This microcontroller illustrates a couple of important points:

1. Not all microcontrollers are created equal. Even though this chip is called a microcontroller it does in fact contain an 80586 Pentium class microprocessor core.
2. Not all chips are created equal. Even though this may look like a single chip it is in fact a system on a chip. Not only does it contain the AMD x586 core, it also includes these features:

Programmable Interrupt Controller

Enhanced DMA controller

Dual 16550 compatible UARTS

Programmable Interval Timer (PIT)

And... Real-time Clock (RTC)

As you can see here.

Regards,

-- Anonymous, October 29, 1999

Answers

And.....???

-- Anonymous, October 30, 1999

How did that old poem go, Flies have fleas upon 'em and so ad infinatum" or something like that.

-- Anonymous, October 30, 1999

Walt,

And... it's just another data point in the ongoing embedded controls arena. Check out the embedded controls section of this forum. Also, you may want to do a search on "ajedgar" to find some more threads to do with embedded PCs and the Y2K problems.

Very briefly, we know that millions of embedded systems ship every month. We know that some of them will have problems and glitches due to the century date change issue. We know that the problem is not zero, yet we have no satisfactory measure of how bad the problem will be. We know that persons who say anything about Y2K with categorical certainty simply can't see the big picture or just don't, "get it".

The real-time clock (RTC) in that embedded microcontroller is a standard IBM PC/AT compatible (circa 1984) RTC. Which means that it only stores the last two digits of the year -- in Binary Coded Decimal (BCD) -- format. This is a brand new (or at least very recent), highpowered (x586 Pentium Class CPU based), highly integrated (lots of built in extra features, including RTC) microcontroller chip with, as far as I'm concerned, a broken RTC in it. This type of RTC is called "Y2K compatible", because as long as all the software above it "knows" that it only stores the last two digits of the year the system as a whole can be made to look "Y2K ready".

Many people just don't understand how fragile much of our technological infrastructure really is. It's very big, very complex, and surprisingly less robust then you would hope or expect. Do you remember when much of the paging system across the US went out in May 1998? Or the Jun 1998 outage of Pacbell's PCS service? Or the December 1998 PG&E electric utility outage in San Fransisco? All of these were major failures that should not have happened, or at least should not have been of the scale and/or the duration that each was.

In the case of the paging system a major satellite failed and it took several days to bring a backup into position. For Pacbell's PCS service outage a corrupted database was uploaded and it took two days to sort out (what? no backup database? no automatic switch-over? can't run it manually?). And in the case of the electric outage in San Fransisco the fault cascaded much further than it "should" have.

Each one of these failures could have been prevented or at least made transparent to the end user with better network design, better Reliability, Availability, and Servise (RAS) design, and most importantly proper staffing and expertise. All across North American industry, companies have been cutting support staff, cutting corners on important product life cycle phases -- such as design review and product validation, and letting their most experienced people walk out the door. The type of people who would pay for themselves ten or a hundred times over if just once a year or so they were to raise their hand and say, "wait a minute, you can't do that...". All of this is happening directly or indirectly as a result of the ever increasing push for cost-cutting, profits, new services and time-to-market.

The above is no surprise to anybody who has been employed in the engineering field through the eighties and nineties. We have become much too sanguine about it all.

Now take this little thing called Y2K. Y2K is a Common Mode Failure (CMF). A CMF is a fault in an otherwise robust system that causes multiple faults in common and at the same time (hence "common mode"). A simple example is an airplane. A Boeing 747 has four engines, it can lose one or two engines and still stay in the air. These are called "single fault" and "double fault" respectively. However, if the aircraft runs out of gas it's not going to stay in the air too much longer. This is called a Common Mode Failure because running out of gas causes all four engines to shut down.

Come January 1st a lot of potential glitches are going to happen to systems all over the world within 24 hours of each other, and many of them will be trying to communicate their slightly confused view of reality to other slightly confused systems. Also, the problems and glitches are almost certainly going to persist for months following January 1st. Y2K will take on a whole new meaning in the new year, it will be, "Year 2000, the year we spent pruning the outgrowth of our hubris."

With Y2K there is the potential that we are going to experience the world's largest, globally distributed, co-incident, common-mode failure scenario of modern times.

Let us hope we really have fixed enough of it to squeak by.

Time will tell.

Regards,

-- Anonymous, November 01, 1999

AJ,

Excellent article by a Dallas Semiconductor engineer that deals with RTC's is at:

http://www.devel.penton.com/ed/pages/magpages/oct0198/ti/1001t12.htm.

Sorry about the lack of hot-link, I'm html clue-less. This article describes the Y2K/RTC issue much better than Dr. Frautachi (sp?)paper (at least it resonated better with me).

-- Anonymous, November 17, 1999

cl, the link doesn't work, and penton.com seems to be some kind of portal that goes all over the show. Do you have anything more than a URL?

-- Anonymous, November 17, 1999

Sorry,

Try

1. http://www.dallassemiconductor.com/Prod_info/Time_Keep/y2k.html

2. then select "Solving the Year 2000 Real Time Clock Dilemma" (sic) by Jim Lott - Dallas Semiconductor

OR

1. go to http://www.dallassemiconductor.com. 2. Select graphic link in upper right hand corner of page "It's Cominig - January 2000". 3. Select the article by Jim Lott.

Sorry. Hope this helps you find it. It really is interesting in my opinion.

-- Anonymous, November 17, 1999

CL, Yes, that was a pretty good overview, thanks for the pointer.

What the article doesn't mention is the sheer magnitude of the problem. The huge numbers of open, closed, niche, and proprietary operating systems that go into literally hundreds of thousands of new designs each year, where each new design averages ten thousand units per year. Each of these designs has to take into account all the little subtilities mentioned in the article, and be precisely tested.

I've been in this industry too long to think that every design engineer specified it correctly, that every application engineer implemented correctly, that every test engineer tested it correctly, and that every QA engineer validated correctly. As the last paragraph states:

In a nutshell, developing systems expected to function properly before, during, and after the century date change requires thorough testing of the hardware platform and the platform-development tools. Year-2000 failures can occur in the hardware, operating system, compiler, date-conversion routines, and application programs. Everything needs to be checked for correct date handling.

There are going to be problems, all through 1998 and 1999 we have heard about the odd problem here and there. In six weeks and a day we will begin to experience a rash of glitches but we really won't know the true magnitude of the consequences of the effects until well into the 2000 calendar year.

Here's an interesting data point from Mon treal. French-fries and Y2K burned the firehouse down. Lord. ;-(

-- Anonymous, November 18, 1999

P.S. This guy knows what he's talking about but he didn't mention firmware. He should have.

-- Anonymous, November 18, 1999

My favorite cartoon is a guy at a palmreader who says you will be fine through Y2k, until a malfunctioning satellite falls on your house. I guess there are no chips in satellites.....no wait....I guess all the chips in all the satellites made by all the different companies over all the different years, are all Y2k compliant (and we know that without looking). Whew, what a relief.

-- Anonymous, November 18, 1999