'98. Author: kevin sugrue ( triona )
Date: Jun. 15 8:22 AM 1998
Hi Everybody. I've read most of your letters
with great interest.
I'm in charge of the Y2k Project in relation to
the embedded systems in my company here in
Ireland.
Would I be correct in summarising as follows
1. Embedded chips should not create a problem if
they are not in any way linked/connected to date
recording devices.
2. As there is difficulty in getting reliable
info from the vendors it is simpler and posibly
cheaper to replace embedded chips.
Problem here of course is that if you do not get
reliable info from the vendor how do you know
what to replace it with?
I'm not a member of IEEE so I can't log on to get a
copy of the two articles Bart Zoltan referred to
in the June 1998 issue - "Brooding on the year
2000 Issue" and Making embedded systems year
2000 compliant".
Could anyone be so kind as to send me a copy - my
email address is ksugrue@indigo.ie.
Keep up the good work and lets hope we get
everything 2000 compliant on time.
Kevin Sugrue
99. Author: Bart J. Zoltan ( bartzoltan )
Date: Jun. 15 9:06 AM 1998
Method for Determining if an Embedded System is
Susceptible to
Year 2000 Problems
Electronic sensors can determine if a circuit
contains an energized
oscillator, which could be the source of an
error which will arise
on or about the date change from December 31,
1999 to January 1,
2000, also known as "the Millennium Bug" and the
"Y2K" problem.
Of prime importance to the owners and
manufacturers of
microprocessor-based systems is knowledge
whether their system is
potentially prone to a Y2K problem. Many
systems which are not
obviously date-driven or date-sensitive can
be susceptible to
Y2K problems because of embedded circuitry.
For an electronic system to be prone to a Y2K
problem, it must
have some active element which continues to
keep time even when
the system is powered down or unplugged.
Electronic systems which
are powered and have an oscillating element
will always emit a
level of electro-magnetic radiation. A system
without an
oscillating element independent of other
components cannot be
date aware. The system would require an
external update from some
other internal element of the system, which
itself must have an
oscillating element.
When the current in a conductor is changed, the
electromagnetic
field about the conductor is also changed. Any
computer or
timekeeping device must have a time base, which
is usually a
square wave. The existence of the time base
gives rise to an
electromagnetic signal, which can be detected
by suitable
electronics.
Apparatus sensitive to electromagnetic
fields can be used to
sense if any part of a circuit is "active". While
turning off all
external power, an antenna, especially a
near-field antenna, can
be used to scan the system under test for active
elements. To test
a system for Y2K vulnerability, the complete
system must be
unplugged or unpowered. Using a near-field
probe, each area of
the circuitry is scanned for an
electromagnetic signal, which
indicates of an active circuit element, such as
an oscillator or
time-base.
The base frequency of the oscillator may be one
frequency to
search. Much higher frequency components will
also be radiated,
these frequencies being inversely
proportional to the rise time
and fall time of the active components. Note
that these
radio-frequency signals will be present at
high frequencies,
even when the base square wave signal is quite
slow.
Only the dv/dt of the rise and fall determine the
width of the
power spectrum. Frequently these rise and fall
times are not
spelled out in clock component
specifications.
Fmax= 1/( p * trise)
Rise/fall time (nSec) Fmax (MHz)
1 320
2 160
3 106
4 80
5 64
6 53
7 45
8 40
9 35
If the system is completely scanned and there is
no source of
electromagnetic radiation, then the system is
very unlikely to
have a Y2K fault, since the system is
time-unaware. If there is
a source of electromagnetic radiation in the
system, then no
conclusion can be made to the Y2K status of the
system.
Background: A major technical challenge of the
millennium
transition is the problem of the failure of
computer-based
systems, because embedded systems engineers
represented the year
field with two digits instead of the full four
digits.
Representations of date and time in
microprocessor-based embedded
control systems or system components are
capable of causing
product failure, system shutdown or data
corruption at and about
the calendar change from December 31, 1999;
23:59:59 to January 1,
2000; 00:00:01. This Y2K problem occurs
because most embedded
systems use an electronic calendar means which
represent the
year in a two digit year format, YY, such as 96,
97, 98, 99,
instead of the full date format, YYYY, such as
1996, 1997, 1998
and 1999. Most of the integrated circuits
designed for datekeeping
and time function reset the year register to 00
after 99 instead
of the full 2000.
Experimental Data:
Using a Hewlett Packard 54600 oscilloscope,
with a P6109B 10X
probe, a Macintosh computer, Model LC as the
device under test.
The computer was probed for its real time clock
in various states
of power. The case was opened, and probed with
the open probe,
with the ground probe connected to case
ground.
The table below summarizes the results for the
computer:
110 VAC Internal 3.6V Probe p-p measured
Power Lithium battery Status voltage
frequency
ON IN contacting 309.4 mv 32.84 kHz
ON IN non-contacting 4.25 mv 34.86 kHz
OFF IN contacting 182.8 mv 32.73 kHz
OFF IN non-contacting 6.0 mv 32. kHz
OFF OUT non-contacting 1.00 mv none found
The experiment was then done using a modified
digital time-piece
described in US Patent 4,419,016 and
manufactured by Seiko, Japan.
