PLC processors break in 2017?

Hi there,

I was just speaking to my boss and he's asked me to look into a potential problem in some of our older equipment.

He seems to remember that, many moons ago, when the older equipment was installed (late 80's early 90's) someone told him that there would be an issue with the processors in 2017.

There are S5, S7 and Mitsi processors in different machines and he's not sure which PLCs they were talking about but he's asked me to look into this (obviously back then it seemed like a million years away so he didn't really bother about it).

Has anyone heard of anything like this? I've Googled to no avail?

I'm guessing it could potentially be a bit like the (so called) millennium bug? Maybe where the 'clock' is running out or rolling over or whatever?

Anyone else heard of anything like this with any of the older processors?

Cheers

I've never heard of that for a PLC, but I do remember that UNIX is supposedly going to have a big problem in 2038:

https://en.wikipedia.org/wiki/Year_2038_problem

There might be a rollover of the free running clock, counting milliseconds from some start date in say 1980, that when it rolls over the PLC may think it's 1980 again, but again I haven't heard of it.

Hmm, I haven't heard of any issue with 2017. But there is one for 2038. Having to do with time and 32 bit hitting it's limit.

https://en.wikipedia.org/wiki/Year_2038_problem

But it's been awhile since I've played with that generation of PLCs.

EDIT - Beaten to the punch while replying.

A 32 bit clock will be hitting *its limit.

ja dass konte sein weil 2017 ist die helfte von 2038 (na ja von 2000) also da konnte die hochste bit schon umfallen und dan gibts negative zahlen,also dass soll unsigned sein, also
ja dass konnte ein problem sein, nicht im RTC aber im programm.

By mentioning 2017, I'd guess this is related to RoHS regulations that will take effect that year. Basically it means that no manufactures will be allowed to sell any electronics that do not meet current RoHS rules after (I think) July of 2017. That will apply even to replacement parts, which are still exempt so far. So for older electronics that were not designed to those standards, it's not going to be economically feasible for mfrs to redesign older devices to conform, and they will cease supporting them.

In the case of S5 Siemens processors, I believe that is already the case. Most likely S7s will continue, but you may be forced to upgrade to a newer compatible processor.

http://www.google.com/search?q=rohs+2017+deadline&ie=UTF-8&oe=UTF-8&hl=en&client=safari

RoTaTech said:

A 32 bit clock will be hitting *its limit.

Click to expand...

While we are correcting one another...

"...*32-bit clock will be hitting its limit..."

"...The processor uses a 32-bit architecture..."

"...The 32-bit integer contains 32 bits..."

"...The 32-bit image has 32 bits per pixel..."

Any references to being of a "32-bit" nature are hyphenated in the singular.

Any references to using "32 bits" are not hyphenated and are used in the plural.

It should not be used unhyphenated in the singular as "32 bit".

This goes for all "x-bit" references...8-bit, 16-bit, 64-bit, 128-bit, etc.

shooter said:

ja dass konte sein weil 2017 ist die helfte von 2038 (na ja von 2000)...

Click to expand...

Translation...

shooter said:

yes this could be because 2017 is half of 2038 (well 2000)...

Click to expand...

shooter,

What are you getting at here? 2017 is not half of 2038 (1019) and 17 is not half of 38 (19)?

To my knowledge, I don't think there is any issue with the year 2017 for PLC clocks?

There is certainly no issue with the year 2017 in relation to the year 2038 bug.

Interesting to note - Rockwell used to have the processor's Wall Clock, or real time clock range set to between January 1, 1972 and January 17, 2038. This is two days shy of the expected January 19, 2038 bug. So it appeared that someone at Rockwell might have thought that the 2038 bug would be an issue for their processors in the future?

1972 was the introduction of UTC as we know it today. That is why they used it as their base date.

But...

"...Recent developments with the Common Industrial Protocol (CIP) specification have resulted in the selection of a different base date of January 1, 1970 by the Open DeviceNet Vendors Association (ODVA)..."

So the ODVA specified a new base date of January 1, 1970. This is known as the "epoch" which means reference point.

Rockwell, at v16, changed their base date to January 1, 1970 12:00:00 A.M. UTC to conform to the new CIP standard.

The 2038 bug is set to hit at 03:14:07 UTC on 19 January 2038 as it increments to 03:14:08 seconds. This is derived from the fact that 03:14:08 UTC on 19 January 2038 is exactly 2,147,483,647 seconds after the January 1, 1970 epoch. 2,147,483,647 is the maximum positive number that a 32-bit signed integer can hold. For systems with 32-bit signed integers, used for time values interpreted as the number of seconds since the epoch, they will rollover to a negative number and so on...

However, at v16 Rockwell also changed the higher end of the range...

The 1/1/1970 12:00:00 A.M. UTC to 12/31/3000 11:59:59 P.M. UTC

Hmm, the year 3000? So what happened to the thinking that the year 2038 bug was going to hit their processors?

"Someone" must have realized that the Logix processors use a 64-bit signed integer to count microsecond ticks since the epoch, and that the 32-bit signed integer problem is not of their concern because...

