 Van Dusen presenting Enabling High-Speed Connectivity in Modern Sensor Systems. Herb? Hello, my name is Herb Van Dusen from WL Gorin Associates. Today I'll be talking about the importance of high-speed connectivity in sensor systems. I have been working for Gor for over 40 years and have a BFCE degree from Rensselaer Polytechnic Institute. I am currently a co-chair of the SOSA Electromechanical Subcommittee, and my co-authors are Grant Lawton, Greg Powers, and Jeff Woods, all from Gor Associates. Sorry about that. Got it. Okay, so I missed the introductory slide, but it pretty much said what I said. All right, so to start with, talk about sensor systems in general. I think that everyone here would agree that the field of sensor systems is one of the most important and challenging engineering endeavors today. To say that sensor systems are mission critical is an understatement. Vehicles today have become more dependent on sensors for basic operations, and the entire mission itself might be just to gather data using these sensors. With such a large number of sensors all working together, reliability is becoming more and more of a concern with these systems. Failure with any of the components of a sensor system could degrade the mission or cause risk to the crew. Reliability concerns increase exponentially with the number of sensors. With each sensor possibly generating enormous amounts of data, the demands for data processing and control are reaching new heights and becomes harder and harder to keep up with the information. To achieve a true open systems architecture, sensor elements must operate seamlessly with the data processing and other functions of the system, including software, activators, and other sensors. Providing real-time feedback and sensors enhance the use of artificial intelligence, the future will see more pilotless vehicles as they become more effective at performing complex missions, thereby minimizing the risk to highly trained pilot. A key to all of this is to have a reliable means of communicating electronic signals throughout the vehicle. Connectivity is the lifeline of keeping the system running effectively and efficiently. This discussion is focused on the challenge and what the SOSA Electromechanical Subcommittee is doing to enhance error-free connectivity. My colleague and co-author on this paper, Grant Lawton, first suggested the idea that a smartphone is a common everyday example of a sensor system. The elements of a complete sensor system are all there. Higher and higher speed processors make these little units what used to be called supercomputers. More and more memory is required to run sophisticated applications and store data from its sensors. Power is the driving factor in smartphones to extend battery life. Finally, open architecture software needs to drive the innovation and enable artificial intelligence. Sensors are numerous. Some are low speed like microphones, touch screens and switches. Others can be very high frequency including Bluetooth, RF antennas and other high speed cameras including video. They even have temperature sensors to protect the devices from overheating. The support of vehicle sensors are pretty... Support elements of vehicle sensors are pretty much the same as those found in the smartphone including the high speed processing, terabytes of memory capacity, efficient power generation and open architecture software with an eye toward artificial intelligence. Some of the sensor elements are the same while others are much more sophisticated however. Multiple camera systems must be integrated into these systems providing 360 degree situational awareness. Complex IR sensors detect and identify targets and threats. Advanced gyros or IMUs, inertial measurement units working with the high speed processors facilitate greater accuracy for dead reckoning. There are probably other sensors at this time that are highly confidential and maintaining secrecy and security is an important capability. Although there are similarities... I'm sorry I skipped the slide. Although there are similarities between a smartphone and a vehicle sensor system, there are significant differences. Data rates tend to be higher with the sensor systems. The interconnect differences are remarkably different. The smartphone signals are transmitted over highly integrated printed wiring boards and flex circuits. While sensor systems must use discrete interconnects made from copper wire or fiber optic assemblies needing rugged connectors. While these differences exist, the connectivity requirements is still the same. Communications between the sensors and the rest of the system must be seamless and error free and this is such a challenging task. The challenges of connectivity occur on many fronts. There is IO from a multiple of sensors, multitude of sensors, each needing attention on a regular basis and must be serviced in real time. For some sensors, low latency is critical. Latency can be defined in many ways, but for this discussion, I would just say that latency is the time it takes for a signal to be detected by a sensor before it's acted upon by a sensor element. Latency can be due to interconnect media itself, but is most often due to a function of the signal protocol being used. Ethernet is great for sending streams of data at high speeds over twisted pairs, but signal conditioning can increase latency and response time. Although there are ways to overcome this within Ethernet, it may be worth considering a direct attach approach. Although that may put more demands on the signal integrity of a link. Systems should allow sensors to be easily swapped out as the mission changes. Connector systems must be robust mechanically and compatible between different manufacturers in order to allow for this swapping. A major consideration for interconnect design is EMI immunity. We can never know what kinds of electromagnetic fields may be encountered during an operation, so we must do whatever is possible to protect the signals from being corrupted. As I mentioned, data rates keep increasing, so signal integrity is more demanding as these capabilities increase. Interconnects must be designed with the potential for upgrades in mind so that new modules won't require interconnects to be replaced after each enhancement. Field repairability is also a concern, particularly when higher speeds demand better controlled impedance connectors, which can be more difficult to assemble. With multiple sensors being added, there must be a strategy to make these interconnects scalable so that new systems design approach is not needed as the number of connections go up or as the data rates