JPH0213368B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention is suitable for, for example, recording color video signals in the form of digital signals at high density.
This invention relates to a rotating magnetic head device for a VTR.

First, we will consider what is required of a VTR when recording digital video signals at high density.

When transmitting a digital signal in binary, the S/N of the transmission path (the signal is peak-to-peak,
If the peak value (noise is the effective value) is 20 dB or more, the bit error rate is approximately 1×10 ^-7
It is known that the allowable error rate in digital transmission of video signals is approximately 1×10 ^-7 .

Therefore, in a digital VTR, the S/N of the digital signal from the reproduction equalizer is approximately
Must be 20dB or higher.

In order to achieve high-density recording in a digital VTR, it is necessary to increase the number of recording bits per unit area of tape.

If the number of recording bits per unit area is S, then the recording density S is: S=L・T L: density (number of recording bits per unit length in the track length direction) T: track density ( (number of tracks per unit length in the track width direction).

As for the fiber density L, generally speaking, increasing the recording density along the track results in shorter wavelength recording. If the magnetic layer of the tape is sufficiently thick, the number of magnetic particles that provide magnetic flux to the playback head is approximately proportional to the square of the wavelength.The signal voltage of the playback head is proportional to the number of magnetic particles, and the noise voltage is proportional to the square root of the number of magnetic particles. Therefore, if the only noise source is the tape, the S/N of the reproduced digital signal is proportional to the wavelength. The S/N of the amplifier system is also approximately proportional to the wavelength. Therefore, if the track width is constant, the longer the recording wavelength (if the relative speed of the head and tape is constant, the narrower the frequency band), the better the S/N will be.

Also, regarding the track density T, if the track width is narrowed, the signal voltage of the playback head and the tape noise power will decrease in proportion to the track width.If all noise is generated on the tape, the noise voltage will be Since it is proportional to the square root of the width, the S/ of the reproduced digital signal is
N is proportional to the square root of the track width. The inductance of the reproducing head is approximately proportional to the head piece thickness (track width). If the inductance of the reproducing head is constant, the number of turns is inversely proportional to the square root of the track width. This winding Since the magnetic flux interlinking with the line is proportional to the track width, the voltage induced in the read head is proportional to the square root of the track width.If the inductance of the read head is constant, the noise generated by the head amplifier is constant. Therefore, if the only noise source is the head amplifier, the S/
N is proportional to the square root of the track width. Therefore, if tape noise and amplifier noise are independent, the S/N of the reproduced digital signal is
is proportional to the square root of the track width.

From the above, in order to increase the recording density S, it is necessary to narrow the track width to make the track density T as high as possible, to keep the recording wavelength as low as possible, and to avoid increasing the linear density L unnecessarily.

If the track density T is increased in order to increase the recording density S, the following two problems arise.

As the guard band between tracks becomes narrower, crosstalk due to magnetic flux leakage from adjacent tracks increases.As the track width becomes narrower, tracking during playback becomes difficult.Firstly, the crosstalk from adjacent tracks in the term , in Figure 1, (1) is the playback head, (2)
When the track is E: the level of the main signal E _c : the level of crosstalk λ: the wavelength of the signal W: the track width of head 1 x: the width of the guard band ΔW: the blur of the magnetized region (track) Crosstalk Ct is Ct=20logEc/E =A+Bx/λ [dB] x≫ΔW ΔW0.67λ b=6.9 B=-60 This is an experimental approximate value.

Therefore, for example, if W = 40 μm and x = 20 μm, the relative speed between the head and tape is
In the case of 25.59 m/sec (in the case of SMPTE "C" format), the theoretical frequency characteristic of crosstalk according to the above equation is as shown by curve _C1 in FIG.

Next, regarding the tracking accuracy, as the track width becomes narrower, track deviation is more likely to occur in the reproducing head, and crosstalk from adjacent tracks becomes even larger. Although this tracking accuracy can be improved by various servo techniques, it is basically determined by mechanical accuracy and is a major factor that hinders the improvement of recording density.

Therefore, as long as the normal recording method is adopted based on the above-mentioned section, the required minimum width of the track and guard band is determined, and higher density recording cannot be performed.

