KR940010171B1 - Transversal equalizer - Google Patents

Abstract

No content.

Description

Equalizer

1 is a circuit diagram showing one embodiment of the equalizing device of the present invention.

FIG. 2 is a circuit diagram showing an embodiment in which the equalizer of FIG. 1 is applied to a ghost reducing device. FIG.

2 is a circuit diagram showing another embodiment of the present invention.

4 is a circuit diagram showing a transversal filter.

5 is a circuit diagram showing a conventional equalizer for ghost reduction.

* Explanation of symbols for main parts of the drawings

1, 34: Input terminal 4: Cascade input terminal

31 switch 33 adder

35: output terminal TF1, TF2: transversal filter

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an equalizer and, more particularly, to an assimilation device, which is constituted by serial and non-circular series connections, suitable for filtering an input signal in real time.

In recent years, in television broadcasting, a GCR (ghost cancellor reference) signal enters a vertical retrace period as a reference signal for ghost removal in response to a demand for higher image quality. A television receiver or the like is configured to remove ghosts by performing waveform equalization using a GCR signal by employing an equalizing device.

4 is a circuit diagram showing a transverse filter for performing waveform equalization in real time. The circuit of FIG. 4 is described, for example, in Document 1 ("Parformance Evaluations of Selected Automatic Deghosting Systems for Television" IEEE, Transactions on Consumer Electroncs, Vol CE-26, February 1980), with a six-tap transverse filter. It is called.

The input signal is sampled every T seconds, and this input sampling value {Xi} is provided to the input terminal 1. The input sampling value {Xi} is given to the counters M1 to M6, respectively, and multiplied by the coefficient values (hereinafter referred to as tap coefficients) C1 to C6. The outputs of the counters M1 to M6 are given to the adders A1 to A6, respectively. The outputs of the adders A1 to A6 are configured to be delayed by delays D1 to D6 and given to the output terminals 2 and the next adders A1 to A5, each counter The delay signals of the outputs of M1 to M6 are added to be output. The delayers D1 to D6 are driven by the clock CK of the period T seconds supplied via the input terminal 3, and are delayed and outputted by the T signal. An output according to the tap coefficient is shown on the output terminal 2, and the transmission path can be equalized by setting the tap coefficient. A signal from the front end is input to the cascade input terminal 4, which is added to the output of the counter A6 in the adder A6.

Now assume that the input sample value {Xi} is a function δi}. The impulse response {ai} is then a1 = C1, a2 = C2, ..., a6 = C6, and the tap coefficient itself. The length of the impulse response string is equal to the number of taps, and longer pulse trains can be obtained by increasing the number of taps.

That is, the filtering time can be lengthened by increasing the number of taps.

The ghost cancelable delay time range due to the GCR signal is 44.7 의 of the width of the GCR signal, as shown in Document 2 ([Ghost Elimination Reference Signal Method] Journal of the Korean Television Society, pp.239-240). . As described above, the filtering time length is determined according to the number of taps, and the number of taps for obtaining a delay time range of 44.7 ms is represented by 44.7 ms / T (clock period). Usually, the clock frequency in the digital processing of a television signal is set to four times the color part carrier frequency (14.31313 MHz). Clock period T is 69.84ns.

In other words, 640 taps require many taps for ghost removal using the GCR signal.

Transversal filters are usually integrated circuits (TCs), and according to the improvement of the degree of integration, for example, 64 taps per chip are integrated such as TF-ICs employed in Toshiba's Ghost Cleaning TV Tuner (TT-GC9). have.

5 is a block diagram showing a conventional equalizing device employing such a transverse filter. The apparatus of FIG. 5 is disclosed by Document 3 (Ghost Mitigation Tuner and Its Functions) magazine Cromer, December, 1989, pp. 48-51.

The input terminal 5 receives a video signal subjected to ghost interference. This input video signal contains a GCR signal. The input video signal is input to the input terminal 6 of the transversal filter 11. The cascade input terminal 7 of the transversal filter 11 is connected to the reference potential point, and the input terminals 6 and 7 correspond to the input terminals 1 and 4 in FIG.

