JPH03284081A - Skew correction device - Google Patents

Abstract

PURPOSE:To attain normal skew correction by providing an extrapolation circuit receiving skew information from a skew arithmetic circuit and filling skew information between two data when the information is missing. CONSTITUTION:A reproduction FM signal is led respectively from heads A, B to input terminals 1a, 1b. Each FM signal is inputted to variable delay circuits 4a, 4b via band pass filters 2a, 2b and demodulation circuits 3a, 3b and inputted to a skew detection circuit 6. The skew detection circuit 6 detects a skew quantity between signals A1, B1 at a specific period of an overlap period and gives a detected skew quantity output to an extrapolation circuit 11. The extrapolation circuit 11 gives skew quantity information to a delay control circuit 7 when no skew is detected by the skew detection circuit 6 and the skew information controls variable delay circuits 4a, 4b independently. Thus, a demodulation output whose skew is corrected is outputted from an output selector switch.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application Field) The present invention provides a skew correction device used in a magnetic recording and reproducing device capable of recording and reproducing high-quality audio signals, such as an audio reproducing system of a so-called HiFi VTR. Involved in

(Prior Art) A helical scan type VTR that records and reproduces an audio signal on the same track as a video track has been developed. In this system, when reproducing a Yin track, the reproduced signal traces different tracks by combining these first and second oscillations into one continuous ff'i.
You need to connect it to signal number 3. Therefore, the first and second
The signal 952 is recorded and played back so that it has an overlap period, and within this overlap period, the signal 952 changes from the first signal to the second signal, or from the signal 952 to the first signal. , and a third signal is formed.

The problem here is that signal dropouts occur as a result of switching selection, resulting in so-called switching noise, which remains in the third signal.

Therefore, in order to solve this problem, the present invention proposed a skew correction device as shown in FIG. 10 in the specification of Japanese Patent Application No. 1-21061. In Figure 10, la
, lb are inputs h'Er to which FM signals from A and B heads of different azimuths are introduced. F from terminal 1a
The M signal is passed through a band-limiting filter 2a to a demodulation circuit 3a.
The FM signal from terminal 1b is input to demodulation circuit 3b via band-limiting filter 2b. The demodulated output of the demodulating circuit 3a is transmitted through a variable delay circuit 4 whose delay level can be controlled.
a to the first pane P1 of the output selection switch 5. The demodulated output of the demodulation circuit 3b is guided to the second terminal P2 of the output selection switch 5 via a variable delay circuit 4a whose delay group is controllable. These variable delay circuits 4a and 4b are
The delay l can be controlled independently of each other. The output 7J selection switch 5 is operated by the head 1 moss switching pulse SWP from the terminal 8 to select the first and second terminals P.
The signals introduced into I and P2 are switched and selectively outputted to the output terminal 9 to form a third signal.

In the above, in the case of a system in which skew does not occur during the overlap period, the variable delay circuits 4a and 4b are unnecessary, and a continuous third It is possible to generate the output of the output terminal 9 at the output terminal 9.

The discontinuous signal generated in the third signal due to skew is A
- Switching from the first signal to the ff12 signal or from the second signal to the first signal within the overlap period 11
．． 'L's Sue 1' - Detects the skew and controls the delay m of the variable R delay circuits 4a and 4b according to the detected skew 6! 1
By adding +, the first. It is possible to maintain continuity of the signals obtained at the second terminals P1 and P2. Here each demodulation circuit 3
The outputs of a and 3b are input to a skew detection circuit 6 which detects skew hanging from these demodulated outputs. The skew detection circuit 6 generates skew information representing each skew by a calculation described later in a specific section of the overlap period, and sends the information to the skew calculation circuit 7. ! The i extension calculation circuit 7 controls the U extension claws of the variable delay circuits 4a and 4b based on the skew information. By controlling the J and U variable delay circuits 4a and 4b, the It is possible to connect signals q continuously and eliminate switching noise when switching heads.

The skew detection circuit 6 has a configuration similar to that shown in FIG. In FIG. 11, the terminal 101a.

101b, the demodulation circuit 3a in FIG.
, 3b supply demodulating horns. The gap detection circuit 6 has a configuration shown within the dotted line frame, and the skew calculation circuit 61 calculates the skew amount based on the demodulated outputs from the terminals 101a and 101b.