Therefore, when recording a digital video signal, the digital signal converted from the video signal is divided into multiple channels, and the signal of each channel is recorded as a multitrack, and adjacent tracks are in contact with each other, and A possible method is to record the azimuth angles so that they are different from each other.

That is, if the azimuth angle between the reproduction head 1 and the track 2 is θ, then the azimuth loss La is: La=20log|sin(πw/λtanθ/(πw/λ)tanθ|
[dB]. Therefore, if the relative speed between the head 1 and the tape is constant, the higher the frequency, the greater the azimuth loss La.

As an example, as shown in Fig. 3A, when the crosstalk from the adjacent track is actually measured when the track width W is 60 μm, there is no guard band, and the azimuth angle θ is 14°, the curve C ₂ in Fig. 2 is obtained. As shown in Figure 3B, the track width W
When the crosstalk from the adjacent track was actually measured when the width of the guard band was 40 μm, the width x of the guard band was 20 μm, and there was no azimuth angle θ, it was as shown in curve _C3 in Figure 2 (the relative velocity between the head and tape teeth,
(same as for curve C ₁ ).

According to these measurement results, in the case of azimuth recording (curve C ₂ ), in the low frequency range of approximately 2 MHz or less, crosstalk from adjacent tracks decreases as the frequency increases due to azimuth loss. . However, in high frequencies above 2 MHz, crosstalk becomes large due to coupling with other heads and interference from other channels, that is, crosstalk between channels.

Also, a normal record with a guard band (curve C ₃ )
In the case of , in the low range where the frequency is approximately 200 kHz or less, the theoretical value of crosstalk in the above term is curve C ₁
The crosstalk corresponds to , and at higher frequencies, it becomes inter-channel crosstalk.

Comparing the two, the frequency is approximately
At low frequencies below 1MHz, the crosstalk of azimuth recording is 4 to 6 dB lower than the crosstalk of normal recording.
It's just a little more, and in the higher ranges it's the same.

Therefore, when the track pitch is the same, there is no significant difference in inter-track crosstalk between azimuth recording and normal recording.

However, in terms of the signal reproduction level, if the track pitch is the same, if azimuth recording is used, the reproduction level will be increased by the guard band width, and the S/N will be advantageous. For example, in the example shown in Figure 3, However, azimuth recording has a better S/N.

Also, if a tracking error occurs during playback, for example, as shown in Figure 3, if head 1 deviates by 1/2 of the track pitch, in the case of azimuth recording (Figure 3A), head 1 will be Even when a track is scanned, the deterioration of the S/N of the reproduced signal due to azimuth loss is reduced. However, during normal recording (FIG. 3B), the S/N is 0 dB.

Therefore, azimuth recording is also advantageous for tracking errors. Alternatively, if the deterioration in S/N due to tracking errors is the same as that in normal recording, the track pitch can be reduced in azimuth recording, that is, high-density recording can be achieved.

However, when performing azimuth recording, if the azimuth angle θ is too large, the effective recording wavelength λ _e
is small as λ _e =λcosθ, which violates the above term and becomes susceptible to spacing loss and gear loss. Therefore, the azimuth angle θ cannot be made too large. According to experiments, it was found that the angle should be selected between 10° and 30°.

From the above, for high-density recording and tracking, guardbandless azimuth recording using an appropriate azimuth angle is suitable.

However, when the recording frequency is low, the azimuth loss becomes small, such as the curve in Figure 2.
Track-to-track crosstalk increases as frequency decreases, as shown as _C2 . and,
This inter-track crosstalk can also be considered as noise to the main signal, and this crosstalk and other noises reduce the S/N of the reproduced digital signal. According to the above section, the S/N required for the reproduced digital signal is
It is 20dB or more.

Therefore, for these reasons, crosstalk is required to be approximately -30dB or less, and -30dB
Recording and reproduction of low-frequency digital signals that cause the above-mentioned crosstalk is not preferable. For example, the second
In the case of azimuth recording (curve C ₂ ) shown in the figure, crosstalk is less than -30 dB when the frequency is approximately
Since the frequency is 1MHz or higher, recording and reproduction of digital signal components whose frequency is 1MHz or lower cannot be performed. However, the digital signal immediately after A/D conversion from the video signal includes a large amount of components with frequencies of 1 MHz or less corresponding to the video signal.

