Abstract: The present disclosure is directed to a circuit layout of a dual port static random-access-memory (SRAM) cell. The memory cell includes active regions in a substrate, with polysilicon gate electrodes on the active regions to define transistors of the memory cell. The eight transistor (8T) memory cell layout includes a reduced aspect ratio and non-polysilicon bit line discharge path routing by positioning an active region for the first port opposite an active region for the second port and consolidating power line nodes at a central portion of the memory cell.

Abstract: An integrated circuit includes a non-volatile memory, a charge pump that generates high voltages for programming operations of the non-volatile memory array, and a charge pump regulator that controls a slew rate of the charge pump. The charge pump regulator generates a sense current indicative of the slew rate and adjusts a frequency of a clock signal provided to the charge pump based on the sense current.

Abstract: Embodiments of an electronic device include an integrated circuit, a reconfigurable stream switch formed in the integrated circuit along with a plurality of convolution accelerators and a decompression unit coupled to the reconfigurable stream switch. The decompression unit decompresses encoded kernel data in real time during operation of convolutional neural network.

Abstract: A convolutional accelerator includes a feature line buffer, a kernel buffer, a multiply-accumulate cluster, and iteration control circuitry. The convolutional accelerator, in operation, convolves a kernel with a streaming feature data tensor. The convolving includes decomposing the kernel into a plurality of sub-kernels and iteratively convolving the sub-kernels with respective sub-tensors of the streamed feature data tensor. The iteration control circuitry, in operation, defines respective windows of the streamed feature data tensors, the windows corresponding to the sub-tensors.

Abstract: A learning method detects anomalies and is implemented on a microcontroller including at least one memory, the microcontroller being configured to receive data sets coming from at least one sensor, the memory being configured to store a maximum number of categories, a category including at least a signature and an occurrence.

Abstract: A method of operating an electronic device includes generating scramble control codes. The scramble codes are generated by generating a random number, shifting the random number to produce a shifted random number, generating control signals by selecting different subsets of the shifted random number, and generating scramble control words by selecting different subsets of the random number based upon the control signals. The method further includes receiving a password comprised of sub-words and scrambling those sub-words according to the scramble control codes, retrieving a verification word comprised of sub-words and scrambling those sub-words according to the scramble control codes, and comparing the scrambled sub-words of the password to the scrambled sub-words of the verification word to thereby authenticate an external device that provided the password.

Abstract: A voltage supply circuit and a method for controlling a voltage supply circuit are provided. The voltage supply circuit includes a positive charge pump stage that generates a positive voltage and a negative charge pump stage that generates a negative voltage. The voltage supply circuit also includes a control stage that compares a voltage representative of the negative voltage with a reference voltage and causes a slope of the positive voltage to decrease when the voltage representative of the negative voltage exceeds the reference voltage.

Abstract: A convolutional accelerator includes a feature line buffer, a kernel buffer, a multiply-accumulate cluster, and mode control circuitry. In a first mode of operation, the mode control circuitry stores feature data in a feature line buffer and stores kernel data in a kernel buffer. The data stored in the buffers is transferred to the MAC cluster of the convolutional accelerator for processing. In a second mode of operation the mode control circuitry stores feature data in the kernel buffer and stores kernel data in the feature line buffer. The data stored in the buffers is transferred to the MAC cluster of the convolutional accelerator for processing. The second mode of operation may be employed to efficiently process 1×N kernels, where N is an integer greater than or equal to 1.

Abstract: A low power crystal oscillator circuit having a high power part and a low power part. Oscillation is initialized using the high power part. Once the crystal is under stable oscillation, the circuit switches to the low power part and continue operation for a long duration.

Abstract: A PFC correction circuit includes first, second, and third phase inputs coupled to three-phase power mains, with a three-phase full-wave rectifying bridge connected to an input node. First, second, and third boost inductors are respectively connected between first, second, and third phase inputs and first, second, and third taps of the three-phase full-wave rectifying bridge. A boost switch is connected between the input node and ground, and a boost diode is connected between the input node and an output node. A multiplier input driver generates a single-phase input signal as a replica of a signal at the three-phase power mains after rectification. A single-phase power factor controller generates a PWM signal from the single-phase input signal. A gate driver generates a gate drive signal from the PWM signal. The boost switch is operated by the gate drive signal.

Abstract: A successive approximation analog-to-digital converter includes a digital-to-analog converter DAC configured to receive a digital signal. First conversion units of the DAC are configured to sample an analog signal via a first switch and provide a first level voltage. Each first conversion unit includes a first capacitor array and a first switch array controlled from the digital signal. A single second conversion unit of the DAC is configured to provide a second level voltage. The second conversion unit includes a second capacitor array and a second switch array. A comparator operates to compare each of the first level voltages to the second level voltage and to provide a comparison signal based on each comparison and actuation of a set of third switches. A control circuit closes the first switches simultaneously and closes the third switches successively for the conversion of each sampled analog signal.

