Load chip fault monitoring circuit for power management chip
By sending the duty cycle of the rectangular wave signal on the ERR pin to indicate the fault level, combined with homologous clock and clock frequency division technology, the problem of difficulty in identifying fault level in the existing technology is solved, and the processor chip failure is achieved Multi-dimensional monitoring and refined operations improve the timeliness and accuracy of fault identification and processing.
Patent Information
- Application Number
- CN202411076204.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-07
- Publication Date
- 2024-11-15
AI Technical Summary
When monitoring and handling load chip failures, existing power management chips are difficult to quickly and accurately identify fault levels, which limits the safe operation and fault handling capabilities of the system, especially in complex application scenarios such as automotive autonomous driving and artificial intelligence.
By sending the duty cycle of the rectangular wave signal on the ERR pin to indicate the fault level, combined with homologous clock and clock frequency division technology, the power management chip allows the power management chip to identify multiple fault level information sent by the processor chip, thereby taking different follow-ups operate.
Multi-dimensional monitoring and refined operation of processor chip failures is realized, timeliness and accuracy of fault identification and processing is improved, information identification process is simplified, and fault tolerance is increased.
Smart Images

Figure CN118962401A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power management chips, and in particular to a load chip fault monitoring circuit for a power management chip. Background Art
[0002] In addition to supplying power to various load chips in electronic equipment systems, traditional power management chips (PMIC) can also monitor whether the load chips have overvoltage, undervoltage, short circuit, overcurrent and other faults through built-in voltage and current monitoring circuits.
[0003] As the complexity and functionality of load chips increase, power management chips are also facing new demands in monitoring and handling load chip failures, especially in application scenarios such as digital signal processor (DSP) chips, microcontroller chips (MCU) or any other type of processor-based chips in application fields such as autonomous driving and artificial intelligence. How PMIC can quickly and accurately identify faults in load chips and the type of faults, and make fast and safe processing, is also a technical problem that needs to be solved to ensure the safe operation of the entire system.
[0004] Therefore, a module called Safety Management Unit (SMU) is implemented in some processor chips. It is responsible for collecting alarm signals (Alarm) after hardware or software failures in other modules, and each Alarm can be configured separately to trigger internal operations and / or notify the outside. Of course, when notifying the outside, a fault signaling protocol (FSP) is used to send a notification to the outside through a pin. The content of the notification is whether there is a fault in the processor. Without loss of generality, this pin is called the ERR pin. According to this FSP, if the processor has no fault, the ERR pin should send a flip signal to the outside and each high level time t high和 Low level time t low are all within the prescribed range of FSP, that is, t min ≤t high ≤t max , and t min ≤t low ≤t max , where t min and t maxIt is the shortest and longest duration of the high and low levels of the fault-free flip signal specified by FSP. The most common way is that the processor sends a square wave flip signal with a frequency that meets the above conditions and a duty cycle of 50%. If a certain fault occurs in the processor and the fault is configured to be notified externally, the ERR pin should stop flipping, that is, keep the ERR pin at a high level or a low level. When the time of maintaining the high level or the low level is greater than tmax, that is, t high >t max or low >t max , it means that the processor is successfully notified of a fault; if the high or low level duration of the flip signal sent out by the processor through the ERR pin is too short, that is, t high <t min or low <t min When the processor sends a fault notification, it is also regarded as a fault notification sent by the processor. Accordingly, the PMIC should monitor and analyze the fault-free or fault-existing notification sent by the processor through the ERR pin in real time. Without loss of generality, this module is called the Error Monitoring Unit (EMU). If the processor is faulty, further corresponding processing can be performed.
[0005] In the prior art, the fault monitoring unit EMU of the PMIC can analyze in real time the signal indicating whether a fault has occurred, which is sent by the safety management unit SMU in the processor chip through the ERR pin according to a specific fault signal protocol FSP. However, the notification sent by the ERR pin in the prior art only contains two types of information, namely, fault and no fault, which means that there are only two subsequent corresponding operations of the PMIC, namely, maintaining normal operation and solving the processor fault in a unique way. This has constrained the possible multi-dimensional fault monitoring in future higher reliability applications and the more refined subsequent corresponding operations of the PMIC. Summary of the invention
[0006] In view of the shortcomings of the prior art, the present invention provides a load chip fault monitoring circuit for a power management chip, which overcomes the shortcomings of the prior art. Through an ERR pin, in addition to indicating whether a processor fails, it can also indicate the severity of the fault that occurs in the processor, thereby allowing the power management chip to take different subsequent operations.
