A multipath sampling circuit includes an input line electrically having an input voltage, a plurality of voltage amplifiers in parallel electrically with one another, each voltage amplifier having a respective input electrically coupled in series with the input line, each voltage amplifier having a different gain and a different saturation voltage; and a plurality of track-and-hold circuits. The track-and-hold circuits have a first state in which a respective input of each track-and-hold circuit is electrically coupled to an output of a respective amplifier. The track-and-hold circuits have a second state in which the respective input of each track-and-hold circuit is electrically decoupled from the output of the respective amplifier. The track-and-hold circuits can be electrically coupled to a summing circuit, a buffer amplifier, or an operational amplifier.

An apparatus is provided comprising a first t-switch, which includes an input port arranged to be connected to a first voltage source, a center-tap port, and an output port arranged to be connected to a load. The first t-switch is configured to connect the input port to the output port in an on mode and disconnect the input port from the output port in an off mode. The apparatus further comprises a bias voltage generation circuit configured to generate a bias voltage, the generated bias voltage coupled to the center-tap port of the first t-switch, the bias voltage determined based upon an output port voltage.

An AD converter includes a plurality of analog input terminals, a reference signal generation circuit that generates an analog reference signal, a sample-and-hold unit that includes a plurality of sample-and-hold circuits sampling the analog reference signal or one of analog input signals from the analog input terminals, a control unit that controls the sample-and-hold unit, and a conversion unit that converts an output signal from the sample-and-hold unit into a digital signal. The control unit controls the sample-and-hold unit to perform the output operation for analog input signal and the sampling operation for the analog reference signal.

A sensor interface circuit comprises an input arranged to receive a sensor signal being an electrical signal representative of an electrical quantity. The electrical quantity includes a physical quantity converted in a sensor. The sensor interface circuit includes conversion means arranged for converting the sensor signal into a digital signal, memory to store sensor characterisation data, signal processing means adapted to obtain a first sensor result by processing the digital signal using the sensor characterisation data, and an output unit arranged to receive and output the first sensor result, the digital signal and the sensor characterisation data.

An MRAM-based vector multiplication device, such as can be used for inferencing in a neural network, is presented that is ultralow power, low cost, and does not require special on-chip programming. A crosspoint array has an MRAM cell at each crosspoint junction and periphery array circuitry capable of supplying independent input voltages to each word line and reading current on each bit line. Vector multiplication is performed as an in-array multiplication of a vector of input voltages with matrix weight values encoded by the MRAM cell states. The MRAM cells can be individually programmed using a combination of input voltages and an external magnetic field. The external magnetic field is chosen so that a write voltage of one polarity reduces the anisotropy sufficiently to align the cell state with the external field, but is insufficient to align the cell if only half of the write voltage is applied.

The present invention provides an analog-digital hybrid architecture, which performs 256 multiplications and additions at a time. The system comprises 256 Processing Elements (PE) (108), which are arranged in a matrix form (16 rows and 16 columns). The digital inputs (110) are converted to analog signal (114) using digital to analog converters (DAC) (102). One PE (108) produces one analog output (115) which is nothing but the multiplication of the analog input (114) and the digital weight input (112). The implementation of PE is done by using i) capacitors and switches and ii) resistor and switches. The outputs from multiple PEs (108) in a column are connected together to produce one analog MAC output (116). In the similar manner, the system produces 16 MAC outputs (118) corresponding to 16 columns. Analog to digital converters (ADC) (104) are used to convert the analog MAC output (116) to digital form (118).

A multi-branch analog multiplexer (anamux) includes protection circuitry to help dissipate both positive and negative injected current without increasing the size of hardening transistors in each branch, thereby avoiding increased leakage current and enabling an analog to digital converter to operate with the required accuracy. The protection circuitry is tied to the body of the hardening transistor to lower the threshold voltage of the hardening device, thereby enabling the hardening device to handle more of the injected current.

The inventive concept relates to a method and system for cost-effectively predicting the dynamic nonlinearities of on-chip segmented digital-to-analog converter (DAC) and analog-to-digital-converter (ADC), by looping a DAC to an ADC, using a programmable-gain-amplifier (PGA) and an external load board. The method may include a first loopback step of supplying an output signal from a coarse DAC, to which a sinusoidal signal is supplied, to a coarse ADC and a fine ADC through an external load board, a second loopback step of supplying an output signal from a fine DAC, to which a sinusoidal signal is supplied, to the fine ADC and the coarse ADC through the load board, and a step of predicting dynamic nonlinearity of each of a DAC and an ADC by processing equations exhibiting dynamic nonlinearity of a sub-DAC and a sub-ADC, which are obtained in the first loopback step and the second loopback step.

A method and circuit for performing multi-layer vector-matrix multiplication operations may include, at a first multiplier-accumulator (MAC) layer, converting a digital input vector using one-bit digital to analog converters (DACs); sequentially performing vector-matrix multiplication operations for the analog DAC signals; and sequentially performing an analog-to-digital (ADC) operation on outputs of the vector-matrix multiplication operations to generate binary partial output vectors. At a second MAC layer, the method and circuit may sequentially receive the binary partial output vectors from the first MAC layer at multi-bit DACs; and sequentially perform vector-matrix multiplication operations to generate a summed binary output for the second MAC layer.

A method for performing vector-matrix multiplication may include converting a digital input vector comprising a plurality of binary-encoded values into a plurality of analog signals using a plurality of one-bit digital to analog converters (DACs); sequentially performing, using an analog vector matrix multiplier and based on bit-order, vector-matrix multiplication operations using a weighting matrix for the plurality of analog signals to generate analog outputs of the analog vector matrix multiplier; sequentially performing an analog-to-digital (ADC) operation on the analog outputs of the analog vector matrix multiplier to generate binary partial output vectors; and combining the binary partial output vectors to generate a result of the vector-matrix multiplication.