Abstract: A channelized active bandpass filter having only two branches which provide respective frequency-selective feed-forward signal paths. The two signal aths have overlapping frequency response bands such that the combination of the two paths provides a composite filter with a bandpass response. The two branches may be provided with bandpass transfer characteristics of different orders and shapes, such as a second-order response and a fourth-order response. Two-way signal splitting and combining to define the two channels may be performed with in-phase splitters and combiners, for example, or diplexer circuits, each composed of two bandpass filters with different characteristics but overlapping frequency responses and preferably approximately equal center frequencies. Combinations of the two splitting and combining arrangements are also usable.

Abstract: A programmable signal translating device, praticularly a transversal filter, includes a wave transport body for transporting a wave along its length, a first channel of electrical conductors having a plurality of code sections spaced along its length for generating phase-coded electrical signals from the wave propagated by the transport body, a second channel of electrical conductors connected in parallel to the first channel and of identical code sections as, but of opposite polarity to, the first channel, and programmable electrical connections between the code sections of the first channel and the code sections of the second channel and the output circuit, for connecting selected code sections of the first and second channels to the output circuit.

Abstract: Digital signal processing (e.g. filtering) apparatus comprises pipelined digital signal processing means for generating each output sample of an output digital signal from input samples of an input digital signal by combining the input samples with a plurality of subsets of a set of filter coefficients associated with that output sample during a respective plurality of sample periods of the input digital signal; and coefficient generating means for supplying sets of filter coefficients to the signal processing means in response to the current state of a filter control signal. The coefficient generating means is operable to supply the plurality of subsets of each set of filter coefficients to the signal processing means during the respective plurality of sample periods of the input digital signal.

Abstract: A set of parameters (W) is calculated for an equalizer which equalizes a transmission line impulse response (LIR) in such a way that the equalized impulse response (EIR) approximates a target impulse response (TIR) with predetermined length (L). An error function (E) is minimized via eigenvalue-eigenvector calculation of a channel dependent matrix (O). This channel dependent matrix (O) is built up from a signal (X), a distorted signal (Y), and length (L) and delay (D) of the target impulse response (TIR). The error function (E) comprises a first contribution which represents the difference between the equalized impulse response (EIR) and the target impulse response (TIR) and a second contribution which represents the energy (E.sub.DC) that is transmitted in unused frequency bands. The eigenvector corresponding to the minimum eigenvalue of the channel dependent matrix (O) represents the set of equalizer parameters (W).

Abstract: An integrator-multiplier-integrator circuit scheme usable in transverse fers, a transverse filter employing such a circuit, and a method for using each. The multiplier-integrator-multiplier has a capacitatively loaded integrating amplifier fed by a transistor. The current through the transistor, and hence the time it takes to charge the integrating capacitor, depends largely on the bias of the transistor, not the size of the capacitor, permitting one to set and control integration time by setting the transistor's parameters, and controlling its bias, effectively controlling integration time by us of only one semiconductor device. An additional circuit for auto-zeroing (i.e. canceling quiescent offset) increases adaptivity of the circuit. Preferably the phase of inputs to the first multiplier is made selectably variable to minimize phase difference at the multiplier, thus increasing circuit stability.

Abstract: A simulator for simulating the attenuation characteristics of a length of coaxial cable in a cable television (CATV) network employs a diplex filter to separate the CATV signal and AC power into at least two separate signal paths. The AC power path has a resistance selected to simulate the DC loop resistance of the desire length of cable. The CATV signal path is in parallel with the AC power path, and has an attenuation selected to simulate the characteristics of the cable at CATV frequencies. In the preferred embodiment, the simulator is equipped with separate signal paths for the upstream and downstream CATV signals that have attenuations selected to simulate the characteristics of the cable within the frequency bands used for upstream and downstream CATV signals, respectively. The resistor in the AC power path has a sufficiently high power rating to handle realistic AC power levels.

Abstract: A continuous-time analog filter receiving a continuous-time analog input signal is provided. The filter utilizes a plurality of delay cells for receiving the input signal, and then processes the input signal through taps having specific weight values. The signal is multiplied by the weight factors and then summed to produce a continuous-time analog output signal.

Abstract: In an equalizer with n number of frequency ranges for digitized signals in which by means of n-1 cascaded filter circuits containing parallel connected digital low-pass filters and digital high pass filters whose respective frequency responses are linked by complementary transfer functions. A desired amplitude response is implemented via an increase or decrease in the amplitudes of adjacent frequency ranges in the associated filter circuit by weighting the outputs of the digital high-pass and low-pass filters, and a control unit forms the respective weighting factor from an equalizer control signal which contains the desired amplitude response as information.

