[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gpsd-users] ✘Some GPS to fail on April 6 GPS rollover

From: Mike Tubby
Subject: Re: [gpsd-users] ✘Some GPS to fail on April 6 GPS rollover
Date: Tue, 2 Apr 2019 20:49:21 +0100
User-agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:60.0) Gecko/20100101 Thunderbird/60.6.1


Its worse than that ...

Firstly, some older receivers will stop working altogether, while others will output wrong week numbers forever. We're talking old Rockwell receivers and old Trimble receivers like the S-Vee-Six, ACE, ACE-II.

Another class of GPS receivers fails to work in week zero but recovers when we get back to week no. 1 - some of these are used in Spectracom frequency and time master clocks and run for a whole week in 'hold over'.

Another class of GPS receivers have manufacturer applied software fixes (in some cases 'hacks') that apply a local date (often the firmware compile date) to un-roll or offset the WNRO problem to a different place ... you'll find I contributed a hack like this to NTP's 'parse' driver for Trimble TSIP back in 1999-ish - basically it does:

    if (week_no <= 990)
        week_no += 1024;

which you can sort-of get away with on the first roll-over but doesn't really hold for the second and subsequent ones!  So what manufacturers have typically done is to pivot around the compile time of firmware or provide a configurable register to allow the user to set the pivot point ... which is great in as far as it goes - it just kicks the can down the street for a few years, but gives us a bigger problem of knowing when/where the WRNO problem has been deferred to.  Some Trimble GPS receivers state they would break in 2017 (but didn't) others are allegedly good to 2024 - but we'll have to see.

Another class of GPS receivers can be bricked by testing them! The test goes like this - build a 5 min stream of GPS data using Linux 'gpssim' and play it out using a Hack-RF SDR with a good 10MHz clock (taken from a GPSDO with an outside antenna shielded from the SDR of course!).  What some receivers do is implement "extended week number" (16-bit week number) by nudging their own week number counter forwards each time they detect a week number change - they store this in FLASH.  The problem is that the WRNO test played out by the SDR appears to work - once - and then when you attempt to use the receiver for real its lost, can never recover lock and is effectively bricked.  Trimble SK8 and ACE-III are both bricked by testing them - I know - I've done it. Consider one with a 1PPS output discipling a TCXO on some important system or critical national infrastructure.

All-in-all this second WRNO event is worse than the first because the first could be handled with an absolute hack and now we're using relative hacks and they're all different.

People are going to think they got away with the second WRNO event only to be bitten, unexpectedly, at some point in the future.

We have upgraded all of our customer's critical systems and tested them.

Your Mileage [or Week Number] May Vary ;-)


On 02/04/2019 20:11, Gary E. Miller wrote:
Yo All!

Looks like some GPS will give bad dates after the April 6 rollover.

Here is another list:

Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
        address@hidden  Tel:+1 541 382 8588

            Veritas liberabit vos. -- Quid est veritas?
     "If you can’t measure it, you can’t improve it." - Lord Kelvin

reply via email to

[Prev in Thread] Current Thread [Next in Thread]