Eric J. Basham

Inductor Modeling in Wireless Links for Implantable Electronics

This paper describes the ac power dissipation of coils as well as their self-capacitance, self-resonant frequency, and quality factor Q. In the past, self-resonant frequency was rarely calculated during design because of the lack of suitable closed-form design equations. However, coils are widely used in biomedical applications as inductive links for both power and data, and the power transfer capacity and the data rate of inductive links are determined by the operating frequency of the coils. The maximum operating frequency is limited by the self-resonant frequency of the coil. We present here an analytical express for the optimal frequency of a coil in terms of the design parameters. By varying the design parameters, we can move the optimal frequency close to the operating frequency, thereby boosting the efficiency of the inductive link. We have verified the derivation experimentally and shown it to be useful in optimizing coil Index performance.

An efficient wireless power link for high voltage retinal implant

The design and analysis of a power telemetry system with multiple output voltages is presented in this paper. The system includes a Class E power transmitter, cascaded resonant tank, diode rectifiers, reverse data telemetry and regulators. The power loss associated with voltage conversion is minimized by using a proposed cascaded resonant tank. Reverse telemetry senses the power level fluctuation due to the coil movement and reliably provides minimally required power for the implant. The power telemetry system provides 5 voltage levels (tunable up to plusmn 12 Vplusmn2.5 V and ground), each supported by a dedicated integrated regulator except the ground. The maximum deliverable power is in excess of 100 mW when power coils are separated by 1 cm.

Circuit and Coil Design for In-Vitro Magnetic Neural Stimulation Systems

Magnetic stimulation of neural tissue is an attractive technology because neural excitation may be affected without requiring implantation of electrodes. Pulsed discharge circuits are typically implemented for clinical magnetic stimulation systems. However, pulsed discharge systems can confound in-vitro experimentation. As an alternative to pulsed discharge circuits, we present a circuit to deliver asymmetric current pulses for generation of the magnetic field. We scaled the system down by using ferrite cores for the excitation coil. The scaled system allows observation using electrophysiological techniques and preparations not commonly used for investigation of magnetic stimulation. The design was refined using a comprehensive set of design equations. Circuit modeling and simulation demonstrate that the proposed system is effective for stimulating neural tissue with electric-field gradients generated by time-varying magnetic fields. System performance is verified through electrical test.

A test microchip for evaluation of hermetic packaging technology for biomedical prosthetic implants

The development of a test chip that will be used to evaluate a hermetic and biocompatible package for the driving CMOS circuitry of a retinal prosthesis is described. The package design is estimated to be about 2 /spl times/ 2 /spl times/ 0.3 mm/sup 3/ and will be formed by conformal layers of parylene and a metal (e.g. titanium) as inner and outer protections, respectively. The test chip has been specifically designed for evaluation of the packaging technology. It consists of many blocks of analog and digital components as well as relative humidity and temperature sensors. The test chip has more probe points than a typical chip, allowing a more thorough evaluation of circuit behavior during the testing. This chip will first be coated in a layer of parylene C and soaked in heated isotonic saline for an extended period of time. Every block in the chip will then be tested for functionality using the surface probe points. The next step is to coat the surface of another test chip with parylene and a metal and repeat these soak tests. The results will then be analyzed and mean time-to-failure for the different samples will then be computed. Using the accelerated testing paradigm, these results will then be extrapolated to mean time-to-failure in the operating intraocular environment. Parylene test structures have already undergone an accelerated lifetime test and results have been analyzed.

Hafnium transistor design for neural interfacing

A design methodology is presented that uses the EKV model and the gm/ID biasing technique to design hafnium oxide field effect transistors that are suitable for neural recording circuitry. The DC gain of a common source amplifier is correlated to the structural properties of a Field Effect Transistor (FET) and a Metal Insulator Semiconductor (MIS) capacitor. This approach allows a transistor designer to use a design flow that starts with simple and intuitive 1-D equations for gain that can be verified in 1-D MIS capacitor TCAD simulations, before final TCAD process verification of transistor properties. The DC gain of a common source amplifier is optimized by using fast 1-D simulations and using slower, complex 2-D simulations only for verification. The 1-D equations are used to show that the increased dielectric constant of hafnium oxide allows a higher DC gain for a given oxide thickness. An additional benefit is that the MIS capacitor can be employed to test additional performance parameters important to an open gate transistor such as dielectric stability and ionic penetration.

An analog circuit implementation of a quadratic integrate and fire neuron

Silicon neurons are of importance both to implement hybrid electronic-biological system as well as to develop fundamental understanding of the neurobiological systems they emulate. We have implemented a hardware version of the quadratic integrate and fire neural model. The quadratic integrate and fire neuron differs from the more common integrate and fire neuron in that the model, and thus the hardware, intrinsically generate spikes. Readily available discrete surface mount components are used to make the hardware available to a wider audience and facilitate experimentation.

A Course for Designing Transistors for High Gain Analog Applications

A laboratory based course for Electrical Engineering Masters students that teaches students how to design a process flow for metal oxide semiconductor field effect transistors (MOSFETs) that are optimized a high DC gain is presented. This course was developed, because while much of analog circuit design has to use processes that are optimized for high speed digital operation, there still exists design space for transistors that are optimized for analog operation in the low frequency domain. An example of this design space would be circuits that interface with neurons.

Hafnium transistor process design for neural interfacing

A design methodology is presented that uses 1-D process simulations of Metal Insulator Semiconductor (MIS) structures to design the threshold voltage of hafnium oxide based transistors used for neural recording. The methodology is comprised of 1-D analytical equations for threshold voltage specification, and doping profiles, and 1-D MIS Technical Computer Aided Design (TCAD) to design a process to implement a specific threshold voltage, which minimized simulation time. The process was then verified with a 2-D process/electrical TCAD simulation. Hafnium oxide films (HfO) were grown and characterized for dielectric constant and fixed oxide charge for various annealing temperatures, two important design variables in threshold voltage design.

High-k dielectric fabrication process to minimize mobile ionic penetration

A process for fabricating hafnium oxide (HfO) films to minimize ionic penetration was developed and tested. A 333Å HfO film was successfully deposited by thermal evaporation. The film was characterized through capacitance versus time (C-T) and capacitance versus voltage (C-V) measurements. The films were exposed to a solution of 0.1M NaCl physiological saline and preliminary results showed that the ionic species did not alter the electrical characteristics. The relative effective dielectric constant of the hafnium oxide layer and SiO2 interfacial layer was 10.5, while the relative dielectric constant of the hafnium oxide layer was 18.

In vitro magnetic stimulation of unmyelinated nerves

Magnetic stimulation of neural tissue is an intriguing technology because excitation may be affected without a direct interface between the stimulator and the tissue. Current methods of magnetic stimulation use large air core coils, limiting the size of the neural preparations available for study. We use ferrite cores to reduce the stimulation area. Scaling allows the use of unmyelinated neural preparations with a range of space and length constants. Results are shown using two well studied neural preparations under varying test conditions.