Just a super quick post to give a little bit of context on a problem. Some (but not all) people see where your recorded GPS track is shifted by more than a hundred meters or more. In general, your route pattern is correct, but it may have shifted to one side or the other. In many cases, however, it sorts itself through the completion of the workout. In other words, it is usually a one-time problem.
The problem seems to be largely affecting companies using the Sony GPS chipset in their devices, which are basically all new Garmin devices from recent years, most new Suunto devices from past years, all Polar devices over the years, all COROS devices, the Wahoo RIVAL and more. Pretty much everyone except Apple.
Very rough list of affected devices (I’m missing a ton here, these are just the most popular)
– COROS GPS watches (however, must double check the original Pace)
Garmin Forerunner 45/245/745/945 / Fenix 6 / MARQ / Vivoactive 3/4 / Venu
– Garmin Edge 130 Plus 530/830/1030 Plus
– Polar Vantage V / M / V2, Grit X, Ignite
– Suunto 5 and Suunto 9
– Wahoo RIVAL
Again, I’m missing a boatload here, but those are the big tickets by heart.
All of these companies have switched to the Sony GPS chipset since about 2018, with Polar and Suunto jumping on board initially, followed soon after by Garmin, COROS and more. The Sony GPS chipset is widely used by these companies for its energy savings that provide longer battery life.
But – that’s all but the point, and not the root of the problem. The problem is related to the ephemeris data file, also known as the Extended Prediction Orbit (EPO) or Connected Predictive Ephemeris (CPE) file. Or just the pre-cache file from the satellite. That’s the file that is delivered to your device regularly (usually every few days). This file makes your watch find GPS satellites almost immediately when you go outside. It’s basically a cheat sheet of where the satellites are for the next few days, or up to a week or so.
Your watch or cycling computer will automatically retrieve this file from your phone, Wi-Fi or USB via Bluetooth Smart, depending on how you connect your watch. Most companies will deliver it anytime your watch is synced and needs a new version. So for your part you never do anything – it just happens quietly in the background.
At this point, the data in this file is wrong, so the data that your watch uses during those first few minutes is also wrong, leading to the offsets. Below is an example of this, from a DCR reader:
Since the watch maintains its connection to GPS satellites during training, it will generally correct itself again while it fetches updated data from above. Most people report it will happen resolves itself by the end of the workout or the next workout. You can see that from yet another DCR reader here:
Same here – identical errors at the starting point on my Fenix 6X Pro and my wife’s Forerunner 45S then they fix themselves – I walked, she ran, so my route is more accurate as it took more time to fix I think … pic.twitter.com/tZ4ysgYPtW
– ac72 (@ ac72) January 2, 2021
To solve this, You can just stick your watch / cycling computer outside and have it recorded for an hour of training, then just throw it away. It has received the required satellite information to probably correct itself.
But it won’t affect everyone. For example, I tried today with two Sony based watches (Garmin FR745 and FR945) to get it to record an incorrect track – and it wouldn’t happen. I suspect this is because in my case yesterday and today I bumped the watch a few times while playing with my kids so I guess it had already re-downloaded what it needed to be correct from the port. I have also tried it with two non-Sony chipsets (Fenix 5S and Apple Watch SE) and saw no issues either.
If you talk to Garmin today, they are working on a solution to the problem, which they think is most likely only updating the server’s satellite pre-cache data, which in turn picks up your watch as always. But first they (and others) work to figure out exactly what is wrong with the data causing this.
Speaking to another person in the industry who addressed the issue, they noted that 2020 technically had 53 weeks and that this is the 53rd week. As such, the suspicious problem with Sony data files could actually be related to that complexity.
Look … just when you thought you had escaped 2020, it’s still waiting for you at the start of your run.
With that – thanks for reading!