Vern, Cyrus, Jim and Dave.

The problem: we are running PEM front ends in the mid stations, but dont want to install a complete timing sytem (fanout and IRIG-B units) to service just one front end computer per building.

One possible solution is to use the atomic clock/MCA distribution system to provide a local IRIG-B signal at the mid stations. To test this, we took a Time Code Translator (TCT) which used to be installed at the mid station and fired it up in the MSR. It was connected to the MCA via a short LC-LC single mode fiber optics cable. Its rear IRIG-B (the square wave version) was viewed on a scope by Vern, and then plugged into the front end computer h1susauxb123. When restarted, this machine's GPS time was off by 1year and 1day. Cyrus tracked this to a setting on the TCT card which was not outputting the year info in the IRIG-B code, verified by the scope. This was fixed and when h1susauxb123 was again rebooted it was 16 seconds off. This is the accumulated leap seconds between UTC and GPS time.

Thinking about it, we realized we should have expected this. The LIGO Timing System is GPS based. It is actually GPS time which is provided by the IRIG-B unit. Whereas all commercial timing systems, including  the Symmetricom NTP server and the Timing Solutions MCA, use UTC time encoded into IRIG-B. Therefore the independent Symmetricom/Datum/TimingSolutions will differ from GPS by the leap seconds. We certainly dont wish to maintain a leap seconds lookup table, so we are abandoning the TCT as an IRIG-B source for the mid stations. This closes out WP#4175.

To proceed onwards, Rolf has suggested an RCG code change whereby a front end which is missing an IRIG-B card will use its system time instead. 

When a solution has been found, the timing units in MX can be removed and MY PEM can be constructed.