Abstract: Methods, systems, and devices for thermometer coding for driving non-binary signals are described. A set of drivers may be used to drive a signal line, with each of the drivers calibrated to have different individual drive strengths. To drive a signal line to successive voltages in accordance with a non-binary modulation scheme, additional individual drivers of the set may be used. The different drive strengths of the individual drivers of the set may scale in non-linear fashion, which may offset non-linearities associated with the individual drivers as additional individual drivers of the set are activated.

Abstract: The number of bits in the quantizer can be decoupled from the number of bits in the feedback digital-to-analog converter (DAC) A delta-sigma analog-to-digital converter circuit can include a first quantizer to generate an output having a first number of bits and a second quantizer coupled to an output of the first quantizer, where the second quantizer can receive the output of the first quantizer and generate an output having a second number of bits. The feedback DAC can be coupled to the second quantizer to receive a representation of the output of the second quantizer, where the output of the feedback digital-to-analog converter circuit has the second number of bits. These techniques can reduce the area of the feedback DAC, e.g., 4 or 5 bits, and the techniques can achieve a higher maximum stable amplitude (MSA) because it is effectively a second order loop.

Abstract: A series-connected delta-sigma modulator (DSM) comprises a first DSM, configured to receive an input signal, comprising a first loop filter, configured to generate a first processed signal; and a first quantizer, coupled to the first loop filter, configured to generate a first quantized signal, and to feed back the first quantized signal to the first loop filter, wherein the first quantized signal comprises a clipping error smaller than a first predetermined value; and a second DSM, coupled to the first DSM, configured to receive the first quantized signal from the first DSM, comprising a second loop filter, configured to generate a second processed signal; and a second quantizer, coupled to the second loop filter, configured to generate a second quantized signal, and to feed back the second quantized signal to the second loop filter, wherein the second quantized signal comprises a quantization error smaller than a second predetermined value.

Abstract: A dynamic element matching (DEM) encoder is provided that converts an N-bit digital codeword into a pattern of 1-bit values. The DEM encoder includes a binary switching tree that includes plurality of switching blocks interconnected between an encoder input and a plurality of encoder outputs. The plurality of switching blocks are configured to receive a plurality of first control signals such that each switching block receives a respective first control signal and is independently programmable based on the respective first control signal into a first mode or a second mode. Each switching block includes a splitting circuit programmed into the first mode or the second mode to split a digital input into two digital outputs using either both a first splitting operation and a second splitting operation that is different from the first splitting operation or the first splitting operation over the plurality of sampling intervals.

Abstract: In described examples, a sample and hold circuit is configured to periodically connect one input of an op-amp to a reference voltage through a switch while a second input of the op-amp is connected to an output of the op-amp. Offset cancellation is performed by storing a sampled offset on a sampling capacitor coupled to the second input of the op-amp.

Abstract: An integrated circuit includes an analog-to-digital converter (ADC) having selectable first and second analog channel inputs and a digital output. A window comparator coupled to the digital output. The window comparator configured to compare a digital value on the digital output to first and second threshold values. A programmable clock circuit configured to provide a clock signal to the ADC. A controller that, response to assertion of the trigger signal, is configured to generate a sample rate control signal to the clock circuit to cause the clock circuit to increase the frequency of the clock signal and toggle selection between the first and second analog channel inputs. A result comparison circuit having a comparison input coupled to the digital output. The result comparison circuit is configured to compare a first digital conversion output from the ADC to a second digital conversion output from the ADC.

Abstract: A sigma delta modulator comprises an input configured to receive an input analog signal; a summing junction configured to subtract a feedback analog signal from the input analog signal; a first stage including a low pass filter coupled to the summing junction, wherein the low pass filter is configured to generate a first filtered signal; a second stage coupled to the low pass filter, configured to generate a second filtered signal by an active filter; a back-end stage coupled to the second stage, wherein the back-end stage comprises an analog to digital converter configured to convert the 2nd filtered signal to a digital output signal by sampling at a predetermined sampling frequency (fs); and a feedback path for routing the digital output signal to the summing junction, wherein the feedback path comprises a digital to analog converters, DAC, converting the digital output signal to the feedback analog signal.

Abstract: Some embodiments of the present disclosure are directed to an antenna system for an airplane. The antenna system including an antenna system enclosure and a desiccant enclosure within the antenna system enclosure and having desiccant material positioned between an outside air pipe and an inside air pipe, the inside air pipe extending from the desiccant enclosure to open to an interior air volume of at least one component of the antenna system within the antenna system enclosure, the outside air pipe extending from the desiccant enclosure to ambient air outside the antenna system enclosure, and the desiccant material is configured to absorb moisture in the ambient air flowing from the outside air pipe to the inside air pipe through the desiccant enclosure while air pressure is being equalized from outside to inside the antenna system enclosure.

Abstract: A method for selectively dropping out feature elements from a tensor in a neural network includes receiving a first tensor from a first layer of a neural network and obtaining a compressed mask for the first tensor. N mask bits of the compressed mask are received at each of N lanes of a reconfigurable computing unit and feature elements of the first tensor are respectively received at the N lanes. Feature elements are selectively dropped out from the first tensor to generate feature elements to use as at least part of a second tensor by selecting, based on a single mask bit of the compressed mask selected based on the lane, either a zero value or a feature element received at the lane for a feature element of the second tensor. The second tensor is propagated to a second layer of the neural network.

