TRANSCUTANEOUS SIGNAL AND POWER TRANSMISSION FOR VENTRICULAR ASSIST DEVICE

Transcription

1 Multi-Disciplinary Engineering Design Conference Kate Gleason College of Engineering Rochester Institute of Technology Rochester, New York TRANSCUTANEOUS SIGNAL AND POWER TRANSMISSION FOR VENTRICULAR ASSIST DEVICE Sara Carr, EE, Lead Engineer; Carl Hoge, ME, Robert McGregor, IE; Keith Lesser, CE; Oxana Petritchenko, ME, Project Manager. Rochester Institute of Technology Project Number: ABSTRACT The objective of the Transcutaneous Signal Transmission project was to develop a system that could transmit signals to and from a Left Ventricular Assist Device (LVAD) with a decreased amount of physical wiring through the skin. The original design of the LVAD called for a percutaneous cable consisting of 23 wires with a diameter of 8mm. Through the senior design process 16 wires were eliminated, decreasing the cable diameter to 2.7 mm. This provided 370% improvement in flexibility and less skin penetration, minimizing the risk of infection. The number of percutaneous wires was reduced by changing the placement of components as well as implementing a communication protocol, SPI, capable of transmitting and receiving the required signals through fewer wires. It was found that all signals could accurately be transmitted through the system except for the motor controller duty cycle. In addition, a transcutaneous energy transmission system was developed in order to further reduce required wires by wirelessly delivering power. The power efficiency was found to be 10% over the system and 27% over the coils. NOMENCLATURE ABS: Acrylonitrile Butadiene Styrene AC: Alternating Current DC: Direct Current HDMI: High-Definition Multimedia Interface HESA: Hall Effect Sensor Array LVAD: Left Ventricular Assist Device PCB: Printed Circuit Board PWM: Pulse Width Modulation PIC: Programmable Interface controller RIT: Rochester Institute of Technology SPI: Serial Peripheral Interface TET: Transcutaneous Energy Transmission VAD: Ventricular Assist Device INTRODUCTION A ventricular assist device is a mechanical device used to partially replace the function of a failing heart. Some VADs are used short-term, typically with patients recovering from heart attacks or heart surgery, while others are used long-term, typically with patients suffering from congestive heart failure. Long term VADs are intended to keep patients alive with a good quality of life while they wait for a heart transplant. The first versions of the RIT LVAD blood pump used a large cable approximately 8mm in diameter to transmit all power and control signals through 23 wires. This cable leads from the control unit outside the body of the patient, entering through skin, toward the LVAD blood pump. The lack of flexibility in this cable causes discomfort, limited range of motion, and is associated with many health risks to the patient because the exposure of the tissue to the cable increases the risk of infections, which often can be fatal to the patient. Approximately forty percent of patients with an implanted VAD die from infections alone, developed from wires penetrating skin. 1 Therefore, the flexibility, size, and material type of the cable penetrating the skin are critical parameters in reducing these infections. The main objective of this senior design group was to reduce the required number of wires going from the exterior system, presented on the left side of Figure 1, through the dermis of the patient to the LVAD blood pump, located on the right side of Figure 1. This was to be achieved using a signal protocol, SPI, capable of transmitting multiple signals using only three wires. Copyright 2010 by Rochester Institute of Technology

2 Figure 1. Original LVAD control electronics schematic. PROCESS The process of improving the design of the signal transmission of the RIT LVAD device took a systematic approach. The general strategy was to improve flexibility of the cable by eliminating signal wires. Then, consideration of each technology available was conducted by taking into account various factors that are important to the performance and reliability of the LVAD. With the technology chosen, the group had to ensure functions of the LVAD were not impaired. Most importantly, some components of the divide had to be safe to implant inside the human body. The first step was to obtain a full understanding of the current LVAD system in order to determine how to effectively achieve the objective. The team assumed certain electronics could be placed inside the body in order to eliminate larger wires passing through the skin. Figure 2. Proposed LVAD control electronics layout. The final product was integrated with the subsystem of the Electronics Miniaturization group, senior design project number P10021, using DisplayPort and two-pin power connectors. All necessary signals would be transmitted from the LVAD to the inner electronics box, through this signal transmission system, and to the external electronics box of group P The proposed electronics reconfiguration can be observed in Figure 2 and 3. Figure 3. Proposed placement of electronics. Copyright 2010 by Rochester Institute of Technology Page 2 of 8

