JPS6252968B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a transversal filter using switched capacitors as an LSI filter.

As a conventional transversal filter,
Some use CCDs (charge-coupled devices). This is obtained by detecting the amount of charge at each node of the CCD delay line, weighting the value with a certain coefficient, and adding them. The principle is shown in FIG.

In the figure, 1 is the input terminal, D ₁ to _{D o} are CCD delay lines, and V ₁ to
V _o is the node voltage of each CCD delay line, 2-1 to 2-
n indicates a weighting coefficient unit, 3 indicates an adder, and 4 indicates an output end. This is an example of an n-stage delay line filter, which is obtained by an input signal.

Assuming that the voltage at each node is V _K , the weighting coefficient applied to it is h _K (K=1, 2...n), and the sampling frequency is _C , which is the same as the period T _C of the clock pulse, the output from output terminal 4 is Vout
teeth becomes.

An example of a CCD transversal filter using this principle is a split-electrode transversal filter. This means that the Kth electrode is (1+h _K ): (1-
h _K ), and the output from each electrode is detected by an Op-Amp.

In addition, CODECs using switched capacitors (hereinafter referred to as S・C・F) as LSI filters
Research and development is being carried out on filters for

This filter is composed of a switched capacitor, an operational amplifier (hereinafter referred to as Op-Amp), and the like. In the same way, Op-Amp is used in both S・C・F and split electrode CCD transversal filters. When Op-Amp is incorporated into an LSI filter, it is subject to the following restrictions.

(1) When using at high gain, the frequency band used cannot be widened.

(2) In the configuration of a high-order filter, Op-Amp also increases by the order, so the number of Op-Amp also increases, and the power consumption by Op-Amp increases, making it difficult to create a high-order filter.

The present invention provides an LSI filter that does not use Op-Amp in order to eliminate the above-mentioned runner-up.

In the filter configuration of the present invention, n×n capacitances are arranged in n rows and n columns, each capacitance in the row is weighted to a value according to the coefficient of the transversal filter, and each capacitance is The input signal is charged by connecting a signal every T time for each capacitor to the input side for τ (<T) time using a switch, and one capacitance for each row is sequentially charged for the remaining T−τ time of the cycle. , are selected in the order according to the transversal filter configuration, and are connected to the output end to obtain an output signal.

Since this method does not require an Op-Amp in the filter configuration, (1) the frequency of use can be extended to the high frequency band;

(2) Since C/MOS switches are used as active elements, power consumption is extremely low compared to Op-Amp. Therefore, high-order filters can be designed.

An LSI filter with the following advantages can be realized.

The invention will now be explained with reference to the figures.

FIG. 2 shows a time chart of clock pulses of the transversal filter of the present invention.

In the figure, 2-1 is the write signal clock pulse.
2-2 represents a read clock pulse, and the horizontal axis represents time t.

Of the write signals 2-1, 2-1-1 operates the switch S ₁ j, and the capacitance is increased by the clock pulse.
C ₁₁ , C ₁₂ , C ₁₃ , . . . C _1o is charged with an input signal.

2-1-2 is a clock pulse that operates switch S ₂ j, 2-1-3 is a clock pulse that operates switch S ₃ j, and 2-1
-4 is a clock pulse that operates switch S ₄ j, ... 2-1-n is a clock pulse that operates switch S _o j, respectively.
Capacitance C ₂₁ , C ₂₂ , C ₂₃ ,...C _2o , C ₃₁ , respectively
C ₃₂ , C ₃₃ ... C _3o , C ₄₁ , C ₄₂ , C ₄₃ ... C _4o , ... C _o
1 , _Co2 , _Co3 ... _Coo is charged.

2-2 of read clock pulses 2-2
-1 is a clock pulse that operates switch Sj (n-j+2), and 2-2-2 is a clock pulse that operates switch Sj (n-j+2).
3), 2-2-3 is a clock pulse that operates switch Sj (n-j+4), . . . 2-2-n is a clock pulse that operates switch Sj (n-j+1), respectively.

FIG. 3 diagrammatically explains the writing and reading of input signals of the transversal filter of the present invention.

Writing and reading are performed using the clock pulses shown in FIG.