Abstract: Digital signal processing (e.g. filtering) apparatus comprises pipelined digital signal processing means for generating each output sample of an output digital signal from input samples of an input digital signal by combining the input samples with a plurality of subsets of a set of filter coefficients associated with that output sample during a respective plurality of sample periods of the input digital signal; and coefficient generating means for supplying sets of filter coefficients to the signal processing means in response to the current state of a filter control signal. The coefficient generating means is operable to supply the plurality of subsets of each set of filter coefficients to the signal processing means during the respective plurality of sample periods of the input digital signal.

Abstract: An analog, adaptive generalized transversal equalizer for use in the filtering system of a disc drive PRML read channel, the transversal equalizer employing the use of non-ideal delay elements. The filtering system comprises the equalizer connected in series with an adaptive, analog prefilter. The prefilter is comprised of a plurality of serially connected, adaptive, analog filter stages having variable transfer functions determined by adaptive parameter signals received by the filter stages. The generalized transversal equalizer comprises a plurality of serially connected, adaptive, analog low pass filters, having taps on either side of each low pass filter, a plurality of multipliers that receive signals at the tap locations of the delay circuit, and a summing circuit that receives the outputs of the multipliers.

Abstract: A high speed signal transport system (10) provides data transmission across a balanced channel (14). The balanced channel (14) carries differential signals from the output of a differential transmitter (12). An equalizer (18) compensates for distortion in the balanced channel in order to reduce timing jitter of the signal. The signal is applied to the input of a differential receiver (20) which exploits the precise quantization threshold inherent in differential data to accurately convert the signal to digital data. The differential receiver (20) also employs a known frequency reference along with digital transition detection to rephase the digital data to the local clock. No duty-cycle limitation is placed on the signal carried by the balanced channel (14) and therefore the differential transmitter (12) and differential receiver (20) are not required to encode the signal.

Abstract: A signal-to-noise enhancer 10 includes a first 90 degree hybrid set 20. A first output end of the first 90 degree hybrid set 20 is connected to an input end of a limiter used a magnetostatic wave element 62a utilized the magnetostatic surface wave mode. Also, a second output end of the first 90 degree hybrid set 20 is connected to an input end of a filter used a magnetostatic wave element 62b similar to the magnetostatic wave element 62a in structure, via a resistor 32 as a first attenuator. Furthermore, an output end of the limiter is connected to a first input end of a second 90 degree hybrid set 40 via a resistor 52 as a second attenuator. Also, an output end of the filter is connected to a second input end of the second 90 degree hybrid set 40.

Abstract: The relationship between the input and output of a signal transfer circuit, such as a tune circuit, a filter, a phase shifter, an amplifier, etc., is determined by the circuit response to the input signal. The value of the circuit response (R) is controlled by the value (V) of a control signal supplied thereto. According to the invention, the relationship between the latter values is approximated by a linear regression function based on a number of measured first pairs of response/control signal values and based on such regression function a transcoder determines control signal values for providing response values of the circuit in accordance with second pairs of response/control signal values.

Abstract: A noisy input signal x is filtered with a restoration filter of median type to generate a filtered input signal y. The sum of the absolute differences between filtered and unfiltered signal is calculated for each position of a sliding window within the input signal representing a local estimate of the noise, and is combined with a global measure of the input signal noise to compute two coefficients a and b which are respectively applied to the unfiltered and filtered signal to generate the output signal z=a*x+b*y which is both globally and locally adapted to the structure of displayed images. Advantageously different kinds of filters operate in parallel, whereby the kind of filter elected is locally adapted to the picture activity.

Abstract: The sweep rate limitations that heretofore have constrained the maximum sweep rates of swept analysis instruments are obviated by optimizing filter circuitry and post-processing the IF signal using various techniques to compensate for errors caused by fast sweeping. Some selective windowing is also used to compensate for distortions due to fast sweep rates.

Abstract: A sigma-delta (.SIGMA..DELTA.) analog-to-digital converter (ADC) accepts band-limited analog signals, and subtracts an analog replica of an output pulse- or amplitude-density modulated (ADM) signal therefrom to produce an error signal. The error signal is processed by an analog filter or resonator with a nondelayed forward path and a tapped nonaccumulating delay line, and summed feedback and feedforward weights coupled to the taps, to thereby produce a resonated signal. An ADC processes the resonated signal, and produces the ADM signal. The ADC undesirably produces quantization noise. A digital-to-analog converter (DAC) noiselessly converts the PDM signal into the analog replica, to aid in forming the error signal. In a particular embodiment of the invention, the resonator includes a recursive analog transversal filter with delays and linear weighting elements for linearity and high operating speed. The ADC may be in a high-speed system such as a radar.

