EP1723523A1

EP1723523A1 - Method and integrated switching circuit for increasing the immunity to interference

Info

Publication number: EP1723523A1
Application number: EP05716727A
Authority: EP
Inventors: Wolfgang Fey; Micha Heinz; Adrian Traskov; Frank Michel
Original assignee: Continental Teves AG and Co OHG
Current assignee: Continental Teves AG and Co OHG
Priority date: 2004-02-20
Filing date: 2005-02-17
Publication date: 2006-11-22
Also published as: US20070205930A1; WO2005081107A1; CN1922581A; JP2007535234A; US8578258B2; KR20060131842A

Abstract

The invention relates to a method for increasing the immunity to interference of an integrated switching circuit (16). According to said method, error signals are transmitted between at least one microprocessor chip or multiple microcontroller (1) and at least one additional component (2) in the form of one or more such error signals. A minimum pulse length, which is independent of the clock pulse frequency of the microprocessor or microprocessors, is defined for said transmission. A signal on an error line with a specific pulse length that is in excess of said minimum length is interpreted as an error. The invention also relates to an integrated switching circuit, which is configured in particular to carry out the aforementioned method and comprises at least one microprocessor chip or multiple microcontroller (1) and at least one additional component (2), which contains in particular separate power components and one or more pulse spreading units and/or signal delay units for the sequential emission of error pulses (6, 6') via at least one error line (3, 4)

Description

Method and integrated circuit to increase immunity to interference

The invention relates to integrated circuits, in particular in electronic control units, which are preferably components of motor vehicle control units. In particular, the control units are control units for motor vehicle brake systems. Such control devices are suitable, among other things, for carrying out safety-critical regulations. The motor vehicle control units preferably contain control programs such as ABS, ASR, ESP and the like. Due to the high safety requirements, the control systems which are formed by the integrated circuits include safety circuits which help to identify a failure or an error and initiate suitable measures, such as switching off the entire system or emergency operation, by measures which partially maintain the operation of the controlled system.

Control systems with integrated circuits that contain microprocessors are common for the tasks outlined above. In order to prevent malfunctions or to recognize them, it makes sense to provide at least two processor cores. Although three or more processor cores would further increase security, this is not always desirable for reasons of cost in connection with the very high number of pieces common in the field of automotive technology. There is therefore a need for inexpensive circuits with a high level of security.

For example, a concept of a control system consisting of two integrated circuits, which are accommodated in separate chip housings, has proven extremely successful. This has the advantage of a spatial separation of power components (PowerFETs etc.) and highly integrated microprocessor components (μC, memory, etc.).

In the above example, error-relevant communication is preferably carried out via two error lines ERR and ERR_N.

The invention is based on and improves on the above system in order to further improve the immunity to interference.

This object is achieved according to the invention by a method according to claims 1 and 2 and an integrated circuit according to claim 15.

The control system can also have one or more monitoring circuits, which are accommodated in particular on one or more further separate chips (watchdog).

In the known electronic controller, the processor chip and power chip are preferably connected via the lines ERR and ERR_N ^~ . The power chip must be able to recognize pulses from the processor chip that are transmitted to ERR or ERR_N with a minimum pulse width of, for example, Tmi _n = 30 ns. It is possible that external electrostatic, magnetic or electromagnetic interference (e.g. static charges, ESD) can interfere with the system and have undesirable effects.

According to the invention, methods and integrated circuits are described in accordance with the independent patent claims in order to increase the interference immunity, in particular the interference immunity of the signals ERR and ERR_N. This leads to increased system availability. According to a basic idea of the invention, the following measures, in particular or in any combination, are carried out:

• Significant extension of the minimum pulse width T _min • filtering the signals ERR and ERR_N in the power chip • redefinition of atchdog time window in the power chip with the introduction of delay time T _W i _ndow - Delay • • introduction of the block "Toggle and Delay ,, in the Processor chip to guarantee the minimum pulse width T _M in.

This has the following advantages:

• The filtering of the power chip input signals ERR and ERR_N significantly improves the immunity to interference of the system. In particular, the influence of electrostatic, magnetic or electromagnetic interference is considerably reduced, since these are generally of short duration and can therefore not be mistakenly interpreted as error signals due to the new, larger minimum pulse length.

• The inventive method w: j.rd ensures that there is no restriction on the error monitoring of the processor chip by the power chip compared to the previous, unfiltered solution. This means that no error pulses are lost.

The integrated circuit comprises at least one microprocessor chip or multiprocessor microcontroller and at least one further module, in particular one is an integrated module. The integrated component preferably comprises power components and is therefore, in particular, a mixed-signal component.

Problems arise due to the generally significantly higher clock frequency of the microprocessor components compared to the mixed-signal components. As a result, successive error events can be so close to one another that they can no longer be easily distinguished from other faults.