Abstract: A low power crystal oscillator circuit has a high power part and a low power part. Crystal oscillation is initialized using the high power part. An automatic amplitude control circuit includes a current subtractor that decreases current in the high power part as an amplitude of the crystal oscillation increases. A current limiting circuit may limit current in the low power part in order to further reduce power consumption by the low power crystal oscillator circuit. Additionally, an automatic amplitude detection circuit may turn off the high power part after the amplitude of the crystal oscillation reaches a predetermined level in order to further reduce power consumption of the low power crystal oscillator circuit, and may turn back on the high power part after the amplitude of the crystal oscillation reaches a second predetermined level in order to maintain the crystal oscillation.

Abstract: An in-memory computation circuit includes a memory array with SRAM cells connected in rows by word lines and in columns by bit lines. A row controller circuit simultaneously actuates word lines in parallel for an in-memory compute operation. A column processing circuit includes a current mirroring circuit that mirrors the read current developed on each bit line in response to the simultaneous actuation to generate a decision output for the in-memory compute operation. A bias voltage for word line driver and a configuration of the current mirroring circuit to inhibit drop of a voltage on the bit line below a bit flip voltage during execution of the in-memory compute operation. The mirrored read current is integrated by an integration capacitor to generate an output voltage that is converted to a digital signal by an analog-to-digital converter circuit.

Abstract: An in-memory computation circuit includes a memory array including sub-arrays of with SRAM cells connected in rows by word lines and in columns by local bit lines. A row controller circuit selectively actuates one word line per sub-array for an in-memory compute operation. A global bit line is capacitively coupled to many local bit lines in either a column direction or row direction. An analog global output voltage on each global bit line is an average of local bit line voltages on the capacitively coupled local bit lines. The analog global output voltage is sampled and converted by an analog-to-digital converter (ADC) circuit to generate a digital decision signal output for the in-memory compute operation.

Abstract: In an embodiment, an integrated circuit includes: a voting circuit including N scan flip-flops, where N is an odd number greater than or equal to 3, and where the N scan flip-flops includes a first scan flip-flop and a second scan flip-flop, where an output of the first scan flip-flop is coupled to a scan input of the second scan flip-flop; a scan chain including the N scan flip-flops of the voting circuit, and third and fourth scan flip-flops, the scan chain configured to receive a scan enable signal; and a scan enable control circuit configured to control a scan enable input of the first or second scan flip-flops based on the scan enable signal and based on a scan input of the third scan flip-flop or an output of the fourth scan flip-flop.

Abstract: A memory cell including a set of active regions that overlay a set of gate regions to form a pair of cross-coupled inverters. A first active region extends along a first axis. A first gate region extends transversely to the first active region and overlays the first active region to form a first transistor of the pair of cross-coupled inverters. A second gate region extends transversely to the first active region and overlays the first active region to form a second transistor of the pair of cross-coupled inverters. A second active region extends along a second axis and overlays the first gate region to form a third transistor of the pair of cross-coupled inverters. A fourth active region extending along a third axis and overlays a gate region to form a transistor of a read port.

Abstract: A leadframe including a metal oxide layer on at least a portion of the leadframe are disclosed. More specifically, leadframes with a metal layer and a metal oxide layer formed on one or more leads before a tin finish plating layer is formed are described. The layers of metal and metal oxide between the one or more leads and the tin finish plating layer reduce the formation of tin whiskers, thus reducing the likelihood of shorting and improving the overall reliability of the package structure and device produced.

Abstract: A PFC correction circuit includes first, second, and third phase inputs coupled to three-phase power mains, with a three-phase full-wave rectifying bridge connected to an input node. First, second, and third boost inductors are respectively connected between first, second, and third phase inputs and first, second, and third taps of the three-phase full-wave rectifying bridge. A boost switch is connected between the input node and ground, and a boost diode is connected between the input node and an output node. A multiplier input driver generates a single-phase input signal as a replica of a signal at the three-phase power mains after rectification. A single-phase power factor controller generates a PWM signal from the single-phase input signal. A gate driver generates a gate drive signal from the PWM signal. The boost switch is operated by the gate drive signal.

Abstract: A memory array arranged as a plurality of memory cells. The memory cells are configured to operate at a determined voltage. A memory management circuitry coupled to the plurality of memory cells tags a first set of the plurality of memory cells as low-voltage cells and tags a second set of the plurality of memory cells as high-voltage cells. A power source provides a low voltage to the first set of memory cells and provides a high voltage to the second set of memory cells based on the tags.

Abstract: Techniques and systems are provided for implementing a convolutional neural network. One or more convolution accelerators are provided that each include a feature line buffer memory, a kernel buffer memory, and a plurality of multiply-accumulate (MAC) circuits arranged to multiply and accumulate data. In a first operational mode the convolutional accelerator stores feature data in the feature line buffer memory and stores kernel data in the kernel data buffer memory. In a second mode of operation, the convolutional accelerator stores kernel decompression tables in the feature line buffer memory.