Showing papers presented at "Electrical Design of Advanced Packaging and Systems Symposium in 2016"

Machine learning based MoM (ML-MoM) for parasitic capacitance extractions

Abstract: This paper is a rethinking of the conventional method of moments (MoM) using the modern machine learning (ML) technology. By repositioning the MoM matrix and unknowns in an artificial neural network (ANN), the conventional linear algebra MoM solving is changed into a machine learning training process. The trained result is the solution. As an application, the parasitic capacitance extraction broadly needed by VLSI modeling is solved through the proposed new machine learning based method of moments (ML-MoM). The multiple linear regression (MLR) is employed to train the model. The computations are done on Amazon Web Service (AWS). Benchmarks demonstrated the interesting feasibility and efficiency of the proposed approach. According to our knowledge, this is the first MoM truly powered by machine learning methods. It opens enormous software and hardware resources for MoM and related algorithms that can be applied to signal integrity and power integrity simulations.

An experimental study of EMI reduction of DC-DC converter with frequency hopping technique

Abstract: In this paper, the EMI problem of the DC-DC converter is discussed. The power spectrums of the DC-DC converter's nodes are analyzed and the noise scanner method is used to prove that the switching node of the DC-DC converter is the dominant source of EMI. To reduce EMI of DC-DC converters, frequency hopping technique can be applied. The DC-DC converter with frequency hopping technique is fabricated using 180nm CMOS process. The test chip results by IC-stripline and noise scanner methods show a significant improvement in EMI. The EMI reduction amount is 6.87 dB at fundamental switching frequency by applying the frequency hopping technique with IC-stripline measurement.

Diagnosis and tuning of filtering antenna based on extracted coupling matrix

Abstract: A general method is proposed in this paper to extract the two-port coupling matrix of filtering antennas. By using the passivity and the realizability properties of a filter network, the rational polynomial forms of S-parameters can be obtained from the reflection and realized radiation efficiency of a filtering antenna. The coupling matrix is then obtained with transformation and optimization. After the coupling matrix is reestablished, the diagnosis and tuning of filtering antenna can easily be carried out. Our method has been validated with the successful tuning of a stacked patch antenna.

Delayed-rational Green's-function-based method of transmission lines and the Heaviside condition

Abstract: In electronics as well as in communications, it is highly desirable to be able to transmit a signal that is preserved from any distortions that is due to the lossy nature of the medium in which the signal travels, normally a transmission line. In this paper, we exploit the connections between the delayed Green's-function-based method from the authors and the well-known distortionless Heaviside condition. It is found that, in the method, an important results was already present but its importance not yet understood. In particular, we prove that we are able to identify the distortionless transmission line associated to a generic transmission line. We consider only 1-conductor transmission lines, with frequency-independent per-unit-length parameters. The multiconductor transmission-line case will be addressed in future works.

Modeling of power distribution networks for path finding

Abstract: In this paper an algorithm is described for obtaining the response of Power Distribution Networks (PDN) arising in chip, package or pcb during an early design phase. Results are provided for power grids arising in silicon interposers to validate the approach.

Schur complement preconditioner for fast 3D full-wave MoM package-board extraction

Abstract: In this paper, a preconditioning technique based on the Schur complement method is presented to accelerate the convergence of fast near-linear complexity iterative Method of Moments (MoM) solution. The Schur complement method diagonalizes the near field blocks to a block-diagonal form which can be used as an effective preconditioner to expedite the iterative solver convergence. Numerical experiments demonstrate a significant advantage over ILUT or recently published null-field based methods.

High-speed serial link challenges using multi-level signaling

Abstract: The paper discusses challenges of high-speed serial links using multi-level signaling and gives comparison of the performance of the various channels with different lossy materials and equalization options. Advantages and problems of multi-level signaling are shown based on the test results.

Balanced bandpass filter with reflectionless common-mode suppression

Abstract: A compact balanced bandpass filter using two types of LC resonator and quarter-wavelength resonator is proposed in this paper. By taking the advantage of the property of quarter-wavelength resonator and symmetry of the proposed circuit, a bandpass response can be achieved while the band stop response can be achieved at the same frequency band. Parallel LC resonator shunt to the ground is used for the differential bandpass response and series LC resonator shunt to the ground via a resistor is used for the common-mode suppression response. For the purpose of a non-reflected suppression of the common mode noise, a resistor matching to the port impedance is connected to the end of a series LC resonator. By doing so, a single frequency point common-mode noise suppression can be achieved at the desired band. The proposed circuit is analyzed and experimented carefully. The validity of this structure is demonstrated by the experiment result on a 2.4GHz filter.