Alternatively, the invention can also be carried out in particular in the following way:

• Digital filtering can be replaced by analog filtering.

• If the microcontroller (processor chip) works much slower than the mixed signal chip (power chip), digital filtering can be carried out without the "toggle & delay" block in the microcontroller.

Further preferred embodiments result from the subclaims and the following description of the figures.

The invention is explained in more detail below with the aid of examples.

Show it

1 shows a system overview of the interconnected chips, Fig. 2 is an illustration of the filter characteristic in connection with the error lines and

Fig. 3 is a diagram showing the time behavior of the signals (timing diagram).

A system overview is given in FIG. 1:

In the electronic controller of a motor vehicle brake system, the microcontroller 1 (processor chip) is connected to the mixed signal IC 2 (power chip) via the error lines 3 and 4 and via the SPI interface 5.

The microcontroller 1 consists of two independent processor cores (Core A and Core B), the operations of which are continuously checked in terms of hardware by the comparison blocks Compare A and Compare B. If an error is detected during this comparison operation, then this error information must be reliably transmitted to power chip 2 via lines 3 ERR and 4 ERR_N.

According to the method, an error is preferably signaled by changing the level on an error line (e.g. from logic "high" to "low").

In particular, it is provided that each of the lines 3 ERR and 4 ERR_N changes the signal level once for each error. The levels at ERR and ERR_N are preferably opposite or complementary.

The mixed signal IC 2 generally works and preferably considerably slower than the microcontroller. For this reason, it must be ensured that the minimum pulse width T _min on the signal lines ERR and ERR_N is not is exceeded. Otherwise it is possible that module 2 "overlooks" an error.

These two tasks are performed by the blocks "Toggle & Delay ,, 6 and 6 '. If there are several errors in the processor chip, the level changes to ERR and ERR_N are delayed ("delay") in order to guarantee the minimum pulse width T _min .

The mixed signal IC filters the ERR / ERR_N signals each with a filter device 7, 7 '. This filtering is done digitally in particular. A filter time constant Tpii _te r is preferably provided. The minimum pulse width T _min is preferably much larger than the internal system clock of the power chip 2.

The filtering of the signals ERR / ERR_N advantageously suppresses external interference (electrostatic, magnetic or electromagnetic interference), thereby increasing the reliability and availability of the system.

In the "Level Compare ,, 8, 8 'blocks, a redundant check is carried out to determine whether both error signals have opposite levels. Equal levels on ERR and ERR_N lead to an error and are output outside of a test routine on the signals "Error Detected ,, 9, 9 '. In addition, the "Pulse Detect ,, 10, 10 'block searches for edges on the filtered ERR / ERR_N signals. The signal outputs of block "Level Compare" 8 and "Pulse Detect" 10 are logically linked by an OR block and form the output "ERR Detected A" 9. The same applies to the redundant path 11 '. In the electronic controller according to the example, a watchdog test routine with artificial error transmission is signaled at regular intervals T _oop via the SPI interface 5 from processor chip 1 to power chip 2. During this test routine, the error detection in the microcontroller and the connections via the error lines between the power chip and the processor chip are tested. An error is generated in the microcontroller via software-controlled test structures, which leads to a single level change on the ERR / ERRN lines at the output of the "Toggle and Delay" blocks.

While the test routine ("watchdog transmission") is active, power chip 1 defines a time window in block 50 via bus 5. Within the time window, a valid error signal for testing is considered in particular a signal which consists of exactly one edge change on the filtered ERR and ERR_N- Management exists. An error is detected by the block "Pulse Detect" 10 (or 10 'in redundant branch B) if no or more than one edge occurs on the filtered signals ERR filter or ERR_N filter within the time window (see timing diagram in Fig. 2). The level monitor 8 is switched off during the watchdog transmission by means of line 18 (18 'in the redundant branch B), since the level change to ERR and ERR_N cannot take place absolutely synchronously due to different signal propagation times.

At the end of the time window defined by block 50, the signals ERR / ERR_N must again have opposite levels, since level monitoring of the filtered signals now starts again. If there is a level change on the filtered signals ERR or ERR_N outside the watchdog time window, this is recognized directly as an error by the block "Pulse Detect" 10 and output on line "Err Detected" 9. To achieve this functionality, block 50 is connected to block 10 via a control line.

Block "Level Compare" 8 is also provided redundantly (see block 8 '). Block 8 checks whether the error signals of the complementary error lines are both present. If this is not the case, for example if one of the error lines is defective, an error is output.

