I exchanged several emails with Toyota on the Y2K problem. Their response is that it is not possible for there to be date sensitive chips in any vehicle. Their rationale is that a date sensitive chip needs constand power supply or else it has to be "reset" after a power outage. Here is a direct quote: "THIS PRINCIPLE IS DEMONSTRATED WHENEVER A VEHICLE'S BATTERY IS DISCONNECTED OR CHANGED. WHEN THE VEHICLE IS RECONNECTED TO A BATTERY, THE DRIVER DOES NOT NEED TO HAVE THE ON-BOARD COMPUTERS REPROGRAMMED OR RESET." They go on to express regrets that this claim or explanation is not "in writing" for dealers or the general public. It is also absent from their web page's FAQs. Any comments? The explanation seems to be sensible and, if true, would mean ALL cars must have no date sensitive chips. Thanks, Charlie

-- Charles Michael (charliecmt@hotmail.com), April 24, 1998

Answers

The explanation is bogus: it's trivially easy to store information in a CMOS memory powered off a rechargeable battery or capacitor while the car's main battery is disconnected. It's also possible that the car contains an IC with an integrated realtime clock and battery, and a silicon design bug such that it'll Y2K fail EVEN IF THE REALTIME CLOCK IS NOT USED by the system! Such chips are known to exist.

Whether anything important in a car IS so designed is quite another matter. Neglecting trivia like clocks and trip recorders, the only place I can concieve the real time being of sensible use is for managing the "service needed" lamp. Does this system disable the car if it thinks it has not been serviced for a century? Seems most unlikely to me, though not impossible. My guess is that most cars will continue to function.

-- Nigel Arnot (nra@maxwell.ph.kcl.ac.uk), April 24, 1998.

Although I'm not an automotive engineer, I would think that any embedded system tracking service would not be date oriented but rather length-of-service oriented. It would track miles driven and/or days since last service. Date would be, I suspect, largely irrelevant to that proceess.

-- Paul Neuhardt (neuhardt@compuserve.com), April 24, 1998.

In order to track the length of time in service it needs to be able to count time - days / hours and uses the date/time function built into most chips. One of the problems is that this function is built into most chips and may or may not be accessed by the software/firmware. Whether it is accessed or not, it can still cause problems. Another problem is no one knows what the default start day is for most of these chips. It might be the right (current) day or some arbitrary day, which means no one knows when it will really roll over.

-- Rebecca Kutcher (kutcher@pionet.net), April 24, 1998.

There is no absolute requirement that there be a start date involved at all. For an example of this, check out the rollover problem associated with the Global Positioning System. The satellites simply count number of weeks since service started and are about to overflow the storage allocated for that information. The fact that the system went into service on a particular date in 1980 (I think) isn't relevant. What matters is that in September of 1999 it will have been 1025 weeks since operation began, and that will be transitted by the satellites as 0 weeks.

If this is the case with cars, and I see no reason why it shouldn't be, then any rollover problems would occur simply by operating the car for a certain period of time, and dates would mean nothing. If such systems have been in place for more than a couple of years, we should have started seeing problems already. If such systems haven't been in place long enough for this to have been an issue yet, there is still nothing magical about the year 2000 that would cause problems to occur.

-- Paul Neuhardt (neuhardt@compuserve.com), April 24, 1998.

Just a note....The service timer on a car is lit by using an implied milage timer in most cases. Thats for the emmision system service light. No dates involoved.

On the higher line cars that have a timer for oil changes and such it's driven off the odometer.

All this said, I would not be the least surprised to find they used a generic board/chip that had date capabilities. Car manufacturers are $$ driven and will do whatever it takes to sell cars with the highest profit possible. If a generic chip can be bought for .05 cents less than a specially built one, guess which gets used?

-- art welling (artw@lancnews.infi.net), April 26, 1998.

My concern is with fuel injected cars. At first glance, the fuel injection system of your car might not appear to be date sensitive. However, the injectors are controlled by a chip which regulates the fuel/air mixture ratio, and times its delivery into the cylinders with microsecond accuracy. The concern here is the timing issue. Isn't this essentially a time/date sensitive embedded system that could potentially fail? Will some owners of fuel injected vehicles be unpleasantly surprised on 1/1/2000? Any ideas?

-- Jon O'Dette (jodette@erols.com), April 27, 1998.

