Classifications

• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
  • G06F11/14—Error detection or correction of the data by redundancy in operation
  • G06F11/1497—Details of time redundant execution on a single processing unit
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
  • G06F11/08—Error detection or correction by redundancy in data representation, e.g. by using checking codes
  • G06F11/10—Adding special bits or symbols to the coded information, e.g. parity check, casting out 9's or 11's
• • H—ELECTRICITY
  • H03—ELECTRONIC CIRCUITRY
  • H03K—PULSE TECHNIQUE
  • H03K19/00—Logic circuits, i.e. having at least two inputs acting on one output; Inverting circuits
  • H03K19/003—Modifications for increasing the reliability for protection
  • H03K19/0033—Radiation hardening
  • H03K19/00338—In field effect transistor circuits

Definitions

• the present invention relates to digital circuits insensitive to external disturbances, in particular localized disturbances originating in particular from heavy ion bombardments. Such a disturbance is liable to untimely change the state of a memory point, and special structures of memory point must be adopted to overcome this drawback.
• a memory point was likely to fail only if the disturbance directly affected this memory point. For example, a heavy ion had to reach one of the transistors constituting the memory point. Disturbances occurring outside of memory points, that is to say in combinational logic circuits, had a very low probability of modifying the states of memory points. Indeed, these disturbances resulted in very brief pulses which were filtered almost immediately by the high capacities of the conductors. Even if such a disturbance caused a parasitic pulse reaching the input of a memory cell, this pulse had a low probability of modifying the state of the memory cell.
• the generation of a parasitic pulse by a combinational logic circuit can be considered as an error which could be corrected by a conventional solution.
• FIG. 1 illustrates a conventional solution that could be used to correct the errors produced by a combinational logic circuit. It is a triple redundant error correction system. The same combinational logic circuit 10 is duplicated twice, respectively at 11 and at 12.
• the outputs of circuits 10 to 12 are supplied to a majority voting circuit 14 which selects for its output the value which is supplied by at least two of the redundant circuits 10 to 12.
• the output of the majority voting circuit 14 is therefore free of error in the event of failure of at most one of the redundant circuits 10 to 12, even if this failure is permanent.
• An object of the present invention is to provide a solution for eliminating at the output of a combined logic circuit any parasitic pulse caused by a localized disturbance, occupying a relatively small silicon surface, is known.
• the present invention provides a circuit protected against transient disturbances, comprising a combinational logic circuit having at least one output; a circuit for generating an error control code for said output, - and a memory element disposed at said output, controlled by the control code generation circuit to be transparent when the control code is correct, and for keep its state when the control code is incorrect.
• the circuit for generating error control code comprises a circuit for calculating a parity bit for said output and a circuit for verifying the parity of the output and the bit parity.
• the circuit for generating error control code comprises a duplicated logic circuit, said memory element being designed to be transparent when the outputs of the logic circuit and of the duplicated circuit are identical, and, while said outputs are separate, keep its state.
• the error control code generation circuit comprises an element for delaying said output by a predetermined duration greater than the maximum duration of transient errors, said memory element being designed to be transparent when the outputs of the logic circuit and of the delay element are identical, and for, while said outputs are separate, to keep its state.
• said memory element is produced from a logic gate providing said output of the logic circuit, this logic gate comprising at least two first transistors controlled by a signal from the logic circuit and at least two second transistors controlled by the corresponding signal from the duplicated circuit, each of the second transistors being connected in series with a respective one of the first transistors.
• the present invention also relates to a circuit protected against transient disturbances, comprising a combinational logic circuit having at least one output connected to a first synchronization flip-flop clocked by a clock, a second flip-flop connected to said output and clocked by the delayed clock d 'a predetermined duration, and a circuit for analyzing the outputs of the scales.
• the analysis circuit signals an error if the outputs of the flip-flops are different.
• the circuit comprises a third flip-flop connected to said output and clocked by the clock delayed by twice the predetermined duration, the analysis circuit being a majority voting circuit.