[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions:
[0008] A load chip fault monitoring circuit for a power management chip comprises a fault detection unit EMU, the output CLK of the fault detection unit EMU REFPin and the input CLK of the security management unit SMU in the processor chip REF The pins are connected, and the input ERR pin of the fault detection unit EMU is connected to the output ERR pin of the safety management unit SMU;
[0009] The clock CLK used by the security management unit SMU to send the ERR pin waveform SMU CLK is the working clock of the fault detection unit EMU EMU or CLK EMU A clock after a certain frequency division;
[0010] Each complete rectangular wave period T sent WAVE为 Several CLK SMU The clock period T SMU ; That is, T WAVE =N RW ·T SMU , where N RW The CLK required for a rectangular wave SMU The number of clock cycles;
[0011] The safety management unit SMU is connected to the external control unit through the ERR pin, and is used to output a rectangular wave signal corresponding to the duty cycle; when the processor chip is notified by the safety management unit SMU through the ERR pin, the signal is T WAVE Conducted for reference period;
[0012] After the processor chip completes power-on reset and before the fault detection unit EMU officially starts the fault monitoring function, the safety management unit SMU sends a set of training waveforms to the fault detection unit EMU through the ERR pin to train the fault detection unit EMU to recognize the waveform of the reference period TWAVE;
[0013] When the processor chip has no fault or the fault has been resolved, the safety management unit SMU sends a rectangular wave signal with a duty cycle corresponding to the fault through the ERR pin;
[0014] When the highest level of fault occurs in the processor, the safety management unit SMU will maintain the current level of the ERR pin until the current fault is resolved and then restore to sending a rectangular wave signal with a duty cycle corresponding to the fault-free state;
[0015] When a lower level fault occurs in the processor, the safety management unit SMU will send a rectangular wave with a duty cycle to load the fault level through the ERR pin;
[0016] When the safety management unit SMU is sending a fault notification and the fault has not been resolved, a new fault occurs and the level of the new fault is compared with that of the currently occurring fault. When it is determined that the new fault level is higher, the safety management unit SMU immediately sends the waveform corresponding to the new fault level after it has sent the fault level waveform corresponding to the currently occurring fault. If it is determined that the new fault level is the same as the currently occurring fault level, or is lower than the currently occurring fault level, the safety management unit SMU continues to send the corresponding waveform in the original manner of the currently occurring fault.
[0017] Preferably, the reference period T WAVE =N DIV1 ·N DIV2 ·N RW ·T EMU =N T ·T EMU ;
[0018] Among them, N DIV1 Indicated as CLK REF The clock cycle is CLK EMU The multiple of the clock cycle, N DIV2 Indicated as CLK SMU The clock cycle is CLK REF A multiple of the clock period, that is,
[0019] N T =N DIV1 ·N DIV2 ·N RW ;
[0020] The processor chip connects N DIV2 and N RW or N DIV2 ·N RW The product of is written into the PMIC register. After the training is completed, the fault detection unit EMU will obtain the expression T WAVE =N T ·T EMU Medium T The value is compared with the value of the PMIC register. When they are consistent, the fault monitoring function is officially enabled.
[0021] Preferably, when the processor has a fault of the highest level, the safety management unit SMU needs to recover to send a rectangular wave signal with a duty cycle corresponding to the fault-free state; the current level is determined, and if it is a low level, the safety management unit SMU starts from the next reference period T WAVE The rising edge of starts to resume flipping; if it is currently high, the safety management unit SMU starts from the next reference cycle T WAVE The falling edge starts to resume flipping.
[0022] Preferably, when a lower level fault occurs in the processor, the safety management unit SMU will send a rectangular wave loaded with fault level information through the ERR pin, and then continue to send a filling rectangular wave during the fault until the current fault is resolved and then restore to sending a rectangular wave signal with a duty cycle corresponding to a fault-free state.