Timing margin analysis and Power measurement with DDR4 memory

Abstract: DIMMs built with DDR4 (Double Data Rate 4th-generation) SDRAM (Synchronous Dynamic Random-Access Memory) are the current memory components used on HPC (High Performance Computing) systems. The DDR4 signal interfaces operate up to a 3200 Mbps data rate and at 1.2 V. This is a higher frequency at a lower voltage, therefore lower power, than the third generation DDR3 DIMMs. The higher frequency and lower voltage results in decreased timing margins. The characterization of the timing margins and power usage is of significantly increased importance in DDR4. In this paper, a methodology for experimentally quantifying timing margins and power is applied at bounding voltage and frequency corners to plan, design, and architect HPC systems optimized for power consumption and with timing margin.

SMPS noise managing methodology for RFI solution in mobile platforms

Abstract: In mobile platforms, RFI problems caused by the noises on switched-mode power supplies (SMPSs) have arisen. The SMPS noise which affects RF bands is generated by the voltage ringing at the switch transitions in the SMPS operation. This voltage ringing is caused by the parallel resonance between the chip-level capacitance and the PCB routing inductance in the off-chip level. In this paper, we propose a new SMPS noise managing methodology to reduce the SMPS noise in RF bands and optimize the PCB designs for SMPSs. Additionally, we consider the possible deleterious effect of the proposed managing methodology on the SMPS efficiency.

Simulation based generation of X-parameters using the harmonic balance first moment

Abstract: As the complexity of radio frequency circuits increases, the generation of macromodels for circuit blocks that can be used in system simulations has become increasingly useful and even necessary. Recently X-parameters1 have been proposed as an extension to the concept of S-Parameters in order to create a macromodel that captures some of the key nonlinearities in the circuit. In this paper, an efficient semi-analytical approach for the computation of X-parameters is presented. More specifically, a clear relation between the X-parameters and the circuit moments is developed, which in turn facilitates the application of moments to further improve the efficiency of X-parameter generation.

Design and signal integrity analysis of high bandwidth memory (HBM) interposer in 2.5D terabyte/s bandwidth graphics module

Abstract: Spurred by the industrial demands for terabyte/s bandwidth graphics module, high bandwidth memory (HBM) has been emerged to overcome the limitations of conventional DRAMs. Additionally, due to the fine pitch and high density interconnect routing between GPU and 4 HBMs in 2.5D terabyte/s bandwidth graphics module, HBM interposer has also been to the force. However, several signal integrity issues of the HBM interposer occur due to the manufacturing process constraints. In this paper, we design the HBM interposer using 6 layers redistribution layer (RDL) and TSVs in 2.5D terabyte/s bandwidth graphics module. And then, in the designed HBM interposer, electrical performance of the HBM interposer channels using M1, M3, and M5 layer is analyzed by simulation in the frequency-and time-domain. With the simulation results, it is observed that the designed HBM interposer shows good signal integrity.

Audio frequency ground integrity modeling and measurement for a TDMA smartphone system

Abstract: In this paper, we first propose a ground network model of an arbitrary shaped multi-layer printed circuit board (PCB)/chassis for accurate and efficient analysis of audio frequency ground noise coupled from a time division multiple access (TMDA) RF power amplifier (PA) to an audio circuit in a smartphone system. We designed test vehicles with varied extent in the ground noise coupling. We successfully verified the proposed model by comparing the ground noise coupling levels obtained from the model, 3-D electromagnetic (EM) simulation and measurement in time and frequency domain. We further discussed the performance of the proposed model by comparing the accuracy of its transfer ground impedance (ZG12) and analysis time with those of from the EM simulation. The proposed model showed high performance with the ZG12 agreed 91.7 % with the EM simulation, and the analysis time 95.5 % reduced compared to the simulation.

Prediction of near-field shielding effectiveness for conformal-shielded SiP and measurement with magnetic probe

Abstract: This paper presents a fast prediction method to evaluate the near-field shielding effectiveness(SE) for conformal-shielded System-in-Packages (SiP). By revising the typical EM method to meet SiP circumstances, the near-fieldSE can be quickly estimated. A simple conformal-shielded SiP module is designed. To experimentally verify the proposed method, two kinds of metallic shielding materials, copper and nickel, are sputtered on the module with different coating thicknesses. In the near-field, the measured SE reaches upto 52 dB for 4-μm copper coating as well as 54 dB for 15-μm nickel coating. The predicted results are closely agreed with simulations and measurements from 0.01 and 1 GHZ.

Design and modeling of an absorptive frequency selective surface with several transmissive bands

Abstract: An absorptive frequency selective surface (AFSS) which allows signals transmitted in some specific bands is proposed. The transmitting characteristic is rare among AFSS design in the past and can be used to shield the whole package of mixed signal systems. The proposed AFSS provides above 97% absorption at 5.6 GHz in the Wi-Fi band and allows at least half power transmission below 1.84 GHz and above 8.95 GHz, which covers GPS band, most of GSM bands and X-band. Additionally, the proposed work is polarization independent and stable for oblique incident waves up to 45 degrees for both TE and TM polarizations. It is worth to note that clear design concepts and closed-form formulae based on a simple equivalent circuit model are provided, as well as useful design procedures.