...a 64-bit signed integer can hold a positive number up to...

9,223,372,036,854,775,807

Which in funny terms is...

nine quintillion, two hundred twenty-three quadrillion, three hundred seventy-two trillion, thirty-six billion, eight hundred fifty-four million, seven hundred seventy-five thousand, eight hundred seven

This means that systems using 64-bit signed integers to hold epoch, or Unix time, can stay ticking along for up to 292 billion years into the future.

I'm sure that will see "most" of us out.

So Rockwell set a maximum year (3000) into the future that they reckoned far enough to never see a present Logix processor still beating.

985 years away. Who knows?

For Siemens, Mitsubishi, or any other controllers you like; whether the 2038 bug it will affect them or not depends on whether they implement 32-bit signed integers to reference epoch time.

Now, you'll have to do that particular research for yourself.

Regards,
George

jraef,

A recent search might have thrown up the year 2017 for the RoHS Hazardous Substance Directive and Industrial monitoring and control instruments, but I would doubt that what uptown47's boss "heard" in the '80s / '90s had anything to do with a Directive that only began in 2003?

But I can see how a confused link between the two may have been made here.

G.

Maybe the original machine builder had just plucked a date of the air. .. "you will need a new machine by this date or disaster happens". How else do you sell new machines if you build them to be reliable?

If the issue is really due to date, then provided that your machines don't actually need to use real time clocks for anything, just setting the date back a couple decades should sort you.

The fact that you're still using S5s is of some concern though given that support for these ended over a decade ago.

Saffa said:

Maybe the original machine builder had just plucked a date of the air. .. "you will need a new machine by this date or disaster happens". How else do you sell new machines if you build them to be reliable?...

Click to expand...

Saffa,

While I can understand the financial benefits that such underhanded tactics might produce for an unscrupulous OEM, I would imagine they'd be expecting a return on their endeavours in at least the same decade?

If an OEM had supplied systems to a customer, such as uptown47's boss, around the 1990 mark, while leaving them with the expectation of it somehow catastrophically failing in 2017, then that would be some 27 years later. That's a pretty good service life, and that's a long time to be rubbing your hands gleefully.

Underhanded? Yes. Unscrupulous? Yes. Bright? No.

(I know you were joking, so am I)

Saffa said:

...The fact that you're still using S5s is of some concern though given that support for these ended over a decade ago.

Click to expand...

We're still using S5s and yes, I am concerned. But we're down to just 2 after migrating them out over the last few years. I just used the PG 720P again on Friday and the screen is acting up. It's going to fail anytime soon. We have a laptop ready to be delivered with Step 5 migrated to it and licensed, just in case.

G.

uptown47 said:

Hi there,

I was just speaking to my boss and he's asked me to look into a potential problem in some of our older equipment.

He seems to remember that, many moons ago, when the older equipment was installed (late 80's early 90's) someone told him that there would be an issue with the processors in 2017.

There are S5, S7 and Mitsi processors in different machines and he's not sure which PLCs they were talking about but he's asked me to look into this (obviously back then it seemed like a million years away so he didn't really bother about it).

Has anyone heard of anything like this? I've Googled to no avail?

I'm guessing it could potentially be a bit like the (so called) millennium bug? Maybe where the 'clock' is running out or rolling over or whatever?

Anyone else heard of anything like this with any of the older processors?

Cheers

Click to expand...

I don't know about S5 or Mitsi, but in S7 the Date_and_Time data type maxes out at 2089. The other data types I found went even longer.

Let's try another line of thought here. Could it be that your boss is going senile and 2017 is when he plans to retire?

uptown47 said:

Hi there,

I was just speaking to my boss and he's asked me to look into a potential problem in some of our older equipment.

He seems to remember that, many moons ago, when the older equipment was installed (late 80's early 90's) someone told him that there would be an issue with the processors in 2017.

There are S5, S7 and Mitsi processors in different machines and he's not sure which PLCs they were talking about but he's asked me to look into this (obviously back then it seemed like a million years away so he didn't really bother about it).

Has anyone heard of anything like this? I've Googled to no avail?

I'm guessing it could potentially be a bit like the (so called) millennium bug? Maybe where the 'clock' is running out or rolling over or whatever?

Anyone else heard of anything like this with any of the older processors?

Cheers

Click to expand...

RussB said:

Let's try another line of thought here. Could it be that your boss is going senile and 2017 is when he plans to retire?

Click to expand...

A big thanks for everyone's response on here. I've been away for a few days hence the late reply.

Very strange you say that Russ as he is intending to retire at the end of 2016.

All this info has set my mind at ease at least that there isn't going to be an issue.

We have quite a few S5's and a couple of old Mitsi's. I've been trying to get these upgraded for a few years now but keep getting my requests for money kicked back at me. Maybe a bigger plan to replace the machines entirely. Or maybe just no appetite to 'risk' it when the machines are so critical to process.

Every year after Christmas I have my fingers firmly crossed when we have to power everything back up again.

Thanks again guys!