3 The communication between the inner case and the outer case was provided by two MicroChip dspic33f integrated circuits. One transceiver was implanted inside the body and connected to the P10021 electronics by a 20 pin DisplayPort connector. Three wire SPI was used to transmit the LVAD data to the exterior controller at a frequency up to 40MHz. The voltages from the HESA were scaled to the tolerance of the microcontroller using a voltage divider. A voltage divider is a simple circuit that uses resistors to change the potential of the signal, but is scaled in a way that can regenerate the original signals. A 5V to 3.3V voltage regulator was used to deliver a voltage source to the microprocessor; since the processor draws varying amounts of current a resistor network was not sufficient for this voltage source. These position signals were sampled by the dspic33f using the built in 12 bit analog to digital converter. Once digitized, these signals were sent along the SPI interface. The PWM duty cycles were received on the SPI interface, sampled on the dspic33f, and sent via another DisplayPort connector to project P10021 at a rate of up to 20Mbps. The group had chosen to use two PIC33F microcontrollers to interpret and transmit the LVAD control signals because it was capable of converting necessary analog signals to 12 bit digital signals and produce all four of the required PWM signals. Figures 2 and 3 represent the proposed high level design, reconfiguration of control electronics and integration with the miniaturization team s project. ENCLOSURE The enclosure was designed to be implanted inside a human body, and therefore had to protect the inner electronics from water, shock and pressure damage. In addition, material selection was dependent on compatibility with biological tissue. The design of the case is presented in Figure 5. The enclosures were constructed from ABSPlus, a strong plastic material that is printed from a DimensionElite 3D printer. This material was chosen for its mechanical characteristics and ease of fabrication. To provide mechanical shock protection for the PCBs, rubber grommets were selected to be placed in the PCB mounting holes. Loctite 5248 biocompatible silicone was used to cover the internal enclosure and wire. This silicone is a biologically inert material, and can adhere to the surface of the plastic, providing protection against water and body fluids. TET SYSTEM In order to power the LVAD system located inside the body, four percutaneous 26 AWG wires were required. A TET system was designed in hopes of eliminating this required connection. Using inductive coupling, a primary coil located on the outside of the body would transfer the required energy to a secondary coil located on the inside of the body. Figure 5 represents required design elements for the system. Figure 5. Protective case design. Dimensions in inches. Figure 6. Schematic of TETS. In order to induce a magnetic field between the two coils, an AC power source was required. The only power supply available was a DC battery; therefore an inverter was designed to convert the energy. A PWM controller was used to create a pulse and based on a resistor and capacitor network applied to the controller, a different frequency square wave could be formed. Copyright 2010 by Rochester Institute of Technology Page 3 of 8

