RU2017207C1 - Method for diagnostics of combination logical circuits - Google Patents

Abstract

FIELD: computer engineering. SUBSTANCE: method is based on measuring number of logical changes at output terminals of logical circuits. Coefficients of transparency of i-th input to output are determined for units corresponding to different errors of tested circuit. Key pattern of frequency X_i, which corresponds to resolution of monitor of logical changes, is applied to F_hep input of tested circuit. Key patterns of different frequencies are applied to other N-1 inputs. Duration of pulse half-period of maximal applied frequency should be at least twice as pulse half-period of frequency F_hep. Transparency coefficient is determined by the equation Z/2(N-1), where Z is number of time intervals where more than two logical periods were detected, N is number of input terminals of tested circuit. Comparison of transparency coefficients for circuits with different faults and transparency coefficient which was get from experiment provides necessary information. EFFECT: improved diagnostics. 5 dwg

Description

 The invention relates to automation and computer technology and can be used in monitoring and diagnosing failures of combinational logic circuits (CLS) and nodes.

 The aim of the invention is to increase the reliability of the control of CLS.

 Figure 1 presents the structural diagram of the connection of devices for the diagnostic procedure; figure 2 presents the connection diagram of the experimental device at different stages of the diagnostic procedure; in FIG. 3 presents the diagnosed circuit (object of experiment); figure 4 presents the timing diagrams obtained using a logical analyzer (LA) during the experimental connection of the device shown in fig.2b, a; figure 5 - timing diagrams obtained using LA during experimental connection of the device shown in figv,

 The devices shown in FIG. 1, with which the method is implemented, comprise an exposure unit (BV) 1, consisting of a generator (G) 2, a simple generator of rectangular pulses, which can be performed using simple logic elements using positive feedback, and with the use of standard microcircuits of the GG1 type, and divider counters 3, 4, 5 — groups of four-digit binary counters satisfying the frequency requirements of the signal counter 3 fed to the synchronization input and connected to I have accounts for increasing (decreasing) in such a way that the Overflow signal of the previous counter is fed to the subsequent synchronization input. The number of counters depends on the number of input poles of the experiment 6 object (OE) so that for every one to four input poles there is one counter. OE 6 can be any CLS, where X are the input poles of the OE, and Y is the output pole. LA 7 - a device for tracking digital signals (logger of logical differences), where L are the input channels of the aircraft. You can use any aircraft with the number of input channels of at least three, capable of tracking in asynchronous mode (i.e. with an internal clock frequency).

 Figure 3 presents the CLS, it is also OE 6, each element of which can be logical elements of the type AND, OR, AND-NOT, OR-NOT, NOT. Moreover, if OE 6 has several output poles Y, then the entire diagnostic procedure is divided into stages, each of which includes working with only one of these poles.

 An example of a specific diagnostic procedure.

 For the experiment, the OE 6 circuit is taken with a Const 1 malfunction at point B, shown in FIG. 3, executed on 155 series chips, where the OR-NOT 8 logic element is K155LE1, the I 9 logic element is K155LE1, the NAND logic element 10 - K155LA3, logic element I 11 - K155IL1, having four input poles X1-X4 and one output pole Y. BV 1 is made of a generator 2 - K531GG1 and two counters 3, 4 - K155IE7. As the LA 7, an experimental logical analysis system (SELA-1) is used, developed and created at the Research Institute of Automation and Electromechanics, Tomsk. The transparency coefficients are determined for models corresponding to various malfunctions of a given CLS.

 From discrete mathematics, the concept of a Boolean derivative or a Boolean difference is known. It is denoted as D (i / F) and is in mathematical form a record of sets on which the variable from the i-th input is transmitted to the output or, as they say, which “sense” the path from the i-th input to the output.

 The introduced integral diagnostic parameter, the transparency coefficient of the i-th input to the output, is related to the Boolean derivative by the following relation: it is equal to the ratio of unit sets of the Boolean derivative to the number of possible sets of 2 ** (N-1) supplied. They mean that the Boolean derivative takes the value 1 on the set of N-1 inputs, provided that the modulating signal of the i-th input at the output of the circuit appears.

 In the future, for CLS, vector, i.e. multidimensional, transparency coefficient, consisting of transparency coefficients of all inputs of the circuit.

