Patents by Inventor Xiao Huo
Xiao Huo has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Publication number: 20220147792Abstract: A processor for generating binarized weights for a neural network. The processor comprises a binarization scheme generation module configured to generate, for a group of weights taken from a set of input weights for one or more layers of a neural network, one or more potential binary weight strings representing said group of weights; a binarization scheme selection module configured to select a binary weight string to represent said group of weights, from among the one or more potential binary weight strings, based at least in part on a number of data bits required to represent the one or more potential binary weight strings according to a predetermined encoding method; and a weight generation module configured to output data representing the selected binary weight string for representing the group of weights.Type: ApplicationFiled: November 12, 2021Publication date: May 12, 2022Applicant: UNITED MICROELECTRONICS CENTRE (HONG KONG) LIMITEDInventors: Yuzhong JIAO, Xiao HUO, Yuan LEI
-
Patent number: 11176447Abstract: A deep neural network models semiconductor devices. Measurements of test transistors are gathered into training data including gate and drain voltages and transistor width and length, and target data such as the drain current measured under the input conditions. The training data is converted by an input pre-processor that can apply logarithms of the inputs or perform a Principal Component Analysis (PCA). Rather than use measured drain current as the target when training the deep neural network, a target transformer transforms the drain current into a transformed drain current, such as a derivative of the drain current with respect to gate or drain voltages, or a logarithm of the derivative. Weights in the deep neural network are adjusted during training by comparing the deep neural network's output to the transformed drain current and generating a loss function that is minimized over the training data.Type: GrantFiled: June 19, 2018Date of Patent: November 16, 2021Assignee: Hong Kong Applied Science and Technology Research Institute Company LimitedInventors: Yuan Lei, Xiao Huo
-
Patent number: 10665584Abstract: A well-less Transient Voltage Suppressor (TVS) Silicon-Controlled Rectifier (SCR) has a P+ anode region that is not in an N-well. The P+ anode region 20 is surrounded by N+ isolation regions near the surface, and a deep N+ region underneath that is formed in a p-substrate. A N+ cathode region is formed in the p-substrate. The deep N+ region has a doping of 5×1018 to 5×1019/cm3, compared to a doping of 1×1016/cm3 for a typical N-well, or a doping of 1×1013 to 1×1015/cm3 for the p-substrate. The high doping in the deep N+ region causes a recombination current that can shunt half of the anode current. Since the deep N+ region is much shallower than an N-well, the sidewall capacitance is greatly reduced, allowing for higher speed applications.Type: GrantFiled: March 7, 2019Date of Patent: May 26, 2020Assignee: Hong Kong Applied Science and Technology Research Insstitute Company, LimitedInventors: Chenyue Ma, Chun-Kit Yam, Xiao Huo
-
Publication number: 20190385047Abstract: A deep neural network models semiconductor devices. Measurements of test transistors are gathered into training data including gate and drain voltages and transistor width and length, and target data such as the drain current measured under the input conditions. The training data is converted by an input pre-processor that can apply logarithms of the inputs or perform a Principal Component Analysis (PCA). Rather than use measured drain current as the target when training the deep neural network, a target transformer transforms the drain current into a transformed drain current, such as a derivative of the drain current with respect to gate or drain voltages, or a logarithm of the derivative. Weights in the deep neural network are adjusted during training by comparing the deep neural network's output to the transformed drain current and generating a loss function that is minimized over the training data.Type: ApplicationFiled: June 19, 2018Publication date: December 19, 2019Inventors: Yuan LEI, Xiao HUO
-
Patent number: 10510743Abstract: An Electro-Static-Discharge (ESD) protection device has a Fin Field-Effect Transistor (FinFET) with a silicon fin with a step separating a top fin and a bottom fin. The gate wraps around the top fin but not the bottom fin. Normal gate-controlled channel conduction occurs in the top fin between a source and a drain in the top fin. Underneath the conducting channel is a buried conducting region in the bottom fin that conducts after a breakdown voltage is reached during ESD. A ledge, abrupt slope change in the sidewalls of the fin, or a doping increase occurs at the step between the top fin and bottom fin. The bottom fin is 2-3 times wider than the top fin, causing the resistance of the buried conducting region to be 2-3 times less than the resistance of the conducting channel, steering breakdown current away from the channel, reducing failures during breakdown.Type: GrantFiled: July 18, 2017Date of Patent: December 17, 2019Assignee: Hong Kong Applied Science and Technology Research Institute Company, LimitedInventors: Xiaoyong Han, Xiao Huo, Shuli Pan
