Designing SoCs For Hybrids

Hybrid vehicle sales are growing, driven by a global concern for lower vehicle emissions and consumer demand for better economy. This has set off a rush by semiconductor companies to provide key components for those vehicles because they are much more reliant on electronics than regular gasoline-powered vehicles.

But the changeover is not as straightforward as it might sound. Hybrid vehicles have unique requirements and issues, and suppliers throughout the automotive ecosystem are scrambling to deal with them. While they require more SoCs, those SoCs are required to do much more complex tasks in more constrained environments.

To begin with, power is a key delineation between hybrid electric and traditional combustion engines.

“The whole power train and available power for electric vehicles makes the design substantially different from combustion engines,” said Adam Sherer, product management group director for automotive safety in the Systems Verification Group (SVG) at Cadence. “In the power train, we may have a distributed motor system at each wheel without a transmission. These ASIL-D (automotive safety integrity level D) components operate in a networked environment, so the functional verification task inherently must scale from the individual ICs to this complex subsystem. Furthermore, in an electric vehicle, the power consumed by the electronics directly affects driving range which remains a key metric today where battery-based vehicles trail combustion engines.”

Consequently, electronics providers are encountering stringent requirements for power consumption, which translates into an increased need for requirements tracing and rigorous verification as the new systems take to the road.

According to Richard York, vice president of embedded at ARM, the conventional power train segment is one of the most cautious bits of the industry because the volumes are high, and therefore they’re not keen on taking risk. “When you’re in a high-volume market, you don’t want to take risk unless you have to. The market up until now hasn’t changed dramatically year on year. It’s been a constant, steady change. And that means they are less likely to look around for new solutions. Conventional power train up until now has been fairly steady, and until recently hasn’t had enough change to force them to look to other solutions. But that is changing.”

Functional safety requirements for the conventional power train world, and the electric motor world, are quite different, he said. “I wouldn’t say it’s easy to make a conventional powertrain controller safe, but typically with a gasoline or diesel engine if you do detect faults it’s relatively easy to cut off the fuel supply to make that system inherently, reasonably safe. When you do that, the conventional engines don’t tend to lock up. They don’t tend to misbehave. And cutting off the fuel supply is a relatively straightforward thing to do. You just shut down the fuel pump. That’s made those systems straightforward to achieve the right levels of safety. In the electric space, it’s much harder because, for a start, there are big batteries. The amount of current and fire risk in those is much greater. Electric motors, if you stop controlling them properly, won’t tend to behave near as nicely as a gasoline or diesel engine in terms of shutting down. If you short circuit some of the things in the drive electronics, these things can really misbehave. As such, the functional safety requirements are quite a bit higher for these engine controllers, for the motor controllers than they are in a gasoline engine.”

Another challenge in the design of a hybrid vehicle is the blending of power from different sources. “As soon as you blend any of these systems again, the system that blends the power is quite a critical component and the safety requirements of that blending bit are higher — and this makes the system as a whole, a whole lot more complex,” York continued. “You want a smooth transfer of power between the two bits. Some of the hybrids are actually not a hybrid. They are entirely electric, and there’s a little engine in the background to top off the battery. Most hybrids, though, have two power sources being blended together in a transmission system to give a good smooth amount of power throughout the range. If you get that blending wrong and there’s a fault in that, you could have a fairly catastrophic problem.”

From ARM’s point of view, this translates to the SoC providers as far as how much support is given in terms of functional safety, he said. “If we are interacting with a conventional powertrain SoC vendor, their demands are much more manageable. They like the functional safety support we give them but they’re not placing very high levels of requirements on us because it’s much more straightforward to manage safety in that sort of system. Hybrids change the conversation. They are much more worried about the documentation we’re providing, how much testing have we done on our processes, than in the conventional world.”

Vehicle system complexity on the rise
All of the new requirements for hybrid vehicles are in addition to existing requirements of traditional vehicles with combustion engines, and this makes the system level challenge that much more complex.

“If you think about a traditional car with a combustion engine, the electronics started creeping into those cars little by little,” said Mick Tegethoff, director, AMS product marketing at Mentor Graphics. “The first place you saw electronics was for the timing control, what controls the sparks in the spark plugs. Then you had computers helping with the fuel injection, automatic door locks, ABS brakes, traction control. But even here in the traditional automotive space, there are still plenty of challenges. What we find in that category is that they build ICs that on one hand have to communicate with a controller in the 5V world, and in the other, it has to actually drive a mechanical assembly in the brake system. Those are very complex designs.”

And then there are hybrids. What makes a vehicle a hybrid is that in addition to a combustion engine making the wheels turn, there will be an electric motor doing that.