Tap coefficients C-29 to C0 to C34 are given to each of the 64 counters, not shown in the transverse filter 11, respectively. The subscript of the tap coefficient indicates how many times the clock period T corresponds to the ghost of the delay time. The tap coefficient of the tap corresponding to the winning part of the GCR signal which is not affected by ghost is C0. In the initial setting, 0 is set to a tap coefficient different from 1 in C0. Therefore, in the initial setting state, the transversal filter 11 outputs the video signal input to the input terminal 6 from the output terminal 8 as it is.

The transversal filter 11 has a non-circular configuration and corresponds to a delay range of -2 dB (pre-ghost) to 2.4 dB (post-ghost) by tap coefficients C-29 to C34. That is, the transverse filter 11 removes the waveform distortion (waveform equalization) and removes the ghost (close ghost) having a short delay time. The output of the transversal filter 11 is given to the output terminal 10 via the subtractor 9 at the output terminal 8 and is also given to the delayer 21.

The output of the retarder 21 is given to each input terminal 6 of the transversal filters 12 to 20 having the same configuration as the transversal filter 11. The number of taps of the transversal filters 12 to 20 is all 64 and the tap coefficients C35 C610 are given to the transversal filters 12 to 20. In other words, the transversal filters 12 to 20 correspond to the post-ghost with a delay time of 2.4 to 42.6 ms. The output terminal 8 of the transversal filter 20 to 12 is given to each cascade input terminal 8 and subtract 9 of the transversal filter 19 to 12 between It is made up of. The tap coefficients C35 to C610 are set to zero in the initial state, and the ghost cancel signal is output from the transversal filter 12 by modifying the tap coefficients. The subtractor 9 outputs the video signal from which the ghost component is to be removed to the output terminal 10 by subtracting the output of the transversal filter 11 and this ghost cancellation signal. Each tap coefficient C-29 to C610 is obtained by calculation of the GCR signal extracted from the input / output video signal and the reference signal, and is configured to be corrected in sequence every predetermined time. That is, the GCR signal included in the video signal from the input terminal 5 and the GCR signal included in the video signal from the output terminal 10 are extracted. In addition, the tap coefficient is modified so that a correlation operation between the error signal and the GCR signal included in the input video signal is performed to converge the error signal. In addition, for the cost, the same IC is used for all of the above-mentioned transversal filters 11 to 20.

In this way, the non-circular transversal filter 11 performs waveform equalization and near ghost removal, and the hand ghost (see Document 3) and the distant ghost generated by the transversal filter 11 are circulated. It is removed by the transversal filters 12 to 20. The series connection with this non-circulating type is a configuration most suitable for ghost removal as shown in Document 3.

In general, for all ghosts, there is no practical problem as long as the delay time is removed within about -2 ms. On the other hand, the delay time of a postghost may be 40 microseconds or more. As described above, since the ghost elimination delay time range is 44.7 ms, the corresponding range of all ghosts is set to -2 ms in the example of FIG. That is, the tap coefficient of the transversal filter 11 is set to C-29 to C0 to C34.

Since the output of the transversal filter 11 and the output of the transversal filter 12 are subtracted from the subtractor 9, the ghost component is removed. The delay is shifted by T, and the delay unit 21 operates with a delay time of 34T.

However, the integration of digital ICs is expected to increase, and the number of taps that can be integrated in one chip is expected to increase. As a result, the tab for post-studding of the transversal filter 11 increases. Therefore, in this case, it is necessary to lengthen the delay time (the number of stages of delay) of the delay unit 21 in accordance with the increase in the number of taps in order to shift the output of the transversal filter 11 and the transverse filter 12 by T in the time axis. In addition, there is a problem that the circuit size increases and the cost increases.

As described above, in the conventional equalizer, when the number of taps in one chip increases by increasing the degree of integration of the TF-IC, it is necessary to extend the delay time of the delayer as the number of taps increases, and the circuit size increases. There is this.

The present invention has been studied in view of the above problems, and it is an object of the present invention to provide an assimilation device capable of reducing the circuit size without using a delay in the case of employing a circulation type or a non-circulation series connection. do.

