[time-nuts] Parity Errors

[Andy Talbot]

This morning, I stumbled out of bed still half asleep to make my early
morning coffee and glanced up at the MSF controlled wall clock. 'Wed 3 Jan'
it said.  Shuffled into kitchen and while pouring beans into the grinder
...  " 'ang one, it's not Wednesday!"

Checking the clock again, comparing the time against my GPS clock and
another MSF one it was also showing one minute error in the time.  Clearly
there had been a burst error during the update period and an even number of
bits corrupted.   I've had this before on another MSF clock with various
parts of the decoded time and it's been easy to see which possible pairs of
bits had failed.

The timecode for the minute error must have corrupted when it was receiving
01 or 02, and the units and 2s bit got switched,  (or the units and the
parity, but they're separated so less likely).

The day of week is simpler to visualise the error mechanism.   Sunday is
transmitted as zero, Wednesday as 3.   Two adjacent bits corrupted, so it
will pass the parity check.

Andy
www.g4jnt.com