Abstract: A digitally controlled programmable transversal filter (DCPTF) employing a lithium niobate surface acoustic wave (SAW) delay line and two large scale integration (LSI) gallium arsenide integrated circuits to digitally control the magnitude and sign of the 32 tap weights from the delay line. The DCPTF results in a significant reduction in size over the prior art with little sacrifice in performance. The DCPTF is completely programmable and is constrained only by the bandwidth and the number of taps.

Abstract: A transverse filler for use in a CCD camera eliminates KTC noise even at very fast pixel rates with relatively relaxed constraints on component accuracy and timing control, The circuit includes summation, divider and subtraction circuits which add the plural outputs of a tapped delay line to which a video signal is applied and generate average KTC noise and KTC noise plus signal representations. The output of the circuits is applied to the input of an capacitor and the output of the capacitor is connected to an electronic switch and to a Sample-and-Hold amplifier. The capacitor is operative to subtract the average KTC noise from the average KTC plus Video signal, The circuit is able to run on-board binning and does not have pixel feedthrough problems.

Abstract: The use of one or more multi-gate (e.g., dual-gate) FETs are employed in an RF or microwave delay line. The carrier drift velocity in each multi-gate FET is controlled in accordance with the variable magnitude of a delay-control voltage applied between its drain and source, thereby controlling the time delay experienced by an RF or microwave signal traveling between spaced first and second gates of a multi-gate FET. The gates of a plurality of multi-gate FETs may be serially-coupled through amplifying circuits to produce a delay chain in which the total delay is the sum of the delays of all the multi-gate FETs in the chain. A single delay-control voltage, which can be continuously variable, may be used to control the total delay provided by all multi-gate FETs in the chain. Alternatively, a separate delay-control voltage, which can be independently continuously variable, may be used for independently controlling the delay provided by each individual multi-gate FET in the chain.

Abstract: A limited-bandwidth microwave transversal or recursive filter, compatible with microwave monolithic integrated circuit (MMIC) design requirements is described in this invention. The device comprises a means to split an incident signal into multiple parts for distribution among a plurality of input ports of frequency-selective feedforward and feedback network branches comprising filter elements and active devices so designed as to provide a desired degree and type of signal filtration. After filtration, the resulting signals from these branches are combined to form a composite filter output signal.

Abstract: An input-weighted transversal filter having a first input circuit for inputting a sampled input signal which is sampled at every predetermined period, registers for storing a predetermined number of tap weights, the register selectively supplying the tap weights in a predetermined order, a plurality of tap weight multipliers for multiplying the sampled input signal and a tap weight supplied from the registers, a pipeline processing adder for adding the multiplied outputs from the tap weight multipliers with each other and a second input circuit for inputting signals to the pipeline processing adder, the second input circuit having at least one input coupled to the output of the pipeline processing adder.

Abstract: A waveform equalizing filter unit includes a waveform equalizing filter with a plurality of taps each having a unitary delay element and a coefficient multiplier, a clock switching circuit for selecting a clock signal from at least two different frequencies, with the selected clock signal being fed to each of the unitary delay elements of the waveform equalizing filter, and a control circuit for selectively switching the tap coefficients of the coefficient multiplier in the waveform equalizing filter based upon which one of the different clock signals has been selected. A signal time adjuster is further included in a stage situated before the waveform equalizing filter that receives the input signal prior to being fed to the waveform equalizing filter to adjust the timing of the input signal in synchronism with the operation of selecting one of the different clock signal frequencies.

Abstract: In this digital signal measurement apparatus, an approach is employed to deliver a measurement signal from a measurement signal generator to a measured circuit to transform the signal on the time base through the measured circuit to a signal on the frequency base by a frequency base transform circuit and to further obtain a signal on the time base by a time base transform circuit. Thus, a difference between the signal on the time base through the measured circuit and the signal on the time base from the time base transform circuit is employed. Thus, for example, even if the measured circuit is a linear system, a measured result in the digital region and a result of the analog measurement can be in correspondence with each other. Accordingly, gain correction of a measurement signal is unnecessary, thus making it possible to prevent an increase in an error of a measured result of S/N.