 Since the function implemented by this circuit changes with a significant malfunction in the circuit, it is natural to expect that different transparency coefficients correspond to different functions. This assumption is confirmed experimentally.

 Based on the foregoing, it is proposed to use this characteristic to distinguish between significant faults in the circuit. The advantage of this approach is, on the one hand, that it is possible to calculate this transparency coefficient directly from the model, and on the other hand, this parameter is quite easy to obtain experimentally.

 The procedure for determining the transparency coefficient for models corresponding to various malfunctions of the investigated circuit and filling in the reference table is as follows.

 The orthogonal disjunctively normal form (DNF) is constructed according to the structural scheme, which is obtained by superposition of functions that describe each element as orthogonal DNF (DNF, in which all conjunctions are pairwise orthogonal).

F $\genfrac{}{}{0ex}{}{\text{1}}{\text{i}}$ VF

.Find the Boolean derivative of the function D (i / F) with respect to the ith input (and F can be either a healthy or a faulty function), according to the formula
D (i / F) = F $\genfrac{}{}{0ex}{}{\text{1}}{\text{i}}$ ⊕ F $\genfrac{}{}{0ex}{}{\text{o}}{\text{i}}$ =

F $\genfrac{}{}{0ex}{}{\text{1}}{\text{i}}$ Vf

.

 Using the properties of orthogonal DNF, the number of units of DNF of this Boolean derivative is determined.

 By dividing the number of units of the Boolean derivative by 2 ** (N-1) - the number of possible input sets at N-1 input poles determines the transparency coefficient of the i-th input to the output. Thus, the entire dimensional vector of transparency coefficients is determined.

 For each new fault introduced into the model, a multidimensional transparency coefficient is determined, which is entered in the reference table.

 A specific example of obtaining a transparency coefficient for the CLS model shown in Fig.3.

)=x₁(

)((

)V

)=

По формуле D(i/F)=

F $\genfrac{}{}{0ex}{}{\text{1}}{\text{i}}$ VF

определяют булевы производные D (i/F):
_D(1/F)=

;
_D(2/F)=

;
_D(3/F)=

;
_D(4/F)=

.Determine the DNF of this healthy function
F = BΛC = (x ₁ ΛA) Λ (

) = x ₁ (

) ((

) V

) =

By the formula D (i / F) =

F $\genfrac{}{}{0ex}{}{\text{1}}{\text{i}}$ Vf

define the Boolean derivatives D (i / F):
_{D (1 / F) =}

;
_{D (2 / F) =}

;
_{D (3 / F) =}

;
_{D (4 / F) =}

.

The number of units of each Boolean derivative is determined:
D ₁ (F) = 1;
D ₂ (F) = 1;
D ₃ (F) = 1;
D ₄ (F) = 1 and the vector transparency coefficient for a working scheme
[1/8, 1/8, 1/8, 1/8].

 This procedure can be automated and performed in a short period of time.

 The transparency coefficient is determined for the CLS model of the corresponding Const 1 fault at point B.

V

=(

)V

=x₂Vx₃V

..Determine DNF function
F ^′ =

V

= (

) V

= x ₂ Vx ₃ V

..

F $\genfrac{}{}{0ex}{}{\text{o}}{\text{i}}$ VF

:
D(1/F

=0;
D(2/F

=(

)=

x₄;
D(2/F

= x

=x₁x

V

x

;
D(3/F

=(

)=x

=x

x₄V

x₄;

.Boolean derivatives are determined by the formula
D (i / F) =

F $\genfrac{}{}{0ex}{}{\text{o}}{\text{i}}$ Vf

:
D (1 / F

= 0;
D (2 / F

= (

) =

x ₄ ;
D (2 / F

= x

= x ₁ x

V

x

;
D (3 / F

= (

) = x

= x

x ₄ V

x ₄ ;

.

Get transparency
[0, 1/4, 1/4, 1/4].

 Consider the equivalence classes of faults in this circuit (faults for which there are no separating tests).

 Faults Const 0 at point C, Const 0 at point X1, Const 1 at point X2, Const 1 at point X3, Const 1 at point X4, Const 0 at point F, Const 1 at point F, Const 0 at point A, Const 0 at point B there is no transparency coefficient [0, 0, 0, 0].