• the present invention further relates to a circuit protected against transient disturbances comprising a combinational logic circuit having at least one output connected to a first synchronization flip-flop clocked by a clock, a second flip-flop clocked by the clock and receiving said output delayed by a predetermined duration, and a circuit for analyzing the outputs of the flip-flops.
• the analysis circuit signals an error if the outputs of the flip-flops are different.
• the circuit comprises a third flip-flop clocked by the clock and receiving said delayed output of twice the predetermined duration, the analysis circuit being a majority voting circuit.
• the present invention further relates to a circuit protected against transient disturbances, comprising three identical logic circuits. Each of the logic circuits is preceded by a memory element with two inputs respectively receiving the outputs of the other two logic circuits, each memory element being designed to be transparent when its two inputs are identical, and to maintain its state when the two inputs are different.
• the logic circuits are inverters and the memory elements comprise, in series, two P-channel MOS transistors and two N-channel MOS transistors, a first of the inputs of the memory element being connected to the gates of a first of the P-channel MOS transistors and of a first of the N-channel MOS transistors, and the second input of the memory element being connected to the gates of the other two transistors.
• FIG. 1 previously described , illustrates a conventional solution for correcting errors produced by a faulty circuit
• FIG. 2A schematically represents a first embodiment of a circuit according to the invention making it possible to suppress parasitic pulses generated by localized disturbances in a combinational logic circuit
• - Figure 2B shows a timing diagram illustrating the operation of the circuit of Figure 2A
• Figures 3A and 3B show two examples of state-keeping elements used in the circuit of Figure 2A
• FIG. 4 schematically represents a second embodiment of the circuit according to the invention making it possible to suppress parasitic pulses
• FIG. 5 represents an example of a state-conserving element used in the circuit of FIG. 4
• FIGS. 6A, 6B and 6C represent other examples of state preservation elements of the type used in the circuit of FIG. 4, •
• FIG. 7A schematically represents a third embodiment of the circuit according to the invention making it possible to suppress unwanted pulses;
• FIG. 7B represents a timing diagram illustrating the operation of the circuit of FIG. 7A, •
• FIG. 8A represents a variant of the embodiment of FIG. 7A, • FIG. 8B represents a timing diagram illustrating the operation of the circuit of FIG. 8A ;
• FIG. 9A represents a fourth embodiment of the circuit according to the invention making it possible to suppress parasitic pulses;
• FIG. 9B represents a timing diagram illustrating the operation of the circuit of FIG.
• FIG. 10A schematically represents a fifth embodiment of the circuit according to the invention making it possible to suppress parasitic pulses
• FIG. 10B represents a timing diagram illustrating the operation of the circuit of FIG. 10A
• Figure 11 shows an application of the embodiment of Figure 4 to an asynchronous loop
• FIG. 12 represents an improvement of the structure of FIG. 11, - FIG. 13 represents a simplification of the structure of FIG. 12, •
• FIG. 14 represents an application of the principle of FIG. 13 to a static memory cell
• FIG. 15 represents a majority voting circuit produced from the structure of FIG. 14.
• FIG. 2A schematically represents a first embodiment of the invention exploiting this characteristic.
• a combinational logic circuit 10 is associated with a verification circuit 20 which provides an error control code P for the output A of the logic circuit 10.
• the circuit 20 is for example a circuit which conventionally calculates a parity bit P for output A of logic circuit 10, this output A can of course be multiple.
• the parity bit P is combined by OU-Exclusive with the output A of the logic circuit 10, which provides an error signal E which is active when the parity is bad, that is to say when the output A or the parity bit P contains an error.
• the error signal E and the output A are supplied to an element which one will say "with conservation of state" 24. It is in fact a memory element similar to a rocker with controlled transparency, c ' that is to say having a first mode, selected when the error signal E is inactive, where the output A is transmitted as it is to the output S of the element 24. In a second mode, selected when the signal E is active, element 24 retains the state of output A, as it was before the activation of error signal E.
• a flip-flop 26, conventionally provided for locking the output of the logic circuit 10, receives the output S of the state-conserving element 24 instead of directly receiving the output A of the circuit 10.
