STUDY OF THE NOISE IMMUNITY OF A TWO-CHANNEL ELECTROMAGNETIC FLAW DETECTOR

V. K. ZHUKOV, B. A. DOBNER

(Presented by the joint scientific and technical seminar of the Department of Information and Measurement Technology and the Sector of Electromagnetic Flaw Detection of the Scientific Research Institute of Electronic Introscopy)

The question of the noise immunity of electromagnetic flaw-detection equipment is currently becoming of primary importance in connection with the introduction of automatic flaw detectors. In the absence of special measures to increase noise immunity, even a single interference pulse can cause the false rejection of a good product.

The effect of interference on the signal-transmission channel is the most dangerous. Mechanical interference in the form of periodic vibrations and individual impacts of the item being inspected acts on the flaw-detector sensor. This causes the appearance of a false signal in the receiving part of the flaw detector and triggers its automatic circuitry. Random electrical interference in the form of spikes of fluctuation noise, as well as 50 Hz pickup (mains interference), acts mainly on the input circuits of the amplifiers. Voltage surges in the flaw detector’s power supply, caused by switching on and off powerful technological equipment in the shop, are also dangerous.

Fig. 1. Block diagram of the flaw detector

Figure 1 shows a block diagram of an electromagnetic flaw detector for monitoring the surface quality of turbine-blade blanks, which has increased noise immunity against the above-listed disturbances.

The flaw detector uses two through-type eddy-current self-comparison sensors \((D_1, D_2)\), having a common magnetizing winding and two pairs of measuring windings located at a quite definite distance from one another and each connected to its own channel. The operation of such a circuit is explained with the aid of the diagrams (Fig. 2, a, b).

Fig. 2. Timing diagrams explaining the operation of the instrument: a — under the action of the useful signal; b — under the simultaneous action of interference on both channels

If the inspected item, passing through the sensors at a quite definite speed, has a defect, then at the outputs of both channels pulses are obtained, shifted by the time

\[ t_3=\frac{l}{v}, \]

where \(l\) is the distance between the measuring sections of the sensors;
\(v\) is the speed of motion of the defect.

At the output of the flaw detector a time selector of pulses is used, consisting of three monostable multivibrators ZhM1—ZhM3 and a coincidence circuit CC.

The appearance of a pulse in the first channel will cause ZhM1 to operate; it produces a pulse of duration \(\tau_1 \leq t_3\). After a time \(t_3\), the monostable multivibrator ZhM2 is triggered in the second channel; it produces a pulse of duration \(\tau_2 = 0.5\tau_1\). This pulse is applied to one of the inputs of the coincidence circuit CC. On the trailing edge of the ZhM1 pulse, the monostable multivibrator ZhM3 is started; it produces a pulse of duration \(\tau_3 \geq t_3 - \tau_1 + \tau_2\), applied to the second input of the coincidence circuit. Thus, two pulses will act simultaneously at both inputs of the coincidence circuit. In this case, a rectangular pulse will appear at the output of the coincidence circuit, which, entering the automation unit, will cause the rejection of the given item.

In the case where interference acts simultaneously through two channels, two pulses will appear simultaneously at the outputs of the pulse selector (Fig. 2, b). As a result, the pulses arriving at the coincidence circuit turn out to be separated in time, and the coincidence circuit does not produce a signal. In this case no rejection will occur.

The study of noise immunity was carried out on the block diagram (Fig. 3). To determine the influence of changes in the supply voltage, jumps in the supply voltage were artificially created in the flaw detector, with provision for both changes in the magnitude of the jumps and changes in their frequency. Figure 4, a presents the probability distribution of false operation of the flaw detector versus the magnitude of the relative change in supply voltage, with the total number of jumps \(n = 100\). The graph shows that appreciable over-rejection appears when the relative change in supply voltage is 30%.

Figure 4, b gives the probability distribution of false operation of the flaw detector versus the frequency of jumps in the supply voltage.

To study the influence of industrial-frequency pickup, the flaw-detector sensor was placed in a solenoid powered through an autotransformer from a 50 Hz mains supply. Owing to the use of self-comparison sensors,

Fig. 3. Test block diagram

Fig. 4. Distribution of the probability of false triggering of the flaw detector:
a — as a function of the magnitude of jumps in the supply voltage;
b — as a function of the frequency of jumps in the supply voltage

Owing to the selective properties of the carrier-frequency amplifiers CFA and the threshold devices TD, at the inputs of the pulse selector, under the action of a single pickup (without a useful signal), triggering of the flaw detector is not observed over a comparatively wide range of pickup-field intensities (0–25 oersteds).

Figure 5 shows the dependence of the number of triggerings of a two-channel flaw detector on the magnitude of the interference (curve 1), and the same dependence for one channel (curve 2), under simultaneous action on the measur—

Fig. 5. Dependence of the number of false triggerings of the flaw detector on the magnitude of the pickup

measuring circuit of the signal and interference. In this case, the useful signal was simulated by means of a pulse generator and a polarized relay (Fig. 3), and the number of operations was recorded by the counting device PP.

The effect of vibrations was studied by means of a special vibrating rocker arm, to which a turbine-blade blank placed in the sensor was attached. The vibration amplitude was varied by the appropriate selection of eccentrics, which caused the rocker arm to execute reciprocating motion. The vibration frequency was varied by regulating the number of revolutions of the motor, on whose shaft the eccentric was mounted. Figure 6 shows the probability distribution

Fig. 6. Probability distribution of false triggering of the flaw detector as a function of the vibration frequency of the item

of false triggering of the two-channel flaw detector as a function of vibration frequency for three values of the vibration amplitude. From the graph it may be concluded that, when the vibration period does not exceed the duration of the pulses of the waiting multivibrators, the probability of false triggering of the two-channel flaw detector is considerably lower than that of a single-channel flaw detector, which justifies the use of the second channel.

The investigations carried out confirmed the assumed substantial reduction in false triggerings of the flaw detector under the action of various kinds of interference.