Shedding Pounds In Automotive Electronics

Weight is emerging as a key concern for carmakers as more electronic circuitry is added into vehicles that are either fully or partially powered by batteries. As a result, chipmakers and OEMs are exploring alternative substrate materials, different types of sensor fusion, and new ways to reduce the number of wires.

Adding pounds reduces driving range for electric or hybrid vehicles. The automotive chip industry needs to shed weight and still perform its juggling act to hit all functional safety, reliability and longevity targets. And it has to accomplish this while experimenting with AI at 7nm and beyond. As a result, OEMs are starting to explore different sensor architectures and flexible circuit boards and chips.

“Every gram matters,” said Dragos Maciuca, executive director of Ford Motor Co.’s Research and Innovation Center in Palo Alto, Calif. Speaking at the recent MSTC FLEX 2019, organized by MEMS & Sensors Industry Group (MSIG) and FlexTech, SEMI technology communities, he said, “What we want from this audience is a $5 LiDAR and 50% weight savings.”

Put simply, the goal is less weight and 360° perception.

“Since the microelectronic systems content of automobiles is growing almost exponentially, the weight and space requirements for these traditional system components will no longer be manageable,” said William Stone, director of printed electronics, integrations and operations at Brewer Science. “It’s easy to point to the wiring and circuit board side and challenge them to find weight savings. Traditional wiring harnesses are exceptionally heavy and bulky due to heavy copper cores and thick insulators. Likewise, printed circuit boards are bulky and require rigid bracketing for mounting and stability.”

But looming large is the increasing number of sensors in automobiles. Cars already have more than 100 sensors in them, said Maciuca. That number is expected to grow to about 600 for electric vehicles, and it will increase again for fully autonomous vehicles.

Why weight? How much and where?
Ford breaks technology into four parts—autonomous, connected, electrified and shared—each of which can affect weight and cost. The more autonomous the car, the more sensors are needed to make the robot car drive.

Just the in-cabin and maintenance monitoring will become important in fleets of driverless taxis, said Nicolas Sauvage, senior director of ecosystem for the MEMS Sensor Business Group at TDK-InvenSense. “Suddenly, you can’t trust that someone else in the car has been checking everything. I believe you will start to see even more sensors to monitor the health of the car itself.”

For example, he foresees in-cabin gas sensing to detect problems with the air in the cabin. That can include alerts for fire, someone smoking, or high CO₂ levels because the window has not opened in days. And that’s just inside the cabin. The car also will monitor driving and numerous functions within the car. If you can’t trust a human to check if a car is ready to drive, the rule of thumb is to add more sensors.

Still, the radar, LiDAR and cameras need to be wiped clean every now and again. “We have several autonomous cars,” said Srikanth Saripalli, an associate professor at the Unmanned Systems Lab of Texas A&M University. “They were all working fine, and one day we took one out for testing and it stopped working. We were not sure what happened. It took us a couple of days. We tested everything and still didn’t understand what had happened. And then suddenly we looked on the lasers that we have—the LiDAR we have on top of the cars. We saw a lot of bugs sticking on them. In Texas you have lots of flying bugs. You have to clean your laser every day.”

Some weight and cost will be shed when self-driving cars rule the road. Taking a driver out of the car saves on the steering wheel weight and cost, as well as everything the driver looks through. “You may not need it anymore,” said InvenSense’s Sauvage. “You may actually start to have the costs of these cars reduced a lot—maybe to a point when it’s all driverless cars where you start to not need bumpers because these vehicles won’t bump into each other anymore.”

Another example, said Sauvage, is adding a very good LiDAR/radar solution with precise time of flight. “As soon as you start to have all the driverless cars knowing about each other, and maybe you have infrastructures being built for them, at some point you don’t have accidents. If you don’t have accidents any more, the cost of this car could be reduced even further.”

While the overall weight will continue to drop, not everything is lighter. “If we move to an electric powertrain, all of a sudden we have fewer types of sensors. But we are actually going to have a lot more number of them, because now each battery cell needs to be monitored separately, we need temperature, voltage,” said Maciuca.

One thing we do know: the rate of weight increase is accelerating, said Doug Burcicki, director of Automotive Market Development at Mentor, a Siemens Business “The BMW 5 Series has increased total vehicle weight by 50% over roughly 40 years. In another case, in only one generation of evolution of the Golf (Golf SE >> e-Golf), the cost of the EE system increased by 70% and the weight 32%.”

Other weight factors
Vehicle electrification creates a power challenge, too, which has a direct bearing on weight. “As we move to vehicle electrification, instead of talking small amounts of currents, we are now talking watts and kilowatts,” said Lance Williams, vice president automotive strategy, ON Semiconductor. “That’s a big change for the automotive industry, where you are talking milliamps and amps.”

The higher the power, the more batteries that are required. This is one of the reasons that FPGAs are so popular in automotive. They use less power than multiple components because they can all be integrated onto a single chip.