• FIG. 2B represents a timing diagram illustrating the operation of the circuit of FIG. 2A.
• the output A of the logic circuit 10 is in any state X.
• the error signal E being inactive, the element 24 is in " transparent "and transmits state X on its output S. This state X is locked in flip-flop 26.
• the output of flip-flop 26, being possibly re-supplied to logic circuit 10, this logic circuit generates a new output A after a delay t c corresponding to the propagation time in the "critical path" of circuit 10.
• the output A of the circuit 10 changes state, for example, goes to 0. It is the same for the output S of the element 24 which is always put in "transparent" mode by the signal E
• a parasitic pulse starts on the output A, which ends at an instant t 3 .
• FIG. 2B illustrates an unfavorable case where the parasitic pulse on the signal A risks causing an untimely modification of the state of the flip-flop 26.
• the end of the parasitic pulse coincides with the following active edge of the clock CK, which front causes the storage in the flip-flop 26 of the state of the output S immediately preceding the instant t 3 .
• the error signal E becomes active during the duration t p of the parasitic pulse, making the element 24 "opaque" to the evolution of the signal A between the instants t 2 and t 3 .
• signal S does not change state during the parasitic pulse and the flip-flop 26 stores a correct value.
• a flip-flop only changes state if the new state has been presented to it long enough before the corresponding clock active front, for a period known as initialization.
• a risk of erroneous value storage by the flip-flop 26 occurs in a range of variation of the position of the parasitic pulse, this range going from a position where the end of the pulse precedes the active edge of the clock CK of the initialization duration, at a position where the start of the pulse occurs at the time of the active edge of the clock CK.
• the state conservation element 24 is also a storage cell, the state which it must store must have been presented at least for an initialization time before the storage order (l 'activation of signal E). Thus, it is necessary that the duration separating the instants t x and t 2 is greater than this initialization time. In addition, it must also be guaranteed that a complete initialization time of the element 24 has elapsed before or after the parasitic pulse between the times t x and t 3 , this to be sure that the element 24 takes into account counts the level out of the pulse.
• a state preservation element 24 will generally be produced from logic gates supplying the outputs of the combinational logic circuit 10 to flip-flops 26.
• FIGS. 3A and 3B represent two examples of state preservation elements performing a NAND function with two inputs.
• the two inputs a and b are supplied to an AND gate 30, the output of which is connected to a first input of a NOR gate 32.
• An Exclusive OR gate 22 ′ equivalent to the Exclusive OR gate 22 in the figure 2A, receives the inputs a and b, as well as the parity bit P.
• the output E of the gate 22 ' is supplied to a second input of the NOR gate 32 and to a first input of an AND gate 34.
• the outputs of doors 32 and 34 are supplied to an OR gate 36 which provides the output S of the state-conserving element, which output is looped back to a second input of AND gate 34.
• the signal E is at 1, causing a memorization of the state of the output S in a memory point consisting of the gates 34 and 36.
• the circuit of Figure 3A requires four doors to perform the function of a single door.
• FIG. 3B represents a less costly solution in terms of hardware for producing a state conservation element performing a NAND function.
• the input signals a and b are supplied to the two inputs of a NAND gate 38, the output of which is connected to a capacitor C by means of a switch K.
• switch K is controlled by the signal error E supplied by the OU-Exclusive door 22 '.
• the switch K When the error signal E is inactive, the switch K is closed and the capacitor C charges at the level supplied by the gate 38. When the error signal E is activated, the switch K is open, but the state of the output S of the element is kept by the capacitor C for the duration of the parasitic pulse. It will be noted that the capacitor C can be constituted by the simple capacity of the output line S. State conservation elements carrying out other logical functions can be produced by a person skilled in the art. For example, to perform the identity function using the solution of FIG. 3B, the single input signal is directly supplied to the switch K. The embodiment of FIG.
• FIG. 4 represents an embodiment making it possible to avoid this drawback.
• the combinational logic circuit 10 is duplicated once at 11.
• the output A of circuit 10 and the duplicated output A * of circuit 11 are supplied to a state preservation element 24 ′ which transmits on its output S the state of its input A or A * when the inputs A and A * are identical and which retains its state when the inputs A and A * become different.
• FIG. 5 represents a state conservation element
