Yien-Tien Chou

LTCC Layer-to-Layer Misalignment-Tolerant Coupled Inductors and Their Application to Bandpass Filter and Helical Inductors

Mutual inductance between coupled inductors can provide large equivalent series inductance required by filter design. As mutual inductance is dependent on the distance between coupled metal lines within coupled inductors, mutual inductance in traditional straight line coupled inductors (SLCI) is susceptible to layer-to-layer misalignment during fabrication if two coupled lines are located at different substrate layers. Misalignment-tolerant coupled inductors (MaTCI) are proposed to alleviate the effect of misalignment on mutual inductance. Circuit model is also proposed to describe their mechanism. Their performance is verified and compared with traditional SLCI using coupled inductors based transmission zero circuits and bandpass filters. Measurement results of three samples of a transmission zero circuit based on proposed MaTCI show only 0.56% frequency variation at 1.9 GHz, while traditional SLCI give about 3.2% variation from samples at the same low temperature cofired ceramic substrate. Measured results of three samples of a bandpass filter using MaTCI also show more stable performance at transmission zero frequency, insertion loss and passband bandwidth as compared with the filter using traditional SLCI. Electromagnetic simulation also shows similar performance improvement for MaTCI under misalignment. The same concept is also applied to helical inductor and two misalignment tolerant helical inductors are proposed. Simulation results show that they are more stable than the traditional stacked helical inductor.

Magnetic Near-Field Probes With High-Pass and Notch Filters for Electric Field Suppression

Several new types of low-cost and robust magnetic near-field probes manufactured in low-temperature co-fired ceramics (LTCC) are presented in this paper. Parallel C-shaped strips and their variations are inserted into the loop area in the front end of probes to achieve common-mode high-pass and notch filters for electric-field noise suppression. These probes with this kind of filter have excellent wideband electric field suppression. They are called high electric field suppression probes type A ~ D. The size of loop aperture in all probes is 100 μm long and 400 μm wide. The signal received from the loop is routed to a measurement apparatus through a semi-rigid coaxial cable with an outer diameter of 0.047 in. The flip-chip junction with low loss and good shielding is used between the probe head in LTCC and the semi-rigid coaxial cable. We take the probes over a 2000-μm-wide microstrip line as device-under-test to measure the probe characteristics. The isolation between electric and magnetic fields for a reference probe based on an old design using the same LTCC process is better than 30 dB from 0.05 to 12.65 GHz. The type A probe has two parallel C-shaped strips, it has better isolation of 35 dB from 0.1 to 11.05 GHz. Type C has one end of its strip shorted to ground, its 30-dB isolation frequency range can be extended to 0.05 ~ 17.8 GHz. With additional layout variation in type D, isolation can be improved to 40 dB up to 10.9 GHz. The spatial resolution for these probes is 140 μm when the distance between the metal surface of the microstrip line and the nearest edge of the loop is held at 120 μm. The calibration factors of the proposed probes are only slightly increased as compared with reference probe.

Space Difference Magnetic Near-Field Probe With Spatial Resolution Improvement

To achieve good spatial resolution, small loop size is required in the traditional loop probe. However, the smaller loop size will lead to lower sensitivity for the probe. In addition, loop size is always limited by the minimum line spacing of the fabrication process. Another problem is that the asymmetric electric field coupling into a probe will not be canceled perfectly even if the structure of this probe is symmetric. To circumvent these problems, a space difference magnetic near-field probe with three kinds of spatial resolutions is proposed in this paper. The probe head including a one-turn loop and a two-turn loop is manufactured in low temperature co-fired ceramics (LTCC). The one-turn loop is clamped with the two-turn loop. Two loops are covered with two shielding ground plates to form a tri-plate structure. The received signals from these two loops are outputted with two SMA connectors through two striplines. The flip-chip junctions with low loss and good shielding capability are used between the probe head and the striplines. The probe characteristics are measured using a 436- μm-wide microstrip line with impedance of 50 Ω. Two output ports have different spatial resolutions because two different loops are located above the microstrip line at different height. The proposed probe will have higher spatial resolution when the received signals are outputted in difference. These results are also verified by measuring a 2000- μm-wide straight and a 500- μm-wide meander microstrip line. The experiment results have good agreement with the simulation results and show the resolution improvement of the difference output of the proposed probe for different lines.