-
Publication number: 20190027470Abstract: An Electro-Static-Discharge (ESD) protection device has a Fin Field-Effect Transistor (FinFET) with a silicon fin with a step separating a top fin and a bottom fin. The gate wraps around the top fin but not the bottom fin. Normal gate-controlled channel conduction occurs in the top fin between a source and a drain in the top fin. Underneath the conducting channel is a buried conducting region in the bottom fin that conducts after a breakdown voltage is reached during ESD. A ledge, abrupt slope change in the sidewalls of the fin, or a doping increase occurs at the step between the top fin and bottom fin. The bottom fin is 2-3 times wider than the top fin, causing the resistance of the buried conducting region to be 2-3 times less than the resistance of the conducting channel, steering breakdown current away from the channel, reducing failures during breakdown.Type: ApplicationFiled: July 18, 2017Publication date: January 24, 2019Inventors: Xiaoyong HAN, Xiao HUO, Shuli PAN
-
Patent number: 10134722Abstract: An Electro-Static-Discharge (ESD) protection device has a Silicon-Controlled Rectifier (SCR) with a triggering PMOS transistor. The SCR is a PNPN structure with a P+ anode/source within a center N-well, a P-substrate, and an outer N-well that connects to a cathode using N+ well taps. The P+ anode/source is both the source of the triggering PMOS transistor and the anode of the SCR. A trigger circuit drives the gate of the triggering PMOS transistor low, turning it on to charge the P+ drain. Since the P+ drain straddles the well boundary, making physical contact with both the center N-well and the P-substrate, holes flow into the P-substrate. The P+ drain is located near guard rings that suppress latch-up. The holes from the P+ drain flood the region under the guard rings, temporarily weakening their effect and reducing the trigger voltage.Type: GrantFiled: April 12, 2017Date of Patent: November 20, 2018Assignee: Hong Kong Applied Science and Technology Research Institute Company LimitedInventors: Chun-Kit Yam, Xiao Huo
-
Publication number: 20180301445Abstract: An Electro-Static-Discharge (ESD) protection device has a Silicon-Controlled Rectifier (SCR) with a triggering PMOS transistor. The SCR is a PNPN structure with a P+ anode/source within a center N-well, a P-substrate, and an outer N-well that connects to a cathode using N+ well taps. The P+ anode/source is both the source of the triggering PMOS transistor and the anode of the SCR. A trigger circuit drives the gate of the triggering PMOS transistor low, turning it on to charge the P+ drain. Since the P+ drain straddles the well boundary, making physical contact with both the center N-well and the P-substrate, holes flow into the P-substrate. The P+ drain is located near guard rings that suppress latch-up. The holes from the P+ drain flood the region under the guard rings, temporarily weakening their effect and reducing the trigger voltage.Type: ApplicationFiled: April 12, 2017Publication date: October 18, 2018Inventors: Chun-Kit YAM, Xiao HUO
-
Publication number: 20170285116Abstract: A magnetic sensor disclosed according to the present application is provided, which includes at least two magnetic sensing elements forming at least one magnetic sensing element pair, and currents in the magnetic sensing elements in each of the at least one magnetic sensing element pair are reverse and the magnetic sensing elements are arranged symmetrically.Type: ApplicationFiled: March 31, 2017Publication date: October 5, 2017Inventors: Hui Min GUO, Shu Zuo LOU, Xiao Ming CHEN, Guang Jie CAI, Chun Fai WONG, Xiao HUO
-
Patent number: 9356442Abstract: Electrostatic discharge (ESD) protection is provided by a charge-latching power-to-ground clamp circuit. A filter capacitor and resistor generate a filter voltage that is buffered by three stages to drive the gate of a BigFET such as a large n-channel transistor. A transmission gate between the stages turns off when BigFET turns on, causing charge to be latched. The filter capacitor can then discharge while the BigFET remains on. A leaker resistor slowly discharges the gate of the large BigFET and turns the transmission gate back on when the BigFET turns off after shunting the ESD current. The length of time that the clamp shunts the ESD current is determined by the leaker resistor and gate capacitance of the BigFET, not by the filter capacitor, so a small filter capacitor may be used.Type: GrantFiled: July 8, 2014Date of Patent: May 31, 2016Assignee: Hong Kong Applied Science and Technology Research Institute Company, LimitedInventors: Xiaowu Cai, Beiping Yan, Xiao Huo
