I just talked to the service manager of my city's Chevy dealership. After telling him of my concern about whether his mechanics can replace microprocessors, he said they've been doing this for years. In mentioning Y2K, and what Ed says in his book, the man said, "There are no microprocessors in automobiles that recognize dates." So, I said there was no problem then. He said that was right, and that the author of the book was "blowing sunshine out of his ear." Anyone out there "in-the-know" about this, or who has talked with their mechanics?

-- Holly Allen (Holly3325@juno.com), June 11, 1998

Answers

I am a technician with about 20 years experience working in diagnostics on automobiles. At one time I was certified By GM as a master technician, after extensive training by GM at their center in Morrestown, NJ.

The man you spoke to is mistaken. His techs have never replaced a microprocessor except when they changed an ECM or a PCM, or perhaps a body control module or an AC control. They changed CONTROL MODULES. It would be a rare tech that had ever even opened a control module to look at the board.

A modern automobile has upwards of 100 processors in it. Are any date sensative? I don't know. On newer high end luxury cars I am sure there are.

Since 1996 we have been getting into OBDII on board systems that are MUCH more sophisticated than they used to be. These systems run active tests on the car AS YOU DRIVE. Yes, they can track time quite closely. They measure the velocity of the crankshaft from one cylinder firing to the next.

I am NOT a processor expert, nor am I up on how these modules are internally programed. That info is not released to us peons who actually work on these beasts.

Can a processor track time without a Real Time clock on the board? I don't think so. Is the RTC date sensative? I hear that they are.

Will your car run on 1-1-2000? I don't know. Chances are I have more experience actually working on cars than the guy you asked in the dealership, and I DON'T KNOW. I have authored articles on diagnostic techniques and I DON'T KNOW.

The more I learn about 'registers' and 'counters' the more I wonder if they will.

Personally, all our vehicle have old style V-8's that are fuel injected, but can be switched over to carborators and old style distributors in about a day. I plan on having the parts on hand to do just that long before 2000.

I am cetain that some of my test equipment will not survive the rollover. The manufacturers won't talk to me about Y2K.

-- art welling (artw@lancnews.infi.net), June 11, 1998.

I read an article on "press clippings" not too long ago that said a car from one of the Big Three was tested for Y2K problems. They set the clock ahead on the dash and when it flipped to 2000 the car would not start. They would not name the vehicle. I'll look for the article and post it here if I can find it.

-- Annie (anniegaff@mailexcite.com), June 12, 1998.

The "right" way to do a timer in a safety-critical system is to have a counter that is incremented every (second, millisecond, whatever). It resets to zero every time the machine is stopped. It either can't wrap around because it would take longer to reach the wrap-around count than the machine can run for, or it wraps around every few minutes and the code is written to handle this.

This has an advantage in that calculating a time-difference is easy, you just subtract (and correct for wrap-around if you get negative when positive is the norm). Using a realtime clock for the same purpose is a pain; you have to do long chains of arithmetic on seconds, minutes, hours, days, etc. (which of course slows down your chip).

This isn't to say that there aren't some things that did it this way for economic reasons, and will exhibit a Y2K bug. It just seems implausible that any such systems will be critical to the continuing operation of the vehicle, and widely used, and that there is an effective conspiracy to prevent any of the engineers who know this from letting us know. My guess is that there will be very few automotive Y2K failures that actually disable the vehicle.

-- Nigel Arnot (nra@maxwell.ph.kcl.ac.uk), June 12, 1998.

The key issue here is the difference between tracking time and tracking a date. Don't confuse the two.

Tracking time might be "Gee, it's been 100 milliseconds since I last checked the throttle setting. Is it still correct?" This happens in modern cars all the time. However, this process is not very Y2K-error prone.

Tracking a date would be something like "If today's date is more than six month's since the date of my last service, I'm going to flash a dashboard light at the driver, and after eight months I'm not going to start." While this is the situation that Y2K would cause problems in, it happens in modern cars seldom if ever.

-- Paul Neuhardt (neuhardt@compuserve.com), June 12, 1998.

When a person replaces the battery in their car where and how would they enter the date and year??? If you can't do this how would the year 2000 be a problem?

-- Patrick Marcinko (pmmarcinko@juno.com), June 12, 1998.

Another thought to add. Modern cars (96 and up) HAVE the capability to shut down fuel to one or more cylinders ON PURPOSE. If it 'see's' a missfire it just may do that to protect the 'environment'. SO.......A control module certainly can and will shut a car down if the parameters are correct.

An offer....I will procure a modern automotive control module and ship it to an expert on embedded systems to examine, if someone here is an expert and would like to look one over. Any takers?

E-mail if interested

-- art welling (artw@lancnews.infi.net), June 12, 1998.

Paul,

-- Art Scott (Art.Scott@marist.edu), June 15, 1998.

Paul, You said: "Tracking time might be "Gee, it's been 100 milliseconds since I last checked the throttle setting. Is it still correct?" This happens in modern cars all the time. However, this process is not very Y2K-error prone."