Abstract: A tapped delay line uses a varistor network for tap weighting. The varistors may be used in a resistive summing network or as amplifier gain control elements. In addition, the varistors can be programmed by a voltage programming circuit comprising a switch array, an input voltage programming bus, and sets of master and slave digital programming signals.

Abstract: An equalizer (10) for removing parabolic phase distortion from an analog signal (3), utilizing a pair of series connected transversal filters. The parabolic phase distortion is cancelled by generating an inverse parabolic approximation using a sinusoidal phase control filter (18). The signal (3) is then passed through an amplitude control filter (21) to remove magnitude ripple components.

Abstract: A signal modification circuit receives an input signal and produces an output having a modified characteristic. When the signal modification circuit is embodied in an audio circuit receiving an analog audio signal having a first variant level at its input, the audio circuit produces an output signal having a second variant level in which a frequency characteristic in a high frequency range is modified so that the second variant level decreases in accordance with a function of l/fm (f being frequency and m being an integer determined depending upon the nature of the audio signal) relative to the first variant level while being trapped in selected discrete frequencies. For this purpose a combination of a low-pass filter and a plurality of trap filters connected in series with one another, or a combination of a low-pass filter and a plurality of band-pass filters connected in parallel to one another.

Abstract: A technique for generating the tap-weight coefficients for a signal dispersion cancelling filter utilizes circuitry which is responsive to received signal samples and generates a time-ordered sequence of derived signal samples. The tap-weight coefficient circuitry includes filters which operate in response to this sequence. One such filter is responsive to the derived signal samples in a first portion of the sequence after the time-order of the derived signal samples therein is reversed and another filter is responsive to the derived signal samples in a second predetermined portion of the sequence. This technique is applicable to a variety of communications systems including television and is particularly suitable for integrated circuit implementation.

Abstract: In a high-pass filter circuit comprising a high-pass filter having a transfer function in accordance with the following equation: ##EQU1## a compensation network is connected in series with said high-pass filter, said network comprising a summing circuit, a continuous line and a low-pass filter connected in parallel therewith, said low-pass filter having a transfer function in accordance with the following equation ##EQU2## and the filter coefficient d.sub.K-1 of the low-pass filter being chosen to be equal to the difference of the filter coefficients b.sub.n-1 -a.sub.n-1 of said high-pass filter.

Abstract: A high-frequency integrated circuit continuous time low pass filter. An on-chip oscillator is built into the filter. The filter frequency accuracy is established by trimming the frequency of the on-chip oscillator during wafer probe. The oscillator is off during normal operation of the filter. Therefore, the filter does not produce noise that will degrade the performance of the filter during normal operation. After trimming, the filter design is such that accuracy is maintained even during temperature and power supply changes. The adjustment can be made without the oscillator by direct measurement of the filter response.

Abstract: A SAW device tapped delay line includes an apodized input IDT (inter-digital transducer) and a plurality of substantially identical unapodized output IDTs arranged to receive, after delays which successively increase by an amont T, a SAW propagated from the input IDT. Each output IDT is terminated with a low impedance which is constituted by the low input impedance of a buffer amplifier, to minimize acoustic regeneration by the output IDTs. Grounded dummy fingers with the same metallization ratio as the output IDTs are provided between adjacent output IDTs to provide a constant SAW velocity. The delay line is particularly suited to use in an equalizer of a microwave radio transmission system, in which the delay T is the inverse of the symbol rate of the system. Intermediate frequency automatic time domain equalizer (IF ATDE) arrangements using the SAW device tapped delay line are described.

Abstract: A transversal filter comprising: a series connection of M+L delay cells (12) defining M+L+1 signal samples; M+L+1 branches each having a midpoint connected to receive a respective one of said signal samples and first and second multipliers (10, 11) for multiplying its signal sample by respective first and second coefficients; and first and second summing circuits (13, 14) respectively connected to sum the outputs from said first multipliers (10) and from said second multipliers (11), wherein a wideband phase shifter (17) is associated in series with each delay cell (12), said phase shifter providing a phase shift .delta..phi. such that .delta..phi.=.OMEGA..sub.0 T (modulo 2.pi.) where .OMEGA..sub.0 is the angular frequency of the intermediate carrier and T represents the duration of the delay due to a delay cell (12).

Abstract: A digital filter operating at intermediate frequency, having a transverse filter; a recursive filter; a device for storing the different coefficients of the two filters; a D-A conversion device situated at the output of this storage device; a device for addressing this storage device; and controls for the addressing device.