For convenience of explanation, 64 capacitors are arranged in 8 rows x 8 columns, the capacitances in each row are weighted to suit the desired transmission characteristics, and the capacitances in each column are the same. Suppose there is. In the figure, 7 is the capacitance row and 3
1, 32, . . . 38 show charging status using input signals. 8-1 represents reading of 31, 22, and 13, and 8-2 represents reading of 84, 75, 66, 57, and 48.

The numbers in the figure are capacitance suffixes, for example 11 means C ₁₁ and 12 means C ₁₂ .
Further, V _3T , V _2T , V _1T , V _0T , and V _-T represent the potentials of the capacitances charged by the input signal, and T represents the time interval.

The write clock pulse 2-1 in FIG. 2 causes capacitance rows 11, 12, 13, . . . from the top of the diagram.
. . 18, 21, 22, 23, . . . 28, etc. are charged with input signals for each row at intervals of T time.

The write clock pulse 2-1-3 in FIG. 2 charges the capacitor rows 7 simultaneously for a period of τ (<T), and the read clock pulse charges the capacitors 31, 22 at the next time T-τ. ,13,84,
Charges 75, 66, 57, and 48 are read out.
The readout output Vout (3T) at that time is Vout (3T) = C ₃₁ V (3T) + C ₂₂ V (2T) + C ₁₃ V (1T) + /C ₈₄ V ( _0T ) + C ₇₅ V (-1T) +...+C ₄₈ V (-4T)/ Here C _T =C ₃₁ +C ₂₂ +C ₁₃ +C ₈₄ +C ₇₅ +...+
C ₄₈ .

The next capacitance rows 41, 42, 43...48 are also written and read in the same way as above, and the read output Vout (4T) is Vout (4T) = C ₄₁ V (4T) + C ₃₂ V (3T )+ _C23V (2T)+/ _CTC85V (1T)+ _C76V (0T)+...+ _C58V (-3T)/Writing and reading are performed in the _same manner below.

FIG. 4 shows an embodiment of the present invention, in which capacitors with switches are arranged in n rows and n columns, where 11 is an input terminal, 12 is an output terminal, and 13 is an output terminal.
A coefficient unit with transversal filter weighting coefficients of A ₁ to _{A o} . For each capacitance in a capacitance row, for example 14, C ₁₁ is the weighting coefficient of A ₁ , C ₁₂ is A C ₁₃ is given a weighting coefficient of A ₃ , ... C _1o is given a weighting coefficient of _{A o .} 15 and 16 are also similar capacitance rows. The capacitance of each column is the same capacitance. 17 is a group of switches directly connected to capacitance, and has an input, an output, and a middle point. In the figure, switch S ₁₁ is connected to capacitance C ₁₁ , S ₁₂ is connected to C ₁₂ , and S ₁₃ is connected to C ₁₃ .
...S _1o is connected to C _1o , respectively.

Switch groups 18 and 19 are also switches of the same type as 17; C ₂₁ is connected to S ₂₁ , C ₂₂ is connected to S ₂₂ , and C _2o is connected to S _2o.
, C _o1 is connected to S _o1 , C _o2 is connected to S _o2 , and C _oo is connected to S _oo . 20 is ground.

Figure 5 is a diagram showing a combination of a switch and capacitance to give a positive coefficient to capacitance. In the figure, 21 is a switch, 21-1 is a contact on the input side, and 2
1-2 is the middle point, 21-3 is the contact on the output side, 21-
4 is a contactor connected to each contact, and 22 is a capacitance.

In this case, the input signal from the input terminal IN is 21
When the capacitance 22 is charged through the contact 21-4, the charge becomes positive with respect to the ground.
1-3, the input signal is output from the output terminal OUT in a positive state.

FIG. 6 shows a combination of capacitance and switch used to provide a negative coefficient.

In the figure, 23 and 25 are switches, and 24 is a capacitance.

When an input signal is input to the input terminal IN, the capacitance 24 is charged through the switch 23 23-1. At this time, the charge on the ground 20' side connected to 25-1 is negative.