4 The coils used were 7cm in diameter and made of 20 and 75 turns respectively. This ratio was determined in order to compensate for some of the voltage drop across the coupled coils. The inductance of the coils was calculated using the following approximation. Equation 1 The coils could then be tuned to a specific frequency using a capacitor combination. Equation 2 Although the coils would be sewn into place to ensure continuous coupling, a backup battery was added to the design for unseen power failures. A switch was added to the internal circuitry to measure the input voltage, so that if the TET voltage dropped below 14.8V, the backup battery would power the system until the TET power reached a suitable level. RESULTS AND DISCUSSION SIGNAL TRANSMISSION The integrity of the signals through the interior and exterior transceivers was a customer requirement and required several different tests. With the use of a Hewlett Packard 33120A Waveform Generator, a 10Hz to 600Hz sine wave varying from 0-3.3V was input to the interior transceiver and measured at the output of the exterior transceiver. Through this system, the analog signal was digitized, transmitted through the SPI, and then converted back to an analog signal. Figure 7 displays the comparison of the input and output signal using a Hewlett Packard 54602B oscilloscope. able to work well at the higher frequencies that are required by the current system. Testing showed the system updates at about 170 khz while the ideal update rate would be 5000Hz. Table 1. HESA signal summary for the prototype with programmed PICs showing number of samples per cycle Sine Frequency (Hz) In Samples per period Out Etc - There were two reasons for not being able to meet the target sampling rate. The first is that delays had to be added to the program in order to get SPI communication to work smoothly. Also, the analog to digital converter on the PIC33F s were not optimally configured to scan 8 channels. Further work on the firmware is required in order to achieve the required throughput to function with the entire LVAD system. The PWM signals were tested by placing a 20 khz square wave with an amplitude of 3.3V on the exterior transceiver input. The output on the interior transceiver was probed and the duty cycle was adjusted from 0% to 100%. It was seen that the accuracy of the duty cycle was within one percent, as shown in Table 2. The delay of the PWM signals was found to be 32µs, which was less than the 200µs update requirement. Figure 7. HESA signal results for the prototype with programmed PICs Although the signals were replicated accurately at low frequencies, the system was not Copyright 2010 by Rochester Institute of Technology Page 4 of 8

5 Figure 8. PWM signal results for the prototype with programmed PICs. Table 2. Summary of PWM duty cycle signal accuracy. Duty Cycle % In Out Finally, the duty cycle of the motor controller was measured. Since the carrier of the motor controller was at a lower frequency of 50 Hz, the signal was polled at the input and then recreated at the output. It was seen that the accuracy of this duty cycle was only within 10%, which was not suitable for the control signal. This was again due to delays and the flow of the firmware not being optimal. The results could be much better if a PIC with another PWM generator was used. These results are shown in Table 3. Figure 9. Motor Controller signal result for the prototype with programmed PICs. Table 3. Summary of Motor Controller duty cycle signal accuracy. Duty Cycle % In Out FLEXIBILITY TEST It was a customer requirement for the new cable utilized by the team to be more flexible than the current cable by a marginal value of 150% and ideal value of 200%. Figure 6. Flexibility test set-up. Copyright 2010 by Rochester Institute of Technology Page 5 of 8

6 The test fixture used is shown above in Figure 6. With a fixed distance of 15 cm between the two stands, the wire was simply placed over the stands as shown. Various weights were applied in the middle of each wire for a total of seven deflection measurements not exceeding 10 cm. The flexibility was compared by calculating applied force over measured deflection for each of the measurements. The comparison of flexibility on each cable can be observed in Figure 7. It was seen that more applied force was required to deflect the original cable, indicating that it was less flexible. The improvement in flexibility of the new cable compared with the original cable was calculated to be 370%. With greater than 95% confidence, the flexibility value for the new cable was greater than the current cable by at least 200%. Figure 8. Test stand for wireless energy transfer. Figure 7. Flexibility comparison for old and new cable penetrating skin. TET SYSTEM TEST The TET system was tested to check for system range and power efficiency. A test stand was developed to adjust distance and alignment as seen in Figure 8. With this stand, various frequencies were tested to optimize the energy transferred; the analysis is plotted in Figure 9. Figure 9. Study of optimal frequency for wireless energy transfer. The desired spacing between coils ranged from a minimum of ½cm to 2cm. The TET system was tuned to 10 khz and connected to an adjustable resistor network where the output power was measured at different loads. The system was measured at five different distances from 0cm-2cm. A distance of 0cm was equivalent to a hardwired model, in which the coils were bypassed and the H-bridge was connected directly to the rectifier. The input current was limited to 2A and the input voltage varied depending on the load between 10-15V. The results of power efficiency over the entire system are seen in Figure 10. Copyright 2010 by Rochester Institute of Technology Page 6 of 8