[0023] Preferably, the duty cycle of the rectangular wave sent by the safety management unit SMU through the ERR pin is written into the register of the PMIC by the processor chip through a traditional serial bus interface after power-on.
[0024] Preferably, it also includes a counting register cnt1 and a high level counting register cnt2. When the fault detection unit EMU has locked N T The value of the reference period T WAVE After the start and end boundaries, the count register cnt1 increases from 0 to N T -1 for cyclic increment counting, that is, every reference cycle T WAVE The corresponding count register cnt1 changes from 0 to N T -1; in a reference cycle T WAVE The fault detection unit EMU counts the time when the ERR signal is at a high level by the high level count register cnt2. EMU Count units.
[0025] The present invention provides a load chip fault monitoring circuit for a power management chip. It has the following beneficial effects: by loading information on the duty cycle of the waveform sent by the ERR pin, it is possible to implement the level configuration of various faults occurring in the processor chip, and give priority to notifying faults with high levels, so that the power management chip can monitor, identify and handle faults more timely. This provides the necessary premise for more dimensional fault monitoring that may occur and more refined subsequent corresponding operations of the power management chip. And a specific waveform is sent through the pin of the SMU, especially the duty cycle, rising edge, falling edge, number of pulses and other information of the rectangular wave within the reference period are defined to notify the power management chip of what level of fault has occurred. By adopting the setting of the same source clock scheme and clock division, the EMU of the power management chip can transform the recognition of the duty cycle of the rectangular wave loaded with information sent by the SMU of the processor chip into a simple positive integer comparison, which greatly simplifies the operation of information recognition, and also increases the fault tolerance during transmission, sampling and recognition by utilizing the interval. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] In order to more clearly illustrate the present invention or the technical solutions in the prior art, the drawings required for describing the prior art are briefly introduced below.
[0027] Figure 1 It is a structural schematic diagram of the present invention;
[0028] Figure 2 The security management unit SMU sends the WAVE The waveform diagram of the high level and low level;
[0029] Figure 3 This is the RW375 waveform that the safety management unit SMU sends out through the ERR pin when the processor has no fault or the fault has been resolved;
[0030] Figure 4 This is the waveform diagram sent by the safety management unit SMU when the highest level of fault occurs in the processor;
[0031] Figure 5 This is the waveform diagram sent by the safety management unit SMU when a lower level fault occurs in the processor;
[0032] Figure 6 A flowchart for sending waveforms when multiple faults occur in the processor;
[0033] Figure 7 This is a flow chart before the fault detection unit EMU starts the fault monitoring function;
[0034] Figure 8 Flowchart for starting the fault monitoring function of the fault detection unit EMU. DETAILED DESCRIPTION
[0035] In order to make the purpose, technical solutions and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the present invention.
[0036] Examples, such as Figure 1-8 As shown, the present invention discloses a load chip fault monitoring circuit for a power management chip, comprising a fault detection unit EMU, an output CLK of the fault detection unit EMU REF Pin and the input CLK of the security management unit SMU in the processor chip REF The pins are connected, the input ERR pin of the fault detection unit EMU is connected to the output ERR pin of the safety management unit SMU;
[0037] The safety management unit SMU is connected to the external control unit through the ERR pin to output a rectangular wave signal corresponding to the duty cycle; the safety management unit SMU in the processor chip can send out three types of notifications through the ERR pin, and each type of notification can be represented by a rectangular wave with a significantly different duty cycle or a continuous high level or low level. In this embodiment, in order to facilitate the EMU to sample the waveform sent by the SMU through the ERR pin and parse the duty cycle of the rectangular wave, the clock CLK used by the safety management unit SMU when sending the ERR pin waveform is SMU CLK is the working clock of the fault detection unit EMU EMU or CLK EMU The clock after several frequency divisions is the same source clock. This is achieved by dividing the same source clock as described below:
[0038] 1. Set the output CLK of the fault detection unit EMU REF The pin is set as the working clock CLK of the fault detection unit EMU EMU Or the clock after integer division, that is, CLK REF The clock cycle is CLK EMU N DIV1 times, N DIV1 Is a positive integer.
