GPS Week Rollover Garmin GPS

Equipment
1

This is a follow on from an earlier thread on this subject before the last rollover, see GPS Week Roll Over

For those who don’t know, GPS navigation messages consist of two time fields, one is the number of 1.5 seconds units since week start and the other is the week number, which is the number of weeks since last epoch. To work out the UTC time you take the two and can work out the time and date. This is done inside the GPS which uses the time/date or makes it available for you to use. Simples. Well it would be apart from the week number is 10bits which means after 1023 weeks it wraps back to 0. This happens every 19.6 years and if the firmware in your GPS is designed to handle this then dates will jump back some years at rollover. The first epoch is 00:00:00 Jan 6th 1980. Then first rollover occurred on 00:00:00 Aug 21st 1999 and most recently Apr 6th April 6th 2019.

It’s reasonably simple for firmware writers to handle 1 rollover typically by knowing the week number when they wrote their software. The can then play mod 1024 maths with the week number so their software works OK when the rollover occurs. i.e. the real week number cannot be less than the saved number and if it is the real week number is weeknumber+1024. This means the software should give the correct date for maybe nearly 20 years.

The referenced thread shows people testing devices, some which were reasonably current and some old when the last rollover occurred. I was lucky, my 2006 vintage GPS (Garmin Vista HCx) coped fine. However, it’s nearly 2026 which means the software is nearly 20 years old and that is getting close to the 19.6 year typical lifespan of a rollover fix. This was brought home by a long thread on the QRP Lab forum where the GPS module used did it’s week rollover at the beginning of August and that made me realise the date on my GPS may fail soon.

So I did some digging to see if there was any newer software updates for my GPS. There is a UI update a few releases on from mine that was dated 2008 which may buy me some time. But more deep digging showed Garmin released a big firmware updater program in 2019. Surprisingly this supports my GPS. Garmin haven’t made the Vista HCx since 2011-2012, that’s 13 years yet they are still supporting it so tight-fisted hams like me can squeeze more use out of it and not buy a new unit.

You can find the updater program from here on Garmin’s website: Garmin: WebUpdater Software Update Collection and it’s available for ancient Windows, more modern Windows and MacOS. I am in the process of dumping Windows but still have a Win10 machine. the program worked effortlessly finding my USB connected GPS and then offering me the latest UI and Firmware updates. I made sure there were fresh batteries installed even though it would be running of USB power. It took a total of about 5mins to reflash all the software.

I’ve now got 2019 vintage software which means it should keep the date correct till the next rollover in 2038 when I’ll be coming up to 80 and this may not be an issue.

However, full marks and a bonus to Garmin for supporting really quite old and obsolete units. (Hey it must be obsolete, it has a tiny display and no touch screen.) In today’s planned obsolescence world where you are expected to drop $1500 on a new phone every 18months, Garmin supporting old tech means they bought my loyalty for all future GPS device purchases.

Whether the updater supports even older units, I don’t know. But it’s probably worth trying if you have a crusty old GPS and want to see about giving it a new lease of life.

:+1: :+1: Garmin :+1: :+1:

6 Likes
2

But it is nice when manufacturers continue to support old devices.

3 Likes
3

Shouldn’t this read “This happens every 19.6 years and if the firmware in your GPS is not designed to handle this then dates will jump back some years at rollover.”

4

It should say some years after rollover. If you have some nonvolatile memory you can count the rollovers and make it work a long time. Or do what modern units will do and use the newer CNAV message which has 13bit week number and goes for 157 years.

I’m still impressed that Garmin support such old units to make it even longer before I give them any more of my money.

5

Less so for their line of wearables. More forks in their product line than a cutlery tray at a Happy Eater.

They zoom in and out of press releases like a blue bottle with the two bob bits. New updates every 5 minutes, but held back for current gen for ages then trickle down to one gen back. More new models punted on to the runway than at Paris fashion week.

Now they’ve gone and slapped a subscription on a device that costs an arm and a leg. In the space of about 3 hours and 27 minutes I’ve migrated from a Fenix 5 to a Fenix 6X Pro to a Forerunner 965 to my current squeeze the Enduro 3.

There’s about 927,612 variants of their watches as well so very confusing for customers not familiar with their numerous watch lines. Pick the wrong one, particularly during a sale, and you may end up with a device that for all intents and purposes is, or is fast approaching, end of life. Such is Garmin’s strike rate the past couple of years.

At least their GPS lines in marine and aviation seem to be a bit more settled. InReach (subscription aside) seems fairly stable and well-supported too.

For watches I’d be more looking at a Suunto in the future over the current Garmin circus. Hard to leave though when you’re fairly walled in to their ecosystem with a watch, bike computer, dive watch etc.

1 Like
6

This thread made me get out my Garmin Etrex purchased in 2007 which seems to work fine. The problem with this unit is the bizarre 4 pin ePole connector. I’m now wondering what I did with the lead for it and I will need to check whether I still have a PC with a 9 pin serial port. I might decide to purchase a new USB lead…. or maybe just continue using my GPS64 and sell the Etrex, if indeed anyone will want such old tech. Maybe it’s too much like flogging a dead horse.