• Inputs a and b are supplied to an AND gate 50, the output of which is supplied to a first input of an AND gate 52 and to a first input of an OR gate 54.
• Duplicate inputs a * and b * are supplied to an AND gate 56 whose output is connected to the second input of door 52 and to the second input of door 54.
• the outputs of doors 52 and 54 are connected respectively to doors 36 and 34 similar to doors 36 and 34 in FIG. 3A.
• doors 34, 36, 52, and 54 constitute a state-conserving element having the logical "identity" function. To create any logical function, it suffices to connect two doors, each carrying out this function in a conventional manner, at doors 52 and 54.
• the elements with state conservation are produced from the internal structure of conventional logic gates. Two transistors connected in series are provided for this for each transistor normally required in the conventional gate. The two transistors are controlled to be opened at the same time, so that, if one of them closes due to a disturbance, the second, remaining open, prevents any untimely flow of current.
• Such a configuration lends itself particularly well to a structure of the type of FIG. 4 comprising two redundant logic circuits. Indeed, the two transistors of the series association are then controlled respectively by a signal and its duplicated signal.
• FIG. 6A represents an element with state conservation according to this principle having a function of reverser.
• the output S of the circuit is connected to a high potential by means of two P-channel MOS transistors in series MPI and MP2.
• the output S is also connected to a low potential by two N-channel MOS transistors in series MN1 and MN2.
• N-channel MOS are controlled by a normal signal a while the remaining transistors are controlled by the duplicated signal a *.
• the two MP transistors or the two trans sistors MN are conductive and force the output S to the corresponding potential to fulfill the function of inverter.
• FIG. 6B represents a state preservation element performing a NOR function. Its output S is connected to a high potential via one MOS P-channel MOS transistors in series, controlled respectively by the normal input signals a and b and their duplicated signals a * and b *. The output S is also connected to a low potential by means of two series associations of N-channel MOS transistors, one comprising two transistors controlled respectively by the signals a and a *, the other comprising two transistors controlled respectively by signals b and b *.
• FIG. 6C represents a state preservation element performing a NAND function.
• the output S is connected to the low potential by one through four N-channel MOS transistors in series controlled respectively by the signals a and b and their duplicated signals a * and b *.
• the output S is also connected to the high potential by means of two series associations of P-channel MOS transistors, the first comprising two transistors controlled respectively by the signals a and a *, and the second comprising two transistors controlled respectively by the signals i ⁇ and b *.
• FIGS. 6B and 6C operate on the principle described in relation to FIG. 6A. More generally, this principle of placing duplicate transistors in series applies to any logic gate.
• the circuit of FIG. 6A can be used as a dynamic memory cell insensitive to disturbances. For this, the state of the cell is stored redundantly on the two inputs a and a * by capacitive effect. If one of the inputs is disturbed, the output S retains its previous state 14
• FIG. 7A represents a third embodiment of a combinational circuit according to the invention insensitive to localized disturbances. It includes a single combinational logic circuit 10. The suppression of the parasitic pulses is obtained exclusively thanks to a temporal redundancy, contrary to the previous embodiments combining the temporal and hardware redundancies.
• the output A of circuit 10 is supplied to three flip-flops 70, 71 and 72 clocked respectively by the clock CK, the clock CK delayed by a duration ⁇ and the clock CK delayed by a duration 2 ⁇ .
• the outputs SI, S2 and S3 of these flip-flops are supplied to a majority voting circuit 74 which supplies the corrected output S.
• FIG. 7B represents a timing diagram illustrating the operation of the circuit of FIG. 7A.
• This chronogram represents, in the form of vertical bars, the active edges of the clock signals CK, CK + ⁇ and CK + 2 ⁇ . It is assumed that the signal A has a parasitic pulse straddling the first edge of the clock CK, occurring at an instant t 0 .
• the flip-flop 70 activated at time t 0 , erroneously stores the state of the parasitic pulse.
• the signal A passes to 1 in a normal manner. This transition occurs at a time interval t c after an instant tl where the last edge of the clock CK + 2 ⁇ has occurred.
• the time t c is the propagation time through the voting circuit 74 and the logic circuit 10.
• the state 1 of the signal A is sampled by the following respective edges of the clocks CK, CK + ⁇ and CK + 2 ⁇ .
• the signal SI remains at 1 while the signals S2 and S3 pass to 1, respectively at times t 3 , t 4 and t 5 .
