Serious question on Motorola's RTC'sgreenspun.com : LUSENET : TimeBomb 2000 (Y2000) : One Thread
I went to the Motorola web site to check on their embedded/RTC I.C's back on 12-98, they had a list of 70 that they openly admitted wouldn't work in 2000. Today I went back to the same site(http://www.mot-sps.com ), and the list has grown to 94. I work for an I.C. manufacturer and when we sell product lines, It's between 5,000-1,000,000 pieces per order. If you take the low number of 5,000 and multiply by the 94 known bad I.C's, thats 470,000 bad chips out there, and thats just for 1 maker(Motorola). Am I missing something, or is this as serious as it sounds??
-- Paul (firstname.lastname@example.org), May 14, 1999
The embedded question is the real wildcard for Y2K. Reasonable people can seriously differ. Consider just one small issue. The PDP-11 was a very popular mini-computer of the late 70's/early 80's. Many engineers "grew up" in college using this as their computing/control source for virtually everthing. The instruction set was ported onto a lot of custom boards/apps around the world. (The Russians stole a lot for their military applications.) Guess what? The operating system for the "old" PDP-11 stuff is totally non-y2k compliant. (It WILL fail.) Now, how many of these "control" boards/old PDP-11's are out there? My guess is that there are several thousand around the world. (The New Hampshire nuke plant uses one for safety applications! Sysman has one in his shop as well for some singular application.) Could a large refinery or chemical plant make a big hole in the ground as a result of this one problem? Yeah, its at least possible. If TEOTWAWKI occurs, the embedded/controller failures will be the reason.
-- RD. ->H (email@example.com), May 14, 1999.
Many of Motorola's RTC chips only support 2-digit years. To simplify a bunch, this means the year field will roll over from 99 to 00. The part itself will continue to work correctly, and this 'limitation' is well documented.
IF the RTC is being used to maintain a complete date (probably true in a minority of applications), it is the responsibility of the software reading the date to window it properly to handle the century change. This is pretty trivial to do (although reading the RTC itself properly is very tricky). If the RTC is only being used as an interval timer, then this year rollover limitation is irrelevant.
These are NOT 'bad chips' and nobody 'openly admitted' that they wouldn't work. They have a known, documented limitation, easy to work around (within a 100-year period, anyway).
I'll agree that if your description had been honest, your worries would be justified. Fortunately for all of us, it wasn't honest.
-- Flint (firstname.lastname@example.org), May 14, 1999.
to amplify on what flint just said, suppose you have a system that is built around 2 digit years. The date formats are all fixed in hardware, your client base isn't goin to retool the entire plant just to upgrade, you buy the old format chips even though they aren't "compliant".
such is life in the real world. Besides, people still need compatible replacement parts/modules until they DO swap out the old plant completely. for better or worse, the market for these style chips isn't gone. If y2k turns out to be less than teotwawki then MOTO will be selling them for years to come.
-- hunchback (email@example.com), May 14, 1999.
Flint, Would that it were so!
We know of several chip lines that not only WILL NOT handle the 99 to 00 rollover but will PERMANENTLY lock up! One design in particular was used in HVAC systems for about 15 years. It was redesigned in 1989, just as the long building boom was winding down. Heard about it at a conference here in Phoenix during a Q and A. Those chips are everywhere.
How could this happen? Competitive pressure to keep costs down precluded the cost of the logic to handle negative numbers.
-- K. Stevens (kstevens@come_January_the_infrastructure_will_fail_.com), May 15, 1999.
Can you tell me the make and model of such a chip? This certainly isn't true of any Motorola chip I've been using since about 1980, and I hadn't heard of any such chip before. Indeed, none of the embedded people writing on any y2k forum has ever mentioned such a failure mode for any known chip. And creating a chip that will lock up rather than just start over from 00 would seem to be more expensive!
I think someone is pulling your leg. Chip designers don't do things like that on purpose, and it isn't easy by accident.
-- Flint (firstname.lastname@example.org), May 15, 1999.