Abstract: Transversal and recursive filters having an input propagation lumped element network with a band pass filter response, a low pass filter response, or a high pass filter response, and an output propagation network having a low pass frequency response, a high pass frequency response, or a band pass frequency response are described. Weighting means including a plurality of field effect transistors are used to successively couple the input propagation network and the output propagation network. The general shape or response of the transversal filter is provided by choosing the characteristics of the input and output propagation networks in relation to the desired filter response. With the general shape provided by the propagation networks rather than only by the transversal elements, fewer transversal elements are required to provide a desired response.

Abstract: A new structure for electro-magnetic signal filters is described. The structure of the invention incorporates ideal time lag elements having delay values selected to match an arbitrarily selected set of frequencies matched to a desired frequency response and connected in a cascade array by second filter elements which null the frequency filtered by the preceding array, the output of which is summed to provide a composite output with the desired frequency response. The resultant filters are easily implemented in conventional hardware and the techniques may be applied to a broad range of the electro-magnetic spectrum by the choice of appropriate components.

Abstract: An image processing apparatus is provided which is capable of real time image processing, especially filtering, without generating a delay in a frame memory. A cascade filtering circuit filters digital video signals obtained by digitally converting analog video signals to obtain a time sequential digital pixel data stream for the filtering operation based on a pattern of predetermined constants of a p-rows, q-columns. The cascade filtering circuit uses p basic filter circuits (100) cascade-connected with each other. Each basic filter circuit carries out the filtering for one-row and q columns of pixel data. Each basic filter circuit (100) includes delay circuits (40, 41, 42), multipliers (20, 21, 22), and adders (30, 31, 32).

Abstract: A magnetic article surveillance system utilizing microcomputer control and unique time domain and frequency domain information gathering channels whose information is processed by the microcomputer via preselected time domain and frequency domain criteria.

Abstract: A very rapidly converging adaptive filter which uses a variable scale factor for each weight of the filter. The value of the variable scale factor is chosen for each iteration and is based upon the sign changes of the incremental weight change. The variable scale factor exhibits large values when no sign changes occur and smaller values when sign changes occur. The new filter provides considerable improvement in increase of convergence rate and decrease in residual errors even in the presence of heavy noise while requiring only a modest increase in hardware.

Abstract: An I-Q channel adaptive line enhancer has bandpass filters with passbands that can be widened to well beyond the band that is actually being processed by the adaptive line enhancer. This eliminates the requirement of filtering one of the sidebands as is required in single channel adaptive line enhancers, in which the sum and difference sidebands are generated on up-conversion. Band limiting is done with lowpass filters, and the local oscillator may be made variable so that any part of the band that is passed by the bandpass filters can be processed by a single I-Q channel adaptive line enhancer. This type of I-Q channel adaptive line enhancer can process any one of a large number of possible bands without tunable bandpass filters, or without banks of adaptive line enhancers which have been preset to adjacent sub-bands.

Abstract: Adaptive transversal filter for use with three-signal level signals, arranged to cancel shifts in the two extreme signal levels relative to the central signal level, information signaling whether the signal level is the central signal level being coupled to first delay means (2) having N+1 connecting terminals, and information signaling whether the signal level is either one or the other extreme signal level being coupled to second delay means (4) having N+1 connecting terminals and comprising first and second coefficient registers (C.sub.OO -C.sub.ON ; C.sub.O -C.sub.sN) each containing N+1 coefficients; means (8, 9) for always multiplying one of the coefficients by an associated connecting terminal signal of the delay means, means (10) for adding together all the multiplying results obtained for providing a cancellation signal and means for always blocking the signal from a connecting terminal of the second delay means when this signal represents neither the one nor the other extreme signal level.

Abstract: A correlator apparatus that simultaneously delays two time-varying input signals using just a single transmission line, with the signals being connected to opposite ends of the line. Directional couplers spaced along the transmission line extract delayed versions of the two input signals, for correlation by a bank of correlators, to produce a correlation envelope for the two signals. The apparatus is substantially less massive than prior apparatus of this kind and can be easily constructed and aligned.

Abstract: A hybrid programmable transversal filter (HPTF) is described that employs a LiNbO.sub.3 SAW delay line and two monolithic dual-gate GaAs FET arrays to control magnitude and sign of the 16 tap weights. The HPTF is completely programmable and is constrained only by the bandwidth (100 MHz centered at 250 MHz) and the number of taps. Theoretical calculations of tap weight control range and dynamic range are presented, compared with experiment and used to justify the hybrid LiNbO.sub.3 SAW - GaAs FET combination. Dynamic range of 85 dB and a continuously variable tap weight control range of 70 dB are demonstrated.