Thanks for all the responses. Maybe, since I started this string, I should emphasize again the reasoning behind Toyota's response. If the vehicle has date-sensitive chips in it then ANY time you disconnect the battery (during replacement etc.) then such a chip, being without power to operate its internal calendar, would be all out of sinc when it was eventually reconnected. This would be much the same as the effect of Y2K. Since your car and its subsytems do not go haywire at such times then it would seem logical to assume that no such chips are involved in the car's operations. This would also seem to apply to arguments about chips that count days etc. Maybe someone should disconnect their battery for a few days just to check it out? To be sure, there are chips - just not date sensitive ones. That is my read and I have not seen any arguments on this forum that have changed that judgement on my part. Mind you, I am not playing Pollyanna over Y2K nor do I have any partisan feelings for Toyota. I do find it interesting, however, that some of these postings display an almost religious or ideological determination to not let a fact or logical argument get in the way of a conviction/desire? that there MUST be a problem. Are there some out there who shelter a deeply held doomsday fixation? Y2K is surely a dangerous problem that is worsening rapidly, however we all need, I think, to keep some level of objectivity and healthy skepticism in judging events and information. That's my feeling, anyway. Thanks, Charlie Michael

-- Charles Michael (charliecmt@hotmail.com), April 27, 1998.

Perhaps as a qualifier on my contribution/commentary above I should note that many of the actual postings on this forum are thoughtful and well reasoned. I have received many direct emails, however, which have proven to be less so and many of them, contrary to my assumptions, do not seem to have been posted. Oh well! Thanks, Charlie

-- charles Michael (charliecmt@hotmail.com), April 27, 1998.

Toyota sez: "If the vehicle has date-sensitive chips in it then ANY time you disconnect the battery (during replacement etc.) then such a chip, being without power to operate its internal calendar, would be all out of sinc when it was eventually reconnected."

Interesting theory. Too bad it's meaningless.

Show of hands -- how many of you own computers that lose date & time when you pull the power plug out of the wall?

Dallas Semiconductor and others manufacture date/time chips used on motherboards. Many of these chips have internal lithium batteries. You can pull the plug as may times as you like, but the clock keeps running.

So, the answer remains unknown. Do *any* vehicles depend on date?

-- Ron Schwarz (rs@clubvb.nospam.com), April 27, 1998.

Art,

Good point about the cost of the chip determining which one gets used. However, I would question whether or not the existence of date functionality in the chip would cause a problem if that functionality were never used. In other words, who cares what year the chip thinks it is if you never ask the chip for a date?

-- Paul Neuhardt (neuhardt@compuserve.com), April 27, 1998.

I was told that on some of the fancier models of cars (Lincolns, Cadilacs, tec.), when you take them into the shop for repairs, they plug the car into a special computer and the screen outputs a list of diagnositc information, including a list of events that occured and were detected by the computer. I was told on some cars these events have a time/date stamp on them. Has any one else seen/heard about this?

-- J. Morgan (Jmorgan@dlcnorth.com), April 28, 1998.

Paul,

You said:

"Good point about the cost of the chip determining which one gets used. However, I would question whether or not the existence of date functionality in the chip would cause a problem if that functionality were never used. In other words, who cares what year the chip thinks it is if you never ask the chip for a date? "

It is my understanding that all Intel architecture chips that run the Microsoft DOS style operating systems maintain their real time clocks in the format yymmddhhmmss.ffffff. If this is so (and so far no one has said that it is not) then 0001010000000000001 is less than 991231115959999999. This is relevant if the chips are used to control the staring and stopping of any process because they will likely start a process and not stop it or stop a process and not start it. Either of which could cause significant trouble.

I wish someone would tell me that I am wrong and point me to documents that would convince me, but as of now, it looks like we are in big trouble whether the chips care about what date it is or not.

George

-- George Valentine (GeorgeValentine@usa.net), April 28, 1998.

George,

You may be right about the Intex 80X class of chips, and if you are then there could very well be a problem. My reference materials on such things are buried deep in the bowels of my basement.

However, who is to say that those are the chips used in the cars? A lot of other people make computer chips in this world. Come to think of, Intel makes an awfully large number of chips that aren't 80X class.

Let's go one step further. Assume that Buicks are loaded with 80386 chips. The chip itself only keeps the date in register so long as it has power. No power, no date. Now, you could do what PC manufacturers do, which is make sure that your car's clock chips are always fed a constant supply of power through the use of NiCad batteries that keep a charge going even when normal power is disconnected. However, if you don't care about preserving a date, there are several reasons not to provide batteries and just let the clock reset to default everytime you power up. I'll mention the first three that come to mind.