Therefore, in order to further reduce the low frequency signal components that cause harmful inter-track crosstalk in the digital signal, it has been considered to perform format conversion (encoding) on the digital signal.

Various methods have been proposed for this format conversion, but if the original digital signal is, for example, an NRZ signal as shown in FIG. , mirror signal, and ^M2 signal (modified mirror signal) are shown in Figure 4B~
D, and its frequency spectrum becomes as shown in FIG. However, in Figure 5, τ is the bit period, _fs is the sampling frequency,
f _o is the Nyquist frequency, in this case the A/D
At the time of conversion, the digital signal is a parallel signal, but at the time of recording, it is converted from a parallel signal to a serial signal, so the sampling frequency f _s is the frequency in the serial signal (therefore, the frequency f _s is the frequency of the A/D conversion (It is the value obtained by multiplying the sampling frequency at the time by the number of bits per sample).

In addition, Figure 6 shows that for the original digital signal
The frequency spectrum is shown when 8 and 10 conversions of the LDC encoding method are performed, where the broken line is the theoretical value and the solid line is the measured value.

According to FIGS. 5 and 6, compared to the original signal (NRZ signal), Biphase, M ² ,
When performing 8,10 conversion, the low frequency components are reduced. As an example, in the case of 8,10 conversion (Figure 6)
, apply the specific frequency and calculate f _s
Assuming 38.4MHz (the basis for this value will be explained later), as shown in Figure 6, the cutoff frequency at which the spectrum becomes 1/2 on the reduction side is approximately
The frequency is 1.3MHz, and the spectrum is rapidly attenuated in the frequency band below this.

Therefore, by format conversion, it is possible to reduce low frequency digital signal components where azimuth loss cannot be expected.

Further, in order to perform guard bandless azimuth recording more effectively, the digital signal may be divided into a plurality of channels and recorded as multi-track.

Figure 7 shows a high-density digital image that takes into account the above points.
This shows the recording system of a VTR. That is, in FIG. 7, for example, an NTSC color video signal is input to input terminal 1.
1 to an input processor 12 to separate or remove synchronization pulses and burst signals;
The synchronization pulse and the burst signal are supplied to the master clock forming circuit 21, which forms a clock pulse that is synchronized with the burst signal and has a frequency that is, for example, four times as high as the frequency _fc , and these clock pulses and the synchronization pulse form the control signal. Identification signals for lines, fields, frames, and channels, sampling pulses, and various timing signals are supplied to the circuit 22, and these signals are supplied to predetermined circuits, respectively.

Further, in the processor 12, the color video signal from which synchronization pulses and burst signals have been removed is processed by A/
The signal is supplied to the D converter 13. In this case, the sampling frequency is 4f _c , and f _c = 455/2f _h (f _h : horizontal frequency), so the number of samples in the horizontal period for 1 is 910 samples, but the horizontal blanking period is sampled. Considering that it is not necessary, the 12th
As shown in the figure, the number of samples in the effective video area in each horizontal period is 768 samples. Note that HD is a horizontal synchronizing pulse and BS is a burst signal (these have been removed, but are shown for convenience).

Furthermore, the number of lines in one field is 262.5 lines, of which 10.5 lines are occupied by vertical synchronization pulses and equalization pulses. Test signals such as VIR and VIT are inserted into the vertical blanking interval, and these are also considered valid data.
Therefore, the number of effective video lines in one field period is set to 252, and the 12th to 263rd lines are considered effective video lines in odd-numbered fields, and the 274th to 525th lines in even-numbered fields.

In this way, in the converter 13, the color video signal is sampled based on the above points, and is A/D converted, for example, into an 8-bit parallel digital signal (PCM signal) per sample.
quantized to

Then, this digital signal is supplied to the interface 14 and, for example, A~
It is sequentially and repeatedly distributed to the D channel. In other words, among the 768 samples of one line, the (4n+1)th sample (n=0 to 191) is the A channel,
The (4n+2)th sample is the B channel, (4n
+3)th sample is C channel, (4n+
4) The th sample is distributed to the D channel. Then, in these channels A to D, the digital signals from the interface 14 are supplied to time axis compression circuits 15A to 15D, and as described later, the time axis is compressed to 41/44, and the digital signals of these four compressed channels are The signal is sent to the error correction encoders 16A to 16D and the recording processor 1.
13 and 1.
The signal is converted into a signal in the format shown in Figure 4.