Modeling and measurement of high-bandwidth and high-density silicone rubber socket for 100Gbps transceiver IC test

Abstract: In this paper, an RLGC model of a silicone rubber socket is proposed and compared with simulation results. Due to the reusability of the silicone rubber socket, it is suitable for a reliable test. In addition, its electrical performance is compared with that of conventional solder balls. The performance of the two are compared using measurements in both frequency and time domains. Their eye diagrams are nearly identical at the data rate of 28Gbps, which is the data rate of a single channel in a 100Gbps transceiver IC that consists of 4 parallel channels.

Multiphysics simulation for the reliability analysis of large-scale interconnects

Abstract: A coupled electrical-thermal-mechanical simulation technique is developed for the reliability analysis of interconnects. The multi-physics simulation is based on the finite element method and characterizes electrical, thermal, and mechanical, aspects of interconnects simultaneously. The multi-physics simulation is capable of analyzing large-scale problems with a significantly enhanced computational efficiency. The efficiency enhancement is achieved by using a domain decomposition scheme called the finite element tearing and interconnecting, parallel computing, and the localized nature of thermal stresses. A numerical example is provided to demonstrate both the capability and efficiency of the proposed simulation.

A new pinwheel meander-perforated plane structure for noise suppression in system-on-package

Abstract: As the market of wearable and internet of things (IoT) expands, there is a strong demand of shrinking the size of System-on-Package (SoP). In this paper, we propose a new electromagnetic bandgap structure (EBG) to mitigate the noise coupling within SoPs. EBGs have been widely used in printed circuit boards (PCB). Thus, the EBG size reduction is critical for integrating it in the package. The proposed structure has a unique noise suppressing pinwheel meander perforated plane structure (PMPP) that can be used in small size package. Critical problem of EBG's size reduction is shifting the suppression frequency to higher. Therefore, we propose to combine the EBG structure, meander-line, and pinwheel shaped patch to reduce size of EBG unit cell from 12.2mm×12.2mm to 2.44mm×2.44mm while maintaining the suppression frequency and bandwidth. The stop band bandwidth of proposed structure is 3.58∼12.83GHz.

Efficient time-domain variability analysis of active circuits

Abstract: An advanced technique for numerical inversion of Laplace transform is presented for time-domain statistical analysis of large active circuits with multiple stochastic parameters. The proposed method is based on a model-order reduction technique that guarantees the stability of the reduced active circuit by construction. Pertinent numerical results are presented that validate the efficiency and accuracy of the proposed method.

Novel time domain integral equation method hybridized with the macromodels of circuits

Abstract: This work presents a novel transient field-circuit simulation method based on the time-domain integral equation (TDIE) and circuit macromodels. The traditional field-circuit simulation method constructs the Kirchoff's equations for lumped circuits with SPICE models, but they cannot handle problems with unknown circuit structures. The proposed method creates parametric macromodels to represent the port constitutive relations of the unknown circuits. Then the obtained macromodel is coupled with TDIE to enable the transient field-circuit cosimulation. Numerical examples are given to demonstrate the validity of the proposed method. The proposed method is targeted to solve unknown I/O modeling issues in general signal integrity problems.

A novel 3-D printed loop antenna using flexible NinjaFlex material for wearable and IoT applications

Abstract: A stretchable, flexible loop antenna working at 2.4GHz ISM band was fabricated by the additive manufacturing (AM) 3-D printing technology. NinjaFlex, a flexible 3-D printable material was utilized for the first time as a 3-D hemi-sphere substrate for the loop antenna. A 3-D printer based on the Fused Diffusion Modelling (FDM) technology was employed to 3-D print the substrate material. The resonance frequency of the antenna shifts in response to the applied force which makes the configuration suitable for the wireless strain sensing application. The proposed antenna was designed for wearable electronics application such as health monitoring earrings. Hence it was designed in such a way that it maintains the Specific Absorption Rate (SAR) of the human head tissues within the assigned standard limits when placed near human replicating head. The proposed antenna system could be useful in the additively manufactured wearable packaging and IoT applications.

Modeling of electromagnetic interference in electronic boards using finite difference time domain method

Abstract: This paper addresses calculation and simulation of Electromagnetic Interference (EMI) on the electronic boards. Keysight EMPro Finite Difference Time Domain (FDTD) simulator is utilized for simulation. The simulated radiated emission is studied due to level variations of excitation voltage and the spatial coordinates (angle) of simulated point regarded to the board. The simulation results are compared to the Federall Communications Commission (FCC) Part 15 Class B at three meters limit.

