It’s
Time To Be Realistic About IED Clock Synching
by John Tengdin, Editor in Chief
January 2000
For many years, sequence of events (SOE) recorders have been commercially
available that time stamp status change events to the nearest millisecond. Since
all the status contacts were wired into one box, there was no issue of clock
synchronization. One clock in the SOE recorder time stamped all events.
This one millisecond de facto standard was understood in 1996 when the
utility industry began defining the requirements specifications for integrated
substation protection, control and data acquisition systems. It was clear that
the industry would not accept a backward step, or any relaxation of the de facto
standard, in an integrated system. It was also recognized that to create time
stamps to the nearest millisecond in many IEDs across a LAN, the clocks in the
IEDs must be set an order of magnitude more precise. Thus, the EPRI Requirements
Specification RP3599 has always required IED clock synching to 0.1 millisecond
to satisfy the one millisecond time stamp requirement. RP3599 also defines
additional accuracy classes for other applications. For synch check using time
tagged zero crossings, the time tagging must be to 0.1 ms (with clocks set to
0.01 ms) as 0.1 ms is equivalent to 2.2 ° in a 60 Hz system. For
synchrophasors, the samples must be synchronized to the nearest microsecond. But
the most lenient requirement in RP3599 is for SOE time tagging to one
millisecond – with all IED clocks synchronized to 0.1 ms.
From the time this requirement was defined, many had hoped that such
clock synchronization could be achieved over the substation LAN. When the
decision was made to use either 10 MB or 100 MB Ethernet for the LAN, hopes were
raised even further that such LANs had adequate and repeatable performance to
meet the 0.1 ms clock synch requirement. Unfortunately, to date no one has been
able to demonstrate such performance, even in a laboratory environment. The so
called “backward time synch method” seems to provide clock synching to a
millisecond or two with some special tweaking of the Ethernet interface card,
but this is an order of magnitude worse than the requirement specification.
Another, but so far untested, approach promises to give 0.5 ms clock synching,
but this is five times worse than the requirement.
Now the time is near to install a number of substation automation
systems in real substations. Thus the time is near to make the hard decisions of
what will actually be installed in these substations. It is unrealistic to
believe that those who analyze system events – and who have had true 1 ms
resolution of events in the past – will be persuaded to accept something
poorer. That being the case, it is now time for the Utility Initiative
participants to face up to reality. That reality is that 0.1 ms clock synching
cannot be achieved over the presently available 10 or 100 MB Ethernet LANs. The
only option is to provide a timing source (IRIG or GPS) directly to each IED via
a timing wire or dedicated bus. The time for dawdling over this issue is past.
Let’s get on with it.
These are your
editor’s opinions. If you agree or disagree, we welcome your comments. Make
them directly in the comment box while you are on line.
Where do you want to go now?
WiseOwl™