“A significant amount of the weight is in the board,” said Mike Fitton, senior director of strategic planning at Achronix. “If you integrate more into a module, the board can be simpler. There are fewer layers on the board, and you can use a smaller heat sink. But you also can shrink the size of the FPGA. An FPGA is an expensive way to reduce form factor and weight. The best way to do that is with an ASIC using an embedded FPGA. That reduces the power and allows you to still keep flexibility where you need it.”

More power is needed to run the EV if it weighs more. “Electrified vehicles are critically impacted by weight, which is quickly becoming ever more critical and thus a KPI of increasing importance to vehicle engineers,” said Burcicki “Weight takes on a new meaning in the age of EV—it is the equivalent of fuel efficiency as it has a linear impact on battery range and every OEM is chasing the ultimate ‘extended’ range vehicle—except for the performance minded.”

Sensor weight
A different way to look at this issue involves the accuracy of sensors. Fewer accurate sensors and wires could decrease weight, but more sensors that are less accurate can provide faster results. Both approaches have their tradeoffs.

“It depends on what the final target is and what you want from the sensors,” said Andrea Onetti, AMS Group vice president and general manager of the MEMS Sensor Division at STMicroelectronics. “It is all about energy and heat. Consider one thing—the sensor provides you information. The quality of this information can vary. In the end, you consider it the accuracy of the sensor itself. This is what really makes the difference, because the quality of the input that you have at the end allows you to reduce the need for computing power that you have to process afterwards.”

That becomes another variable in an increasingly complex equation. Where the processing gets done depends upon how much data needs to be moved, how critical that data is for safety, and where that car will operate. In China, for example, the government is pushing to centralize all data, so processing will be done in the cloud or at some mid-point rather than in the car.

“Processing creates heat—it uses energy,” said Onetti. “And that heat needs to be dissipated, and that dissipation may need space or methods or whatever. If we look at this from a very high level, what you want to have ideally is an absolutely accurate sensor to provide ideal data that reduces by far the need for computing. Consider also that if the quality of the data is not good, you go for different iterations. And that iteration at the end is energy, and energy at the end you can reconvert in weight.”

Onetti sees different segments of the sensor markets starting to benefit from more precise sensors, this is a benefit to automotive. “The sensor is becoming key not just because its function, but because of the intrinsic accuracy that it brings. The accelerometer that you have in your phone because you wanted to have the portrait or landscape position. You need accuracy for that? No. If now you consider that inside the phone you need inertial sensor that helps you to walk in a dead reckoning situation, the accuracy even if is the same component, is by far much higher. And what it means, much better sensor despite that the name of the sensor is the same.”

But more accurate sensors also increase the cost of those devices, and the automotive industry has a long history of playing off multiple suppliers against each other to squeeze every penny out the supply chain.

Miles of wire
The traditional wire harnessing takes up a lot of space and weight even in compact cars. “The wiring harness is one of three heaviest subsystems in many vehicles—as much as 150 lbs in highly contented vehicles—and it’s very typical for the average vehicle to have 100–120 lbs of wire harness in the vehicle. These vehicles weigh on average around 3,500 lbs,” said Mentor’s Burcicki. “Today’s luxury cars contain some 1,500–2000 copper wires—totaling over 1 mile in length. To put that into perspective, in 1948, the average family car contained only about 55 wires, amounting to total length of 150 feet.”

Companies such as Yazaki, which supplies wire harness to car manufacturers, are trying different wire materials, such as aluminum versus copper, and multiplexing.

“Weight is one of the “Holy Trinity” of automotive EE systems—weight, space, cost—a fine balance must be maintained,” said Burcicki. “There are a lot of enabling technologies (wireless power transmission, optical fibers…) that would allow for wire harness and subsequent weight reductions, but there are always tradeoffs and right now—it is hard to replace the cost and functionality of copper conductors—for this reason, space limitations, weight and manufacturability (of the wire harness and the vehicle) will be the eventual drivers of technology replacements.”

But what can the electronics industry do to shed weight?

“Flexible circuits and printed wiring have the potential to replace these components and reduce system weight dramatically and at the same time make the integration of electronics with the physical structure of the automobile much more seamless,” said Brewer Science’s Stone. “Photocurable thermoplastics expand the range of polymeric materials that are available for making flexible circuit substrates, and they have the unique ability to be stabilized in a desired shape after they are molded or formed. That creates new possibilities for producing electronic systems that conform to the surfaces within the automobile.”

Testing the new flexible circuits may be another issue, as standards are just being formed for these flexible circuit boards.

However, tools for designing the wiring are improving. “A premium vehicle can have 5,000 meters of cabling, 100+ ECUs, 500 LEDs and so on. Designing these complex E/E systems is now beyond human capability,” said Andrew MacLeod, director of automotive marketing at Mentor, a Siemens Business wrote. “Virtual design of the E/E system allows an optimized architectural platform to be created, correct-by-construction, reducing the cost and weight of the vehicle.”

