JPH0347626B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a circuit for removing noise generated at the time of switching the reproduction head of an audio reproduction circuit that records frequency modulated (FM) audio signals and reproduces them alternately on a plurality of heads. .

Conventionally, when recording and reproducing FM modulated audio signals using, for example, a helical scan type magnetic tape recording and reproducing device, large-amplitude noise is generated because the FM carrier wave becomes discontinuous at the time of switching the reproduction track. During this time, methods have been considered, such as blocking the demodulated audio signal or maintaining the signal level just before the noise occurred.

FIG. 1 is an audio signal waveform diagram showing the operation of a noise removal circuit using the previous value holding method. In Figure 1, a is the reproduced audio FM audio waveform, and b is the FM audio waveform.
The demodulated audio signal waveform, c is the previous value holding instruction signal, and d is the audio signal waveform from which noise has been removed based on the previous value holding instruction signal. At the time of switching the reproduction head shown in FIG. 1a, the FM carrier wave becomes discontinuous, so that large-amplitude noise as shown in FIG. On the other hand, since the time at which the reproduction head is switched is known in advance, it is easy to generate a pulse indicating the period of noise generation as shown in c, and if the signal shown in b is held at its previous value with the pulse shown in c, , d from which the noise has been removed can be obtained. However, simply holding the signal level just before the noise occurs will cause discontinuity in the signal level at the end of the hold, and especially in the case of a high frequency signal, the difference from the original signal will become large and a new signal will be generated. However, the noise could not be completely suppressed and was harsh on the ears.

However, if we focus on the fact that the noise generation period at the time of switching the playback track has a substantially constant time width, we can detect the signal level before and after the noise generation, and calculate the A first-order approximation that linearly interpolates the signal levels before and after noise generation is possible according to the difference.

FIG. 2 shows a signal waveform from which noise has been removed by zero-order approximation (previous value retention) for a time width t, and FIG. 3 shows a signal waveform from which noise has been removed by first-order approximation. Although large amplitude noise can be removed by such means, new noise will be generated due to approximation errors. Approximation error noise can be considered as the difference between the waveform integral value and the original waveform. That is, the approximation error noise can be expressed as the difference between the waveform integral value (DC component) for a certain period including the reproduction track switching portion of the original audio signal and the waveform integral value for the same period of the waveform after approximation. Based on this idea, Figure 4 shows the magnitude of approximation error noise due to differences in approximation methods when the original signal is a sine wave and interpolation is performed over a fixed time width of 50 μs, using zero-order approximation as a reference. It is.
As is clear from this figure, by performing the first-order approximation, the signal frequency can be reduced compared to the zero-order approximation.
It can be confirmed that this has an extremely large noise reduction effect of approximately 26 dB at 1 kHz and approximately 20 dB at 2 kHz.

Since the time width of the above noise is determined by the playback signal passband up to the final demodulated output, the noise time width τ that appears every time the playback track is switched
is always constant. Therefore, it is known in advance that the timing at which the noise occurs is at the time when the reproduction track is switched, and the timing at which the noise ends is delayed by τ from the time at which the reproduction track is switched. From this, first-order approximation is possible with a simple configuration using a demodulated audio signal delayed by a certain time τ. Conversely, if the period of noise generation is not constant, perfect first-order approximation is extremely difficult.