The thru-reflection-unequal-line (TRuL) calibration method for scattering matrix measurement of multi-port networks

The multimode TRL calibration method is known as an approach for the measurement of multi-conductor transmission line devices. The propagation constants of different modes propagating along the multi-conductor transmission line must be different using this method. However, in general multi-port networks, the propagating constants at each port may be equal. We here then present the thru-reflection-unequal-line (TRuL) calibration method for the calibration of the equal propagation constant case. Lines of unequal length are used to make the eigenvalues unequal during calibration process. This will greatly simplify the calculation. The measured scattering matrix of a branch line coupler on FR4 substrate using proposed calibration method shows good agreement with simulation. This method can also be extended for the calibration of on-wafer and multi-port networks.

Electric field coupling suppression using via fences for magnetic near-field shielded-loop coil probes in low temperature co-fired ceramics

Two types of low-cost and robust magnetic near-field probes manufactured in low temperature co-fired ceramics (LTCC) are presented in this paper. The shielded-loop coil and via fences are used in the probes to provide better electric field coupling suppression. Type I probe is designed to receive horizontal magnetic field, via fences are inserted in the loop aperture and along sides of the probe to reduce electric field coupling from sides. The inner size of the loop aperture is 700×380 μm. The flip-chip bonding, which has low insertion loss and the good shielding capability, is also used in this probe. We take this probe over a 2000-μm-wide microstrip line as device under test (DUT) in measurements, the isolation between electric and magnetic field is better than 10 dB up to 11.5 GHz. The spatial resolution of the probe is 300 μm at 11 GHz. Type II probe is designed to receive the vertical magnetic field. The detected signal is passed along a right-angle channel surrounded by cylindrical via fences. Via fences are also set around the loop and give good shielding to reduce the electric field interference. The inner diameter of the loop aperture is 670 μm. For this vertical magnetic field probe, the measured minimum isolation between electric and magnetic field is 15.17 dB at 17.35 GH up to 20 GHz. The spatial resolution of this probe is 600 μm at 16 GHz.

LTCC layer-to-layer misalignment resistant coupled inductor and bandpass filter

The performance of a misalignment resistant coupled inductor (MaRCI) is verified and compared experimentally with traditional straight line coupled inductor using coupled inductor based transmission zero circuits and bandpass filters. Measurement results of 3 samples of a transmission zero circuit based on proposed MaRCI shows only 0.3% frequency variation at 1.9GHz while traditional straight line coupled inductor gives about 3.2% variation from samples at the same LTCC substrate. Measured results of a bandpass filter using MaRCI also show more stable performance at transmission zero frequency, insertion loss and passband bandwidth as compared with filter using traditional straight line coupled inductor. EM simulation also shows similar performance improvement under misalignment.

Coupling coefficient improvement for inductor coupled vertical interconnect in 3D IC die stacking

We proposed a vertical magnetic coupled interconnect design procedure for 3D IC stacking. As the interconnect is modeled as a coupled inductor pair, we can convert it into a bandpass filter by adding series and shunt capacitors. With given center frequency and bandwidth of the bandpass filter, the required self and mutual inductance in magnetic coupled interconnect and serial and parallel capacitance can be calculated quickly. In our design using TSMC 0.18µm CMOS process, the addition of capacitors increased the transmission coefficient by 7dB. The size of loops for magnetic coupled interconnect is 200µm, the communication distance is 30µm. By using an ASK modulation scheme, we can up-covert the baseband signal spectrum to the lower loss passband of the filter. Our bandpass interconnect can achieve 3.5 Gbps data rate and 6.77pJ/bit power consumption.

Transition characterization using TRL calibration method with unequal “R” calibrators

Traditional TRL calibration method assumes equal reflection at both reflection calibrators. However, process variation during circuit fabrication may result in unequal reflection calibrators. Correction equations are proposed under symmetrical transitions situation. More stable scattering parameters are obtained for single transition using proposed equations. Close results are also obtained for combined characterized transitions using correction equations and directly measured transitions.

Unequal Line (uL) Calibrator Input Mismatch Correction for TRuL calibration method

An input mismatch correction method for TRuL calibration method is proposed. The input mismatched uL calibrator is transformed into an equivalent matched uL calibrator for the afterward TRuL calibration method. The eigenvalues of this equivalent uL calibrator are distinct even when line length is equal to 0deg or 180deg. The measured scattering matrix of a branch line coupler on FR4 substrate using proposed correction method shows good agreement with simulation and the errors caused by 0deg or 180deg line length is reduced.