Internal Probe p-p measured
battery Status voltage frequency
IN contacting 130.6 mv 32.79 kHz
IN non-contacting 5.000 mv 30.58 kHz
OUT non-contacting 1.000 mv none found
The instrumentation used for the above
measurements, automatically
indicates the peak-to-peak voltages, and the
calculated frequency.
The much decreased signal-to-noise ratio on
the non-contacting
measurements causes the apparent frequency
shift.
Bart J. Zoltan
Principal Research Scientist
Wyeth-Ayerst Research
100. Author: Dave Bettinger ( dbettinger )
Date: Jun. 18 9:15 AM 1998
Kevin,
Bart's response, while comprehensive, is way
above most of my clients' abilities to perform,
as most do not have the portable oscilloscopes
or similar diagnostic equipment to accomplish
these tests "in the field." Out of respect for
Mr. Zoltan's scientific stature I believe his
suggestions are well-founded and accurate,
but perhaps more applicable to a laboratory or
engineering concern than the "average"
commercial enterprise.
I have learned that a more lay-centered
apporoach to determining the need for further
Y2k research can be realised through the
application of a six-point checklist. The
checklist is discussed in detail in the
May/June 1998 issue of the Year/2000 Journal
(available at www.y2kjournal.com). For
expediency I will provide the two salient
questions here.
Question 3: Can a the equipment display a date?
This should be obvious to the observer, but may
not always be so. It is possible that a piece of
equipment can only display a date when
connected to an external computer or other
diagnostic equipment (like the auto
mechanic's portable, plug-in diagnostic
computer for vehicles of the '90's.) Even if the
equipment can display a date, it must be able to
retain that date when disconnected from a power
source, as pointed out be Mr. Zoltan, to be a
source of y2k concern. A video cassette
recorder (VCR) can display a date, but does not
retain it when unplugged from the wall. The loss
of such a date does not impede operation of the
equipment; only its scheduling function.
Question 6: Does the equipment utilize a date to
schedule events? This question is the last of
the six-point checklist, so it asumes you've
already answered "yes" to the preceding,
qualifying questions. The answer here can
usually be obtained from the person tasked with
the day-to-day operation or maintenance of the
equipment. If the answer is "yes," you have
accomplished all of the pertinent testing
mentioned in the previous posting from Mr.
Zoltan. You now know you have a piece of
equipment that must be internally
investigated for y2k potential.
Given a "no" answer to Question 6, you are free to
cross the item off your list of items needing
full investigation. Even a "yes" answer can
still lead to an item coming off the list. Most
VCRs use a date to schedule events, but if you
lose power to the device it will still work. (How
many of us have seen a VCR with the
perpetually-blinking clock?) Depending upon
the application, a VCR is likely to be a
low-priority item, unworthy of y2k attention
even though it passes all the tests. So triage is
important even in the embedded chip arena.
An "I don't know" answer is not too bad for
Question 6; it simply means the item must remain
on your "to do" list for y2k potential.
Mr. Zoltan makes an important point about a
piece of equipment requiring the ability to
retain date information when disconnected
from a power source. (Question #2 in the
embedded chip checklist.) Without an
uninterruptible power source (onboard
battery, UPS, etc.) the presence of a timing
function in a piece of equipment is
insignificant. Equipment for which date
awareness is critical, will always have an
onboard power supply or some other means of
ensuring uniterruptible power.
If one cannot perform the type of inspection Mr.
Zoltan proposes, a visual inspection can yield
essentially the same results. An
appropriately-qualified technician or
engineer, (or just someone with a little bit of
dangerous printed-circuit-board knowledge
and a grounding strap) can perform the visual
check relatively easily. The only real
requirement is that one knows what an onboard
power supply looks like. The most common types
are easily recognised by even a lay-person
after a five-minute introduction. Since many
businesses have lots of equipment needing
inspection and few people to do it, the
less-critical inspections can be performed by
SWAT teams that have been trained to find such
devices.
I hope this provides some additional insight
for you. As I have said, I fully support Mr.
Zoltan's recommendations. I only feel that
they may not be feasible for all enterprises.
The bottom line is to inventory, inspect,
prioritize and fix all embedded chip
equipments. The method you use to identify
those equipments needing attention will be
whatever method gets you through the process
the quickest and with efficiency.
DaveBettinger
Director of Business Solutions
CST2000,LLC.
<dbettinger@cst2000.com>
101. Author: George Girod ( ggirod )
Date: Jun. 21 6:32 AM 1998
I would like to make a couple of small comments on
the previous excellent article...
The VCR example above is a good example of
another rule I have often applied. Compliance
is a function of usage. The blinking VCR is just
fine on an audio-visual cart but it will not do
for a security center or for an investigator who
records legal evidence.
Secondly, the absence of an uninterruptable
power supply does not imply date independence.
It may well be that the user sets the date/time
when starting it up after an outage. It may not be
able to be set after 2000. Again, compliance is a
function of usage.
102. Author: Francis Lovering ( francis )
Date: Jun. 22 10:24 AM 1998
Of course, if your video does not roll over to
2000, you can still use it by putting the clock
back
Maybe by a multiple of 7 days to get the right day
of the week.
IO certainly do not program mine by date, I
typically wish to record something next
monday.
I believe the same can be applied to the vast
majority of embedded systems, only a very few
need the actual date to be accurate. |
