This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

(Known Issue) GPS accuracy on Jan 01 2021?

Just had a run with my 945. GPS was way off during the run, I mean about 50-100 meters off reality during the whole route.

Has anyone else had this today? Using GPS+Glonass setting.

Janne / Finland 

  • I have done a few more experiments and think I have worked out what is going on.

    Before going for a run today I connected to GC on my PC and did a sync. I verified the CPE had downloaded correctly by checking in the appropriate folder using file explorer. Took it outside and initiated (but not started) an activity with a map screen and I could see my location was about 200m offset. I then left the watch outside with clear view of the sky for about 20 minutes. When I got back out the location was spot on and I went for my run. The resulting track is probably the cleanest I have seen for a while.

    After the run I synced over wifi and immediately checked my position again on the map. The 200 m offset was back.

    So what is going on is that Garmin is feeding us a corrupt (or out of date) CPE file. The CPE file is the GPS almanac data, it contains information on individual satellites. Exact timing corrections, which ones are faulty or degraded, that kind of thing. There is however another way to get this data, which is to download it directly from the GPS satellites. This however takes time. You need to initiated an activity and leave the watch with a good view of the sky for 15 minutes or so. Make sure you select extended timeout in the activity's setting otherwise it will switch back to watch mode after 5 minutes and the complete almanac might not get downloaded.

    The good news is that there is nothing wrong with the device as such., it is just a case of bad data in causing bad data out. The bad news is that you need to repeat the above procedure after every time your device syncs with Garmin Connect and your GPS almanac data gets corrupted by a new CPE download.

  • 935 seems fine 945 is off.

    A garmin express update improved the offset by ~50% but it’s still there.

    Tried all 3 gps modes and same result so it must be a garmin software issue for 945 users. Thankfully the track is the right shape and distance, just not well overlaid, and I assume that means the climbing metrics are off. 

    Maybe I can use this as a reason (excuse) for my slow pace!? Slight smile

  • yep, I get it too.

    others get it on coros and suunto.

    I do NOT get it on my Apple Watch SE with GLONASS but do on my 945 with or without GLONASS./GALILEO

    I can't get rid of the CPE Expired message and have tried several things to get around that.'s probably Sony. I can't see Sony reacting quickly, let's see if Garmin can get a workaround sorted. as obviously I can't do any exercise if my GPS is 't working. Even the stress of an incorrect GPS fix will stop me from doing indoor workouts.

  • What can cause your FR945 to record a weird trail, like through the middle of a lake, but still get the distance correct?

  • Also the CPE file size seems off, 2 kb only. Isn't that usually somewhere 32k-64kb range.

  • Same for me. Gps is way off! The distance is ok but when you look on the map the route is way off! 

  • Forerunner 35, Johannesburg, Working Well .

  • But that's not what's really going to hurt Garmin.  The first woman who's attacked and uses Incidence Detection sending the cops blocks away from her location will seriously damage Garmin's reputation.  The ambulance looking for the heart attack victim blocks away from his actual location will seriously damage Garmin's reputation

    Incident Detection is sending the alert and location from your phone.  You're gonna have to find a different frivolous lawsuit against Garmin Perry Mason.

    The trail runner that gets lost relying on his GPS and has to be found by Search and Rescue will seriously damage Garmin's reputation.

    Don't forget Nike's reputation should their shoelace become untied and they trip and fall. 

  • One thing I noticed is that despite the mapped route being completely off, as well as the paces throughout the run (e.g. 1-mile PR, other bizarre stats), the total distance is spot on (and therefore the average pace for the run). I verified this because I ran a known and measured route and it came in to the dial. I'm not sure how the distance could be so accurate and yet all the data in between, including the map, are way off (and don't add up to the same total distance reported).

    This is the 3rd time this happens, with the first incident on Dec 23rd for me. I've had the FR 945 for 6+ months and no issues prior.

  • Glad to see it isn’t just me.