FIG. 5 shows a circuit configuration diagram of the first-order approximation method, and FIG. 6 shows a signal waveform diagram of each part in FIG. A demodulated audio signal as shown in FIG. 6e is inputted from an input terminal 1 and supplied to a delay circuit 2 with a delay time .tau. and a switch circuit 5. The audio signal as shown in FIG. Switch circuits 3, 4, and 5 are normally conductive. On the other hand, the reproduction track switching signal shown in FIG. 6g is inputted from the input terminal 13, and the monostable multivibrator 14 generates a pulse having a pulse width τ shown in FIG. 6h. The output pulse period of this monostable multivibrator 14 coincides with the period of noise occurrence in the input audio signal shown in FIG. 6e. Furthermore, in monostable multivibrator 15, monostable multivibrator 1
Using the output of 4 as an input, a pulse having the same pulse width τ as shown in FIG. 6i is generated. The output pulse period of this monostable multivibrator 15 coincides with the period of noise generation in the output signal of the delay circuit 2 shown in FIG. 6f. The output of the monostable multivibrator 15 is supplied to the switch circuits 3, 4, and 5 to control the opening and closing of the switches, and the switch circuits 3, 4, and 5 are cut off at the same time as the pulse shown in FIG. 6i rises. Therefore, as shown in FIG. 6j, k, and l, the voltage at the moment the capacitors 8, 7, and 6 are cut off is held on the output side of the switch circuit.
At that time, what is held in the capacitor 8 is shown in Fig. 6j.
As shown in FIG. 6, this is the voltage immediately after the noise ends, and what is held in the capacitor 7 is the voltage immediately before the noise occurs, as shown in FIG. 6k. The difference between the voltages of the capacitors 7 and 8 held in this manner is detected by a subtraction circuit 9, and the output current of the variable current source 10 is controlled according to the difference. For example, if the voltage held in capacitor 7 is higher than the voltage held in capacitor 8, variable current source 1
0, a current corresponding to the potential difference is output, and conversely, when the voltage held in the capacitor 7 is lower than the voltage held in the capacitor 8, a current corresponding to the potential difference flows into the variable current source 10. Become.

On the other hand, the switch circuit 11 is normally cut off,
Since the monostable multivibrator 15 operates to conduct only during the period when the output pulse (FIG. 6i) exists, the charge held in the capacitor 6 is changed depending on the current of the variable current source 10 as described above. For example, the battery is discharged or charged depending on the difference between the voltage immediately before the noise occurs and the voltage immediately after the noise ends. Therefore, by adjusting the correspondence between the potential difference between capacitors 7 and 8 and the output current of variable current source 10, the voltage of capacitor 6, that is, the voltage obtained at audio output terminal 12, can be adjusted as shown in FIG. It becomes possible to interpolate the period before and after the noise generation period using first-order approximation.

Next, a specific circuit example of the subtraction circuit 9 and the variable current source 10 is shown in FIG. 16 and 17 are output current control signal input terminals, and the holding voltages of the capacitors 7 and 8 in FIG. 5 correspond to them. Transistor 19, 2
0 constitutes a differential amplifier, and the collector current I ₁ of the transistor 19 and the collector current of the transistor 2 depend on the base potential difference of the respective transistors 19 and 20.
A collector current I ₂ of 0 flows. Here, if the base currents of the transistors 19 and 20 are ignored, the current flowing through the constant current source 18 has a relationship of I ₀ =I ₁ +I ₂ (=constant). Further, the transistors 21 and 22 constitute a current mirror circuit, and are set so that a current almost equal to the collector current of the transistor 19 flows through the collector of the transistor 22. Therefore, in the case of I ₁ > I ₂ , the difference current I ₁ −I ₂ between the collector current of the transistor 22 and the collector current of the transistor 20 flows out from the current output terminal 23, and when I ₁ <I ₂ In this case, on the contrary, a current of I ₁ −I ₂ will flow from the terminal 23. That is, the current flowing out or flowing into the terminal 23 can be controlled according to the potential difference between the terminals 16 and 17.

As mentioned above, by taking advantage of the fact that the time width of the noise that occurs at the time of switching the playback track is always constant, and by linearly interpolating the signal level before and after the noise generation period, the previous value is maintained. It is possible to suppress extremely large approximation error noise with a simpler circuit configuration than that of the conventional method.

However, when magnetically recording and reproducing by superimposing an FM modulated video signal, due to the nonlinearity of the magnetic characteristics, the FM demodulated reproduced audio signal contains periodicity related to the horizontal synchronization signal period of the reproduced video signal. Noise occurs.

When such periodic noise is removed by an LPF, the band is limited by the LPF, so the large amplitude noise that occurs at the time of switching the reproduction track described above will generate a transient corresponding to the cut-off time constant of the LPF. However, the period during which the noise occurs becomes longer. Therefore, the interpolation period of the above-mentioned first-order approximation circuit has to be lengthened, and the difference from the original audio signal increases, resulting in a significant deterioration of the sound quality of the reproduced audio signal.