7 Figure 10. Effects of spacing on power efficiency over system. Figure 11. Effects of spacing on power efficiency over coils. It was seen that although 150 khz was the optimal frequency for energy transfer, the PWM amplifier performed best at 10 khz. Over 60% of the power was dissipated over the hardwired system, which included the PWM amplifier and a voltage regulator. Efficiency was then calculated over the coils only, as displayed in Figure 11. It was found that power efficiency was between 15-30% at 0.5cm spacing and between 0-10% at 2cm. CONCLUSIONS AND RECOMMENDATIONS Within the designed system, the signal integrity of the motor controller could be improved by adding an additional signal wire or by utilizing a microcontroller that has five PWM generators. The addition of a signal wire would not greatly affect the diameter of the signal cable, still keeping the cable below 3 mm. By changing to the dspic33fj64mc506 microcontroller, the four PWM signals as well as the motor controller signal could be recreated through the microcontroller. Since the microcontroller comes from the same family as the one used in the system implementation, minimal code changes would have to be made. The change in chip would require a new PCB layout to match the 64 pin foot print, which is currently a 44 pins. A hardwired signal protocol was used for this system due to its limited power consumption as well as high data rates and reliability. Although the technology for wireless systems has been developed, the most commonly used frequency range is 2.4GHz, which is an unlicensed band susceptible to interference that would require additional attention during the design process. Medical bands are now available for MHz, but these data rates are still relatively low at 800 khz, which did not meet design requirements for this system. Future iterations of this project could look into alternative signal protocols, including wireless USB, which meets required data rates and eliminates all signal wires. A focus would have to be placed on USB programming protocol, security of wireless network, interference from other devices, and power consumption. The current TET system is inefficient in power transfer due to power dissipated over the PWM amplifier and the voltage regulator. This could be improved by utilizing different components to generate and sustain the AC signal required for the wireless transfer. Additional testing would be required to optimize the power efficiency of the AC signal at higher frequencies, which would improve efficiency of power transfer over the coils. The voltage regulator however is a necessary component to ensure that voltage spikes do not occur, which could harm internal components. The current design for the TET is an open-loop system, which means that the settings can only be adjusted through human intervention. Future implementations could utilize a feedback network to ensure the internal system is delivering the required power. This would require wireless signal transmission of internal vitals as well as a simple controller such as a proportional integral (PI) controller to compare the reference signal with the delivered signal. The ABSplus material used to construct the enclosures was porous which prevented the case from being independently waterproof. However, ingress protection can be achieved by applying silicone over the case and connectors. One layer, less than 1mm thick in most areas, Copyright 2010 by Rochester Institute of Technology Page 7 of 8

8 was applied at took roughly 18 hours to cure. This thin layer did not adequately seal the enclosures but it is expected that a second layer would. The ABSplus had adequate ductility and strength to protect the electronics during impact, as proved by a drop test. Overall, the customer was pleased with the result of the TET and signal transmission systems, except for two specific elements. First, he would like the HESA signals to be sampled at a faster rate to improve functionality. Second, the customer would like the current H-bridge driving the TET to be replaced with one which does not dissipate as much heat. REFERENCES 1.Auckland Bioengineering Institute News articles- The University of Auckland. Auckland Bioengineering Institute - The University of Auckland. [Online] September 15, [Cited: February 6, 2010.] rticles/template/news_item.jsp?cid= ACKNOWLEDGMENTS Special thanks to our Senior Design guides Dr. Day and Dr. Lux, our customer, Dr. Cheng, and cooperation of the electronics miniaturization team, P10021, Lowell Smoger, Evan Sax, Christine Stone, and Mike Calve. Thanks also to all professors who provided advice and direction during this project. We appreciate the advice and assistance of the machine shop staff, Dave Hathaway, Steven Kosciol, and Robert Kraynik. Thanks to John Bonzo in the Brinkman Laboratory for allowing us to use their DimentionElite 3D printer. Also, we are grateful to Christine Fisher and the Senior Design and Mechanical Engineering offices for assistance on purchasing of our components. This project was sponsored and funded by the National Institutes of Health and the Utah Artificial Heart Institute. Copyright 2010 by Rochester Institute of Technology Page 8 of 8