Industry strives for no-fault automotive microprocessors

Here’s what the auto industry wants: Microprocessors that are tailormade for the automotive environment, that can withstand huge temperature swings and near-constant vibration and, most importantly, never ever fail.

To make microprocessors that meet those requirements, experts say a rethink of the entire production process is order. And that’s exactly what is happening.

“Until now, chips were developed, produced and then offered to carmakers who then decided about possible applications,” said Oliver Bringmann, a researcher at the FZI, a research institute in Karlsruhe, Germany. “Now the goal is to achieve a standard that already takes into account the requirements of the carmakers before development even starts.”

That’s a major change and it’s the goal of a new German research project called Rescar 2.0 that started in February.

Rescar, which is the German acronym for “Robust development of new electronic components for electromobility applications,” groups research institutes, universities, automakers Audi and BMW, and suppliers such as Infineon, Robert Bosch and Elmos.

One of the issues the group faces is that chip requirements for the car differ markedly from one application to the next.

“A chip in the powertrain, for example, needs to be able to function fault-free at temperatures of 170 degrees centigrade,” said Ralf Pferdmenges, an Infineon executive. “In the interior, the requirements for functions such as infotainment applications are less demanding.”

The biggest automotive requirement – reliability – still causes microprocessor makers the biggest headaches. Chips that play a key role in steering and braking systems simple cannot fail.

Yet today there are no universal methods for testing robustness and reliability. Hence, chipmakers and auto companies do their own stress testing. Car companies such as premium carmaker BMW want failure rates of less than five per million. The stated goal is zero failures for all semiconductors.

The issue is about to become more important as electric vehicles and networked cars require ever more powerful electronics. Today a premium vehicle already contains 80 or more electronic control units (ECUs) and that number is set to grow.

Accoding to market researchers Strategy Analytics, the value of the microprocessors currently built into the average car is 272 dlrs. That will rise to 322 dlrs in five years, but the researchers predict that in electric cars the value will be 1,000 dlrs.

Without chips, nothing will work

Nothing will work without chips in an electric vehicle and that introduces a whole new range of challenges.

First, traditional hydraulic and mechanical functions will be replaced by electronic components in electric vehicles. That increases the number of chips that will go into a car.

Second, development times will continue to get shorter, which will increase the risk of problems with testing and reliability.

Third, automotive systems will have to do more and will, thus, become more powerful. Microcontroller chips for engine management can contain up to 150 million transistors.

And fourth, the high-voltage electronics in an electric vehicle will raise interference issues that need to be dealt with.

“What is required of these chips is, in part, completely different from anything in the past,” said Infineon’s Pferdmenges. He noted that, whereas in the past, chips only had to function while the car was being used, an electric-vehicle chip may need to be always on. That will be required for battery management, for example.

The industry needs to attack a range of other issues as well. Issues that Rescar 2.0 is addressing include how vibrations affect chips and connectors; what kind of life cycle can be expected; to what extent production variations can be tolerated; and how conductors are affected by more powerful electronics.

Despite all these challenges, the zero tolerance for faults in critical areas stands undisputed. That may mean the auto industry needs to use redundant systems, just to play it safe.

Rescar 2.0 wants to create a development and production process that eliminates faults even before production has started. “The later you find a fault, the more expensive it becomes,” said Bringmann, the FZI researcher.

Which raises the issue of costs, which are, next to safety, the key factor in making electric vehicles attractive to potential buyers.

“When robustness is already taken into account in the design phase, development cycles get shorter and the costs of electronic components drop,” said Julia Lau, an Infineon engineer involved in the Rescar project.

Some experts also expect that such a new design strategy will reduce the amount of testing required, but most robustness checks will continue to be done in a live environment.

-By Gert Reiling

 


Related Articles