The telephone apparatus according to the present invention has a counter group, an adder group, and a delay group, and the delay amount of the delay group of the first and second transverse filters that assimilates and pulls input sample values according to the tap coefficients given to the counter group. Adding means for adding the counter output at the position according to the position and the output of the first transverse filter, and adding the output of the first transverse filter to the counter output at the head end of the second transverse filter or to the adding means. It is provided with the position to provide.

In the present invention, when the first transverse filter is configured in a circulation type and the second transverse filter is configured in a non-circular configuration, the switch gives the output means of the first transverse filter to the adding means. The output of the second transversal filter is supplied to the first transversal filter as an input signal. The last tap coefficient of the second transverse filter and the first tap coefficient of the first transverse filter are adjacent on the time axis. Thereby, unlike a conventional case, it does not require a retarder and can be comprised circularly.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. 1 is a circuit diagram showing one embodiment of a moving device according to the present invention.

및 가산기 Am 내지 An-2에 출력한다.The configuration of the input weighted transverse filter TF1 surrounded by the broken line is the same as that of the transverse filter of FIG. That is, an input sample value xi} is input to the input terminal 1, and the input sample value {xi} is given to the counters Mm to Mn-1 having (nm) taps, where n and m are n-1> m. The counters Mm to Mn-1 are given tap coefficients Cm to Cn-1, respectively, and the counters Mn to Mn-1 multiply the tap coefficients Cm to Cn-1 by the input sample value {xi}, respectively. The outputs of the adders Am to Am-1 are given to the delayers Dm to Dn-1, respectively, and the delayers Dm to Dn-1 are inputted by the clock CK from the pressure terminal 3. Terminal of the switch 31 by delaying the signal by time T, respectively.

And output to the adders Am to An-2.

Moreover, the cascade terminal 4 is comprised so that the output of the front end animation device which is not shown in figure is supplied.

The output of the transversal filter TF1 is provided to terminal a of the switch 31. The switch 31 has a terminal β or r selected by the control signal from the input terminal 32. These terminals β, r function as cascade input terminals of the transversal filter TF2.

The configuration of the Eosan transversal filter TF2 in broken lines is the same as that of the transversal filter TF1, and the (m + r) counters M-1 to Mm-1 and the retarders D-1 to Dm-1 and (m + l + 1). ) Adders A-1 to Am-1 33. The signal from the terminal of switch 34 is given to adder Am-1. L is a natural number.

An input sample value {xi} is input to the input terminal 34, and this input sample value {xi} is given to each of the counters M-1 to Mm-1. The outputs of each of these counters M-1 to Mm-1 are given to adders A-1 to Am-1, respectively, and the outputs of adders A-1 to Am-1 are respectively via retarders D-1 to Dm-1. The next stage of the adder is the same as the transverse filter TF1. In the transversal filter TF2, an adder 33 is provided between the retarder Dm-l-1 and the adder Am-l-2, and the adder 33 has an output and a switch 31 of the retarder Dm-1-1. It adds the signal from terminal r, and outputs to adder Am-l-2. The output of the delay unit D-1 is output to the output terminal 35.

The subscripts of the elements of the transverse filters TF1 and TF2 indicate how many times the clock period T corresponds to the ghost of the delay time. For example, the tap coefficient C0 indicates that the transverse filter TF2 is described later. The coefficients applied to the main signal in the case of non-cyclic connection are displayed. By one tap coefficients C-1 to C-1 given to the multipliers M-1 to M-1, a clear signal of all ghosts is generated and (m-1) tap coefficients given to the multipliers M1 to Mm-1. Cm-1 to C1 generate a post-guest erase signal.