2 shows how the ERR filters 7 and 7 'can be implemented:

As can be seen in FIG. 1, the fault detection for lines 3 ERR and 4 ERR_N in power chip 2 is basically designed redundantly. A separate error detection circuit 11 and 11 ', each with its own error line 9 "ERR-Detected A" and 9'"Err-DetectedB", is therefore preferably provided for each computer core. In order to reduce the manufacturing costs in circuit 11, filter 7 is preferably only available for one of the existing fault lines 3 and 4. In circuit 11 'there is also only one filter circuit 7'. In both circuits 11 and 11 'the output of the filter 7 or 7' with the connection points 12 and 12 'via cross lines 13 and 13' becomes the inputs of the "level compare" circuits 8 and 8 'in addition to the error signals 3 and 4 fed on the input side. For this purpose, the result of the filter operation is therefore fed back into the block ERR-Check B 11 '. Conversely, the same applies to the ERR_N filter, the result of which is fed back into the ERR-Check A 11 block. The feedback is used for the test on inverted levels. ERR filter 7 is preferably designed as a digital up / down counter, which changes its counter reading depending on the input signal. If there is a signal with the digital low level at the input of the filter, the counter reading is reduced by one. If there is a high level, the counter reading is increased by one. The possible meter readings are limited to the range from 0 to Z _Ma χcount. A possible signal curve is shown in FIG. The same applies to the filter 7 '.

If the counter reading of filter 7 reaches zero, the output of the filter at point 12 goes low. If the counter reaches the value ZMaxcount _/ then the output of the filter goes high. Together with the clock frequency used, the filter time is'

where F _{CPU is} the clock frequency of the microcomputer. With this implementation, all disturbances which are shorter than T _fl ιter and which are detected by the scanning are suppressed. The counter reading of the filter is shown in Fig. 2 depending on the input of the filter.

3 shows a timing diagram which shows the logic levels of the existing control lines as a function of time. A possible definition of a watchdog time window in block 50 will now be explained with reference to this figure. The watchdog time window in block "watchdog window" 50 is determined by processor chip 1 through SPI interface 5. Processor chip 1 selects the watchdog transfer using the chip select signals CSVO_N = low and CSWD_N = high. The watchdog time window 17 begins after the first rising edge 14 on the clock signal of the SPI carry SPICLK. The delay TSync includes signal runtimes and the synchronization time between the external clock of the SPI interface and the internal power chip system clock.

The watchdog time window 17 ends after the rising edge 15 of CSWD_N. However, it must be ensured that the edges of the filtered signals ERR or ERR_N are still recognized within the active watchdog time window. For this reason, the time of edge 15 is followed by a delay time TinoDeia, in which the expected edge is still processed. The following condition applies: WindowDelay ^ Tpilter

However, this condition only meets the requirements in a trouble-free system. However, if one takes into account the effect that interference on the input signal causes an additional delay, then the most robust and cost-effective system is obtained if

TwindowDeiay ~ ^ ^* filter;

since the filter delay presented in the faulty system can never be greater than twice the filter time.

The following requirements must be placed on the signal delay within the processor chip 1:

Outside of the time while the error check routine is running (watchdog transfer), a one-time error event within the processor chip 1 always for an error detection within the power chip 2. In this case, processor chip 1 only has to guarantee that the minimum pulse length T _MI T on the signals ERR or ERR_N is not undercut.

During an error check routine (watchdog transfer) in which a time window is defined via bus 5, it must be guaranteed that, in addition to the error caused by the test procedure, any additional error is reliably detected by power chip 2. In this case, processor chip 1 must change the levels on the error lines ERR / ERR_N once for the first error that occurs. If another error occurs in processor chip 1 during the time T _M , then this error must be delayed until the time T _M m has elapsed. Then processor chip 1 must change the signal level again to ERR / ERR_N. This is done by means 6 or 6 '("toggle and delay"). In this case, power chip 2 either detects the double level change during the watchdog time window 17, or a level change is outside the time window 17. In both cases, power chip 2 recognizes the additional error.

A two level change at a distance of T _Min is therefore sufficient in all cases to detect an error in the power chip. The block "Toggle & Delay ,, 6 or 6 'therefore only has to transmit another error delayed by T _Mln to the power _{chip in} addition to the first error that occurs.

Claims

claims

1. A method for improving the immunity to interference of an integrated circuit (16) in which error signals are transmitted between at least one microprocessor chip or multiple processor μC (1) and at least one further module (2) in the form of one or more error signals, characterized that a minimum pulse length, independent of the clock frequency of the microprocessor or microprocessors, is defined for the transmission, from which a signal on an error line with a certain pulse length is interpreted as an error.

2. Method for improving the immunity to interference of an integrated circuit (16), in which error signals are transmitted between at least one microprocessor module or multiple processor module and at least one further mixed signal module in the form of one or more error signals, said modules being on a chip or are integrated in a chip housing, characterized in that a minimum pulse length, which is independent of the clock frequency of the microprocessor or the microprocessors, is defined for the transmission, from which a signal on an error line with a certain pulse length is interpreted as an error.