[0039] 2. The ERR pin in the safety management unit SMU is driven by a CLK SMU The serial output port of the clock register, clock CLK SMU CLK REF Or the clock after integer division, that is, CLK SMU The clock cycle is CLK REF N DIV2 times, N DIV2 Is a positive integer.
[0040] 3. The period T of each complete rectangular wave sent WAVE It must be several T SMU , T WAVE is the reference period, then T WAVE =N RW ·T SMU , where N RW The CLK required for a rectangular wave is SMU At the same time, for the convenience of EMU analysis, when the SMU sends a continuous high or low level through the ERR pin, it should also be based on T WAVE To be carried out in cycles.
[0041] Therefore, it can be concluded that T WAVE =N DIV1 ·NDIV2 ·N RW ·T EMU =N T ·T EMU ,in,
[0042] N T =N DIV1 ·N DIV2 ·N RW , so N T It should be large enough to identify the duty cycle of the rectangular wave as accurately as possible.
[0043] In this embodiment, for the convenience of description, one of the optional modes of the above-mentioned rectangular waves is defined as follows:
[0044] A rectangular wave with a reference period of TWAVE and a duty cycle of 12.5%, denoted as RW125,;
[0045] A rectangular wave with a reference period of TWAVE and a duty cycle of 37.5% is denoted as RW375;
[0046] The reference period of the rectangular wave with a duty cycle of 62.5% is TWAVE, denoted as RW625,;
[0047] A rectangular wave with a reference period of TWAVE and a duty cycle of 87.5% is denoted as RW875;
[0048] The reference period of continuous high level is the waveform of TWAVE, denoted as RH;
[0049] The reference period of the continuous low level is a waveform of TWAVE, denoted as RL.
[0050] As an example, using the above rectangular wave definition, if it is necessary to indicate that the processor has no fault and five fault levels, including one fault defined as the highest level fault ERR_H, and the remaining four lower fault levels are represented by N = 2 binary numbers, that is, defined as ERR_Lx, x represents one of the binary numbers 2'b00, 2'b01, 2'b10, 2'b11, and the larger the x, the higher the error level. That is, RW125 represents the ERR_Lx fault level binary number 0, RW375 represents no fault, RW625 represents the ERR_Lx fault is ongoing, and RW875 represents the ERR_Lx fault level binary number 1.
[0051] In addition, the duty cycle of the rectangular wave sent by the SMU of the processor chip can also be written into the corresponding register in the PMIC by the processor chip through the traditional serial bus interface after power-on. The minimum value of the duty cycle of the rectangular wave of ERR_Lx fault level binary number 0, no fault, ERR_Lx fault continuous, and ERR_Lx fault level binary number 1 is recorded by the register DUTYCYCLEi ,min and the maximum value of DUTYCYCLE i,max , where i = 0, 1, 2, 3 correspond to the above four rectangular waves. The duty cycle intervals of different rectangular waves cannot overlap and should be separated by enough.
[0052] When the identified waveform is a rectangular wave, it should be EMU Count the number of cycles of the ERR signal high level and divide it by NT to get the duty cycle of the rectangular wave, and then further compare it to which DUTYCYCLE it belongs to. i,max and DUTYCYCLE i,min The interval to which it belongs can be analyzed to determine whether the rectangular wave represents no fault, ERR_Lx fault continuation, ERR_Lx fault level binary number 0, or ERR_Lx fault level binary number 1.
[0053] Specifically, the expression N can be used in SMU T =N DIV1 ·N DIV2 ·N RW N RW 、N DIV1 and N DIV2 Set to 2 respectively m , 2 n and 2 p , then N T =2 m+n+p , further, take m+n+p=4,5,6,…, then NT=16,32,64,…; at the same time, the maximum and minimum values of the duty cycle DUTYCYCLE i,max and DUTYCYCLE i,min It is converted from a percentage to less than N T Integer, so only through simple positive integer comparison can we determine which interval the duty cycle of the rectangular wave belongs to and what information it represents.
[0054] As an example, take N T =16, and set the minimum and maximum values of each duty cycle interval to:
[0055] DUTYCYCLE i=0,min =1,DUTYCYCLE i=0,max =3,
[0056] DUTYCYCLE i=1,min =5,DUTYCYCLE i=1,max =7,
[0057] DUTYCYCLE i=2,min =9,DUTYCYCLE i=2,max =11,
[0058] DUTYCYCLE i=3,min =13,DUTYCYCLE i=3,max =15.
