Chuck, Excellent article. It's amazing that high end vehicles could have as many as 100 microcontrollers but it makes sense. Infotainment and new "smart" features just add to the load. Will be interesting to see how quickly there could be some combining of functions. I assume that is a trend for the future?
Yes, combining of functions is the plan going forward, Al. There are two theories: Domain control, in which electronics are clustered by function; and zone control, in which electronics are clustered by geography. I think it will be a combination of the two, with domain control rpobably being more prvalent. Either way, it will take a lot of computing power to bring it all together.
The more complex a processor/MCU the more prone to failure it can become. Every PC I had locked up at one point, every phone eventually failed, even my graphing calculators would experience an error. Cars will need to be reset in the future like our smartphones, I wager.
Simple MCUs that perform one specific task with little outside connectivity are easy to test and ensure they will work the same every time. Not the case with more transistors onboard.
I'm not suggesting living in the past with cars... but better safety is needed. Hope that is the case.
"A single master controller" would be one huge nightmare from a number of different perspectives. First and very obvious is the replacement price, which would undoubtedly be over a thousand dollars. Not bad while the car is under the 3 month new car warranty, but a real killer for the second owner owner of a ten year old used vehicle. Next comes the fact that wswith all of those dozens of functions in one module, the wiring to that module would therefore consist of all the wiring that went to all of those modules. The resulting multipin connectors would be a horrible reliability problem, and a real pain to install and remove. Next comes the millions of lines of code that would undoubtedly have a few errors that would not appear until after the first few thousand cars had been sold. And have you ever heard of software that somehow connected otherwise unrelated processes? Just consider those possibilities.
Yes, William K, I think that wouldn't be such a great idea, either, to have a single master controller. But I can see how this might be one direction the industry might head, having seen examples in other industries with other technologies. Or maybe the idea would be vetoed from the beginning since engineers would anticipate what a disaster it would be, as you describe!
Elizabeth, It is probably correct that things will move in that direction unless something changes to make it more expensive than some other approach. What might be effective would be a boycot of vehicles that were excessivly expensive to repair, or, better yet, a mandatory ten year warranty on vehicular comtrol modules. A law like that could also, possibly, improve quality. But it is far more likely that it will simply mean that vehicles will be scrapped much sooner, since replacing that VERY EXPENSIVE module might nit make economic sense for an eight year old vehicle. But how will it be rationalized for those long-life-still-pricey battery electric cars?
Of course it could also create a whole new industry that produced custom modules for cars, which could offer the benefit of allowing the owners to have completely different software that ignored all government mandated functions. That could be an unanticipated benefit.
Cabe, completely aside from the use of a single chip there exists that reality of the amount of external I/O that is needed for the automotive application. The car-body environment is a good example of the real world in that most of the inputs and outputs connect to real things, not to display screens and touch panels. A single module would put all of that I/O in one box as well, and that one box would only be serviced as an assembly. THAT is the other reason for believing that it would benefit nothing except possibly the makers bottom line.
A ten year warranty would be a good start towards making a single module realization somewhat acceptable, but it would need to be far more inclusive than the warranties that I am familiar with. As far as making the chip "reliable enough", there would need to be 100% testing done at least twice in the prodduction process. Once at chip level and then again at the completed moddule level.
I have to admit, Cabe, I don't know if I would ever have confidence in the arrival of a single-controller design. Even if we could do it with today's electronic complexity, what would we do in the future? Remember, active safety systems are on their way. Adaptive cruise is here; so is lanekeeping. To a small degree, collision avoidance is already here, too. At some point, most vehicles will begin to incorporate all those features. A few years ago, autonomous vehicles had their trunks packed with multiple computers. Rear seats were packed, too. Designers of those vhicles referred to them as "supercomputers on wheels." So my point is, even if we could get down to a single-module design today, the automotive community would just start incorporating all the active safety systems, and we'd be doing the same process all over again.
I am starting to lean toward William K's thoughts, stop adding features. The statement of 1 year olds distracted is what got me. I see them already, texting and driving. So, why add more to the fray. But that is new HMIs.
As for a single chip to drive it all... I am still think eggs in one basket applies here.
Informative article, Chuck. This does indeed sound like a complicated problem. It seems like that the auto industry's microcontroller evolution has been similar to many other types of technology (I'm thinking of IT networks in particular): attack the problem with as many as you can first, then realize you've created quite a big ball of twine and need to scale back and take a bit of a less-is-more approach. I'm sure they will eventually figure it out and at some point there will be one super-intelligent microcontroller for everything. Then people will fret about that type of design, worrying what might happen if that fails...and the process will start all over again.
I like the ball of twine analogy, Liz. This ball of twine has been growing since the late '80s, when I first wrote about electronically-controlled transmissions. In 2007, awareness of the problem grew, but the the numbers are still creeping upward. The only difference today is the complexity is not growing as fast as it was 15 years ago.
Yes, it's interesting that a lot of industries follow the same rule in this regard, especially in the beginning of introducing new technologies, when complexity grows quite quickly and innovation can't really keep up with it. And as you mentioned, it's when complexity slows down that people can take a step back and see what's been created and try to come up with a better way to do than simply building on top of what already is there.
The automotive industry needs to develop a standard similar to MIL-STD-1553 or ARINC 429. A common bus to handle all the data inputs. You could then have a common display(s) for all the systems to readout. You could see GPS information on the backseat DVD screens. The technology is already in-place just needs to be implement into an automotive frame.
One more point about the constant addition of features that serve mostly as profit sources for those who sell them. The wiring will kieep getting more complex, either because all of the wires now run to one module, or because those wires are distributed between about 250 individual modules. Either choice will reduce reliability and make servicing a vehicle a financial disaster. One alternative is the use of some kind of bus structure, which it has been shown that a bus won't be accepted in cars because the cost in both dollars and failures is far greater than the benefits.
So the sensible solution is to STOP ADDING FEATURES! We simply don't need them, and the very small minority of drivers who would benefit from those features should simply not be accomodated by the OEMs. Few enough folks need those features so that they could easily be added by the aftermarket folks. THAT would offer a benefit to a lot of value-added companies, and not force the rest of us to purchase expensive systems that are primarily designed to compensate for the errors made by 17-year-olds driving. The rest of us can think well enough to not need those features.
Conventional wisdom holds that MIT, Cal Tech, and Stanford are three of the country’s best undergraduate engineering schools. Unfortunately, when conventional wisdom visits the topic of best engineering schools, it too often leaves out some of the most distinguished programs that don’t happen to offer PhD-level degrees.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.