Sunday, January 31, 2016

Discrepancy in GPS Timing of 13 Microseconds

Power Delivery systems rely to some degree on Time Synchronization based on global positioning system (GPS).

A time spike in the global positioning system which rippled through the world on January 28, 2016 was caused by a satellite launched in 1990 failing and triggering a software bug!

Although the timing anomaly measured just microseconds, it could have caused significant navigation errors, Richard Easther, head of the University of Auckland's physics department said.
"The rule of thumb is that for every nanosecond of error, you could be out by as much as a foot," Easther said.

"An error of 13 microseconds or 13,000 nanoseconds works out as just under four kilometres."

What would that error mean for Sampled Values? The 13 microseconds are equivalent  to a difference in the angle of 0.234 degrees in a 50 Hz AC system. This seems not to be very critical. But who knows what happens next.

Be aware that our future power system will rely more and more on GPS or other central time sources. So, the power infrastructure does rely on the GPS (or other means) - which by nature does on the power infrastructure. Everything seems to be highly interconnected.

1 comment:

Rod Hughes said...

Sampled Values certainly requires accurate time synch - noting that the SV frame does NOT contain the specific time but rather a counter of what number the sample is from the start of the 1-second window e.g. at 50 Hz, and 80 samples per cycle the SeqNbr counter ranges from 1 to 4000 and it is therefore critical that all the Merging Units start their counter at the same instant.
At 50 Hz, 80 s/c, samples are taken every 250 microseconds, but at 256 s/c that is just 78 microseconds between samples. If we look at traveling wave fault recorders operating at 1200 s/c, the samples are 17 microseconds apart, but at 60 Hz they are 13.9 microseconds apart.
So a 13 microsecond discrepancy between one MU source and another would sound to be on the verge of a problem for traveling wave technology, but equally we could probably assume that all MUs in the same substation would be synchronized to the same GPS signal so coherency is maintained.
But the point is valid - how tolerant subscribing IEDs are to samples from different MUs whose SeqNbr are skewed in real time is a very vendor and algorithm specific issue.