[0059] Because the duty cycle of the rectangular wave received by the EMU in the PMIC is an interval with the duty cycle of the rectangular wave corresponding to the information sent by the SMU as the middle value, for example:
[0060] When the SMU emits a rectangular wave RW125 corresponding to a fault, DUTYCYCLE=2 / 16=12.5%, and its pulse duration is 2T EMU cycle, and 2 is DUTYCYCLE i=0,min and DUTYCYCLE i=0,max That is to say, as long as the EMU can identify that the pulse duration of the received rectangular wave is within the interval [1,3], it can be successfully interpreted as no fault. This interval definition method gives the EMU a certain fault tolerance capability.
[0061] In addition, the duty cycle intervals corresponding to different information are non-overlapping. When the rectangular wave received by EMU does not belong to any of the above four intervals, or the time between two consecutive rising edges is not equal to T WAVE , it should be determined as an illegal waveform. This may be due to a fault in the SMU itself or interference in the transmission process. Appropriate actions should be taken. WAVE When all are high or low, the EMU can parse it as an ERR_H fault and perform corresponding fault handling operations.
[0062] After the processor chip completes power-on reset and before the fault detection unit EMU officially starts the fault monitoring function, the safety management unit SMU sends a set of training waveforms to the fault detection unit EMU through the ERR pin to train the fault detection unit EMU so that it can accurately identify the start and end boundaries of each cycle of the fault-free or various fault level waveforms to be sent, and count to obtain the reference cycle expression T WAV =N T ·T EMU N T value;
[0063] Specifically, Figure 2 The security management unit SMU sends a period of T WAVE The high level waveform (RH) and a period of T WAVE The low level waveform (RL) is used to inform the EMU that the duration of each high level or low level is the reference period T WAVE , and the rising edge or falling edge is the end of the previous reference cycle and the beginning of the current reference cycle. EMU can monitor the transition edge of the ERR signal and count these high and low levels respectively using the EMU clock CLK EMU The number of cycles counted, i.e. N T The value of N T The method of finding multiple averages, or counting several rounds N in a row T The value of N is consistent with T When N T Lock and T WAVE After the boundary is determined, the EMU must sample at least one rectangular wave indicating no fault before the fault monitoring function can be officially enabled.
[0064] Specifically, Figure 7 As shown, the following steps are included:
[0065] Step S71: EMU enters the reference cycle training phase, with CLK EMU Count T WAVE Time N T The counter is cleared;
[0066] Step S72: CLK EMU Sampling ERR;
[0067] Step S73: Determine whether ERR has a transition edge, if so, proceed to step S74, if not, return to step S72;
[0068] Step S74: N T The counter increases by 1;
[0069] Step S75: Determine whether ERR has a transition edge, if so, proceed to step S76, if not, return to step S74;
[0070] Step S76: Record this round N T The value of the counter and N T The counter is cleared;
[0071] Step S77: Determine N T Is the value of the counter consistent in four consecutive rounds? If so, proceed to step S78; if not, return to step S74;
[0072] Step S78: Lock N T And determine T WAVE The starting and ending boundaries of
[0073] Step S79: Determine whether the sampled ERR d2 is a fault-free rectangular wave within the reference period. If so, the EMU formally starts the fault monitoring function. If not, return to step S78.
[0074] In addition, since the security management unit SMU in the processor chip can send notifications through the ERR pin in accordance with T WAVE The reference cycle is carried out. DIV2 and N RW is selected in the security management unit SMU. Therefore, in this embodiment, the processor chip can DIV2 and N RW Two or N DIV2 ·N RW The processor chip writes the product of N into the corresponding register in PMIC through the traditional serial bus interface to inform the EMU, so that the EMU can obtain N by itself. T =N DIV1 ·N DIV2 ·N RW The result is compared with the N counted by the fault detection unit EMU during the reference cycle training phase. T Compare whether they are consistent. Only if the two are consistent can the PMIC be allowed to enter the normal working state, otherwise corresponding operations should be taken.
