Nice post Tom. I was also wondering about the replacement hardware vs software patch issue. Just speculation on my part but perhaps changing the software involves an extensive QA / test process, vs swapping out a piece of hardware ? Again this is just pure speculation on my part.
Mark Spencer On Aug 11, 2017, at 9:26 AM, Tom Van Baak <[email protected]> wrote: >> The E911 installation, in the news, is just one of several. Others are >> hospitals, >> fire stations, etc. using different dispatch systems. > > Hey, at least important things like mobile phones, ISP's, Google, Amazon, > FedEx and Starbucks aren't affected ;-) > >> In a wide-area simulcast-overlap paging system, the transmitters in the same >> coverage area are carefully set to all transmit at exactly the same time. > > That's fine. And very clever. But why is this "life safety" system tied to > GPS, to a particular vendor, to a particular model of receiver (that clearly > states in the documentation that it has a 1024 week / 19.6 year window of > valid UTC times)? > >> So to me "synchronizing transmitters” means the control system sends the >> traffic out to all the transmitters (over satellite) and tells them all to >> hold the >> messages in a buffer until “the big hand points straight up” or whatever data >> command the system uses. (excuse the vernacular) > > Exactly. In most of the precise timing world the "big hand" is the "top of > the second", or the so-called 1 PPS pulse. The idea is that all 1PPS agree > with each other, whether from a cesium clock, or WWVB receiver, or NTP, or > GPS (or any other GNSS system). > > Since the paging system failed it sounds like it was synchronized to some > "hand" other than 1PPS. The rare GPS rollover events tend not to disrupt the > 1PPS output -- it is still perfectly aligned with UTC -- which is why almost > no one else worries about the recent TBolt episode, or any other GPS receiver > for that matter. > >> The problems being experienced right now appear to be the interface of the >> ThunderBolt >> with the Zetron Model 620 simulcast controller over TSIP. The Zetron box is >> also called >> a “wireless data encoder.” > > Ah, ok. So do you or anyone have contact within Zetron? The easy fix would be > for them to upgrade their firmware and send out a patch. Probably cheaper > than supplying new receivers from Trimble. I don't know; for us, a s/w fix is > easy compared to a h/w fix or a h/w swap-out. But in the real world, once > technicians have driven to a remote installation, maybe there's no real > difference between a s/w fix and a h/w swap. > >> It is not our goal to blame a particular piece of equipment for this problem. > > Right, no need to blame. I think many of us would just want to pinpoint the > root cause of the problem, out of engineering curiosity. By root cause I mean > actual schematics or lines of source code. It's always been my hope, after > every one of these widespread infrastructure events, that the actual source > code or design decisions be published eventually so that we can all learn > from it. > >> The facts are the 1024 roll over happened and just about nobody in the paging >> business knew it was coming. > > Ok, now you know about GPS rollovers! Fun, isn't it? Leap seconds are fun too. > > When the dust settles, you may want to look into the more general topic of > life safety infrastructure vs. free-from-the-sky time & frequency. These days > nanosecond precise time is cheaper than water -- but it's also fragile. A lot > has been written about this. It's both a wake-up call for naive vendors of > products based on GPS alone and also an opportunity for vendors who know how > to design and market resilient timing products. > > /tvb > > _______________________________________________ > time-nuts mailing list -- [email protected] > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. > _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