The vXn result from the four-needle sieve circuit 61 is applied to the lean pull bold circuit 63 and the first terminal P1 of the output selection switch 64. 63. Output selection switch 644.i, whether the calculation result is not obtained from the skew calculation circuit 61 or accurate T: If not, instead of the calculation result of the skew calculation circuit 61,
The pre-detected amount from the sample hold circuit 63 is selectively output as the skico ship S. Although the calculation result is not obtained from the skew calculation circuit 61, the judgment as to whether it is inaccurate is made by, for example, using one demodulation circuit (here, the demodulation circuit 3b, that is, the terminal 101b).
) is detected and determined based on the magnitude of the slope of the signal waveform at the zero-crossing point. Tilt detection circuit 62
indicates the above-mentioned slope. According to the -4 information, the output selection switch 6
4 as i, ++ aIlt, . This is because the slope of the alternating current waveform is steepest in the part that passes through the zero-crossing point when the slope is detected using the zero-crossing point.

Note that even when a zero-crossing point is not detected, the output of the sample-and-hold circuit 62 is used instead.

Note that, in reality, the skew calculation circuit 61 detects the slope at the zero cross point in the same way as the slope detection circuit 62;
It is part of 1.

The skew detection means of this method detects the presence or absence of a zero-crossing point in the reproduced signal, and detects whether there is a zero-crossing point or when the slope of the signal waveform at the zero-crossing point is below a certain value. Output the output as skico information. If a zero-crossing point is detected and the slope at the zero-crossing point is small, the previous detection amount from the sample and hold circuit 63 is used as skew information. A signal with a small slope at the zero crossing point is
It is a low-frequency, small-level signal, and in the case of such a signal, it is not possible to accurately obtain the chimpling value between the head signal and the B head (Lil), which have a skew. Of course, the skew detection circuit and the slope are inversely proportional as shown in Figure 12.
The larger the slope, the smaller the error, and the smaller the slope, the larger the error.

Here, the skew calculation circuit 61 calculates the skew σ based on the principle shown in FIG.

As shown in FIG. 13(A), the reproduction signal of the head and 8
If the reproduction signal of the head has a skew S, the third
As shown in FIG.
Obtain slope information ΔY, which is difference information from the signal a (N−n) after x c k.

ΔY = a (N) -a (N-n)Y=a(N)
-b(s) ...■ ...■ Here, the skew ms is ΔY:Y=S: (nxck) , , @
From the relationship, ■ can be found as. (nxck) is the sampling interval from which the slope information was obtained, and this value is
, is preferably sufficiently narrower than the M band. Then, if S is expressed in units of ri[''tsukck, the above equation 0 becomes the following equation.

■ However, as mentioned above, if the reproduction signals from the A head and B head are low frequency and small (No. 8), each sampling value shown on the left side of the 0 formula and the 0 formula cannot be obtained accurately. The detected skew 1 does not match the actual skewness.Especially when a low frequency, small signal continues for a long time, the sample hold period becomes long, resulting in a large deviation from the actual skewness, resulting in malfunction. .

(Problems to be Solved by the Invention) In conventional skew correction devices, when the reproduced signal is a low frequency or small signal, Urocross detection cannot be detected or becomes inaccurate, the slope becomes small, and the skew When such low-frequency, small signals continue frequently, a method of substituting the pre-detection period is used to deal with the problem, which increases skew errors and causes malfunctions.

In addition, the slope becomes small, making it impossible to obtain sufficiently accurate slope information, which is replaced by a pre-hold means, which has the disadvantage of increasing the error in skew Is. Ta.

SUMMARY OF THE INVENTION An object of the present invention is to provide a skew correction device that can perform normal skew correction without causing a large deviation in skew 1 even when no skew amount is detected.

[Structure of the Invention] (Means for Solving the Problem) The invention of claim (1) inputs skinny information from a skew calculation circuit, and when the information is missing, the skinny information is input, and when the information is missing, the skinny information is input.
An extrapolation circuit is provided which performs a predetermined calculation based on at least two pieces of data in which El h<n and fills in the dark skew information of these two pieces of data.

The invention of claim (2) provides a subtraction circuit that inputs skew information from a skew calculation circuit and calculates the difference between this input information and output 1-knee information, and a subtraction circuit that receives difference information Δ3n from this cylinder reduction circuit. It has a variable filter that corrects the difference information according to the difference, and an adder circuit that corrects the output skew information by adding the corrected difference information output from the variable filter.