• a time interval t c after time t 5 the signal
• the output S of the voting circuit 74 is at 1 when at least two of the signals S1, S2 and S3 are at 1. This is the case which occurs from time t 4 , while the signal S2 is at 1.
• circuit of FIG. 7A does not pass to 1 at time t 0 where the parasitic pulse occurs, but passes correctly to 1 at time t 4 in response to a normal passage to 1 of signal A.
• the parasitic pulse must be sampled by only one of the clocks CK, CK + ⁇ and CK + 2 ⁇ .
• the maximum duration t p of the parasitic pulses can therefore reach the value ⁇ - t h , where t h is the initialization time of flip-flops 70 to 72.
• ⁇ p + t h .
• the period of the clocks must be chosen at least equal to t c + 2 ⁇ + t h , which time corresponds to the maximum propagation time of the inputs of circuit 10 to output S.
• FIG. 8A represents a variant of the embodiment of FIG. 7A.
• the same elements as in FIG. 7A are designated by the same references.
• these flip-flops are clocked by the same CK clock.
• the signal A is supplied to two delay lines in cascade 80 and 81 each introducing a delay ⁇ .
• Signal A is directly supplied to flip-flop 70, output A2 of delay line 80 is supplied to flip-flop 71 and output A3 of delay line 81 is supplied to flip-flop 72.
• FIG. 8B represents a timing diagram illustrating the operation of the circuit of FIG. 8A.
• the first edge of the clock CK occurs. It is assumed that the signal A has a parasitic pulse straddling this edge. It follows that the signal SI goes to one at this instant t 0 .
• the signals A2 and A3 have the same parasitic pulse but offset by ⁇ and 2 ⁇ respectively with respect to the instant t 0 .
• the delay ⁇ is chosen to be greater than the duration t p + t h , where t p is the maximum duration of a parasitic pulse and t h the initialization time of flip-flops 70 to 72.
• t p is the maximum duration of a parasitic pulse
• t h the initialization time of flip-flops 70 to 72.
• the signal A passes to 1 in a normal manner during a clock period.
• the duration sepa ⁇ rant the instants t 1 and t 2 corresponds to the time t c of propagation in the critical path of the circuit 10 and the polling circuitry 74.
• the delay t c is such that the edge corresponding signals A2 and A3 occurs even before
• the signals A, A2 and A3 are sampled while they are at 1.
• the signals SI, S2 and S3 pass to 1.
• the signals SI, S2 and S3 remain at 1 until next edge of the clock signal occurring at an instant t 4 .
• the signals A, A2 and A3 have gone to 0.
• the signals SI, S2 and S3 have gone to 0.
• the signal S has a correct shape, staying at 0 between the instants t 0 and t x and passing to 1 between the instants t 3 and t 4 , while the signals SI, S2 and S3 are all at 1.
• FIG. 9A schematically represents a fourth embodiment of the circuit according to the invention making it possible to suppress parasitic pulses.
• a state preservation element 24 ′ of the type of that of FIG. 4 is used here, designed to operate with duplicated signals. This element receives the output A of the logic circuit 10 and this same output is delayed by a delay line 90 introducing a delay ⁇ . The signal supplied by this delay line 90 constitutes the duplicated signal A *.
• the output S of the element 24 ' is supplied to a rocker 26.
• FIG. 9B represents a timing diagram illustrating the operation of the circuit of FIG. 9A.
• the signal A has a parasitic pulse straddling a first edge of the clock CK occurring at an instant to.
• the signal A goes to 1.
• the instants t 0 and t ⁇ are distant from the propagation time t c in the critical path of the circuit 10.
• the delayed signal A * goes to 1.
• the signals A and A * remain at 1 during a clock period and go to 0 at respective instants t. and t 5 before the next clock edge occurring at time t 6 .
• the signal S is at 1 between the instants t 2 and t 5 .
• This state 1 is sampled by the flip-flop 26 at time t 3 , and corresponds to the state which it is actually necessary to sample in the signal A.
• the operation of the circuit is correct if the clock period is at least equal to t c + ⁇ + 2t 24 , + t p + t h , where t 24 , is the propagation time in the element 24 'and t h the initialization time of the flip-flop 26.
• the value ⁇ must be chosen greater than t p - t 24 ,.
• FIG. 10A schematically represents a fifth embodiment of the circuit according to the invention, making it possible to detect in a simple manner an error linked to a parasitic pulse.
• the output A of the logic circuit 10 is supplied to two flip-flops 92 and 93, one clocked by the clock CK and the other by the clock delayed by a duration ⁇ .
• the flip-flop 92 can be controlled by a front or a level of a first type (rising or falling - high or low) of a clock CK, while the flip-flop 93 is controlled by a front or a level opposite type of the same clock (falling or rising edge - low or high).
• FIG. 10B represents a timing diagram illustrating the operation of the circuit of FIG. 10A.
• a parasitic pulse occurs in the signal A straddling an edge of the signal CK occurring at an instant t 0 . It follows that the signal SI goes to 1.
