Issue and Background
Over the first week of using my new Fenix 8 Solar 47mm, I noticed that it would occasionally take a really long time to acquire GPS signal. Sometimes it was 15, 20, or 30s which is inconvenient, but sometimes it was over 2-3 minutes, and one time I gave up at 5 minutes before rebooting the watch and trying again. My Epix Pro 47mm, for comparison, usually takes between 3s and 6s to acquire GPS signal.
This behavior in the Fenix 8 has been documented on Reddit (link) and on the Garmin forums (link, link). I contacted Garmin support twice:
- The first time was before I had the very long, 5+ minute connection, and I explained that the flagship Fenix 8 was taking between 5x and 20x longer than my Epix Pro to connect. The rep told me that this is normal behavior, devices vary in their time-to-connect times, and that, as long as it connects eventually, it’s not faulty.
- After the watch failed to connect to GPS for 5 minutes, I contacted the rep again, asking for a warranty replacement. This rep told me to take it back to the dealer. I purchased from an online dealer, and returning it at my own expense seems inconvenient and not super fair for a clearly faulty device.
By the way, I now believe that the watch probably would have connected eventually in that one 5+ minute trial, as I later saw another instance of 4.5 minutes. But I still think this watch is defective, or at the very least suffering from a serious bug. Garmin support has always been outstanding and I’ve had absolutely stellar experiences with the company in the past. That said, I was left a little disappointed this time. It rankled that they were telling me I either had to suck it up and deal with the issue, or pay to ship the watch to the dealer and wait weeks for a replacement. So I set out to figure out what was going on.
Research and Troubleshooting
After taking the normal troubleshooting steps of rebooting and factory resetting, I dove into the Garmin support pages (link) and Garmin forums for help. It was recommend that I ensure that the watch:
- Was updated to latest firmware (check)
- Had been synced with Garmin Connect (check)
- And had a current “ephemeris” file, sometimes referred to as EPO or CPE (check)
Having satisfied basic troubleshooting protocols, I was disappointed when the problems persisted. To try and see if this is a hardware issue or a software issue, I got my hands on another, identical Fenix model and began some testing.
The Tests
I used two Fenix 8 Solar 47mm models to test the effects on GPS signal acquisition times of:
- A current or missing ephemeris file
- GPS 1.18 vs. GPS 2.08
- Hardware differences between original, potentially defective watch vs. a second, hopefully pristine watch
In the initial troubleshooting of my original Fenix 8, I noticed that it was always, always faster to acquire GPS signal on the very first try after a reboot. Thus I separated out the first-post-reboot trials from subsequent trials in both watches and added an additional comparison:
4. Differences between first acquisition times post-boot and non-first acquisition times.
Test Devices
I will refer to my test devices thus:
- Watch OG is the first Fenix 8 I bought, which has had connection times all over the place and inspired these tests
- When I got it I synced it with my Connect app and allowed it to upgrade to latest firmware fully (11.68, as I recall) before using it for the first time. I would later regret this sequence of events.
- I also upgraded this watch to the beta program eventually, but this made no difference in the behavior, and the current beta firmware still carries GPS firmware version 2.08
- Watch NEW is the additional Fenix 8 I have, which has only been used for these tests.
- It was originally tested with Firmware 10.23, which carried GPS 1.18
- I performed some tests with this firmware before syncing it to Connect so I could test whether a current ephemeris file made a difference.
- I then synced the Watch NEW, but did not allow it to upgrade, to complete the ephemeris test.
- Finally, it was allowed to upgrade all the way to 11.XX.
Results
I completed many, many trials of taking both watches outside and trying to acquire GPS signal. I started my stopwatch just after the activity screen popped up with the GPS signal meter. Most trials were conducted using the Run profile with All + Multi-Band satellites, though some were with the Walk profile and Auto Select satellites. Anecdotally, I didn't notice a difference between the two.
Here are the main results in graph form, establishing the two major points that:
- Upgrading from GPS 1.18 to GPS 2.08 significantly degrades time-to-acquire times and
- Once on GPS 2.08, the first connection to GPS following a reboot is never more than 15s, and the average connection time among non-first post-reboot connections is significantly greater on both watches
Here are all the comparisons I performed. Statistical tests were all t-tests. I forgot to include the sample sizes, but for most comparisons I had at least 24 samples, excepting the "first post-reboot" comparisons, which had fewer samples.