[0075] After that, it also includes the count register cnt1 and the high level count register cnt2. When the EMU has locked N T The value of the reference period T WAVE After the start and end boundaries, there is a counter cnt1 that will increase from 0 to N T -1 for cyclic increment counting, that is, the change of cnt1 is 0→1→2→3→…→N T -1→0→1→2→…, that is, every reference cycle T WAVE The corresponding change of cnt1 is 0→1→2→3→…→N T -1, in this reference cycle, EMU is controlled by the time when ERR is at a high level as T EMU Count in units. (Assume that the high-level counting register is cnt2) and analyze which waveform among RW125, RW375, RW625, RW875, RH and RL is obtained in this reference cycle.
[0076] Specifically, Figure 8As shown, taking NT = 16, the minimum and maximum values of each duty cycle interval are:
[0077] DUTYCYCLEi=0,min=1,DUTYCYCLEi=0,max=3,
[0078] DUTYCYCLEi=1,min=5,DUTYCYCLEi=1,max=7,
[0079] DUTYCYCLEi=2,min=9,DUTYCYCLEi=2,max=11,
[0080] DUTYCYCLEi=3, min=13, DUTYCYCLEi=3, max=15.
[0081] It specifically includes the following steps:
[0082] Step S1: The fault detection unit EMU starts the fault monitoring function;
[0083] Step S2: Count register cnt1 from 0 to N T -1 is used for circular increment, and the high level counting register cnt2 is from N in cnt1 T -1 to 0 is cleared;
[0084] Step S3: When the counting register cnt1 is equal to 0, determine whether the first level of the signal ERR is a low level; if so, proceed to step S4, if not, proceed to step S7;
[0085] Step S4: Determine whether the count register cnt1 is from 0 to N T -1, whether the level of ERR is low; if so, go to step S5, if not, go to step S6;
[0086] Step S5: determine that it is a low level waveform (RL), that is, ERR_H fault, and immediately trigger the corresponding fault operation of ERR_H, and clear the counting register cnt1 and the high level counting register cnt2, and return to step S2;
[0087] Step S6: determine that it is an illegal waveform, and immediately trigger the fault operation corresponding to the illegal waveform, and clear the counting register cnt1, and return to step S2;
[0088] Step S7: both the counting register cnt1 and the high level counting register cnt2 are increased by 1;
[0089] Step S8: Determine whether the count register cnt1 is accumulated to N T-1, if yes, go to step S23; if no, go to step S9;
[0090] Step S9: Determine whether the level of ERR is high; if so, return to step S7; if not, proceed to step S10;
[0091] Step S10: the counting register cnt1 is increased by 1;
[0092] Step S11: Determine whether the count register cnt1 is accumulated to N T -1, if yes, go to step S15; if no, go to step S12;
[0093] Step S12: Determine whether the level of ERR is high; if so, proceed to step S13; if not, return to step S10;
[0094] Step S13: Determine whether the counting register cnt1 is accumulated to NT-1, if so, return to step S6; if not, go to step S14;
[0095] Step S14: the counting register cnt1 is increased by 1, and the process returns to step S13;
[0096] Step S15: determine whether the value of the high-level counting register cnt2 is within the interval [13, 15]. If so, proceed to step S22; if not, proceed to step S16;
[0097] Step S16: Determine whether the value of the high-level counting register cnt2 is within the interval [9, 11]. If so, proceed to step S21; if not, proceed to step S17;
[0098] Step S17: Determine whether the value of the high level counting register cnt2 is within the interval [5,7]. If so, proceed to step S20; if not, proceed to step S18;
[0099] Step S18: Determine whether the value of the high-level counting register cnt2 is within the interval [1,3]. If so, proceed to step S19; if not, return to step S6;
[0100] Step S19: Determine that the fault level is ERR_Lx, which is a binary number of 0, and return to step S2;
[0101] Step S20: determine that there is no fault, and return to step S2;
[0102] Step S21: Determine that the ERR_Lx fault persists, and return to step S2;
[0103] Step S22: Determine that the fault level is ERR_Lx, which is a binary number 1, and return to step S2;
[0104] Step S23: determine that it is a high level waveform (RH), that is, ERR_H fault, and immediately trigger the corresponding fault operation of ERR_H, and clear the counting register cnt1 and the high level counting register cnt2, and return to step S2.
