, #0H

TT1: DJNZ R7, TT1

DJNZ R6, TT1

JNB ００Ｈ，LOＯＰ; Is there any interference?

STOP: LJMP STOP; fall into an infinite loop.

At this time, you can see that the main program stops working (LED0 stops flashing), while the two interrupt subroutines continue to run (LED1 and LED2 continue to flash).

3. Use timer T1 as software WATCHDOG and 30H unit as software WATCHDOG counter. Add a reset instruction of software WATCHDOG in the main program.

LOOP: CPL P1.0

MOV 30H, #0; reset software WATCHDODG counter

LOOP: CPL P1.0

ＭＯＶＲ６，#８Ｈ

MOV R7, #0H

TT1: DJNZ R7, TT1

DJNZ R6, TT1

JNB ００Ｈ，LOＯＰ; Is there any interference?

STOP: LJMP STOP; fall into an infinite loop.

The T1 interrupt subroutine is modified as follows:

PT1: CPL P1.2; Advanced interrupt program LED flashes

ＩＮＣ ３０Ｈ

MOV A, 30H

ADD A，#0FDH

JC ERRR; Has it reached 3 times?

RETI

ERR: LJMP STATUS; software WATCHDODG action

Before the button is pressed, the program runs normally (all three LEDs flash). After the button is pressed, the main program can quickly resume work, but the two interrupt subroutines are blocked and no longer work. The process is as follows: After the main program detects interference, it enters an infinite loop and cannot perform the operation of resetting the 30H unit. The T1 interrupt causes 30H to increase in value continuously. When the count reaches 3, the software WATCHDOG executes an action and executes an LJMP instruction to restart the program. The interference flag is cleared during the MAIN process (indicating that the interference has passed), so that the main program can quickly resume work. It stands to reason that the interrupts are also reset and opened during the MAIN process. Why can't the interrupts resume work? This is because the reset of the interrupt activation flag has been forgotten, because it has no clear bit address to be programmed, and directly turning to the 0000H address cannot complete the real reset. Software reset is a must-do after using software traps and software WATCHDOG. At this time, it is entirely possible that a program error occurs. In the interrupt subroutine, the interrupt activation flag is set, which will prevent the same-level interrupt response. Since software WATCHDOB is a high-level interrupt, it will block all interrupt responses. Therefore, it can be seen that clearing the interrupt activation flag is essential. Many authors of literature may have made mistakes because they did not realize this.

4. Among all the instructions, only RETI instruction can clear the interrupt activation flag. The error handling procedure ERR mainly completes this function, and other follow-up work is completed by the reset system. For this reason, we redesign the T1 interrupt subroutine as follows:

PT1: CPL P1.2; Advanced interrupt program LED flashes

INC 30H; software WATCHDOOG counter increments

MOV A, 30H

ＡＤＤ Ａ，＃０ＦＤ

JC ERRR; Has it reached 3 times?

RETI

ERR: CLR EA; disable interrupt

CLR A; Prepare reset address (0000H)

PUSH ACC

PUSH ACC

RETI ; Clear interrupt activation flag and reset

This program first turns off the interrupt so that the subsequent processing can proceed smoothly, and then replaces the LJMP instruction with the RETI instruction, thereby clearing the interrupt activation flag and completing the task of turning to 0000H. After the program is run again after this modification, the result is still not ideal: after pressing the button, sometimes only the main program and the high-level interrupt subroutine can quickly return to normal, while the low-level interrupt may still be turned off. If the interference is transferred to the low-level interrupt as follows, the low-level interrupt will definitely be turned off after pressing the button:

LOOP: CPL P1.0

ＭＯＶＲ６，#８Ｈ

MOV R7, #0H

TT1: DJNZ R7, TT1

DJNZ R6, TT1

SJMP LOOP

PT0: CPL P1.1

JB 00H，STOP

RETI

STOP: LJMP STOP; fall into an infinite loop.

After careful analysis, we may conclude that when the software WATCHDOB is nested in a low-level interrupt, only the high-level interrupt activation flag is cleared after reset, and the low-level interrupt flag is still set, so that the low-level interrupt is always disabled.

5. Modify the error handling as follows:

ERR: CLR EA; Correct software reset entry

MOV 66H, #0AAH; Rebuild power-on flag

ＭＯＶ ６７Ｈ，#５５Ｈ

MOV DPTR, #ERR1; prepare the first return address

PUSHDPL

PUSHD

RETI ; Clear the advanced interrupt activation flag

ＥＲＲ１： ＣＬＲＡ

PUSH ACC

PUSH ACC

RETI ; Clear low-level interrupt activation flag

At this time, RETI must be executed twice to reach 0000H to ensure that all interrupt activation flags are cleared and achieve the same effect as hardware reset. Similarly, the software trap must also be executed by the following three instructions

NOP

NOP

LJMP STAT

To:

NOP

NOP

LJMP ERRR

To achieve the goal.

When the main program is disturbed and captured by the software trap, the interrupt flag is not set. During the execution of ERR, the RETI instruction is equivalent to the RET instruction, which can also achieve the purpose of software reset. Interested readers can replace the dead loop with the software trap, replace LJMP ERRR with LJMP STAT and LJMP ERRR1 respectively, and then set the interference detection in the low-level interrupt and the main program respectively. The experimental results will inevitably prove that only LJMP ERRR can achieve software reset without any mistakes, so that the system can get rid of interference and return to normal. In the software reset process of the MCS-51 microcontroller, the interrupt return instruction RETI must be executed twice in succession.