Here, FIG. 13 shows the signal of an arbitrary channel among the signals of one field, which is 24×21
Each block Bi has color video signal data for 1/8 line. As shown in FIG. 14, this block Bi is a 24-bit block synchronization signal SYNC.
and 16-bit identification signal ID and address signal AD
, 768 bits (96 samples) of data, and 32
CRC code of bits sequentially.

Here, the synchronization signal SYNC is the signal ID,
Used for synchronization when extracting AD, data, and CRC codes. Also, the identification signal ID indicates whether this channel (track) is A to D, whether the line, field, or frame is an odd number,
Indicates whether the address signal AD is even or
is the address of that block Bi (block number)
shows. Furthermore, the data is an original digitized color video signal, and the CRC code is used to detect data errors during playback.

As mentioned above, the number of effective lines in one field period is 252 lines, so the number of blocks for one field is 504, and these 504 blocks Bi are arranged in a 24 x 21 matrix as shown in Figure 13. In addition, horizontal (row direction) parity data is added to the 25th and 26th columns.
Vertical (column) parity data is added to the 22nd row, making the entire block 26 x 22.

Note that these horizontal and vertical parity data and CRC code are used to improve the data error correction ability during reproduction, and the parity data is also 840 bits.

The signal processing that forms this parity data and CRC code and adds it to the data is
This is performed in encoders 16A to 16D. Further, signal processing for forming a synchronization signal SYNC, an identification signal ID, and an address signal AD and adding them to data is performed in the processors 17A to 17D.

In the processors 17A to 17D, 8 and 10 conversion of digital signals is also performed. In other words, among the 10 bits (2 ¹⁰ codes),
The disparity (DC level) is 0 or close to 0, and "0" and "1" appear almost ^uniformly28 codes are selected, and the original 8-bit codes are made to correspond one-to-one to these codes. Original 1 sample 8
The bit code is converted to 10 bits. Therefore, the digital signal after this 8, 10 conversion is the 6th
As explained in the figure, low frequency signal components are significantly reduced, leaving only signal components above 1.3MHz.

Furthermore, this 8,10 converted digital signal is converted from a parallel signal to a serial signal in order from block _B1 in processors 17A to 17D. Further, a preamble signal and a postampule signal are added before and after the digital signal for one field. The bit rate of the signal after serial conversion is 4f _c × 8 × 1/4 × 44/41 × 10/8 = 38.4 [Mb/s
] (this is the frequency f _s mentioned above).

This serial digital signal is then supplied to one group of rotating magnetic heads 1A-1D through recording amplifiers 18A-18D. This head 1A~
For example, as shown in FIGS. 9 and 10, the tracks 1D have the same track width W, have a step W, and are sequentially shifted in the track length direction. In addition, the heads 1A and 1C have the same azimuth angle θ/2 in one direction, and the heads 1B and 1D have the same azimuth angle θ/2 in the opposite direction from the heads 1A and 1C,
For example, the azimuth angles of heads 1A and 1C are 7 degrees, and the azimuth angles of heads 1B and 1D are 7 degrees in the opposite direction.

These heads 1A to 1D are rotated together with the drum 5 in the direction of the arrow Vh at a field frequency in synchronization with the color video signal, and the magnetic tape 3 is run diagonally in an Ω-shape over an angular range of approximately 360°, and is run at a constant speed in a direction opposite to the head scanning direction Vh, as shown by an arrow Vt.

Therefore, as shown in FIG. 11, the digital signals of channels A to D are recorded as one diagonal track 2A to 2D per field by heads 1A to 1D, respectively. side view).

In this case, the spacing W between heads 1A-1D is equal to their track width W, so tracks 2A-1D are
2D is formed by adjacent tracks touching each other. Also, by selecting the rotation radius and tape speed of heads 1A to 1D, track 2A of the previous field and track 2 of the next field can be set.
D and are formed so as to be in contact with each other.

And in these tracks 2A to 2D,
The azimuth angle of tracks 2A to 2D is the same as that of head 1A.
The directions are alternately opposite corresponding to azimuth angles of ~1D. Note that 4 is a control track.

