Infineon Technologies

Various examples of the disclosure generally pertain to processing radar datasets using a pre-trained neural network. Various examples pertain to a neural network including one or more attention layers. Various examples further pertain to training strategies for training the neural network. Pre-processing strategies are disclosed for pre-processing radar data so that domain mismatches can be reduced.

MEMS device (10) comprising: a transducer element (12) having a first deflectable membrane structure (14), a rigid electrode structure (16) and a second deflectable membrane structure (18) in a vertically spaced configuration, wherein the rigid electrode structure (16) is arranged between the first and second deflectable membrane structure (14, 18), the first and second deflectable membrane structure (14, 18) each comprise a deflectable portion (14-1, 18-1), and wherein the deflectable portion (14-1) of the first deflectable membrane structure (14) and the deflectable portion (18-1) of the second deflectable structure (18) are mechanically coupled by means of mechanical connection elements (20) to each other and are mechanically decoupled from the rigid electrode structure (16); a carrier element (22) for supporting the transducer element (12); a plurality of first clamping structures (24-1, ..., 24-#) for mechanically connecting the transducer element (12) to the carrier element (22) at edge clamping regions (26-1, ..., 26-#) of the transducer element (12); an elongation structure (32) laterally extending from unclamped edge regions (34-1, ..., 34-#) of the transducer element (12); a plurality of second clamping structures (36-1, ..., 36-#) for mechanically connecting the elongation structure (30) to the carrier element (22) at edge clamping regions (38-1, ..., 38-#) of the elongation structure (32); and wherein the elongation structure (32) comprises first elongation elements (40) which extend at least partially in the same plane as the first deflectable membrane structure (14) from the transducer element (12) to the second clamping structures (36-1, ..., 36-#), or wherein the elongation structure (32) comprises second elongation elements (42) which extend at least partially in the same plane as the second deflectable membrane structure (18) from the transducer element (12) to the second clamping structures (36-1, ..., 36-#).

Techniques for populating a dataset for training or validating a machine-learning model are disclosed. One or more measurement samples (811, 812) are obtained, each of the one or more measurement samples comprising respective radar data (120, 140, 420) observing a respective scene (170, 171) comprising one or more respective objects (181, 182). A reflection signal component of a selected object from the respective radar data and synthesized radar data (821) is then generated based on a combination of the this reflection signal component acting as a template and at least a part of the radar data of the one or more measurement samples. Thereby, the radar data (821) emulates an observation of a synthesized scene (172) comprising at least the synthesized object (183). A synthesized sample comprising the synthesized radar data is then added to the dataset.

Provided is a radio frequency, RF, switch device. The RF switch device includes a plurality of transistors arranged in series and a RF switch terminal configured to receive a control voltage for controlling a switch state of the RF switch. The RF switch device further includes decoupling circuitry configured to receive a reference voltage. The decoupling circuitry is configured to block a current path from the RF switch terminal to the decoupling circuitry in response to a voltage at the RF switch and the reference voltage satisfying a predetermined criterion.

A method for increasing overall pin alignment accuracy for a power semiconductor module having a plurality of pins includes attaching the plurality of pins and one or more power semiconductor dies to a first side of a substrate to form a power semiconductor subassembly; aligning a distal end of each of the pins with an alignment feature of an electrically insulative housing using a jig; and after the aligning, attaching the housing to the power semiconductor subassembly. The housing is attached to the subassembly such that the housing and the substrate together define an interior space in which the pins and the one or more power semiconductor dies are disposed. The pins jut out beyond the housing.

A photoacoustic gas sensor (100) comprises a substrate (110) and a measurement cell (120) within the substrate. An emitter (130) is mounted on the substrate (110) and configured to emit light into the measurement cell (120). A sensor (140) is mounted on the substrate (130) and configured to measure pressure variations in the measurement cell (120).

A photoacoustic gas sensor (100) comprises an emitter (110) configured to emit light, a sensor (120) configured to measure pressure variations and a measurement cell (130) enclosing the emitter and the sensor. At least a first ventilation opening (140a) and a second ventilation opening (140b) connect the measurement cell (130) with an environment (150).

A semiconductor device (100) is proposed. The semiconductor device includes a semiconductor substrate (102) including at least a first semiconductor substrate portion (1021) and a second semiconductor substrate portion (1022) separated from one another by a first part (1041) of an intermediate dielectric structure (104) along a first lateral direction (x1). The semiconductor device (100) further includes a patterned first metal layer (106) over a first surface (1081) of the semiconductor substrate (102), wherein a first part (1061) of the patterned first metal layer (106) electrically connects the first and second semiconductor substrate portions (1021, 1022) at the first surface (1081). Each of the first and second semiconductor substrate portions (1021, 1022) includes a main portion (1023) adjoining the first surface (1081) and an auxiliary portion (1024) adjoining a second surface (1082) opposite to the first surface (1081), and, with respect to each of the first and second semiconductor substrate portions (1021, 1022), a width (w12, w22) of the auxiliary portion (1024) along the first lateral direction (x1) is larger than a width (w11, w21) of the main portion (1023) along the first lateral direction (x1).

In accordance with an embodiment, a microelectromechanical system (MEMS) switch device includes: a substrate; a switching membrane disposed above the substrate; a pull-in electrode disposed above the switching membrane; a metal contact disposed on the switching membrane; and a pull-back electrode disposed below the switching membrane, wherein the switching membrane is movable between an open position and a closed position, and wherein in the closed position, the metal contact electrically connects two RF signal lines.

A method for producing a MEMS chip (100) with an environmental barrier structure (150) made from polymer nanofibers (151), the method comprising providing a wafer substrate (110) having plurality of adjacent through holes (131, ..., 134). The environmental barrier structure (150) is created by applying polymer nanofibers (151) onto one of the front and back sides (111, 112) of the wafer substrate (110). A plurality of MEMS chips (100) are singulated from the wafer substrate (110) by applying a two-step laser process comprising: a first laser-processing step including guiding a laser beam (160) over the one side of the wafer substrate (110) at which the polymer nanofibers (151) are applied in order to locally remove the polymer nanofibers (151) from the wafer substrate (110) and to locally uncover an underlying surface, and a subsequent second laser-processing step including applying a stealth dicing process for singulating the MEMS chips (100) by guiding a stealth dicing laser beam over the locally uncovered surfaces.