[0105] When the processor chip has no fault or the fault has been resolved, the safety management unit SMU sends a rectangular wave signal with a duty cycle corresponding to the fault through the ERR pin, that is, RW375;
[0106] When the processor has the highest level fault ERR_H, the safety management unit SMU will maintain the current level of the ERR pin, that is, stop the flipping of the ERR pin; until the current fault is resolved and the RW375 rectangular wave signal is restored;
[0107] Specifically, Figure 4 As shown in FIG. 1 , when the safety management unit SMU resumes sending the RW375 rectangular wave signal, the current level is first determined; if the current level is low, the safety management unit SMU starts from the next reference cycle T WAVE The rising edge of starts to resume flipping; if it is currently high, the safety management unit SMU starts from the next reference cycle T WAVE The falling edge starts to resume flipping.
[0108] When one of the lower level faults, i.e. ERR_Lx, occurs in the processor, the safety management unit SMU will send a rectangular wave through the ERR pin that loads the fault level through the duty cycle;
[0109] Specifically, Figure 5 The case of x=2'b01 is shown, where RW125 represents a binary number 0, and RW875 represents a binary number 1. The security management unit SMU sends in a sequence from high to low.
[0110] After sending a rectangular wave loaded with fault level information, that is, N=2 RW125 and RW875 waveforms, the safety management unit SMU should continue to send RW625 as a filling rectangular wave during the fault until the current fault is resolved and then resume sending the RW375 rectangular wave signal.
[0111] When the safety management unit SMU is sending a fault notification and the fault has not been resolved, a new fault occurs and the level of the new fault is compared with that of the current fault. When it is determined that the new fault level is higher, then after the safety management unit SMU has sent the fault level waveform ERR_Lx corresponding to the current fault, regardless of whether RW625 representing the ongoing fault has been sent, the waveform corresponding to the new fault level should be sent immediately. If it is determined that the new fault level is the same as the current fault level, or lower than the current fault level, the safety management unit SMU continues to send the corresponding waveform in the original way of the current fault.
[0112] In addition, if the current highest level fault is resolved, the safety management unit SMU should query whether there are other unresolved faults. If so, the safety management unit SMU will send the fault waveform with the highest level among the unresolved faults. If not, the safety management unit SMU will return to sending RW375.
[0113] In this embodiment, the fault detection unit EMU should have corresponding fault handling operations for all fault levels sent by the safety management unit SMU. Specifically, these fault handling operation modes can be fixed or flexibly configured on demand by the processor chip through a traditional serial bus interface. In addition, the fault detection unit EMU should use the EMU clock CLK to process the signal coming from the ERR pin. EMU Perform two beats to eliminate any glitches that may occur.
[0114] The present invention loads information on the duty cycle of the waveform sent by the ERR pin, thereby realizing the level configuration of various faults occurring in the processor chip, and giving priority to notifying faults with high levels, so that the PMIC can monitor, identify and handle faults more timely. This provides the necessary premise for possible fault monitoring in more dimensions and more refined subsequent corresponding operations of the PMIC. In addition, a specific waveform is sent through the pin of the SMU, especially the duty cycle, rising edge, falling edge, number of pulses and other information of the rectangular wave within the reference period are defined to notify the PMIC processor of what level of fault has occurred. By adopting the setting of the homologous clock scheme and the clock division, the EMU of the power management chip can transform the recognition of the duty cycle of the rectangular wave loaded with information sent by the processor chip SMU into a simple positive integer comparison, which greatly simplifies the operation of information recognition, and also increases the fault tolerance during transmission, sampling and recognition by utilizing intervals.