1. It's expensive to add parts you don't need. They may only be a few dollars per car, but when you multiply that over hundereds of thousands of vehicles, it adds up into some real money.

2. Batteries drain over time and have to be replaced. This would be one more point of failure added to the system with no return in functionality. Bad for your customer service ratings.

3. It's just bad practice to add parts to your machine that aren't used.

-- Paul Neuhardt (neuhardt@compuserve.com), April 28, 1998.

I agree that Toyotas response is technically unfounded. Even the fact that you *don't* have to reset the digital clock on your dashboard or radio disproves their argument 'on its face'.

If their are date vulnerable aspects of the 'service indicator' I think it unlikely that it would prevent the engine systems from functioning.

The engine control system itself (i.e. spark, fuel injection, etc) can be extremely time-critical in terms of milli- and micro- seconds, but dont give a flip about "which second in the course of human civilization" it happens to be.

This is a case in point, that engineers will generally design their systems to depend upon physics and not upon politics. Seconds, Minutes, Hours, and Days have engineering meaning as durations of time. With suitable definitions even Month and Year have engineering meaning.

The calendar, however, is essentially a political construction which has no engineering meaning, and for that reason engineers will avoid having to 'mess with it' as much as possible. And even when they do have to 'mess with it', they will avoid *depending* on it.

So this I guess is a reminder that we/people use the words for units of time in primarily two distinct ways: 1) as units of duration--what I refer to above as having 'useful engineering meaning', and 2)as Names for specific instances along the universal time continuum--The Calendar, which rarely has any 'useful engineering meaning'.

My 2 cents...

John

-- John Clonts (jclont@mastnet.net), April 30, 1998.

John,

I agree that engineers will stay far away from dates if they don't need them. However, they could be using off-the-shelf operating systems that implicitly care about dates. It seems to me that most operating systems designers would feel that keeping a clock in microseconds since the big bang would not be as useful as automatically and behind the scenes converting the thing to a date/time format of some sort, particularly if they expected the information to ever be displayed for human viewing. They may not work that way, but they may, and until someone convinces me that they don't work that way I must assume that they do. If they do, then every process control system that issues commands like start a process, wait for a fixed period, stop the process or the reverse may experience catastrophic failure at or across the 19991231235959 to 20000101000000 boundary.

-- George Valentine (georgevalentine@usa.net), April 30, 1998.

George,

The qestion you have to ask yourself here is "How many off-the-shelf operating systems are there for fuel injectors?" I guarantee that your Buick (or Toyota, or whatever) isn't using Windows NT to run the engine.

-- Paul Neuhardt (neuhardt@compuserve.com), May 01, 1998.

The following excerpt was written by a Todd Day who is a list moderator, and ECU (Engine Control Unit, aka car computer) wizard.

"All you guys with 1989-1994 DSMs might want to take another car to the parties on Y2K eve... or arrange for a tow truck. 1995-98 ECUs seem to be Y2K compliant, as far as I can tell (an OBDII requirement). If you don't want to be seen at the party without your DSM, for $100, I can fix your ECU so the overflow problem won't happen until 2089. I've set my ECU to the bewitching hour, and the results aren't pretty. The overflow causes a mask bit to be set which prevents the spark plug in cylinder 3 from firing, but doesn't stop the fuel flow. The fuel flow in that cylinder actually doubles due to a side-effect of the spark bug. This creates some pretty spectacular backfires, I must say... One hell of a way to welcome in the new century.

"Oh yeah - this bug happens at December 31st, 1999 at midnight (or January 1st, 2000, depending on how you look at it) in the CENTRAL time zone. That's because the cars were all built in Illinois. So for you East Coasters, it will occur at 1am, and West Coasters will experience it at 10pm."

-- Markus Fromherz (Markus_Fromherz@yahoo.com), May 20, 1998.

I learned that a Dodge Caravan has an engine control system, which has no calendar inside, and a body control system (for radio, etc.), which has a calendar. The dealer is waiting for information back from Chrysler whether the body control system is Y2K-compliant, and what, if any, issues might arise.

-- Markus Fromherz (Markus_Fromherz@yahoo.com), May 20, 1998.

Markus,

Can you give more pointers to whatever list it is that you got this posting from? I would be very interested to see if I could find out why the ECU's are year sensitive, as I have yet to be able to find anything suggesting that they were. I would be interested in discovering why they were made so, especially if it is some sort of regulatory requirement.

-- Paul Neuhardt (neuhardt@compuserve.com), May 20, 1998.