One way to reduce wires is to daisy-chain sensors together, suggested TDK-Invensense’s Sauvage. “Now the way you see interfacing of the sensors with a central unit will actually really make a difference. As you start to have more sensors and you start to have—I would like to call them control units, and they could be microcontrollers put together, or it could be a GPU or CPU—all of that is going to create fewer wires. That means you could even have a system where all the sensors are connected to each other like a daisy system. You connect a sensor that is connected to a sensor that is connected to a sensor that is connected to the controller. So the wires, rather than going from many-to-one, would be one-to-one-to-one-to-one-to-one to reduce weight.”

Reliability and redundancy
While the automotive industry works to lower the weight and power of electronics, those circuits also must be reliable enough to use in safety-critical systems. This has created a conundrum for carmakers, because the best way to ensure reliability is through redundancy. This approach has been proven in extreme environments, such as in satellites where tin whisker growth has been known to cause shorts, as well as in automotive electronics today using far less sophisticated technology.

“For the adaptive cruise control in my car, I have four cameras detecting data,” said Gert Jørgensen, senior vice president of sales and marketing for Delta’s ASIC Division. “I can put the cruise control on, and if I have a car in front of me the car will drop in speed until I have the right distance. And when the distance gets longer, the car goes up to normal cruise speed again. In order to get this working, you have very complex systems—cameras, algorithms, and redundancy things. This is how they do it. They make more. They do it like the old days in the Space Shuttle. They make redundant systems and then they vote. So, if four are saying one thing and one is saying something else, then they take what the four is saying and hope that this is good. This is just a way to improve quality—to make redundant systems.”

But redundancy also adds weight to the vehicle, so just duplicating components everywhere isn’t feasible.

“When you get to the time in their life when ECUs begin wearing out, that’s going to be problematic,” said Jay Rathert, senior director of strategic collaborations at KLA. “Plus, you have the weight with all of the cable harnesses. But if you can get rid of that and network the actuators to these domain controllers, if they wear out you can swap them out like a laptop. On top of that, the software is simpler.”

This is more complicated than it might appear because use cases and designs will vary significantly. “The mission profile may be 45 minutes to work, and then it’s off all day,” Rathert said. “But if you have a Tesla and it’s plugged in, there are parts that are never off. And if you’re an Uber or you’re doing long-haul trucking, you’re running up to 7 x 24. If the chip is designed to last 12,000 hours, that’s not 15 years anymore. It’s about 1.5 years. There is a tyranny of numbers, and already electronics is the No. 1 failure item. There are a lot of people who have made commitments to having this ready in the next few years, and there are forecasts this will have a $7 trillion impact on the global economy. There are a whole lot of people who want this to be successful.”

That puts more onus on design and verification, but it also increases the need to test these devices more thoroughly. At this point in time, test in automotive is far from perfect.

“This will become much more important for determining how suitable the leading-edge nodes are for automotive,” said Joey Tun, principal market development manager at National Instruments. “To do that you need to know how silicon fails. Does it fail by vibration or temperature? You also need to test various areas of a vehicle differently, and you need highly accurate impedance measurements.”

That, in turn, adds cost and time to the development process. One of the main goals of ISO 26262 is for increasingly autonomous features to fail gracefully. If an engine control system encounters an error, another electronic control unit in a vehicle should be able to take over to get the vehicle off the road and potentially to a repair facility.

“In the past, when something failed, you would look for a pattern and go find the problem,” said Anil Bhalla, director of marketing and business development at Advantest. “The challenge is that now we’re talking about system-level failures. The first problem is how you diagnose a failure when you get it. Then you have to decide what action you need to take. If you change the flow, how do you decide what to change? If you’re doing a system-level test, you can’t just increase the cost, so now you are adding other tests along with that. And then you’re trying to figure out how to fine tune the flow for more reliability. The goal is to get more data from each insertion point for test.”

The devices need to be able to be tested and inspected, too, which has put some restrictions on how components such as sensors are packaged together.

“Packaging is in all cases it’s equally as important as vehicle electrification, but it’s challenging,” said Lance Williams, vice president for automotive strategy at ON Semiconductor. “We can go very, very small, but it is limited by the ability to inspect and the ability for solder joints to stay intact through all this cycling. If you look at a cellphone, you say, ‘Well, why can’t we make all the modules that size?’ It’s impossible, relative to the size of components that are needed as a result of life expectancy and the ability to get that heat out. And you’re also not just dealing with ambient temperatures that vary. I’m not sure what everybody would do if their car came on and said, ‘Hey, I’ve been sitting in the sun too much. You’ve got to cool me off before I can start.’ I see that occasionally on my Apple phone when I lay it on the dash. So those are challenges, but we’re looking at different interconnects, different metallurgical compounds and makeup relative to devices.”

Conclusion
Weight, reliability, power and advanced features are all intricately linked in the automotive world, and those dependencies and tradeoffs will become much more apparent as the electronics inside of vehicles continues to take on more intelligence.

“This is the decade of inflection,” said Ford’s Maciuca. “We will see more change in the automotive in this decade than we have seen in the last 100 years.”

—Ed Sperling contributed to this report.
Story updated Friday, March 15, 2019 to add comments from Mentor, a Siemens Business.