 Failures Const 1 - at point B, Const 1 - at point X1 corresponds to a transparency coefficient [0, 1/4, 1/4, 1/4], etc.

 Proceed to determine the transparency coefficient of the real scheme.

The internal frequency of the aircraft 7 is set equal to 1 MHz. The frequency of the generator 2 BV 1 set equal to 0.5 MHz, therefore, the signal F _m has a frequency of 0.25 MHz, which satisfies the condition for the resolution of the asynchronous mode of the aircraft to be at least 4 times less than the internal frequency of the aircraft and does not exceed the maximum operating frequency of the 155 series chips equal to 100 MHz. The signal F _m modulating frequency is removed from the first output of the counter 3, its frequency is 2 times less than the frequency of the generator 2, which is fed to the synchronization input of this counter. The remaining frequencies F ₁ are taken, starting from the first output of counter 4. Thus, the frequency of the signal F ₁ (1 = 1) is 16 times less than the frequency of the signal F _m , which allows analysis of the number of drops at the output pole of OE 6, equal to at least 16 , for a time equal to the half-period of the frequency signal F ₁ (1 = 1). The frequency of the signal F ₁ (1 = 1) does not have to be 16 times less than the frequency of the signal F _m . In order for the estimate of the number of differences to be reliable, the frequency of the signal F ₁ (1 = 1) must be at least 4 times less than the frequency of the signal F _m The condition for the resolution of the asynchronous mode of the aircraft to be at least 4 times less than the internal frequency of the aircraft follows from the following considerations.

 The essence of the asynchronous LA mode is to write a sequence of logical signals to the LA memory synchronously with the internal clock frequency of the SI aircraft, i.e., for each SI clock pulse, an information bit is recorded in the aircraft memory (one input channel). The recorded signal, as it were, is "cut" by the SR frequency, and the information that was at the input of the aircraft at the time of arrival of the SI clock pulse will be recorded. Obviously, the more often the input signal is “cut”, the more fully they have information about it. But too high a frequency of "cutting" requires large amounts of memory of the aircraft and the use of high-frequency elemental base to create the aircraft. To obtain the minimum complete reliable information about the tracked signal, which has the form of a meander, tracking is performed in asynchronous mode, the input signal must be "cut" 4 times - at the beginning and end of the zero half-cycle and at the beginning and end of a single half-cycle. Hence the condition is to satisfy the resolving power of the asynchronous mode of the aircraft, i.e. be at least 4 times less than the internal frequency of the aircraft.

Connect the experimental device, as shown in figa-d so that a complete exhaustive search of the signal F _m alternately to each of the four input poles OE 6, and the remaining three free poles of the signals F ₁ (1 = 1, 2, 3 ) Moreover, in each experiment, the signals F _m , F ₁ (1 = 1), Y are monitored by LA 7.

The signal F _{m is} fed to the input X1 OE 7 and L1 LA 7, the signal F _{1 is} fed to the input X2 OE 6 and L1 LA 7, the signal F ₂ is fed to the input pole X3 OE 6, the signal F ₃ is fed to the input pole X4 OE 6, signal Y - to the input pole L3 LA 7.

 Power is supplied to BV 1 and OE 6 and an experiment is conducted, the result of which is to obtain a quasi-temporal diagram in Fig. 4a.

An experimental device is connected as shown in FIG. 2b. The signal F _{m is} fed to the input X2 OE 6 and L1 LA 7, the signal F ₁ to the input X1 OE 6 and L1 LA 7, the signal F ₂ to the input pole X3 OE 6, the signal F ₃ to the input pole X4 OE 6, OE 6 Y signal - to the input pole L3 of the LA 7.

 Power is supplied to BV 1 and OE 6 and an experiment is conducted, the result of which is to obtain a quasi-temporal diagram in Fig. 4b.

An experimental device is connected as shown in FIG. 2c. The signal F _{m is} fed to the input X3 OE 6 and L1 LA 7, the signal F ₁ to the input X1 OE 6 and L1 LA 7, the signal F ₂ to the input pole X2 OE 6, the signal F ₃ to the input pole X4 OE 6, OE 6 Y signal - to the input pole L3 of the LA 7.