The invention of claim (3) provides shift registers that input skew information from a skew calculation circuit and output the skew information for n periods in parallel, and each shift register that outputs the skew information from these shift registers, for example. The current coefficients have a weighting circuit that weights them according to JIJ coefficients set too high, and an averaging circuit that outputs data obtained by averaging the total sum of each output from this weighting circuit.

(Operation) The skew detection calculation is determined based on the slope of the waveform at the zero-crossing point during the overlap period and the level difference between two signals having different phases due to S1 Nico. For this reason, it is necessary to detect the zero cross point, but the pyrocross point may not exist or cannot be detected accurately at low frequencies or small signals. If the zero-crossing point cannot be detected accurately or is inaccurate, the amount calculated by calculation will also have an error.

According to the invention of claim (1), when the skew amount is not detected, the previous skew foot is held until the next skew amount is detected, and the change trend of the skew amount is determined based on the two obtained skew amounts. Estimate. This makes it possible to compensate for the amount of skew even if there are consecutive periods in which no skew is detected.

According to the invention as claimed in claim (2), when the error in the detected skew amount is small, the skew amount is assumed to be inaccurate, and the influence of this is reduced to the extent of the skew.

According to the invention of claim (3), the errors in the skew foot are averaged out. In this case, M is set so that the current 11.1-point detection value 1:coe amount appears largely at output value i2], -m.

(Example) Hereinafter, the present invention will be explained in detail with reference to illustrated examples.

FIG. 1 is a block diagram showing an embodiment of a skew correction device according to the present invention, and the same reference numerals are used for the same components as those of the conventional device shown in FIG.

In FIG. 1, reproduced FM signals from the Δ and B heads with different azimuths, similar to those in FIG. 10, are led to input terminals la and lb, respectively. Each FM signal is input to variable delay circuits 4a and 4b via bandpass filters 2a and 2b and demodulation circuits 3a and 3b, respectively, and the outputs AI and B1 of the demodulation circuits 3a and 3b are respectively input to a skew detection circuit 6. be done. The skew detection circuit 6 detects the amount of skew between the signals A1 and 81 in a specific section of the overlap period, and supplies the output of the detected skew to the extrapolation circuit 11 at the next stage. The extrapolation circuit 11 corrects and overturns the sew mushrooms in that section when the skew amount is not detected by the skew detection circuit 6. Whether or not the skew amount is detected is determined by the zero cross detection signal 10a from the zero cross detection circuit 10. It is judged based on whether it is given or not. If the zero-crossing detection signal 10a is not obtained, the skew amount is extrapolated using the next skew amount obtained and the previous skew amount. The skew output from the extrapolation circuit 11 is supplied to a delay 1 arithmetic circuit 7 that independently controls the delay amount of the variable delay circuits 4a and 4b.

As a result, the demodulated output AI is skew-corrected.

B2 is led out to the output terminal 9 as a third continuous signal to the output selection switch 5, which is subjected to switching control by the switching pulse SWP from the terminal 8.

The U rotas detection circuit 10 detects the first zero cross point that occurs from both edges of the window signal set within the overlap period, and uses the zero cross detection point J310a to output the zero cross point to the skew detection circuit 6 and the extrapolation circuit 11. is supplied to provide timing for skew detection and extrapolation processing.

FIG. 2 shows an outline of the operation of the above configuration.

In FIG. 2, nt (t=1.2.3...) represents the skew amount at each skew amount detection time l. In the figure, it is set at time t=0.3°7. The X mark indicates a point where the skew amount was not obtained by the skew detection circuit 6, and in the figure, the time L = 1.
2.4 and 5.6 cannot be obtained. Also, the difference between the skew amount no. 03 (n3
-no) is [, and the difference between the two adjacent skew amounts is S[
shall be.

Now, when n3 is detected, L can be determined by M, and St can be determined from this L using the following equation.

S=1/3...■
If the estimated value of ``)4 that could not be detected is expressed as n4', n4' is nA' = n3 + st
...It can be extrapolated as ■. The amount of skew actually changes gradually, so the actual skew at t=4! The difference between 1n4 and r14', that is, the error due to extrapolation, can be made extremely small.

