To: Mama Bear who wrote (18527 ) 6/17/1998 12:14:00 PM From: Judge Read Replies (1) | Respond to of 31646
>>>Can you fill me in on why the embedded chips on a factory floor are date dependent? I've heard ridiculous assertions vis a vis the Y2K problem. In line at a food store I heard the checker warn the customer in front of me not to ride elevators on 1/1/00. Why would an elevator care on what day it was going up and down? Likewise, why would a robot on a factory floor care what day it was canning Coca Cola? <<< What confuses many people about this -- because it IS counter-intuitive -- is that although a robot or other machine on a factory floor may not "care what date it is", it may nonetheless perform its functions at time intervals ranging from milliseconds to minutes. In each case, depending on how the embedded system controlling the function is designed and programmed, the interval may be calculated as a fraction of a date, ie, YYMMDD/hour/minute/second..... When the date rollover to "00" occurs, the embedded system is unable to calculate the next interval as a fraction of a date because it cannot determine whether "00" means 1900 or 2000. Not all embedded systems are configured this way, but it's almost impossible to determine which are without testing. In addition, "manual overrides" are often an illusory solution because a single factory floor machine may contain dozens of individual embedded systems performing coordinated but independent functions, the failure of any one of which may cause the machine to fail. I recommend that you visit the Automotive Industry Action Group's website, aiag.org and click through some of the articles and letters they have distributed to suppliers of Ford, GM and Chrysler. They make quite clear that problems with such embedded systems are pervasive, difficult to find and fix, and potentially catastrophic to factories. Why are revenues from factory-floor remediation of embedded systems still so low? If the industrial companies I talk to are at all typical, it's because most of them are only just beginning to figure out that need they need to do more than write to the vendors of the machines on their factory floors to find out whether the embedded systems they contain are Y2K compliant. In the vast majority of cases, the machine-manufacturers' initial reaction is "the machine I sold you doesn't care what date it is, it only needs to calculate milliseconds, etc." Then the education process lurches ahead (and sometimes backwards). Only as plant engineers talk to counterparts who are testing similar systems and finding problems do they begin to accept that maybe their machines do "care what the date is" after all. None of this is terribly shocking if you bear in mind that as recently as 18 months ago, many responsible and well-respected I.T. people were smugly assuring their business people that Y2K wasn't an issue for recently installed PCs and applications. Boy, were they wrong! What many people fail to understand is that in many organizations there is fairly little leverage by plant engineers onto the experience of the I.T. department. They inhabit different world and territorial boundaries are jealously guarded. For the most part, they don't even read the same technical publications. The plant today is where MIS was 2 years ago....