By the way, in this case, if we look at each channel, recording will be done using a one-head system, so one head
There is a missing period in records A to 1D, and track 2
The time that can be recorded in A to 2D is approximately 250 horizontal periods, and if we take a margin into account, it is 246 horizontal periods.

On the other hand, as shown in FIGS. 13 and 14,
The number of samples (number of bits) in one block is 105 samples (840 bits), and the number of blocks in one field period is 572. Therefore, the number of samples in one field period is 105×572=60060 [samples], which from FIG. 12 is 60060910/4=264, which corresponds to 264 horizontal periods. Therefore, data for 264 horizontal periods is recorded in 246 horizontal periods.

Therefore, the time axis of the signal is compressed in the time axis compression circuits 15A to 15D, that is, the time axis is compressed to 246/264=41/44.

In addition, as described above, the subsequent circuits 16A to 17D
Since various signals are added in the time axis compression circuits 15A to 15A, the gaps for these additional signals are also
Formed at 15D.

Color video signals are digitally recorded in the above manner.

FIG. 8 shows an example of a reproduction system. That is, tracks 2A to 2D are controlled by heads 1A to 1D.
The digital signals of each channel are played back simultaneously. In this case, adjacent heads 1A to 1D and tracks 2A to 2D have different azimuth angles, and tracks 2A to 2D have different azimuth angles.
Since the low frequency signal components of the digital signal recorded in the head D have been significantly attenuated by 8,10 conversion, the inter-track crosstalk of the digital signal obtained from the heads 2A to 2D is sufficiently small.

Then, this digital signal is transmitted to the reproduction amplifier 31A.
~31D to playback processors 32A to 32D.
The serial signal is converted into a parallel signal, and the 10-bit code is block-decoded into the original 8-bit code signal.
A clock is also formed from the digital signal reproduced by the PLL.

Then, this parallel 8-bit digital signal is
The signal is supplied to TBC (time base collectors) 33A to 33D, and time axis fluctuations are removed. in this case,
The TBCs 33A to 33D have a memory, and the block synchronization signal SYNC is used to locate the beginning of the following signal, and writing to the memory is performed by the clock from the processors 32A to 32D, and by the clock generated by the internal sync. Reading from memory is performed to remove time base variations.

Signals from the TBCs 33A to 33D are then supplied to error correction decoders 34A to 34D. The decoders 34A to 34D each have a field memory, and write data into the field memory in accordance with the address signal AD for each block Bi.
Data errors are corrected using the CRC code and horizontal and vertical parity data for each. Note that when there are many errors and cannot be corrected using the CRC code and parity data, the data of the block Bi is not written to the field memory, and therefore the data one field before is read.

The error-corrected data is then supplied to the time axis expansion circuits 35A to 35D to become the original time axis data, and this output is supplied to the interface 36 to convert it into the original one-channel digital signal. Furthermore, this digital signal is supplied to the D/A converter 37 and converted into an analog color video signal. This color video signal is then supplied to the output processor 38, where sync pulses and burst signals are added to the original signal.
This is an NTSC color video signal, and this is output terminal 3.
It is taken out at 9.

Recording and reproduction of color video signals is performed in the above manner, but during recording, low frequency signal components of the digital video signal are significantly attenuated by 8,10 conversion, and adjacent tracks 2A to 2D are
with different azimuth angles, and these tracks 2A
Since recording is performed so that the 2Ds are in contact with each other, the track density T can be increased, high-density recording can be performed, and the margin for tracking errors during reproduction can be increased.

Therefore, it is possible to record for a long time with a small amount of tape consumption, and to record with stable tracking during playback. Moreover, in this case, the features of digital recording are not lost at all.

Furthermore, during recording, the digital signal is divided into four channels and recorded as multi-track, so guard bandless azimuth recording can be performed more effectively and high-density recording is ensured.

Furthermore, when the heads 1A to 1D are arranged as shown in FIGS. 9 and 10, the starting points of the tracks 2A to 2D are located on the tape 3 as shown in FIG.
By keeping this distance y small, the utilization rate of the tape 3 can be increased and higher density recording can be achieved.