Abstract: This document describes techniques and systems of an exposed portion of a PCB configured to provide isolation among radar antennas. The described radar system includes an exposed portion of a surface of a printed circuit board (PCB) positioned between a first antenna and a second antenna. The PCB includes a metal plating on the surface of the PCB. A width of the exposed portion can delay a phase of electromagnetic (EM) energy conducted by the metal plating relative to a phase of EM energy that does not traverse the exposed portion. A height of the exposed portion can cause an amount of the EM energy conducted by the metal plating to be approximately equal to an amount of EM energy that traverses the exposed portion. In this way, the described systems and techniques can reduce signal-coupling among radar antennas without additional hardware costs and distance between the antennas.

Abstract: A method for selectively dropping out feature elements from a tensor in a neural network is disclosed. The method includes receiving a first tensor from a first layer of a neural network. The first tensor includes multiple feature elements arranged in a first order. A compressed mask for the first tensor is obtained. The compressed mask includes single-bit mask elements respectively corresponding to the multiple feature elements of the first tensor and has a second order that is different than the first order of their corresponding feature elements in the first tensor. Feature elements from the first tensor are selectively dropped out based on the compressed mask to form a second tensor which is propagated to a second layer of the neural network.

Abstract: A device configured to compress a tensor including a plurality of cells includes: a quadtree generator configured to generate a quadtree searching for a non-zero cell included in the tensor and extract at least one parameter value from the quadtree; a mode selector configured to determine a compression mode based on the at least one parameter; and a bitstream generator configured to generate a bitstream by compressing the tensor based on the compression mode.

Abstract: An analog to digital converter (ADC) device includes ADC circuits, a calibration circuit and a controlling circuit. The ADC circuits are configured to generate first quantized outputs according to clock signals. The calibration circuit is configured to perform at least one error operation according to the first quantized outputs to generate second quantized outputs, and is configured to analyze time difference information of the clock signals according to the second quantized outputs to generate adjustment signals. The controlling circuit is configured to analyze the first quantized outputs to generate at least one control signal to the calibration circuit, wherein the at least one control signal is configured to control the calibration circuit to selectively perform the at least one error operation and selectively analyze the time difference information of the clock signals.

Abstract: An analog-to-digital converter has a first digital signal generator that generates a first digital signal based on whether or not a sampling signal of an input signal is equal to or lower than a signal corresponding to a second reference signal higher than a first reference signal, a first slope generator to generate a first slope signal that changes with time from the sampled and held signal equal to or lower than the first reference signal, a second slope generator to generate a second slope signal that changes with time from the sampled and held signal to a signal level equal to or lower than the second reference signal, and a second digital signal generator that generates a second digital signal based on a time at which the first slope signal matches the first reference signal or a time at which the second slope signal matches the second reference signal.

Abstract: Described herein is a method and apparatus for enhancing the dynamic range of a digital-to-analog conversion circuit. Dynamic range enhancement (DRE) is accomplished by modifying the gain of components of the circuit so that the gain of components generating noise is effectively reduced. In a circuit utilizing a plurality of 1-bit DACs, analog signal gain is decreased when the full nominal gain of the analog portion of the circuit is not needed to obtain a desired peak output amplitude. The reduction is accomplished by effectively “disconnecting” some of the plurality of 1-bit DACs. Some or all of the 1-bit DACs are configured to have a third or “tri-state” in which there is no connection to the normal two reference levels thus providing no output. If some portion of the 1-bit DACs is placed in the tri-state, both the signal and noise gain will be reduced.

Abstract: Digital to analog converter generates an analog output corresponding to a digital input by controlling DAC cells using bits of the digital input. The DAC cells individually make a contribution to the analog output. Due to process, voltage, and temperature variations, the DAC cells may have duty cycle error or mismatches. To compensate for the duty cycle error of a DAC cell, a small amount of charge is injected into a low-impedance node of a DAC cell when the data signal driving the DAC cell transitions, or changes state. The small amount of charge is generated using a capacitive T-network, and the polarity of the charge injected is opposite of the error charge caused by duty cycle error. The opposite amount of charge thus compensates or cancels out the duty cycle error, and duty cycle error present at the output of the DAC cell is reduced.

Abstract: A technique for generating analog waveforms includes combining a desired, in-band signal with a randomizing, out-of-band signal at an input of a DAC, operating the DAC to generate DAC output based on a combination of the desired signal and the randomizing signal, and filtering the DAC output to pass the desired signal while removing the randomizing signal.

Abstract: A processing device for executing a machine learning neural network operation includes memory and a processor. The processor is configured to receive input data at a layer of the machine learning neural network operation, receive a plurality of sorted filters to be applied to the input data, apply the plurality of sorted filters to the input data to produce a plurality of different feature maps, compress the plurality of different feature maps according to a sparsity of the feature maps and store the plurality of different feature maps in the memory.

Abstract: The present invention provides a fractional-N frequency synthesizer comprising a divider controller comprising a multistage noise Shaping (MASH) digital delta-sigma modulator comprising L error feedback modulator (EFM) stages, wherein the jth EFM stage is configured to receive as an input the sum of the error of the preceding EFM stage and a high amplitude dither signal derived from the error of the kth EFM stage, where 1?j?k?L.

Abstract: The present description relates to a comparator (2) comprising a ring of gates (110A, 110B, 110A?, 110B?, 106, 108) in series, wherein: each gate implements an inverting function between a first input (100) and an output (102) of the gate; at least one (110A?, 110B?) gate is controllable and is associated with another gate; each controllable gate (110A?, 110B?) comprises a control input (116) coupled with the output (102) of said associated gate, and prevents switching of its output (102) to a high state if its control input (116) is in the high state, and to a low state otherwise; and the control input (116) of each controllable gate (110A?, 110B?) receives the output (102) of said associated gate if an even number of gates separates these two gates, and receives the complement of said output if not.