FIG. 3 shows the actual circuit configuration of the extrapolation circuit 11 that performs the above-mentioned extrapolation. In FIG.
is supplied to The delay circuit 32 is a switching pulse SW
ilJ ill by P, switching pulse S
Delay WP by 1/2 cycle.

The output C of the delay circuit 32 becomes a signal f3D via the sampling switch 33 and the delay circuit 34, which are controlled by the output terminal M of the zero cross detection circuit 1G. The signal is
The skew output sampled by the switch 33 is a delayed signal until the next skew amount is detected. The signal ri and the signal C are reduced in sieve by the sieve reduction circuit 35, and the difference between the skews m is determined ((see A, E).
. The signal E is supplied to a delay circuit 4 controlled by an output JF of a counter 43 that counts switching pulses SWP.
divided by the output G from 2. This allows S[
A belief @1-1 is required. Output 1 of divider 36
is latched until the next skew amount is output from the delay circuit 32. The output I of the latch circuit 37 is added to the output C from the delay circuit 32 in an adder circuit 44, and output
It becomes J. Output J is an interpolated output with good skew.

The interpolated output J is outputted by the delay circuit 38 during the 1/2 cycle period 2! of the switching pulse SWP. if, and returned to the switch 31. The switch 31 turns on the delay circuit 38 and the skew detection circuit 6 to form a signal B as a skew detection output. The signal B is supplied to the delay calculation circuit 7 via the sample and hold circuit 39. The sample hold circuit 39 includes a counter 40 that counts the output M from the zero cross detection circuit 10.
and 1 detection circuit 41, and holds the first skew amount from the time of the first zero cross detection when ST cannot be obtained until the next time when the zero point is detected and the skew amount is obtained. At the same time, from the second time onward, a switching composite output of the skew hanging horn interpolated by st and the detected skew hanging horn is supplied to the delay amount calculating circuit 7.

FIG. 4 is a time chart for explaining the operation of the circuit in FIG. 3 using the lifting platform in FIG. 2 as an example. correspond to the symbols shown in FIG.

The first skew amount output no is input to the delay circuit 32 via the switch 31 (tl), i[The skew 1flnO supplied to the delay circuit 31 is delayed by 1/2 cycle of the switching pulse SWP, and 33. The switch 33 samples the skew ff1nO (12 times) using the zero cross detection output M, and the delay circuit 34 obtains the next skew n3.
It is delayed by one time t2 (to be exact, to time t3, which is 1/2 cycle from t2).

That is, at time t3, the τ reduction/stop circuit 35 causes ")〇-
The calculation n3=L is performed. The counter 43 is no and 0.
An output F indicating how many v cycles is apart from -C is applied to the delay circuit 42, and the delay circuit 42 supplies the number of cycles of I to the remover 36 at time t3. The first data 81 obtained thereby is input to the latch circuit 37 at time t3. The latch circuit 37 next receives the skew amount n7 (t4
) is held until data S1 is obtained and supplied to the adder circuit 44. 7111 'C'J circuit 44 receives the output C from the extension circuit 32 which holds the skew amount "13" at time t3, and thereby the interpolated skew amount n4'~ between times t3 and t4. n6' is reduced by 1q.

As described above, in the above embodiment, skew detection is performed once [1 to
The skew amount is not compensated for the second time, and the skew amount estimated from the second time onward is corrected.

Next, another embodiment of the invention will be described.

FIG. 5 is a block diagram showing an embodiment of the skew detection circuit used in the embodiment of FIG. 1. In FIG. 5, the same components as in FIG. The FM signal from the bandpass filter 2b is input to the demodulation circuit 3b via the bandpass filter 1b'.

The circuit within the dotted line is the skew detection circuit. This skew detection circuit includes a skew calculation circuit 51 that calculates the 1-1-hang based on Ll according to the equation (2), and an output skew 1i1 that corrects the detected skew amount Sn from the skew performance circuit 51.
3r+'. Sue 1] Coe M correction circuit 52 includes a subtraction circuit 53 that calculates the difference between the detected skew Sn detected by the skew ffi calculation circuit 51 and the output skew information 3 n l , and difference information ΔSn from the subtraction circuit 53. A variable filter 54 corrects this difference information Δ3n according to the magnitude of
and the modified difference information 3 from this variable filter 54
an adder circuit 55 that corrects the output skew information Sr+' by addition at the output of nJ;
It consists of a latch circuit 56 connected between the output terminal 57 of S n ' and the addition n circuit 56.