In addition, in order to prevent the deterioration of the playback sound quality as described above, the transient width of the noise that occurs when switching the playback track is shortened without band limiting, and the large amplitude when switching the playback track is detected using the first-order approximation circuit as described above. One possibility is to remove noise. In this case, during silent periods, pulse noise occurs at the playback track switching period, which is unpleasant to the ears. This pulse noise generation will be explained using FIG. 8.

FIG. 8 shows that periodic noise related to the horizontal synchronization signal period of the reproduced video signal is present in the reproduced audio signal,
In addition, it is a reproduced signal waveform showing the operation of a conventional noise removal circuit when there is no audio signal. In FIG. 8, m is a reproduction track switching signal, n is a periodic noise waveform related to the period of the FM demodulated reproduction horizontal synchronization signal, o is a first-order approximation instruction signal, and p is a signal based on the first-order approximation instruction signal o. The approximated noise waveform, q is 1
This is a waveform obtained by de-emphasizing the approximated noise waveform.

During a certain period from the time when the playback track is switched as shown in FIG.
Large-amplitude noise as shown in Figure 2 is generated. A signal indicating a certain period during which this large-amplitude noise occurs is o.

Here, if the noise generation period is linearly interpolated using the signal shown in o, the waveform will be as shown in Figure 8 p, so the period of the above-mentioned reproduced horizontal synchronization signal outside the band can be used as the fundamental wave using a de-emphasis circuit LPF or the like. Even after removing the periodic noise, pulse noise as shown in FIG. 8Q occurs due to the error portion shaded in FIG. 8P.

Therefore, when the above-mentioned periodic noise exists in the reproduced audio signal, the conventional noise removal circuit does not deteriorate the reproduced sound quality and eliminates the pulse noise generated at the time of switching the reproduction track without deteriorating the reproduction sound quality and without the pulse noise occurring at the reproduction track switching period. It was not possible to remove large amplitude noise.

SUMMARY OF THE INVENTION An object of the present invention is to provide a noise removal circuit that eliminates the above-described drawbacks of the prior art and can reduce noise to a level that poses no problem in practice.

The present invention prevents the generation of pulse noise in the reproduction track switching period by controlling the reproduction track switching point using a horizontal synchronization signal of the reproduction video signal so that the integral value of the periodic noise becomes zero. , and is characterized in that large amplitude noise at the time of switching reproduction tracks can be removed without deteriorating the reproduction sound quality.

The cause of this noise, which can be called erroneous interpolation noise, is due to the imbalance of positive and negative energy on the waveform caused by first-order approximation interpolation of periodic noise, in other words, the integral value does not become zero. In other words, periodic noise can be regarded as a sine wave with a frequency of approximately _H (horizontal scanning frequency = 15.734kHz), but
As shown in Fig. 9, if linear approximation interpolation is performed from point Q for T seconds, an imbalance of positive and negative energy will occur on the waveform corresponding to the area difference between the shaded area A and the shaded area B, and the LPF etc. will cause a Even if the sexual noise is attenuated, the noise remains. Therefore, in order to prevent the occurrence of such erroneous interpolation noise, it is best to control the interpolation start point so that the difference in area between the shaded areas A and B becomes zero so that the positive and negative energy imbalance on the waveform does not occur. . Now, as shown in Fig. 9, the interpolation start point Q point is set to θ with respect to the S point.
If the phase is 360 degrees=1/ _H seconds≈63.5 μs, the interpolation starting point where the area difference between the shaded portions A and B becomes zero is determined by the following equation.

θ＝tan ^-1 cos360 _H T＋π _H Tsin360 _H T−1/sin36
0 _H T−π _H Tcos360 _H T−π _H T …(1) This formula gives the phase θ at the start of interpolation at which the false interpolation noise is minimized for the interpolation period T, and T and θ The relationship is as shown in FIG. Therefore, it can be seen that, for example, when T=63.5 μs, the erroneous interpolation noise is the smallest at θ=0 degrees, that is, at the point where the Q point and the S point coincide. In this way, by controlling the interpolation start time so that the positive and negative energy imbalance on the waveform, that is, the integral value becomes zero, it is possible to effectively prevent the occurrence of erroneous interpolation noise.