Scenario 1 |
Scenario 2 |
Comparison |
Result |
Conclusion |
NEW, 10.23, GPS 1.18, No ephemeris |
NEW, 10.23, GPS 1.18, Ephemeris |
Current ephemeris file vs. non-current ephemeris file |
Significantly faster with current ephemeris, 18s vs. 8.8s, p = 0.0080 |
Ephemeris file cuts acquisition time in half. |
OG, 11.XX, GPS 2.08, Ephemeris |
NEW, 10.23, GPS 1.18, Ephemeris |
GPS firmware 1.18 vs. 2.08, two different watches |
Watch NEW with GPS 1.18 significantly faster, 9.3s vs. 44s, p = 0.0054 |
GPS 1.18 on Watch NEW is much faster than GPS 2.08 on Watch OG |
OG, 11.XX, GPS 2.08, Ephemeris, First Acquisition |
OG, 11.XX, GPS 2.08, Ephemeris, Non-first Acquisitions |
After a reboot: first trial vs. non-first trials |
First acquisition is much faster: 11s vs. 46s, p = 1.9 x 10-6 |
On Watch OG, first GPS signal acquisition following reboot is significantly and dramatically faster than subsequent acquisition times. |
OG, 11.XX, GPS 2.08, Ephemeris, First Acquisition |
NEW, 10.23, GPS 1.18, Ephemeris |
Best-case scenario for Watch OG (first post-reboot) vs. Watch NEW on GPS 1.18 |
Very similar (12s vs. 9.3s), though statistically different, p = 0.00011 |
Considering only first-post-reboot acquisition times, Watch OG is very similar (though statistically slower) to Watch NEW on GPS 1.18 |
NEW, 10.23, GPS 1.18 |
NEW, 11.68, GPS 2.08 |
GPS firmware, same watch |
Acquisition times using GPS 1.18 on NEW were significantly faster: 9.3s vs. 19s, p = 3.0 x 10-7 |
Upgrading from GPS 1.18 to GPS 2.08 caused GPS signal acquisition times to double in Watch NEW. |
OG, 11.XX, GPS 2.08 |
NEW, 11.68, GPS 2.08 |
All connection times between watches when both are on GPS 2.08 |
Watch NEW is noticeably and significantly faster, 19s vs. 39s, p = 0.00032 |
Watch OG potentially has some kind of problem beyond just the firmware, or is more strongly affected by whatever flaw might be in GPS 2.08 |
OG, 11.XX, GPS 2.08, Ephemeris, First Acquisition |
NEW, 11.XX, GPS 2.08, Ephemeris, All Acquisitions |
Best-case scenario for Watch OG (first post-reboot) vs. Watch NEW on GPS 2.08 |
First post-reboot acquisition times on Watch OG are noticeably and statistically faster than average time on Watch NEW: 11s vs. 19s, p = 8.7 x 10-6 |
Watch OG first-post-reboot acquisition times are modestly and significantly faster than Watch NEW when both watches run GPS 2.08 |
NEW, 10.23, GPS 1.18, Ephemeris, First Acquisition |
NEW, 10.23, GPS 1.18, Ephemeris, Non-First Acquisitions |
After a reboot: first trial vs. non-first trials, GPS 1.18 on Watch NEW |
No significant difference in first-post-reboot times vs. non-first post-reboot signal acquisition times, p = 0.34 (small sample size) |
When Watch NEW had GPS 1.18, there was no significant difference between first-post-reboot and non-first acquisition times |
NEW, 11.XX, GPS 2.08, Ephemeris, First Acquisition |
NEW, 11.XX, GPS 2.08, Ephemeris, Non-First Acquisitions |
After a reboot: first trial vs. non-first trials, GPS 2.08 on Watch NEW |
When Watch NEW was upgraded, first post-reboot times are significantly faster than non-first, p = 0.00010 (small sample size) |
But when it was upgraded to GPS 2.08, non-first-post-reboot acquisition times became significantly slower! |
OG, 11.XX, GPS 2.08, Ephemeris, First Acquisition |
NEW, 11.XX, GPS 2.08, Ephemeris, First Acquisition |
Different watches, same firmware, first acquisitions only |
No significant difference: 12s vs. 13s, p = 0.094 |
The watches take the same amount of time to acquire GPS signal on the first connection post-reboot when on same GPS firmware |
Discussion
Broadly speaking, the following observations can be made:
- Watch OG is slower to connect on average, even when both devices have the same GPS firmware installed
- GPS connection times seem to be genuinely faster on GPS 1.18 vs. GPS 2.08
- For both watches, the first connection post-reboot is significantly faster than non-first connections on GPS 2.08. This was not true for Watch NEW on GPS 1.18, though the sample size was admittedly small on the GPS 1.18 tests; although I performed 24 connection trials, only 4 were following a reboot.
- If we consider that Watch NEW on GPS 1.18 was how fast a normal Fenix 8 ought to connect, roughly 8-10 seconds, then we can say that Watch OG, and Watch NEW for that matter, behave “normally” on the first connection post-reboot and then subsequently “forget” how to connect normally on subsequent connections.
My theory is that there is some flaw in the newer GPS firmware that causes the watch to inappropriately reference the Ephemeris file, or perhaps to always “cold connect” to satellites, except for that first connection following a reboot. It’s interesting that Watch NEW on GPS 2.08 seems to only connect in the 13-15s interval OR the 28-32s interval, and rarely in between. In other words, the connection times are “binned,” as if it’s using two separate protocols to connect, a fast one and a not-so-fast one.
It’s also interesting that Watch OG continues to have really long connection times that seem to occupy the whole spectrum between 15s and 2-3 minutes, although they do cluster around 45s.10s and 3+ minutes. I'm really sad I didn't have an opportunity to test Watch OG on the older GPS firmware. If it had behaved normally, I could be sure that it's a software issue and will likely be resolved. Hence my regret that I unknowingly updated it straightaway without testing first.
Conclusion
Sadly, I still don't really know what I'm going to do with my Fenix 8. My thoughts:
- It seems reasonable to me that a GPS firmware update might come down the pike which would make it always connect as fast as it does on first boot, roughly 10-12 seconds. I could absolutely live with that, even if it's still slower than my Epix Pro.
- I suppose I could always just reboot the watch in the morning so that it connects quickly for my workout that day and I might never encounter the super long wait times.
- On the other hand, Garmin might never fix it, or this watch might simply have some kind of hardware issue. The hassle of mailing it back to the dealer is annoying, but I might have to deal with it.
Weirdly, this all has me kind of appreciating my Epix Pro, which has always worked great and done everything I need without being problematic. Apart from the GPS issue, I have had a few other annoying little bugs on the Fenix 8 relating to touch and the backlight. These have been largely resolved by updates, but it's still kind of galling for a $1,200 watch.
Hopefully this post is helpful for others, maybe even the Garmin engineers, in developing a fix. I'm going to wait and see if a fix comes down the line before deciding whether to return the watch.