In the case of j3 above, the characteristic of the variable filter 54 has a nonlinear characteristic as shown in FIG. The vertical axis in FIG. 6 represents the output response y of the variable filter 54, and the horizontal axis represents the μ person norm X (difference information ΔSO'). This characteristic is
Difference information △3n' is within a certain variation range + x 1 to -
What is the characteristic of y=x in x1'? The section where the damping characteristic occurs in the range exceeding +X, the section where the damping characteristic occurs in the range exceeding -X, and the section where the damping characteristic occurs in the range exceeding -X.

In such a configuration, the skew correction circuit 52 has S
Input 1: n, Sn+l, Sn+2...
- The amount of code enters and outputs from the terminal 57 with a delay of one clock.
From 5n-1', Sn', 3n+1'-
It is assumed that the output 4-ko is derived as follows.

Input scrap A knee suspension 5n11, output skew fiSn'
Consider the case of

At this time, the input skew 1 knee Sn+1 and the output skew jfi3n' are input to the subtraction circuit 54. (The difference information is ΔSr+−+1 ==Sn+1−3n′.

Here, if the characteristic of the variable filter 54 is expressed as f (X), 3n+1' = f (Sn+1-3n') +
3n -...■ When ΔSn→1 is within the range of +x1 to -x1, the variable filter 54 adjusts y=x in FIG.
It exhibits the characteristics of Sn+1' = '(Sn+1-8n') +S
n=Sn+1...■ Tomuru. According to Equation 0, if the value of 1ΔS1 is small,
The input skew amount and the output skew amount are approximately equal, and the value of ΔS is reflected in the output skew.

When Δ3n+1 exceeds the range of +x1 to -x1, for example, f (X) becomes a product of a certain constant, and Sn+1' = 0.1 (Sn+1-8n')
+Sn is written as Sn+1' = 0.9Sn' i
o.

1Sn+1, and the input skew amount is not reflected in the output skew amount.

When the difference information ΔS is within a certain value, it is considered to be operating normally, and when it is above a certain value, the greater the degree, the more likely it is an error, so it should not be reflected in the output. -ing The variable filter 54 automatically determines whether the input data is an error, and sets the characteristics of X according to this probability distribution.
This is reflected in the output data accordingly.

The relationship between the input skew pattern and the output skew amount based on the above operating principle is not shown in FIG. In FIG. 7, the vertical axis represents the output skew peak, and the horizontal axis represents time. However, since they match, there is an output skew of A3 G at time t11! LISn' matches the input skew filsn. However, there is an input preference at time t12] -1f
Since i311)1 is significantly larger in the positive direction than the previous skew E+isn', the difference information ΔS between this and 811' is not reflected in the output skew m, and the output skew ff1sn+1' at time t12 is smaller than Sn'. This is a group of 6 people.

From another perspective, the variable filter 54 becomes an all-pass filter when ΔS is smaller than a certain level,
When ΔS is larger than a certain level, the result is a 0-pass filter with a lower cutoff point.

In this way, the present skew detection circuit improves accuracy by smoothing data when the error is large, and maintains response speed by outputting data as is when the error is small. If the accuracy is improved by simply using a low-pass filter that matches the probability of knee occurrence, the response speed will be slow and accurate correction will not be possible.

FIG. 8 shows another embodiment of the skew detection circuit according to the present invention.

In this embodiment, the skew nails obtained by the skew calculation circuit 51 are transferred to 0-stage shift registers 58-1, 58-2...5.
8-(n-1), input to 58-n. Each register 58-1.58-2...58-(n-1), 5
8-11 is the detection signal I from the zero cross detection circuit 10.
Data is shifted by Qa. The output from the skew calculation circuit 51 is the current skew@
n seasoning circuits 59-1, 59-2, etc., in which the weighting coefficient is set larger as the register 58-1 inputs
・Unit processing is performed using 59(n-2), 59-(row 1), and 59-n, and the weighted outputs of these weights (j t:J are added in an adder circuit 60.The adder circuit 60 The output skew M is obtained through the division n circuit 61 which averages the input by 1/1.
0 supplies the zero-crossing detection signal 10a to the skew calculation circuit 51 to control the timing of the start of the skew calculation. Jit register 58-1.58-2...
58-(n-1).