How do you go about measuring 100 ms? If you use an inexpensive chip because it's cheaper than designing one to measure 100 ms, a and that chip includes year, then you may very well have a y2k problem.

I've posted my latest draft of my paper on embedded systems at: http://www.academic.marist.edu/arts/Year2000/embedded.htm I'd appreciate your (and others' comments).

-- Art Scott (Art.Scott@marist.edu), June 15, 1998.

Art,

As I've posted before, having a chip that tracks years only matters if both of the following apply:

a) You bother reading the year from it, and

b) You provide some mechanism for storing the data when the car is not operating.

Let's use the hypothetical throttle control chip I mentioned in my last post as an example. Let's assume that the chip in use does in fact keep track of years, but only in two digit format so that if a Y2K problem exists we are most likely to experience it. Now, ask yourself the following question: "Does a throttle controller care how many milliseconds it's been since it checked the throttle when the car is not running?" Answer: Of course not. Furthermore, for the purposes of throttle control, do we care what the date is? Probably not, as I doubt that anyone is bulding cars that are designed to have different fuel mixtures on 1/1/1998 and 1/1/1999. (I can envision controls sensative to seasons, but I haven't heard of any.) Lastly, do we need to know how long the car has been parked and turned off? Again, the answer is no. We only need to track elapsed time while the car is running.

So, we don't need to control the throttle when the car is not running. Does it then make sense to go to the extra trouble and expense of keeping power to the chip and it's associated clock (which might or might not be part of the chip itself) in order to track information we have no intention of using? Of course not. So, we can get by letting the chip poser up at it's default settings, including letting the clock start out an elapsed time of 0.

So, when would our Y2K problem crop up? Well, only when the car had been in continious operation long enough to cause the rollover of the year component of the clock. Let's be pessimistic. Assume that the default setting on the clock is a year of 80 (as in 1980). In order to cause a rollover problem (i.e. a Y2K problem), the car would have to be in continious operation, motor chugging and electrical system active, without any interruption whatsoever for 20 years. I think it is safe to say that any car meeting that criteria would be the exception rather than the rule.

This is not to say that there aren't plenty of examples out there in the world where embedded contollers meet both of the criteria listed at the beginning of this post. I just feel that passenger automobilies are an unlikely place to find them.

-- Paul Neuhardt (neuhardt@ultranet.com), June 16, 1998.

Paul, Thanks for you input. My wife and I just returned from a trip to Portugal. Good trip, nice people, good food, wine, good roads, etc. While we were there we rented an Opel, (Vectra, I think), probably a 1998 model. It had a nice electronic display showing such things as: date with 4 digit year, time, instantaneous fuel consumption; and on another page, for the trip such things as: average speed, average fuel consumption, total distance, etc. I should have played with it more but my eyes were glued to the road trying to stay alive. :) The date/time was always correct no matter how long I left the ignition off. I'm guessing but I believe there must be a battery keeping some computer function powered up. Clearly, it keeps track of date. What we don't know is how they measure time: is it new time minus old time (including date?)? or do they only read out part of the date/time counter? or do they add a bunch of little timers that reset each time the car is started? Does anyone on this list know?

I'll surf the I'net and see if I can find out what Opel says about y2k.

"On the other hand, Saturn on their web site says: The year 2000 will pose no difficulties for the computer or Powertrain Control Module (PCM) in your Saturn. The need to adapt to the turn of the century was taken into account by Saturn engineers."

-- Art Scott (Art.Scott@marist.edu), June 18, 1998.

Art,

Remember that there are several, pehaps even dozens of microprocessors in modern cars. The fact that the clock display kept track of dates doesn't mean that the system(s) running the engine do, anymore than your home computer tracking a date would affect your microwave oven.

However, keep investigating and keep asking. That's how we learn. I sure would like more details on whay Saturn is saying they took Y2K "into account." Where did you get that info from?

Paul

-- Paul Neuhardt (neuhardt@ultranet.com), June 18, 1998.

I'm bringing this up again hoping someone has more info. Assuming we eventually are able to pump gas into our vehicles again, will my '88 Chevy Corsica be driveable? Or will failed microprocessors "kill" it? The paint is badly chipped, and I'd love to have it repainted. But, if it's going to have to be hauled off to the junk yard in a year and a half because it cannot be repaired???? I'm wondering if I should have kept my old '72 Ford LTD! When we get past this, if microprocessors are replaceable, mechanics are going to be the nuevo riche!

-- Holly Allen (Holly3325@juno.com), September 09, 1998.

Holly: The only car I've heard of to date that has a *confirmed* y2k problem is a single model of a Ferrari from, I think, the 1980s. There's a thread about it on the comp.software.2000 Usenet forum. Ask for Mickey. It's his car.

-- J.D. Clark (yankeejdc@aol.com), September 09, 1998.

The article from the Orlando Sentinel which I posted in another thread states that the "Big 3" have said there are no problems with their cars. Of course, it is only a newspaper article.

Link: http://www.orlandosentinel.com/news/y2k/y2k0829.htm

-- Buddy Y. (buddy@bellatlantic.net), September 10, 1998.