FIG. 11 shows an embodiment of a noise removal circuit in which the present invention is applied to a magnetic recording/reproducing apparatus, particularly a rotating head type magnetic tape recording/reproducing apparatus. 24 is a magnetic recording medium, 25 and 26 are playback heads, 27 and 28
is a preamplifier, 29 is a changeover switch circuit for switching reproduction tracks, 30 is a reproduction signal output terminal, 3
1 is a band pass filter (BPF) for extracting FM audio signal components, 32 is an FM demodulator, 33 is an LPF for removing FM carrier waves, 34 is a de-emphasis circuit, 3
5 is a playback audio signal output terminal, 36 is a playback horizontal synchronization signal input terminal, 37 is a monostable multivibrator, 38 is a playback track switching signal input terminal, 39
40 is a D-type flip-flop circuit, and 40 is a reproduction track switching pulse generation circuit that generates a pulse synchronized with the switching time based on the output signal of the D-type flip-flop 39. Note that the circuits 2 to 11, 14, and 15 are the same as the circuits with the same numbers in FIG.

FIG. 12 is a signal waveform diagram of each part of the circuit shown in FIG. 11. Here, r is a playback track switching signal input from the input terminal 38, and s is a reproduction track switching signal input from the input terminal 36.
t is the output signal waveform of the monostable multivibrator 37, and u is a reproduction track switching signal that is controlled so that the integral value of the periodic noise output from the flip-flop circuit 39 becomes zero. , v is the signal waveform obtained by removing the FM carrier wave from the output signal of the FM demodulator 32 by the LPF 33, w is the signal obtained by delaying the output signal v of the LPF 33 by the delay circuit 2, and x is the switch circuit 3 subjected to linear approximation interpolation. The output signal y is the output signal of the de-emphasis circuit 34.

In addition, h and i are respectively h and i in FIG.
This is the same waveform as shown in .

A reproduction signal extracted from the magnetic recording medium 24 by a reproduction head 25 or 26 passes through a preamplifier 27 or 28, and then a reproduction track changeover switch circuit 29 switches the reproduction track. Changeover switch circuit 29
The BPF 31 extracts only the FM audio signal component from the output signal, and the FM demodulator 32 demodulates it. However, as mentioned above, there is a discontinuous part in the FM audio signal at the time of switching the playback track, so
When FM demodulated, large amplitude noise is generated as shown in Figure 12v. Here, using the fact that the period of the periodic noise is related to the reproduced horizontal synchronization signal,
By controlling the reproduction track switching time point so that the integral value of the periodic noise becomes zero, generation of pulse noise in the reproduction track switching cycle is prevented. That is, the leading edge (or trailing edge) of the reproduced horizontal synchronizing signal shown in FIG. The pulse width of the 37 output (FIG. 12 t) is set, and the phase at the start of interpolation is controlled as a result. Since the interpolation width itself is almost the same as the noise generation period and is uniquely determined, the above formula
(1) As is clearer, the interpolation start point is automatically determined. At this time, periodic noise occurs in a phase related to the horizontal synchronization signal of the video signal, so
As described above, the monostable multivibrator 3 uses the leading edge or trailing edge of the reproduced horizontal synchronization signal as the trigger input.
The pulse width of the 7 output may be set so that the trailing edge of the pulse becomes the interpolation start point. If the pulse width of the monostable multivibrator 37 output is set in this way, the monostable multivibrator 14, 15, which ultimately determines the interpolation timing and interpolation time width,
The operation timing of is controlled so that the integral value of periodic noise becomes zero. That is, the first
The pulse shown in Figure 2 t is used as the clock input, and the 12th
The D-type flip-flop circuit 39 is operated using the reproduction track switching signal shown in FIG. Creates a playback track switching synchronization signal. Based on the reproduction track switching of u obtained in this way,
The reproduction track changeover switch circuit 29 and the first-order approximation interpolation circuit shown in FIG. 5 (the explanation is omitted here)
By operating this, it is possible to always make the integral value of the periodic noise in the signal subjected to first-order interpolation, such as the waveform shown in FIG. 12x, always zero.
At this time, as described in the explanations in FIGS. 5 and 6, the width of the output pulses (h, I in FIG. 12) of the monostable multivibrators 14 and 15 is the interpolation time width, in other words, the width of the output pulse is approximately equal to the noise generation. Set the time width to correspond to the period. The thus obtained signal whose waveform is shown in FIG. Therefore, the waveform outputted from the output terminal 35 becomes as shown in FIG. This can be almost completely prevented, and there is no deterioration in playback quality. In Fig. 5, the signal levels before and after the noise generation period are detected simultaneously using the delayed signal and the non-delayed signal, but the signal levels before and after the noise generation period of the non-delayed signal are detected at separate timings. It is clear that a method of doing so is also possible. Furthermore, in addition to interpolation using analog means such as charging and discharging a capacitor, methods such as digitally calculating and approximating the interpolated voltage level from the detected signal level difference do not depart from the spirit of the present invention. Furthermore, it is clear that the delay time of the delay circuit 2 only needs to be at least longer than the noise generation period, and does not need to match the noise generation period.

