"Eric Germann" <ekgermann@cctec.com> writes:

During the recent attacks, there was a brief discussion as to whether the USG had or would invoke GPS SA. We have a client who is building a broad ATM network and had planned to use GPS based timing solutions from TrueTime for a clocking source, in addition to a feed from the LEC.

If GPS SA is invoked down the road during a crisis, what effect is there, if any, on networks which rely on GPS for timing, not positioning?

If one is running in locked-position mode (supported by many receiver elements that are designed specifically for timing applications such as the Motorola Oncore UT; I'm not sure what is inside the Truetime, but it wouldn't surprise me to find it's one of the Oncore family...), induced error is substantially less, but even in the worst case, you're looking at tens to a couple of hundred nanoseconds. In short, nothing that you would notice if you're using GPS as an NTP source. Be aware though (getting back to your original question) that even in the best of times jitter on GPS is more than you would really like to have for a telecom clocking application. This is mitigated somewhat by the fact that you are not getting raw GPS data in most cases, but rather the output from a crystal oscillator that is "steered" by GPS. The high-end GPS-based units often have an OCXO instead of an el-cheapo crystal, in order to help hold-over performance. The GPS-based timing vendors would have you believe that GPS is every bit as good as cesium for a stratum 1 frequency reference, and it's probably good enough for you too, but if you want to be pedantic about it, it's not as good. Does it really matter? I don't think so... particularly in the context of the price of a Cs standard. If the LEC is willing to give you a feed, though, why worry about it? I've seen applications that only have a LEC feed run for years on end. You could always get a rubidium standard (cheaper than Cs by a long shot) if you're worried about what happens if your feed from the LEC dies... ---Rob