-
Patent number: 9305916Abstract: An Electro-Static-Discharge (ESD) protection circuit uses Silicon-On-Insulator (SOI) transistors with buried oxide but no parasitic substrate diode useable for ESD protection. A filter voltage is generated by a resistor and capacitor. When a VDD-to-VSS ESD positive pulse occurs, the filter voltage passes through an n-channel pass transistor and inverted to drive a gate of a big SOI transistor that shunts ESD current. A second path is used for a VSS-to-VDD ESD positive pulse. The filter voltage passes through a p-channel pass transistor to the gate when the positive ESD pulse is applied to VSS. The big SOI transistor can connect between VDD and VSS for a power clamp, and the gates of the n-channel and p-channel pass transistors connect to VDD. A small diode may be added between VDD and VSS to generate a small triggering current to activate grounded-gate transistors near I/O pads for full-chip Pad-based ESD protection.Type: GrantFiled: December 30, 2014Date of Patent: April 5, 2016Assignee: Hong Kong Applied Science and Technology Research Institute Company, LimitedInventors: Xiaowu Cai, Beiping Yan, Xiao Huo
-
Publication number: 20160013636Abstract: Electrostatic discharge (ESD) protection is provided by a charge-latching power-to-ground clamp circuit. A filter capacitor and resistor generate a filter voltage that is buffered by three stages to drive the gate of a BigFET such as a large n-channel transistor. A transmission gate between the stages turns off when BigFET turns on, causing charge to be latched. The filter capacitor can then discharge while the BigFET remains on. A leaker resistor slowly discharges the gate of the large BigFET and turns the transmission gate back on when the BigFET turns off after shunting the ESD current. The length of time that the clamp shunts the ESD current is determined by the leaker resistor and gate capacitance of the BigFET, not by the filter capacitor, so a small filter capacitor may be used.Type: ApplicationFiled: July 8, 2014Publication date: January 14, 2016Inventors: Xiaowu CAI, Beiping YAN, Xiao HUO
-
Patent number: 9054521Abstract: An electro-static-discharge (ESD) protection circuit has a vertical NPN transistor with a floating p-type base created by a deep p-type implant under an N+ source region. The deep p-type implant may be an ESD implant in a standard CMOS process. The p-type implant provides a low initial snap-back trigger voltage, but the holding voltage may be too low, creating latch-up problems. The holding voltage is raised by about one volt by connecting the emitter of the vertical NPN transistor to parallel resistor and diode paths. When the vertical NPN transistor is triggered, its current initially flows through the resistor, creating an increasing voltage drop through the resistor as current rises. Once the voltage across the resistor reaches 0.5 volt, the diode in parallel with the resistor becomes forward biased and shunts a higher current than the resistor, raising the holding voltage. A clamp transistor may replace the diode.Type: GrantFiled: June 25, 2013Date of Patent: June 9, 2015Assignee: Hong Kong Applied Science & Technology Research Institute Company, Ltd.Inventors: Xiao Huo, Beiping Yan, Xiaowu Cai
-
Patent number: 9021002Abstract: A sine wave generator for a Direct Digital Synthesizer (DDS) converts a digital phase input into a digital sine wave output. Sine values and slopes are stored in read-only memory (ROM) for coarse upper phase bits in a first quadrant. A quadrant folder and phase splitter reflects and inverts values from the first quadrant to generate amplitudes for all four quadrants. Each sine value and slope is stored for a range of lower phase bits. A Delta bit separates upper and lower phase bits. Delta conditionally inverts the lower phase bits, the sine value, and the final polarity. A reduced AND logic array multiplies the slope by the conditionally inverted lower phase bits. A reconstructed ADD logic array then adds the conditionally inverted sine value. The conditionally inverted polarity is added to generate the final sine value. Sine generation logic is streamlined with conditional inversion based on the Delta bit.Type: GrantFiled: February 5, 2013Date of Patent: April 28, 2015Assignee: Hong Kong Applied Science & Technology Research Institute Company, LimitedInventors: Zhongzi Chen, Beiping Yan, Xiao Huo, Xiaowu Cai
-