Abstract: An adaptive filter may be realized by weighting the outputs from a tapped delay line. This, however, allows nulls to corrupt a wanted signal spectral region. A fixed (non-adaptive) filter (2) coupled between the signal input and an input of means (3) for summing the weighted outputs serves to allow nulls only to be formed in the skirt region of the fixed filter response thus preserving a wanted signal.

Abstract: A monolithic programmable signal processor with piezoelectric insulator FETs and an InP substrate. The piezoelectric insulator FETs are capable of being constructed on the same substrate with circuits associated with the signal processor and optical components.

Abstract: A CCD delay line may be tapped using a floating-diffusion electrometer at the tap. Charge sensing is made non-destructive by not resetting the floating diffusion to a reset drain after charge sensing. The resulting charge integration on the floating diffusion causes a smearing of the samples described by successive charge packets. The smearing is in the baseband frequencies of the sample frequency spectrum, but does not appreciably affect the subspectra surrounding harmonics of the delay line clock rate. Consequently, smear-free response to the floating-gate electrometer output signals can be obtained by synchronously detecting them at a harmonic of the delay line clock rate. Using floating-diffusion electrometers, rather than floating-gate electrometers, to sense charge packet amplitudes at taps along a CCD delay line lowers the noise in the output response of output-weighted, charge-coupled-devive transversal fiters.

Abstract: A broadband microwave recursive filter that provides sharp transitions in the frequency domain between adjacent stop and passbands comprising a signal input node; a signal output node; a filter circuit connected between the signal input node and the signal output node for providing a signal flow therebetween which has a predetermined frequency bandwidth characteristic; a microwave transistor circuit, with the microwave transistor circuit being band-limited to provide gain in only a restricted window of frequencies within the predetermined frequency bandwidth and connected for providing amplification to signals flowing in the filter circuit between the signal input node and the signal output node while suppressing out-of-window signals resulting from design approximations.

Abstract: A transversal filter is provided which includes a delay circuit comprising a plurality of cascaded saturating circuit elements. The delay circuit has a series of taps from which signals with varying delays are obtained. The obtained signals are combined to form filtered signal(s).

Abstract: An integrated circuit for filtering signals by having cascaded switched capacitor sampling filters. The circuit includes a transmit section which has an anti-aliasing filter, a core section filter, a highpass filter and an encoder for providing analog-to-digital conversion. Each successive filter is sampled at a lower rate to inhibit anti-aliasing. The circuit also includes a receive section which has a digital-to-analog decoder, an output buffer, a receiver core filter and a power amplifier.

Abstract: A programmable charge coupled device transversal filter 5 includes a charge coupled device register 10 for receiving and delaying incoming analog signals, a series of floating gate charge detectors 15, a corresponding number of sets of binary scaled capacitors C.sub.0, . . . 2C.sub.0 . . . 2.sup.n C.sub.0, an output circuit including a positive and negative bus coupled to a differential amplifier, and mask or otherwise definable electrical connections for connecting selected ones of the scaled sets of capacitors between the floating gate 15 corresponding to that set and one of the positive and negative buses 22 and 23.

Abstract: The invention relates to a charged-coupled transversal filter of the parallel-in series-out (PISO) type having a buried channel of one conductivity type and zones diffused into the channel of the other conductivity type, which zones constitute the clock electrodes. The inputs also comprise a zone of the one conductivity type which is provided in the associated zone of the second conductivity type. The weighting factors are formed by selecting the sizes of these zones, which constitute the emitters of bipolar transistors, whose bases are constituted by the clock electrodes and whose collectors are constituted by the channel. The signal to be filtered is introduced by first converting the signal into a current and by then distributing the latter over the various emitters. Stages having positive and negative weighting factors can be readily combined to form a CCD.

Abstract: A space-time adaptive filter system is provided for eliminating unwanted signals from a radar or communication system. The filter system receives a main channel and several auxiliary channels wherein the target signal is not correlated between the various signal channels. Correlated noise components are eliminated by decorrelating the signals. The adaptive filter includes a Gram-Schmidt processor for sequentially decorrelating the auxiliary signals from the main signal. Each decorrelation element of the Gram-Schmidt processor comprises a transverse orthonormal ladder filter.