58-0 also shifts data based on the zero error detection signal 10a.

According to such a configuration, when the input skew amount is varied due to an error, the influence of the error on the output skew amount is minimized by averaging.
Each register 58-1.58-2...58- (n-1
), 58-weighting coefficient at for II.

A2. ...The register 58-1f11i which inputs the current skew ffi for an-2, an-1, and an is set to a large value (al > A2 >...
an-2>an-1>an), the influence of the current input skew m is most reflected on the output skew m.

Figure 9 shows an example of weighted coefficients, the vertical axis represents the size of the coefficients, and the horizontal axis represents the number of stages of the shift register.
lI3, the characteristics of this coefficient may vary depending on the system.

[Effects of the Invention] As described above, according to the present invention, skew errors are reduced and accurate cedar J-correction becomes possible.

[Brief explanation of the drawing]

FIG. 1 is a configuration diagram showing an embodiment of the skew correction RFf according to the present invention, FIG. 2 is an operation explanatory diagram schematically showing the operation of the embodiment of FIG. 1, and FIG. 3 is an implementation of the embodiment of FIG. An example of the extrapolation circuit used in the example is shown below. Fig. 4 shows the operation of Fig. 3.] Fig. 5 is a configuration diagram showing an embodiment of the skew detection circuit according to the present invention. , Fig. 6 is a characteristic diagram showing the characteristics of the variable filter used in the embodiment of Fig. 5, Fig. 7 is an explanatory diagram showing the operation of the embodiment of Figs. 9 is a block diagram showing another embodiment of such a skew detection circuit. FIG. 9 is a characteristic diagram showing the coefficient characteristics of the market detection circuit used in the embodiment of FIG. 8. FIG. Figure 11 shows the conventional skew detection circuit; Figure 12 shows the characteristics of the 4-knee detection error; Figure 13 shows the principle of skew detection. be. 3a, 3b... Demodulation circuit, 4a, 4b... Variable delay circuit, 5... Output selection switch, 6... Skew detection circuit, 7... fiMff! Arithmetic circuit, 10... Z [Cocross detection circuit, 11... Extrapolation circuit, 51... Skew operation circuit, 52... Skew detection circuit, 53... Subtraction circuit, 54... Variable filter,
55...additive circuit, 58-1.58-2...58-(n-1), 58
-n...Shift register, 59-1.59-2...
59-(n-2), 59(n-11, 59-1+・
... weight publication 1) circuit, 60... sieving circuit, 61...
Exclusion circuit.

Claims (3)

[Claims] (1) A reproducing circuit that reproduces a recording signal recorded with an overlap period on a recording track formed on a recording medium as first and second signals corresponding to the recording track; A controllable delay circuit provided in the path of the first and second signals, and detecting a phase difference between the first and second signals from the reproduction circuit during the overlap period, and generating skew information indicating the amount. A skew calculation circuit to output, and the skew information from this skew calculation circuit are input, and when information is missing, a predetermined calculation is performed based on at least two pieces of data from which skew amounts are obtained.
an extrapolation circuit that fills in skew information between two pieces of data; a delay amount control circuit that controls the delay amounts of the first and second controllable delay circuits based on the skew information from the extrapolation circuit; A third signal in which each output signal from the quantity control circuit is switched to an overlap period and the first and second signals are continuous.
A skew correction device comprising: a selection circuit that outputs a signal. (2) a skew calculation circuit that detects a phase difference between the first and second signals from the reproduction circuit during the overlap period and outputs skew information indicating the amount; and the skew information from the skew calculation circuit. a subtraction circuit that calculates the difference between this input information and output skew information; a variable filter that corrects this difference information according to the magnitude of the difference information from this subtraction circuit; The skew correction device according to claim 1, further comprising: an addition circuit that corrects the output skew information by addition when outputting the difference information. (3) a skew calculation circuit that detects a phase difference between the first and second signals from the reproduction circuit during the overlap period and outputs skew information indicating the amount; Enter, n
A shift register that outputs the skew information for a period in parallel, a weighting circuit that weights each skew information from these shift registers by a set weighting coefficient, and an average of the total of each output from this weighting circuit. The skew correction device according to claim 1, further comprising: an averaging circuit that outputs .