“The minute you have an AC motor, it’s a whole different ballgame,” he said. “You need to somehow take energy from a battery and invert it, so it goes from a DC voltage to an AC voltage to drive a motor. There are all kinds of requirements there that it’s a whole new set of rules for those people. Intelligent control, harsh environment, being able to use semiconductors to control changing from a DC voltage stored in a battery to something that drives a motor — all those things have a whole set of new requirements. You have different battery types that must be accounted for. The nickel-metal-hydride batteries need to be packaged safely. There are DC-to-DC converters, high voltage all over the place. And then there are the extremes of operating conditions, computers running at 5V or less, and huge voltages on the motor drivers.” The list goes on.

Functional safety requirements are significant, noted Ahmed Eisawy, product marketing manager for the Eldo product line at Mentor Graphics. “Components are being addressed at a very specific case. When you move from mechanical systems to electrical systems, safety concerns go through the roof. How are you going to make sure the electronics will work automatically correct? People following the [gettech id=”31076″ t_name=”ISO 26262″] standard on mechanical systems and moving to electrical now have more concerns, typically about the tool confidence. They need confidence about each and every electrical component, not just the system level, as well as how they can trace all the requirements from all the datasheets and the spec of the system and components all the way to actual tests that they can apply, and trace back results to those specs,” he said.

Fault injection becomes a major concern. In a mechanical system, there are standard types of tests, but in an electrical system faults must be injected to see what will happen. What-if scenarios become crucial.

“In general, in those types of systems, you see more concerns around IC packaging because you have more electronics around harsh environments,” Eisawy said. “There is thermal reliability. The engine has to work longer because it won’t be replaced every four or five years, so they have to trace that all these electronics have to work for 10 or even 20 years. There are more requirements on the materials being used. Do we use regular silicon or move to silicon carbide, gallium nitride? So new materials are being considered.”

Finally, there are more concerns about discrete products because with DC-to-DC, and DC-to-AC converters, there are a lot of power devices. The insulated gate bipolar transistor (IGBT) becomes more heavily utilized, and there are more of them than in a conventional engine. And all of them have to be considered in context of thermal requirements.

Challenges across the board
While functional safety is a major focus, Sven Natus, Head of Automotive Business Unit Americas at Cypress Semiconductor, pointed out this can’t be limited just to a comparison between traditional power train and electrical vehicles because the ISO 26262 standard is becoming more important in all application domains. “Certainly this is led by functional safety critical devices like braking or steering-by-wire systems, and certainly motor control when it comes to electrical engines, but even in areas like the instrument cluster, driver information or in-body devices, we see that come in more and more.”

Another issue to be resolved is how to handle over-the-air updates within the vehicle system, Natus said, because carmakers across the industry are looking at this approach to keep technology current. Tesla has set the bar for this approach, setting off a competitive race with some technical hurdles.

“We see that Tesla is doing that quite successfully,” he said. “Tesla has to rely on over-the-air updates to fix software-related problems in their vehicles, as they do not have a huge network of service stations. They also recently deployed new features. With a software update, there was an autopilot feature that was provided to the owners of the Model S, and that is, of course, something amazing. You get in your car the next morning, and all of a sudden there is Autopilot available. From a marketing perspective, that was a great coup.”

Accordingly, the Big Three U.S. automakers — General Motors, Ford and Chrysler — are very interested in implementing this technology, Natus said. “One of them is literally pushing to get this today. However, there is major challenge for the semiconductor industry. You don’t want to have your vehicle inoperable at any point in time, and given that [over-the-air updates] involve hundreds of megabytes of data, along with multiple ECUs because of distributed functions in a vehicle, the process is much more complex than updating an app on a smartphone, for example. In addition, there is the issue of bandwidth to distribute that data in the vehicle through the existing networks like CAN, where the bandwidth is very limited. It takes time to provide the update data to all the different nodes, and for sure, nobody wants to have the situation where they are in desperate need of their vehicle but the update cannot be interrupted.”

As such, the OEMs want a background update where the user doesn’t recognize it, and at a certain point, the driver can switch to the new version. In that case, every ECU literally has to carry two software versions simultaneously. While some virtualization is used, for small ECUs two independent flash macros have to be present in order to be able to execute software out of one internal flash, while updating the other. These need to work independently. But today’s technology doesn’t allow this because most of the microcontrollers on the market have a single flash macro that allow for one or the other.

This situation is preventing OEMs from implementing the over-the-air update solutions today. Cypress, however, has a solution that can solve the problem. While not currently being utilized in design, Natus said one of the OEMs is getting close.

Conclusion
Given the differences between traditional combustion engines and hybrid electric vehicles, the technical and regulatory challenges are giving rise to a new way of thinking in the design of a vehicle. Many of the issues are being worked out today. Solutions will roll out over time, because in the auto industry the design cycle can be as long as a five years. At the end of the day, though, it’s all beneficial for the semiconductor industry in terms of longevity for established manufacturing nodes, and it creates plenty of interesting problems for engineering teams to solve.