3. The method according to claim 1 or 2, characterized in that in the event of an error sequence with an error distance that is smaller than the minimum pulse length, the time of the error sequence output via the at least one error line is temporally extended compared to the actual error sequence time.

4. The method according to at least one of claims 1 to 3, characterized in that the / the error signal (s) in a chip that receives the / the error signal (s) of another chip or chip are not processed exactly when the signals Fall below the minimum duration and be processed precisely when the minimum duration is reached or exceeded, for this purpose the signals are passed in particular through at least one filter, in particular a low-pass filter (7, 7 ').

5. The method according to at least one of the preceding claims, characterized in that at least one watchdog time window (17) is specified in the integrated circuit or in the further module (2) within which at least one artificially generated error signal or error signal pattern is generated and checked, so that the fault detection circuits themselves can be checked.

6. The method according to at least one of the preceding claims, characterized in that the watchdog time window (17) has a delay time T _{win <} j _owDelay , wherein until the delay time T _windowDe ι _ay the time window in which at least one error signal or error signal pattern is expected to remain open.

7. The method according to claim 5, characterized in that the delay time T _W j. _n dowDeiay is greater than the filter time _F ii _ter of the filter or filters (7, 7 ') which processes the error signals of the at least one error line (3, 3').

8. The method according to at least one of the preceding claims, characterized in that the error signals are broadened within the error signal-sending chip (1) and / or are output in succession via the error line (s).

9. The method according to at least one of the preceding claims, characterized in that a test of the at least one error line (3, 4) is carried out with the aid of the interface (5).

10. The method according to at least one of the preceding claims, characterized in that the error signals are filtered by filters (7,7 ') with a certain filter time T _F ii_ _ter .

11. The method according to at least one of the preceding claims, characterized in that the Zeifenster Twin _dow D _e i _a ^m yi another building block is set (2) connected to the chip (1) interface (5).

12. The method according to at least one of the preceding claims, characterized in that the condition T i _n _dowDe i _a y greater than the filter time T _F ii _te r is fulfilled.

13. The method according to at least one of the preceding claims, characterized in that the delay TwindowDeiay corresponds approximately to twice the time T _F ii _te r.

14. The method according to at least one of the preceding claims, characterized in that the pulse width T _{Min is set} to a value of at least about 30 ns becomes.

15. Integrated circuit, in particular in such a way that the above method is carried out, comprising - at least one microprocessor chip or multiple processor microcontroller (1) or microprocessor module and at least one further separate module (2) or a mixed signal module integrated in the same module, which / in particular comprises separately arranged power components, and - one or more pulse broadening devices and / or signal delay devices for the subsequent output of error pulses (6, 6 ') via at least one error line (3, 4)

16. Integrated circuit according to claim 15, characterized by one or more filters (7,7 ') for filtering the error signals transmitted by the error lines (3,4).

17. Integrated circuit, in particular such that the method is carried out according to claims 1 to 14, comprising - at least one microprocessor chip or multiple processor microcontroller (1) and at least one further module (2), which in particular comprises separately arranged power components and - one or several filters (7,7 ') for filtering error pulses (6, 6') via at least one error line (3, 4).

18. Integrated circuit according to at least one of the preceding circuit claims, characterized in that the filter (7,7 ') as a digital forward / reverse up counter is executed.

19. Integrated circuit according to at least one of the preceding circuit claims, characterized in that the chips or modules are connected by means of at least one bus (5) and at least one error line (3, 4).

20. Integrated circuit according to at least one of the preceding circuit claims, characterized in that it comprises hardware-based test structures, with the aid of which a test of the at least one error line (3, 4) can be carried out via an interface (5).

21. Integrated circuit according to at least one of the preceding circuit claims, characterized in that the microprocessor chip (1) or the further component comprises at least one watchdog window circuit (50).

22. Integrated circuit according to claim 21, characterized in that the watchdog window circuit (50) specifies a watchdog time window (17) and the watchdog time window (17) has a delay time _W i _ndowDe ι _a y, whereby up to Expiry of the TwinowDeiay delay time, the time window in which at least one error signal or error signal pattern is expected remains open.

23. Integrated circuit, in particular such that the method is carried out according to claims 1 to 14, comprising - at least one microprocessor chip or multiple processor processor microcontroller (1) and at least one further component (2), which in particular comprises separately arranged power components and - at least one watchdog window circuit (50) which specifies a watchdog time window (17) and the watchdog time window (17 ) has a delay time TwindowDeiay, with the time window in which at least one error signal or error signal pattern is expected remaining open until the delay time T _Win doDei _a y expires.

24. Integrated circuit according to at least one of the preceding circuit claims, characterized in that the delay time T _wi noDeiay is greater than the filter time T _F ii _te of the filter or filters (7, 7 '), which the error signal or errors of the at least one Error line (3, 3 ') processed.