 Power is supplied to BV 1 and OE 6 and an experiment is conducted, the result of which is to obtain a quasi-temporal diagram in Fig. 5a.

An experimental device is connected as shown in FIG. 2g The signal F _{m is} fed to the input X4 OE 6 and L1 LA 7, the signal F ₁ to the input X1 OE 6 and L1 LA 7, the signal F ₂ to the input pole X2 OE 6, the signal F ₃ to the input pole X3 OE 6, OE 6 Y signal - to the input pole L3 of the LA 7.

 Power is supplied to BV 1 and OE 6 and an experiment is conducted, the result of which is to obtain a quasi-temporal diagram in FIG.

 The switching procedure when supplying signals from the BV can be automated.

Estimate the number of differences from zero to one at the output pole Y OE 6 (figa) for a period of time equal to the half-period of the frequency of the signal F ₁ . Fig.4b shows that at the pole Y for two of the eight time intervals equal to the half-period of the frequency of the signal F ₁ , there are more than two drops from zero to one, therefore, we can conclude that the transparency coefficient K ₂ = 1/4.

Estimate the number of differences from zero to one at the output pole Y OE 6 (figa) for a period of time equal to the half-period of the frequency of the signal F ₁ . From Fig. 5a, it can be seen that, at the Y pole, for two of the eight time intervals equal to the half-period of the signal frequency F ₁ , there are more than two drops from zero to one, therefore, it can be concluded that the transparency coefficient K ₃ = 1/4.

Estimate the number of differences from zero to one at the output pole Y OE 6 (Fig.5B) for a period of time equal to the half-period of the signal frequency F ₁ . From FIG. 5b it can be seen that at the Y pole for two out of eight time intervals equal to the half-period of the signal frequency F ₁ , there are more than two drops from zero to unity, therefore, we can conclude that the transparency coefficient K ₄ = 1/4.

 Complex or vector transparency coefficient K = [0, 1/4, 1/4, 1/4].

 It is compared with transparency coefficients for fault models; it coincides with the transparency coefficient corresponding to the class of equivalent faults, which also contains the Const 1 fault introduced in the real circuit at point B.

 Compared with the prototype, the reliability of the control and diagnosis of CLS carried out by the proposed method is increased due to the fact that in order to obtain a reference library using other diagnostic methods (implemented by other diagnostic parameters), a large number of experiments should be carried out, which is due to an increase in the probability of error when large volumes of measurements during the preparatory procedure leads to a decrease in the reliability of subsequent diagnostic procedures, and the proposal The lively method allows, by making minimal calculations, to obtain the diagnosis parameter theoretically from the CLS structure, which eliminates the experimental error and error and allows reducing the time of the preparatory diagnostic procedure (filling in the reference table) for a set of exclusion from this procedure of mechanical operations and human participation.

Claims (1)

DIAGNOSTIC METHOD OF COMBINATION LOGIC SCHEMES, based on measuring the number of logical differences at the output poles of the investigated circuit, characterized in that, in order to increase the reliability of the analysis, for the models corresponding to various faults of the diagnosed combinational logic circuit, the transparency coefficients of the i-th input to the output are determined ( i =

),, each of which is equal to the ratio of unit sets of the Boolean derivative to the number of possible input poles of this set circuit supplied to N-1, the number of which is determined by the formula 2 ** (N-1), fed to the input X _{i of the} studied combinational logic circuit meander frequency F _m , which has a value that satisfies the resolution of the logger of the logical differences and does not exceed the permissible operating frequency of the circuit under study, and on the N-1 remaining inputs - meanders of frequencies, the largest of which F _l (l = 1) must have a half-cycle duration and the pulse is not less than 2 times the duration of the frequency period F _m , and the remaining frequencies Fl (l =

) must be a multiple of the largest, while the output signal is fixed by the logger of logical differences, the transparency coefficient of a real logic circuit is determined from the relation

,,
where Z is the number of time intervals for the output signal Y, equal to the half-period of the signal Fl (l = 1), in which more than two logical differences are detected, N is the number of input poles of the circuit under study,
and diagnostics is carried out by comparing the transparency coefficients obtained for circuit models with various faults with the transparency coefficient obtained experimentally.

Images