[0115] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A load chip fault monitoring circuit for a power management chip, characterized in that: It includes a fault detection unit EMU, the output CLK of the fault detection unit EMU REF Pin and the input CLK of the security management unit SMU in the processor chip REF The pins are connected, and the input ERR pin of the fault detection unit EMU is connected to the output ERR pin of the safety management unit SMU; When the processor chip is notified by the security management unit SMU through the ERR pin, the T WAVE The clock CLK used when the safety management unit SMU sends the ERR pin waveform is SMU CLK is the working clock of the fault detection unit EMU EMU or CLK EMU A clock after a certain frequency division; Each complete rectangular wave period T sent WAVE For several CLK SMU The clock period T SMU ; That is, T WAVE =N RW ·T SMU , where N RW The CLK required for a rectangular wave SMU The number of clock cycles; After the processor chip completes power-on reset and before the fault detection unit EMU officially starts the fault monitoring function, the safety management unit SMU sends a set of training waveforms to the fault detection unit EMU through the ERR pin to train the fault detection unit EMU to recognize the reference period T WAVE The waveform of When the processor chip has no fault or the fault has been resolved, the safety management unit SMU sends a rectangular wave signal with a duty cycle corresponding to the fault through the ERR pin; When the highest level of fault occurs in the processor, the safety management unit SMU will maintain the current level of the ERR pin until the current fault is resolved and then restore to sending a rectangular wave signal with a duty cycle corresponding to the fault-free state; When a lower level fault occurs in the processor, the safety management unit SMU will send a rectangular wave with a duty cycle to load the fault level through the ERR pin; When the safety management unit SMU is sending a fault notification and the fault has not been resolved, a new fault occurs and the level of the new fault is compared with that of the currently occurring fault. When it is determined that the new fault level is higher, the safety management unit SMU immediately sends the waveform corresponding to the new fault level after it has sent the fault level waveform corresponding to the currently occurring fault. If it is determined that the new fault level is the same as the currently occurring fault level, or is lower than the currently occurring fault level, the safety management unit SMU continues to send the corresponding waveform in the original manner of the currently occurring fault.
2. The load chip fault monitoring circuit for a power management chip according to claim 1, characterized in that: Reference period T WAVE =N DIV1 ·N DIV2 ·N RW ·T EMU =N T ·T EMU ; Among them, N DIV1 Indicated as CLK REF The clock cycle is CLK EMU The multiple of the clock cycle, N DIV2 Indicated as CLK SMU The clock cycle is CLK REF A multiple of the clock period, that is, N T =N DIV1 ·N DIV2 ·N RW ; The processor chip connects N DIV2 and N RW or N DIV2 ·N RW The product of is written into the PMIC register. After the training is completed, the fault detection unit EMU will obtain the expression T WAVE =N T ·T EMU Medium T The value is compared with the value of the PMIC register. When they are consistent, the fault monitoring function is officially enabled.
3. The load chip fault monitoring circuit for a power management chip according to claim 1, characterized in that: When the processor has the highest level of fault, the safety management unit SMU needs to recover to send a rectangular wave signal with a duty cycle corresponding to the fault-free state; the current level is judged. If it is a low level, the safety management unit SMU will start from the next reference cycle T WAVE The rising edge of starts to resume flipping; if it is currently high, the safety management unit SMU starts from the next reference cycle T WAVE The falling edge starts to resume flipping.
4. The load chip fault monitoring circuit for a power management chip according to claim 1, characterized in that: When a lower level fault occurs in the processor, the safety management unit SMU will send a rectangular wave loaded with fault level information through the ERR pin, and then continue to send a filling rectangular wave during the fault until the current fault is resolved and then restore to sending a rectangular wave signal with a duty cycle corresponding to the fault-free state.
5. The load chip fault monitoring circuit for a power management chip according to claim 1, characterized in that: The duty cycle of the rectangular wave sent by the safety management unit SMU through the ERR pin is written into the register of the PMIC by the processor chip through the traditional serial bus interface after power-on.
6. The load chip fault monitoring circuit for a power management chip according to claim 1, characterized in that: The fault detection unit EMU includes a counting register cnt1 and a high level counting register cnt2. When the fault detection unit EMU has locked the value of NT and determined the reference period T WAVE After the start and end boundaries, the count register cnt1 increases from 0 to N T -1 for cyclic increment counting, that is, every reference cycle T WAVE The corresponding count register cnt1 changes from 0 to N T -1; in a reference cycle T WAVE The fault detection unit EMU counts the time when the ERR signal is at a high level by the high level count register cnt2. EMU Count in units to determine a reference cycle T WAVE The corresponding fault level waveform.