2 Likes
7

That’s annoying about my new PC, it has a real UART built in, software shows it’s there but there’s no connector on the case… it was an option. A check online shows they can be bought new in packaging ready to fit. Just they’re £30 and a USB cable is very much less. Even an genuine FTDI chipset cable comes in for less.

There’s always plenty of mature units for sale on eBay so punt it on if you can.

1 Like
8

I’d speculate you might be able to pick up an aftermarket cable on Ali Express or eBay in a pinch. I’ve picked up a charging cable for my Garmin watches, for example, in the past no problem, usually at a fraction of the price of OEM cables.

Good thing about those old Garmin GPS units is you can pull them apart and replace the batteries reasonably simply. Modern stuff is nearly all built to be almost unserviceable, like most products these days alas.

Oddly enough I replaced the screen on a 2019 Galaxy Tab A yesterday with a knackered digitiser. Took about an hour all in. The repair was actually reasonably simple, taking about 20 minutes, with most of that time spent heating up the case with a hairdryer to soften the adhesive.

The rest of the hour was spent messing about with isopropyl and cotton buds to try and remove Samsung’s Geoff Capes-grade adhesive.

Total cost of repair? €35 approx for the screen (including shipping from China), which was a brand new OEM like for like screen with digitiser, and my time. Versus €100+ quotes I received from high street phone repair shops.

Happy days. :smiling_face:

EDIT: This the cable you were looking for?

1 Like
9

I have the predecessor of yours, the Vista Cx. It looks essentially the same, but has an inferior GPS chipset. The date and time went haywire at the 2019 rollover. I’m pretty sure I looked for a firmware update at the time, but didn’t find one, and by that time I’d long ago retired it anyway. It was always a bit erratic.

I switched it on this afternoon and it reported a date in 1964 (or maybe 2064) but later warped to 2017. Anyway, with nothing to lose I tried WebUpdater (which I already had installed) and blow me down it found an update. I think it went from 3.40 to 3.50 and it now reports the correct date and time! This is indeed remarkable.

[ Afternote: I just looked at the track files it created while it was running the dodgy firmware. It has recorded tracks dated in 2038, 2058, 2064 and 2078 ! ]

I also have an even older GPS12, so old that it doesn’t have the concept of background maps. I can’t remember exactly when I got it but I have data from it dated in 1998 so it has had to handle two rollovers. I switched it on today and it’s spot on.

Martyn M1MAJ

2 Likes
10

I’m rather more impressed that Garmin are/were supporting such old hardware. It seems unexpected for such a large company to do this.

1 Like
11

It either means the code (or that portion at least) probably hasn’t changed that much over the years making the fix pretty consistent across devices. Or they overcharged in the first place to be able to provide long term updates.

I can understand why something like a phone would need the date/time, but for a navigation tool , what would be the disadvantage of having the wrong date? I presume the positioning would still function and generally for 99.99% of people, who are not subject to 6 months of darkness at an artic/antarctic research station or some such and just out on a day trip to Ecclefechan, the date is irrelevant.

1 Like
12

Positioning is fine but time/date functions are wrong. This means tracks have the wrong date info in them. I did think that should my GPS revert its dates to 2007 I would write a quick Python thingy to “polish” the 2007 dates to 2026 dates etc.

But many functions that use date on the unit will be wrong such as showing time of sunrise and sunset for “today” as today will be months out.

It’s only really needs to be able to save the week number into the software so it can see when week number is less than was saved and doing modulo-1024 number crunching. But the real issue is having build systems in place such they can compile/assemble software using tools they may not have used since 2007 etc. I know this bit us at work when a customer wanted to go back and run some simulations from yonks back. We didn’t have any 32 bit machines with any of the developer tools and had to set up some shonky VM to do the job.

And if your name isn’t Microsoft, you also need to test the software and run regressions before you release it.You don’t want to brick many customer units which you end up having to replace with shiny new equipment.

2 Likes
13

If you are only using GPS for immediate navigation, indeed the display of the date doesn’t matter.

But it matters to me. I preserve and curate all of my GPS tracks as a long term record of where I have been and when (not just SOTA, but also local walks, cycle rides, stately home visits etc). The raw data trivially tells me what I was doing on a particular date. With a bit of geospatial database magic, I can do the inverse and determine when I have visited a particular location, even if it was decades ago.

I also use my GPS tracks to geotag photos taken on cameras which have a clock but no GPS of their own.

All of this depends on correct timestamping, obviously. If the date were wrong by an exact multiple of 1024 weeks I could of course fix up the tracks afterwards. In practice when my Vista CX went wrong the dates just went wild (and this suggests that they tried to cope with the rollover but just got it spectacularly wrong!)

Martyn M1MAJ

1 Like
14

Quickly sold, within an hour. Now I don’t have to feel bad about technology lying dormant and unloved in my drawer. :grinning_face:

2 Likes