When this charge is output, contact 23-4 of switch 23 is connected to 23-3 and connected to ground 20, and contact 25-4 of switch 25 is connected to 23-3 and connected to ground 20.
-3, the sign of the capacitance is inverted, and the input signal is output as a signal given a negative coefficient.

The positive coefficients and negative coefficients explained with reference to FIGS. 5 and 6 are used depending on the transmission characteristics of the transversal filter.

Figure 7 shows an electronic switch used in the present invention.
Terminal 26 with transistor switch by MOS
The switch is turned on by the next clock pulse,
It will be turned off. In the figure, IN is an input terminal, OUT is an output terminal, and 27 is an inverter that inverts the clock pulse. 28 and 29 are P channel MOS transistors, 30 and 31 are N channel MOS transistors, and 28 and 30, 29 and 31 are C channel MOS transistors.
It is a MOS transistor. Cij is capacitance.

In the figure, when a clock pulse is input from 26,
The clock pulse is input to MOS transistors 29 and 30, and at the same time, the clock pulse is inverted by an inverter 27 and input to MOS transistors 28 and 31. 28,3 when clock pulse is high level
0 is conductive and 29 and 31 are non-conductive.
The input signal from IN passes through electronic switches 28 and 30 and charges the capacitance Cij. Next, the clock pulse goes from High Level (Figure 2) to Low Level.
(Fig. 2), 28 and 30 become non-conductive, 29 and 31 become conductive, and switch 2
9 and 31, the charge of Cij is connected to the output terminal OUT side.

Figure 2 shows the operation of the transversal filter.
This will be explained using FIG.

In FIG. 4, an input signal is input from the input terminal 11, and switches S ₁ j (S ₁₁ , S ₁₂ . . . S _1
o ) is connected to the input side, the input signal charges the capacitance rows C ₁₁ , C ₁₂ . . . C _1o for a time τ (<T). Next, after T time from t ₁ , the switch S _2j (S ₂₁ , S ₂₂ , S ₂₃ . . . S _2o ) is operated by the write pulse 2-1-2, and the capacitance row C ₂₁ , C ₂₂ , C ₂₃ …
C _2o is connected to charge the input signal.

The capacitance of each row is charged by the input signal at an interval of time T, and then, after t ₁ −t ₀ =T/2 hours, the switch Sj (n−j+2) charges the capacitance at t ₃ −t ₀ =T+T/2. After _t5 - _t0 = 2T+T/2 hours, switch Sj (n-j+4) operates and connects to the output side, charging T-T/2=T/2 hours. This situation, in which the input signal is output, is as already explained in FIG.

Now, the input charge charged to the capacitance of row i is Qij=Cij Vin[it], j=1~n, and the output voltage Vout(iT) is Here a=i+1-((n-j+i+1)) (( )) is an integer modulo n. Now, in equation (2), l=((n-j+i+1)), hl
If =C ₀・A(l), then This represents a transversal filter. The above can be realized as an LSI filter by capacitor switching a transversal filter that does not use an Op-Amp.

[Brief explanation of the drawing]

Figure 1 is a principle diagram of a transversal filter, Figure 2 is a time chart of clock pulses that operate the transversal, filter write, and read outputs of the present invention, and Figure 3 is a write signal and read output of the present invention. FIG. 4 is a diagram showing an embodiment of the present invention, FIG. 5 is a diagram showing a circuit giving a positive coefficient, FIG. 6 is a diagram showing a circuit giving a negative coefficient, and FIG. 7 is a diagram showing a transistor switch. FIG.

Claims (1)

[Scope of Claims] 1. Capacitors are arranged in a matrix of n rows and n columns, each capacitor has a changeover switch connected to input and output terminals, and each row of each capacitor The capacitance of each capacitance is weighted by the coefficient of the transversal filter, and when charging of the capacitance starts from row I and column J, the capacitance of each row is sequentially increased in the direction in which the number of rows increases by 1. The changeover switch is controlled so that the input signal is charged for τ (<T) time at T time intervals, and the capacitance is accumulated in the direction in which the number of rows decreases by 1 and the number of columns increases by 1. 1. A transversal filter characterized in that the changeover switch is controlled to output charges to an output terminal to obtain an output signal.