High-frequency modeling and signal integrity analysis of high-density silicone rubber socket

Abstract: In this paper, we propose and analyze an equivalent RLGC model of silicone rubber socket with single-ended signaling. A silicone rubber socket consists of highly dense metal powders in elastic silicone rubber. When the silicone rubber is compressed, metal powders form a column which corresponds to a pad of package. Thus, it can be modeled as a pair of cylinders. We have successfully verified the proposed model using a 3D electromagnetic (EM) solver in frequency domain and the eye diagram measurement in time domain. As a result, the verified model can be used to determine whether socket is applicable to the test system in the simulation level concurrently with offering physical insight and reducing time spent 3D EM simulating socket.

Signal integrity improvement design of lossy transmission line based on a single-shot pulse: Design methodology of passive equalizer for PCB traces in the time domain

Abstract: High-speed digital signal propagating in PCB (Printed Circuit Board) traces deteriorates seriously in GHz domain due to the electrical loss in the trace coming from skin-effect and dielectric loss. In order to overcome this problem, we propose a novel deterministic and theoretical design method of the passive equalizer based on a single-shot pulse, in contrast to conventional stochastic designs that need repetitive computations. We also fabricate a prototype PCB board based on the proposed design method and demonstrate its effectiveness in terms of the s-parameter as well as the eye-diagram.

Fabrication challenge, device characterization of high-Q 2D, 3D passives devices on glass

Abstract: It is demonstrated that the high quality factor of thick metal spiral inductors, 3D solenoid inductor and low harmonics power level can be achieved by using a mature glass substrate technology. The quality factor of inductors in a 2D/3D type can achieve 70∼100 in this study. The harmonics power level is measured along transmission line and 2nd /3rd harmonics can be below −110 dBm at 30 dBm input power. It is very critical and helpful with high Q-factor and lower harmonics power level for the FEM development of current and next generation's commercial 4G carrier aggregation system.

A novel methodology to create generative statistical models of interconnects

Abstract: This paper addresses the problem of constructing a generative statistical model for an interconnect starting from a limited set of S-parameter samples, which are obtained by simulating or measuring the interconnect for a few random realizations of its stochastic physical properties. These original samples are first converted into a pole-residue representation with common poles. The corresponding residues are modeled as a correlated stochastic process by means of principal component analysis and kernel density estimation. The obtained model allows generating new samples with similar statistics as the original data. A passivity check is performed over the generated samples to retain only passive data. The proposed approach is applied to a representative coupled microstrip line example.

Investigation of shorting vias for suppressing common-mode radiation in different frequency range

Abstract: The effects of shorting vias on suppressing common-mode radiation in different frequency range are investigated based on a typical wire-bonded ball grid array (WB-BGA) package. Conventional shorting vias provide a short return path for common-mode currents so as to reduce the radiation effectively at low frequencies. However, the suppression is limited at higher frequencies due to the excess parasitic inductance. Therefore, a novel concept of absorptive shorting vias is presented for suppressing radiation in high frequency range.

Differential-to-common mode conversion noise suppression with unit cell EBG structure for bended differential lines

Abstract: In order to reduce differential-to-common mode conversion noise in bended differential lines, we propose a unit cell electromagnetic bandgap (EBG) structure. The proposed structure compensates for mismatches of inductances and capacitances between inner and outer lines of the bended differential lines. Its performances of the common-mode noise suppression in frequency and time domains were verified by 3D full wave simulator, HFSS. Also, in order to verify the simulated results, the bended differential lines with the proposed unit cell EBG structure was fabricated and measured. As a result, suppression level of the differential-to-common mode conversion noise is below −20 dB from DC to 6 GHz, and Time-Domain-Through (TDT) common-mode noise voltage is reduced as compared with that of conventional bended differential lines.

PEEC macromodels for above plane decoupling capacitors

Abstract: Decoupling capacitors perform an important function in the impedance reduction of power distribution systems. Hence, they are a key part of an electrical model required for the design of such systems. In this paper, we construct circuit models for the capacitors which include the local environment such that the overall PDN model is simplified and it can be considered as decoupled macromodels.

Coarse-to-fine malleable Pole/Residue Equivalent System Solver (COMPRESS)

Abstract: Frequency-domain models are often need to be converted to macromodels by approximating them as rational functions. Pole/Residue Equivalent System Solver (PRESS) is a method for such fitting of frequency response as poles/residues. In this paper, a modified PRESS method using non-consecutive sample subset based local fitting is presented. The proposed method reduces the number of iterations required, consequently improving the model order. A test case of package — signal and power plane interconnects — is demonstrated. The results show that the proposed method generates compact macromodels.