Publication number: 20140376135Abstract: An electro-static-discharge (ESD) protection circuit has a vertical NPN transistor with a floating p-type base created by a deep p-type implant under an N+ source region. The deep p-type implant may be an ESD implant in a standard CMOS process. The p-type implant provides a low initial snap-back trigger voltage, but the holding voltage may be too low, creating latch-up problems. The holding voltage is raised by about one volt by connecting the emitter of the vertical NPN transistor to parallel resistor and diode paths. When the vertical NPN transistor is triggered, its current initially flows through the resistor, creating an increasing voltage drop through the resistor as current rises. Once the voltage across the resistor reaches 0.5 volt, the diode in parallel with the resistor becomes forward biased and shunts a higher current than the resistor, raising the holding voltage. A clamp transistor may replace the diode.Type: ApplicationFiled: June 25, 2013Publication date: December 25, 2014Inventors: Xiao HUO, Beiping YAN, Xiaowu CAI
-
Patent number: 8643520Abstract: An equalized-impedance shadowed current cell can be arrayed in a Digital-to-Analog Converter (DAC) or other converters or applications. The Equalized-impedance shadowed current cell has primary differential transistors in parallel with shadow differential transistors that have gates driven inversely to gates of the primary differential transistors. A shadow current from the shadow differential transistors is much smaller than a primary current switched by the primary differential transistors. Cell current is not switched off to zero but to the shadow current. The ON state and OFF state impedances of the current cell may be matched during circuit design so that the impedance is the same regardless of digital input values. The Width and Length of the shadow differential transistors are adjusted so that overall output impedances for the ON and OFF states of the current cell are matched. Since output impedance is input code independent, high-speed performance is improved.Type: GrantFiled: November 27, 2012Date of Patent: February 4, 2014Assignee: Hong Kong Applied Science & Technology Research Institute Company Ltd.Inventors: Xiao Huo, Beiping Yan, Zhongzi Chen, Xiaowu Cai
-
Patent number: 8369054Abstract: A power-to-ground clamp transistor provides electrostatic discharge (ESD) protection. A filter capacitor and resistor generate a filter voltage that is buffered by three stages to drive the gate of the clamp transistor. The filter capacitor is about twenty times smaller than in a conventional clamp circuit. Feedback in the circuit keeps the clamp transistor turned on after the R-C time constant of the capacitor and resistor in the filer has elapsed, allowing for a smaller capacitor to turn on the clamp transistor longer. A sub-threshold-conducting transistor in the first stage conducts only a small sub-threshold current, which extends the discharge time of the first stage. The gate of the sub-threshold-conducting transistor is driven by feedback from the second stage. A feed-forward resistor has a high resistance value to slowly raise the voltage of the second stage from the filter voltage, and thus slowly raise the gate of the sub-threshold-conducting transistor.Type: GrantFiled: June 8, 2010Date of Patent: February 5, 2013Assignee: Hong Kong Applied Science and Technology Research Institute Co., Ltd.Inventors: Xiaowu Cai, Beiping Yan, Xiaoyang Du, Xiao Huo, Xiaoyong Han, Bingyong Yan
-
Publication number: 20110299202Abstract: A power-to-ground clamp transistor provides electrostatic discharge (ESD) protection. A filter capacitor and resistor generate a filter voltage that is buffered by three stages to drive the gate of the clamp transistor. The filter capacitor is about twenty times smaller than in a conventional clamp circuit. Feedback in the circuit keeps the clamp transistor turned on after the R-C time constant of the capacitor and resistor in the filer has elapsed, allowing for a smaller capacitor to turn on the clamp transistor longer. A sub-threshold-conducting transistor in the first stage conducts only a small sub-threshold current, which extends the discharge time of the first stage. The gate of the sub-threshold-conducting transistor is driven by feedback from the second stage. A feed-forward resistor has a high resistance value to slowly raise the voltage of the second stage from the filter voltage, and thus slowly raise the gate of the sub-threshold-conducting transistor.Type: ApplicationFiled: June 8, 2010Publication date: December 8, 2011Applicant: Hong Kong Applied Science & Technology Research Institute Company LimitedInventors: Xaiowu CAI, Beiping YAN, Xiaoyang DU, Xiao HUO, Xiaoyong HAN, Bingyong YAN
-
Publication number: 20070148963Abstract: Methods of utilizing carbon nanotubes or composites thereof as hole plugs in vias or in contact holes for connecting conductive layers in integrated circuits are disclosed. Integrated circuits and integrated circuit layers formed by the methods are also disclosed.Type: ApplicationFiled: December 27, 2005Publication date: June 28, 2007Applicant: The Hong Kong University of Science and TechnologyInventors: Philip Chan, Min Zhang, Xiao Huo