,r를 통해 트랜스버설필터 TF2의 가산기(33)에 부여된다.In the embodiment configured as described above, in the case where the non-circulating type and the circulation type are connected in series, the input signal is applied to the input terminal 34 and the output terminal 35 is connected to the input terminal 1. As a result, the output from the output terminal 35 is fed back to the input terminal 1. The transverse filter TF2 performs waveform equalization of the input signal according to the tap coefficients C-1 to Cm-1 applied to each tap and is output from the output terminal 35. The counters M-1 to M-1 correspond to, for example, the previous ghost, and the counters M1 to Mm-1 correspond to the next ghost in proximity. In this case, the transversal filter TF2 constitutes a non-circulating type. The output from the output terminal 35 is fed back to the input terminal 1, and the transverse filter TF1 generates a post-ghost cancellation signal in accordance with the tap coefficients Cm to Cn-1. The output of the transverse filter TF1 is a terminal of the switch 31.

is applied to the adder 33 of the transversal filter TF2 via, r.

Since the adder 33 is disposed immediately before the adder Am-l-2, the delay amount from the adder 33 (terminal r of the switch 31) to the output terminal 35 is (m-1) T. Therefore, the last tap coefficient Cm-1 of the transverse filter TF2 and the first tap coefficient Cm of the transverse filter TF1 are configured to be delayed by the time T on the time axis. In this way, the transversal filter TF1 is configured in a circulation type, enabling waveform equalization and ghost elimination by non-circulation type and circulation type serial connection.

On the other hand, when the switch 31 selects the terminal β, the transversal filters TF1 and TF2 have a non-circular configuration in which the number of taps is (n + 1).

FIG. 2 is a circuit showing an embodiment applied to the ghost reduction device. In this example, the equalizing ICs 36a, 36b, and 36c having the same configuration as the equalizing device of FIG. 1 are connected in series.

The number of taps (n + 1) of the equalizing ICs 36a, 36b, and 36c is 214, and the total number of taps is 642. In addition, l, m, n are l = 29, m = 35, n = 185, for example. The equalizing ICs 36a and 36b and the switches 31a and 31b select the terminal β, and the switch 31c of the equalizing IC 36c selects the terminal r. An image signal is input to the input terminal 37.

This video signal is added to the delay signal of the counter output up to the front end of the counter of the counter filter TF2c provided to each counter of the transverse filter TF2c via the input terminal 34c of the equalizing IC 36C. Is output to As a result, the transverse filter TF2c is configured to remove all ghosts and near ghosts according to the tap coefficient and output them from the output terminal 35c. The video signal is derived from the output terminal 38 and input terminals 1a and 34a of the equalizing IC 36a, input terminals 1b and 34b of the equalizing IC 36b and input terminals of the equalizing IC 36c. IC).

,β를 통해 트랜스버설필터 TF2a에 부여한다. 트랜스버설필터 TF2a의 출력을 출력단자(35a)를 개재하여 등화 IC(36b)의 캐스케이트 입력단자(4b)에 출력된다. 트랜스버설필터 TF1b의 출력은 스위치(31b)의 단자

, β를 개재하여 트랜스버설필터 TF2b에 부여되고, 트랜스버설필터 TF2b의 출력은 출력단자(35b) 및 등화 IC(36c)의 캐스케이트 입력단자(4c)를 개재하여 트랜스버설필터 TF1c에 부여된다. 트랜스버설필터 TF1c의 출력은 스위치(31c)의 단자

, r를 개재하여 트랜스버설필터 TF2c에 부여되고 있다.The cascade input terminal 4a of the equalizing IC 36a is connected to the reference potential point. The transversal filter TF 1a outputs the output according to the tap coefficient given to each counter.

through β to the transversal filter TF2a. The output of the transversal filter TF2a is output to the cascade input terminal 4b of the equalizing IC 36b via the output terminal 35a. The output of the transversal filter TF1b is a terminal of the switch 31b.

is supplied to the transversal filter TF2b via? The output of the transversal filter TF1c is a terminal of the switch 31c.

and r are provided to the transversal filter TF2c.

The input terminals 3a to 3c and the input terminals 32a to 32c in FIG. 2 correspond to the input terminal 3 and the input terminal 32 in FIG.