As explained above, if the present invention is used, FM
Even if there is periodic noise associated with the reproduction horizontal synchronization signal synchronization in the reproduction signal of the modulated audio signal, the large amplitude noise caused by the discontinuity of the FM signal at the time of reproduction track switching is suppressed by the reproduction track switching period. It can be almost completely removed with a simple circuit configuration without any pulse noise or deterioration of reproduced sound quality, and the effect is great.

[Brief explanation of drawings]

FIG. 1 is a signal waveform diagram illustrating noise removal operation by holding the previous value, FIG. 2 is a diagram showing an example of an interpolated waveform by the previous value holding method, and FIG. 3 is a diagram showing an example of the interpolated waveform by the linear approximation method. Fig. 4 is a characteristic diagram showing the noise suppression effect by the first-order approximation method, Fig. 5 is a circuit configuration diagram showing an example of a noise removal circuit using the first-order approximation method, and Fig. 6
The figure is a signal waveform diagram of each part in Figure 5, Figure 7 is a circuit diagram showing a specific example of the subtraction circuit 9 and variable current source 10 in Figure 5, and Figure 8 is a process of generating incorrect interpolation noise due to periodic noise. FIG. 9 is a signal waveform diagram explaining the false interpolation noise removal method, FIG. 10 is a characteristic diagram of the false interpolation noise elimination method, and FIG. 11 is a signal waveform diagram explaining the false interpolation noise elimination method. FIG. 12 is a circuit configuration diagram of the signal reproducing device, and FIG. 12 is a signal waveform diagram of each part of FIG. 11. 2... Delay circuit, 3, 4, 5, 11... Switch circuit, 9... Subtraction circuit, 10... Variable current source,
14, 15, 37...monostable multivibrator, 39...D type flip-flop.

Claims (1)

[Claims] 1. To reproduce a magnetic tape recorded by frequency-multiplexing a frequency-modulated video signal and a frequency-modulated audio signal on the lower band side than the recording band of the frequency-modulated video signal by sequentially switching a plurality of heads. , a periodic noise with a period related to the period of the reproduced horizontal synchronization signal, and
In a magnetic recording signal reproducing apparatus, track switching noise that occurs in a demodulated audio signal during a predetermined period immediately after switching the reproduction track due to discontinuity of the frequency modulated audio signal carrier wave that occurs each time the reproduction track is switched, a detection circuit that detects the signal level difference before and after the period; a delay circuit that receives the demodulated audio signal and has a delay time of at least the predetermined period; and a detection circuit that detects the signal level difference before and after the period; An interpolation circuit linearly interpolates the signal level immediately before the predetermined period and the signal level immediately after the predetermined period in the delay circuit output signal, and the reproduced horizontal synchronization signal and the track switching period signal are input, and the phase of these input signals is and a track switching signal generation circuit that outputs a signal whose phase is set to a specific relationship as a track switching signal. An audio signal reproducing device equipped with a noise removal circuit, characterized in that the phase is set so that the integral value of the periodic noise becomes zero during a predetermined period immediately after track switching.