• the flip-flop 93 does not yet sample the signal A and its output S2 remains unchanged (at 0).
• the comparator 95 does not yet indicate an inequality of the signals SI and S2, and the signal ERR indicates an absence of error by a state 0.
• the next edge of the clock CK + ⁇ occurs which samples the signal A while it is at 0. It follows that the signal S2 goes to 0.
• the period of the clock must be chosen at least equal to t c + t h + ⁇ , the duration ⁇ being at least equal to the duration p + t h .
• the output SI which is used. So we have to make sure that a transition from the SI output is not propagated to output A before the next edge of the clock CK + ⁇ . In other words, the propagation time t c must be greater than ⁇ . In this case, the clock period will be equal to t c + t h , that is to say equal to the clock period of the conventional circuit without protection against transient errors.
• the error signal provided by the circuit of Figure 10A can be used in various ways to correct the detected error. It can be envisaged, for example, that this error signal triggers a resumption of operation, for example the repetition of a last "instruction" executed by the system.
• the flip-flops 92 and 93 are clocked by the same clock CK and one of them receives the signal A delayed by the duration ⁇ .
• FIG. 11 represents an arrangement according to the invention for protecting such a circuit, using the principle of duplication of FIG. 4.
• the output of a logic circuit 10 and the output of a duplicated logic circuit 11 are connected respectively to the two inputs of a first state-conserving element 24a and of a second state-preserving element 24b, both of the type of that of FIG. 4.
• the output of element 24a is looped back into circuit 10, while the output of element 24b is looped back into circuit 11. It is necessary to use two state-conserving elements, because if we only used one of which the output would be looped back to the two circuits 10 and 11, a disturbance in the element would be transmitted to the two duplicated circuits, causing the same error in the two circuits. This error condition would not be corrected.
• the structure of FIG. 11 is however sensitive to a disturbance occurring on the output of one of the elements with state conservation. If the propagation time in the circuit 10 or 11 concerned is less than the duration of the disturbance, the delayed disturbance arrives at the input of the state-conserving element even before the disturbance has disappeared on its output . As a result, the element tends to keep the erroneous state affected by the disturbance.
• FIG. 12 shows a structure avoiding this problem.
• the circuit 10 and its duplicate 11 are each split into two parts, 10a and 10b for the circuit 10, and 11a and 11b for the circuit 11. Between the two parts of each circuit, an element of conservation of additional state is inserted, 24c between the parts 10a and 10b, and 24d between the parts 11a and 11b, the elements 24c and 24d being connected in the same way as the elements 24a and 24b.
• FIG. 13 represents a simplification of the structure of FIG. 12, made possible if the parts 10a, 10b and their duplicated parts have the same logic function and receive the same inputs. Compared to FIG. 12, the circuit 11b and the element 24d have been omitted.
• the state-conserving elements 24a and 24b respectively receive the output of the circuit 11a and the output of the circuit 10a in place of the output of the circuit 11b in FIG. 12.
• FIG. 14 represents an application of the principle of FIG. 13 for producing a static memory cell.
• the state conservation elements 24a, 24b and 24c are state conservation inverters of the type of FIG. 6A.
• the circuit parts 10a, 11a and 10b are conventional inverters.
• a state conservation inverter followed by a conventional inverter have an identity function. This ensures that the elements 24a, 24b and 24c receive identical input values, which is also valid for the inverters 10a, 11a and 10b.
• the memory cell thus obtained is insensitive to disturbances both in static operation and in dynamic operation.
• FIG. 15 represents a variant of the cell of FIG. 14.
• a P-channel MOS transistor controlled by a clock signal CK has been inserted.
• an N-channel MOS transistor cot connected by the complement of the clock signal CK has been inserted.
• FIG. 15 represents an application of the structure of FIG. 14 to a voting circuit usable in the circuits of FIGS. 7A and 8A.
• the access transistors have been omitted.
• the three input signals SI, S2, S3 of the voting circuit are applied to the inputs of the inverters.
• a voting circuit is thus obtained which is used to memorize the result of the vote in a manner insensitive to disturbances. If this voting circuit is used in FIGS. 7A and 8A, the flip-flops 70 to 72 which precede the voting circuit are simple flip-flops with controlled transparency. It is also possible to connect a conventional memory cell to each of the inputs SI, S2 and S3, controlled by a clock signal. A master-slave rocker is thus formed.