increase. Finally, the connectors and cables that are selected must be approved by a standards body so that multiple manufacturers can provide them. This helps minimize cost and reduce time to implementation. This data connectivity roadmap shows Ethernet connections and other protocols, including included in the paper, and there's a little too detailed to discuss now. I'll just say that it's enough to say that the electromechanical subcommittee is looking into all of these identified high speed needs and is working on ways to address those. Some of these are currently available off the shelf and some are still in development. For instructional purposes, about how we are approaching high speed interconnects, it's worthwhile to look at one example of how engineers might typically approach high speed wiring. This connector diagram shows a pinout that was selected for sending one gigabit per second Ethernet over four twisted pairs as shown in the diagram. The pinout allows signals to be adjacent to each other within the pins of the 38EE9 connector. This connector is a non-controlled impedance variety and this spacing will obviously introduce crosstalk into the link. Crosstalk can be a real nemesis for high speed data transmission which could cause the link to auto negotiate to a lower data rate when more crosstalk can be tolerated. The situation can be improved somewhat by changing the spacing, but there is still a loss in signal density and some electrical parameters will not improve for this new pinout. Additionally, making this switch will not allow the connections to pass specs beyond one gigabit per second, although designers will try to use this approach. This spacing out is commonly done by designers, but it's not in line with many of the SOSA concepts for modularity and robustness. Additionally, each time a new pinout is generated, more testing and qualification steps are required which add uncertainty and risk. So we don't typically recommend using 38EE9 connectors open fin field for high speed signal generation. Instead, we recommend solutions such as this one. This is a purpose built high speed connector, excuse me. It's a purpose built high speed connector which has controlled impedance through a series of high density contacts designated for signal pairs and ground. It is capable of meeting the TIA standard requirements for interconnects up to 10 gigabits per second. This connector scheme can also transmit higher speed data from other protocols which have even more stringent signal integrity requirements than the Ethernet. This connection system is called Hercules produced by Meritech. It has been approved as a standard by the Vita Connect Committee and it has been adopted as J7 by the Electromechanical Subcommittee within SOSA. In addition to connectors for high speed digital data, we are also including coaxial interconnects for RF data. We have identified connectors that can handle various coaxial contact sizes including size 16 for lower speed video signals, size 8 for higher power signals up to 20 gigahertz, and size 12 for high frequencies up to 70 gigahertz. For these contacts, especially with the size 12 version, care must be taken to ensure compatibility between manufacturers for meetability and acceptable electrical performance. Cable robustness is also a consideration and links should be tested after installation, especially for high frequency lines. We see quite often after a line with these higher frequency cables can deteriorate the electrical performance. The current version of the spec includes connections for fiber optics. Currently, this only includes simplex fibers, but with proper configuration of signal lines, data rates of 100 gigabits per second can be achieved. We are also looking into single mode fibers, which can offer longer distances and higher degrees of multiplexing. These connections are made within the environment of a 38999 connector. A new development actively being considered is the fiber optic MT connector series grouped in a 38999 housing. This connector system from TE connectivity is currently being evaluated by the Vita committee and is working through the process as Vita 87. This approach could handle up to 96 fibers within a single connector and allow data rates up to 400 gigabits per second and beyond. A key aspect of connectivity is when to choose fiber optics over copper as data rates increase. This is a complex decision involving the length of the interconnect, data rates, anticipated environment, and field repairability. Latency may also play a role since it takes some finite amount of time to convert electrical signals to optical and optical back to electrical for processing. In general, copper still works quite effectively for Ethernet up to 10 gigabits per second. Developments are occurring with power over Ethernet that could be very interesting for sensors. Power over data lines, which can currently transmit bi-directional Ethernet data at 1 gigabit per second while delivering 50 watts of power seems very interesting in a way to reduce swap. Higher data rates and longer lengths are the domain of fiber optics. Notice in the chart that there are several ways to accomplish higher speed data through scalability. This means choosing how many channels at a given data rate will yield the desired throughput. Transceivers and multiplexing technologies are constantly advancing to provide increased capabilities into the future. Hopefully, you'll now agree that high speed connectivity is critical to sensor system design and should be considered upfront, not left as an afterthought. Both copper and fiber optics assemblies will play a part in meeting the future requirements. Interconnects must be designed for harsh environments, including high levels of EMI. By using standardized solutions, it will help to reduce design and development time and to reduce cost. Finally, increasing data capture creates the need to keep innovating in the interconnect arena. Continuing innovations in both copper and fiber optics will ensure that products are available to meet future needs. So that concludes my presentation. At this point, I'd like to thank my fellow authors who all contributed greatly to this presentation. I'd also like to thank the open group and SOSA consortium for allowing us to present today. And finally, I'd like to ask anyone out there who's interested in sensor system development and high speed interconnects to join our group. We're always looking for help and you will be able to have a significant impact on creating this new standard. Thank you, everyone. Thank you. Do we have any questions for her? I don't have anything in the Q&A box yet.