In the embodiment configured as described above, tap coefficients C-1 to Cn-1 are given to each tap of the equalizing IC 36c. Each tap of the transversal filter TF2b of the equalizing IC 36b is given Cn to Cn + m + l-1, and each tap of TF1b is given a tap coefficient of Cn + m + l to C2n + l-1. The tap coefficients C2n + 1 to C2n + m + 21-1 are assigned to each tap of the transverse filter TF2a of the IC 36a, and the tap coefficients C2n + m + 21 to C3n + 21 are given to each tap of the transverse filter TF1a. -1 is given.

The output of the transversal filter TF2c of the equalizing IC 36c is output to the output terminal 38, whereby all ghosts and near ghosts are removed by the non-circulating type. When the period T of the clock CK is 69.84 ns, the ghost of -2.0 to 2.4 ms of delay time is removed by the transversal filter TF2c.

This output is provided to the equalizing ICs 36a and 36b and the transversal filter TF1c to perform waveform equalization by the circulation type. In this way, the equalization ICs 36a and 36b and the transversal filter TF1c remove ghosts having a delay time of 2.4 to 42.7 GHz.

As described above, in the present embodiment, the transverse filter TF2c is configured in a non-circulatory type to remove all ghosts and near ghosts, and the transverse filters TF1a, TF2a, TF1b, TF2b, and TF1c are configured in a circular type to remove distant ghosts. have. Since the output of the transversal filter TF1c is given to the adder 33 of the transversal filter TF2c, the tap coefficient Cm-1 of the transversal filter TF2c and the tap coefficient Cm of the transversal filter TFic are adjacent to each other on the time axis. Otherwise the retarder 21 is not necessary.

The case where the number of taps of the equalizing IC is 214 has been described. However, when the number of taps is used, for example, when the 320 equalizing IC is used, it is obvious that only the equalizing ICs 36a and 36c need to be omitted.

3 is a circuit diagram showing another embodiment. In FIG. 3, the same code | symbol is attached | subjected to the component same as 2nd figure, and description is abbreviate | omitted.

The present embodiment differs from the embodiment of FIG. 2 in that signals input to the input terminal 37 are input terminals 1a and 34a of the IC 36a, input terminals 1b and 34b and the equalization of the equalizing IC 36b. To the input terminals 1c and 34c of the IC 36c, and the switches 31a, 31b, and 31c all select the terminal β, and to the output terminal 38 without returning the output of the output terminal 35c. It is given.

In such an embodiment, the tap coefficients C3n + 2 l-1 to C-1 are given to each tap of the equalizing ICs 36a to 36a, respectively, and these taps are configured in a non-circulating manner. This can avoid the instability of oscillation that may occur in the circulation type.

In these embodiments, the tap coefficient Cm-1 of the transversal filter TF2c and the tap coefficient Cm of the transversal filter TF1c are set to be shifted by T on the time axis by arranging the adder 33 at the front end of the adder Am-l-2. However, the arrangement of the adders 33 may be set so as to overlap on the time axis.

As described above, according to the present invention, it is possible to configure a series connection of a circulation type and a non-circulation type without requiring a delay device, and the circuit scale can be reduced.

Claims (3)

A transversal comprising a plurality of adder means and a set of multipliers respectively, the first and second transversal filters for equalizing an input sample value according to the tap coefficient applied to the multiplier means and outputting a resultant signal; An equalizing device comprising: coupling means (33) for coupling an output signal of the first transverse filter (TF1) with a signal processed by the second transverse filter (TF2), and the first transverse filter (TF1). And a switching means (31) for sending the first output signal in (2) to the coupling means of the second transverse filter (TF2) or to the input of the second transverse filter (TF2). The method of claim 1, wherein the first and second transverse filters (TF1, TF2) have a predetermined number of adder means, and the coupling means (33) of the second transverse filter (TF2) comprises the first and nth means. And a transversal equalizer positioned between the first adder means. A transversal equalizer comprising a plurality of adders and a set of boosters, each equalizing input sample values according to a tap coefficient applied to the multiplier means and outputting a resultant signal to the first and second transverse filters. A coupling means (33) positioned between the first and nth adder means for coupling an output signal of the first transverse filter (TF1) with a signal processed by the second transverse filter (TF2). Transversal equalizer characterized in that it comprises a).

Images