However, in this case, with the head arrangement shown in FIGS. 9 and 10, errors in the A channel increase during reproduction.

In other words, in an actual VTR, heads 1A to 1
To facilitate machining of D, or head 1A
For reasons such as increasing the contact area with the tape 3 of ~1D and extending the life of the head, the 15th
As shown in the figure, the track widths of the heads 1A to 1D are larger than the size W by α. Then, the track width of heads 1A to 1D is (W+α)
By setting the arrangement pitch of the heads 1A to 1D to the size W, the heads 1A to 1D can be arranged so that the edge of the latest track overlaps the edge of the previously formed track by a size α. Then, the latest track is formed, resulting in tracks 2A to 2.
The pitch and width of D are set to be the size W.

However, in this head arrangement, as shown in FIG. 15, when head 1A forms track 2A, head 1B forms track 2B so as to overlap the edge of track 2A. 2C and 2D do the same thing, so in the end, track 2A is connected to tracks 2B and 2 on both sides.
The edge is scraped by D and the track width is (W
-α), and conversely, the edges of track 2D are not shifted at all, so the track width is (W
+α).

In this way, if the track width of track 2A becomes narrower than the specified value W by the amount α, the S/N of the digital signal of channel A reproduced by head 1A will naturally decrease during reproduction. This results in an increase in errors. According to experiments,
The error rate of the A channel was twice that of other channels.

Further, even if the track width of heads 1A to 1D is the size W (therefore, tracks 2A to 2D
should be in contact with each other without overlapping), as shown as ΔW in Fig. 1, tracks 2A to
Since smearing occurs in 2D, track 2B, 2
Due to the bleeding of D, the track width of track 2A becomes narrower than the specified value W, and the error rate of the A channel also increases during reproduction.

Of course, it is possible to make the track widths or steps of the heads 1A to 1D different and set the track widths of the tracks 2A to 2D to the specified value W, but this would be extremely inconvenient for manufacturing, adjusting, maintaining, and inspecting the heads 1A to 1D. It is.

This invention attempts to solve these problems.

Therefore, in this invention, for example, the 16th
As shown in the figure, the positions of the heads 1A to 1D in the scanning direction (track length direction) Vh are
The head located on the winding side (lower side) is arranged so that it contacts the tape 3 first.

According to such a configuration, as shown in FIG. 17, during recording, head 1D first forms track 2D, then head 1C forms track 2C, and then head 1B forms track 2B. Head 1A finishes recording the field forming track 2A, and then starts recording track 2D again.
They are formed in order from Therefore, track 2A
The track width of ~2D is the specified value W, and the track width of track 2A of the A channel or any other specific channel will not be narrower than the specified value W, so the error rate of the A channel will be reduced during playback. It never increases.

In addition, when the head arrangement shown in FIG. 16 is used, tracks 2A to 2D are arranged as shown in FIG.
However, during recording, for example, by shifting the time axes of the signals of each channel in the time axis compression circuits 15A to 15D, the tracks 2A to 2D are formed shifted in the track length direction.
can be set to the position shown in FIG. 11, so that the utilization rate of the tape 3 will not deteriorate.

Thus, according to the present invention, it is possible to prevent an increase in the error rate of a specific channel. Moreover,
Therefore, the configuration does not become complicated.

In the above description, when the head scanning direction Vh and the tape running direction Vt are in the same direction, the heads 1A to 1D may be arranged as shown in FIG. 9.

Further, the present invention can be applied to any VTR in which one group of a plurality of rotating magnetic heads has a step and is arranged sequentially in the head scanning direction to perform azimuth recording.

[Brief explanation of drawings]

1 to 15, FIG. 17, and FIG. 18 are diagrams for explaining the present invention, and FIG. 16 is a front view of an example of the present invention. 1A to 1D are rotating magnetic heads, 3 is a magnetic tape, 11 to 18D are recording systems, and 31A to 39 are reproduction systems.

Claims (1)

[Claims] 1 A group of a plurality of rotating magnetic heads is arranged such that adjacent rotating magnetic heads have different azimuth angles and form magnetic tracks that touch each other with steps. In the apparatus, the rotating magnetic heads are sequentially arranged in the scanning direction so that the head located on the winding side of the magnetic tape comes into contact with the magnetic tape first.