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Engineering & Computer Science (AREA)
• Quality & Reliability (AREA)
• General Physics & Mathematics (AREA)
• Computing Systems (AREA)
• Mathematical Physics (AREA)
• Computer Hardware Design (AREA)
• Logic Circuits (AREA)
• Lock And Its Accessories (AREA)
• Techniques For Improving Reliability Of Storages (AREA)
• Error Detection And Correction (AREA)

Applications Claiming Priority (3)

Publications (1)

Family

ID=9543078

Family Applications (1)

Country Status (6)

Families Citing this family (23)

Family Cites Families (24)

• 1999
  • 1999-03-09 FR FR9903027A patent/FR2790887B1/fr not_active Expired - Lifetime
• 2000
  • 2000-03-08 CA CA002367151A patent/CA2367151A1/fr not_active Abandoned
  • 2000-03-08 EP EP00910904A patent/EP1159783A1/de not_active Withdrawn
  • 2000-03-08 JP JP2000604527A patent/JP2002539543A/ja active Pending
  • 2000-03-08 WO PCT/FR2000/000573 patent/WO2000054410A1/fr not_active Application Discontinuation
  • 2000-03-08 US US09/936,032 patent/US7380192B1/en not_active Expired - Lifetime
• 2007
  • 2007-06-19 US US11/820,714 patent/US7565590B2/en not_active Expired - Fee Related
• 2009
  • 2009-06-17 US US12/456,477 patent/US7904772B2/en not_active Expired - Lifetime

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events