Garmin & Apple Watch – Best EVER GPS Accuracy Results
Regular readers here will know that I am certainly NOT an apologist of the mediocre accuracy of Garmin devices, especially when it comes to their GNSS/GPS accuracy. Whether or not you think I’m biased in some way, I don’t know. I think I call it how I find it and here I am about to call something I never thought I would have to call. Obviously, there will be caveats 😉
Background
The old Polar V800 and Suunto Ambit 3 were clearly the best-performing GPS sports devices, admittedly you could almost say they were from a different era. I still have the V800 and I do dust it off from time to time and it does still perform sweetly…sometimes…not all the time. Perhaps we have rose-tinted memories of accuracy that was never really there? An example of that was when I borrowed a friend’s Garmin Forerunner 305 a few years ago and it was not as good as I thought it was ‘back in the day’. But you can NOT apply that same logic to the V800. The V800 still performs for me, or at least the GPS does but the battery has seen better days!
The better-performing GNSS chips of yore were branded as SiRFstar. Maybe they were better-performing because they had better aerials or maybe it was because they used so much power to fix a decent lock onto a good signal. Maybe. Maybe there was some other reason.
That era segued into another era where we saw support for multiple constellations of GPS+GLONASS and GPS+Galileo, I seem to remember there was one sportswatch/bikenav that supported all 3 simultaneously but the memory of which one escapes me right now. However these multiple constellations promised much but, in my experience, repeatedly failed to deliver regardless of the watch brand.
Then came the Sony GNSS chip(s) which consumed significantly less energy than previous generations, whilst supporting multiple constellations. Going back a couple of years to when the Sony chips were launched, battery life was a big concern for watch manufacturers and the lure of a low-powered GNSS chip was too much for EVERY sportswatch maker to resist, so they all jumped ship to Sony…Garmin, Polar, Coros, Suunto…all of them. One big jump.
The Sony chip was immediately underwhelming and we first saw it in the Suunto 9 in June 2018. It wasn’t Suunto. Everyone had teething troubles.
Time passed and new firmware from Sony and new tweaks from the watch vendors promised much but delivered little. Polar introduced SBAS support for ground station aerials…any better? Not really.
Garmin tweaked some algorithms, I think it was last year in 2019, and suddenly the Open Water Swim (OWS) GPS tracks were awful. Jeez. GPS-only or GPS+GLONASS…both were ‘not good’.
However.
Things continued to change. After Garmin messed up with the OWS algorithm they soon introduced a fix and it was good. I started to see very good Open Water GPS+GLONASS tracks. Tracks that were as good as they needed to be. Let’s not forget that it is HARD to get a good GPS track when swimming as there is no signal whatsoever underwater. Cycling seemed better too but, well, that’s cycling. Getting a good GPS track is much easier when cycling 10m from buildings.
However, that change did not seem to filter through for me with better running GPS performance. Maybe it was sometimes better, maybe not. I’m not quite sure. I just know that I’d kinda given up on my Garmin Forerunner 945 and instead relied on STRYD for speed and distance and an HR strap for HR. I would often take my 945 on my GPS tests of other watches but it became an expensive data logger for HR, often in my back pocket and so I never bothered to really consider any changes to its GNSS accuracy a) because it takes a LOT of time and b) because it wouldn’t be fair with it in my back pocket.
Then the Forerunner 745 came out a few weeks back and my first standard 10-mile test (in sub-optimal conditions) again proved to me it was the same-old, same-old Garmin. Naturally, to be fair, it had to have another shot at my 10-mile test in optimal conditions. Now, things get hazy here as I performed lots of other tests on the FR745 in other conditions and accuracy seemed to get better but I was simultaneously testing the accuracy of the Apple Watch 6 and that was throwing out some surprisingly decent performances; naturally, therefore, it captured too much of my attention. There was also some new Garmin GPS firmware, v3.1 if memory serves correctly (GPS firmware NOT regular firmware). And around that fuzzy point, it seemed to be the case that Garmin had stopped getting ‘bad’ results. I kept being distracted by the Apple Watch 6 which sadly was GENERALLY not quite as good as it first appeared and distracted too by other tests and other life-related things but FINALLY, I got around to the final 10-mile test with the Garmin Forerunner 745.
10-Mile Test for GNSS/GPS Accuracy
The details of what I do in this ONE specific test are here (10-mile test methodology and results). The great thing is that you can compare all devices against each other over time – sure there are faults to that approach for this one test but there are obvious benefits too. All the source files are there if you want to do the comparisons yourself. Naturally, I do lots of other real-world tests too and always have.
Cut to the chase: The bottom line here is that the Garmin Forerunner 745 and Apple Watch 6 both scored 85%, which is the joint best-ever result, tied with the Ambit 3 Run. It also ties with the Coros Apex 46mm but I take that result as an outlier as it wasn’t so good at most other times in other conditions.
Supporter-only Content: Detailed Accuracy Verdict FR745. Free Content with accuracy-summary here in the Forerunner 745 Review.
Supporter-only Content: Detailed Accuracy Verdict Apple Watch 6.
Caveats
- This could be a one-off result. I’ve not especially looked at Garmin’s accuracy since that test.
- This could be a GPS firmware innovation and equally apply to other Garmin watches also on GPS firmware v3.X. If you have any feedback, below, I’m sure many will be interested to hear it.
- The positive results are almost certainly NOT due to favourable weather or satellite conditions (I record those in the spreadsheet!)
- The positive results are NOT from me being unduly lenient. For these results, I checked and checked again and was as fairly-harsh as I could be.
- For the Apple Watch 6, the prognosis is different. It has a strange smoothing algorithm that is sometimes apparently awesome (like in this test) at other times the smoothing still looks great but it’s applying the smoothing at a shifted location.
False Dawn?
I hope I’m wrong but I do not expect to continue to find consistently excellent results from top-end Garmin devices for running. More realistically there will be subtle overall improvements. Your post-workout, GPS track will be just that tiny bit prettier but your instant running pace could still easily be >10-15 secs/km wrong #BuyAFootpod
There are a couple of things that will be introduced over the next year that should improve GPS accuracy when running. The first being dual-band, dual-constellation setups with a new model of Sony chip, the downside will be a bigger battery hit but I can live with that.
more accurate Sony GNSS/GPS Chip for Garmin sports wearables
Great article! But since we cannot rely on paces provide by the watch’s GPS, is there another alternative like stryd which are precise (just for pace and distance) and not that expensive? I do not use power for running, but the instant pace in my Vantage M is not good either….
yes you cannot rely on the watch instant pace at all
you could try the zwift pod (was called Milestone pod)
https://the5krunner.com/2017/07/10/milestone-pod-review/
or this one from wahoo https://tidd.ly/2T2HrwY
there is the polar stride sensor too https://shrsl.com/2k74v
obviously the garmin one too but that is ANT+ only IIRC
maybe try ebay for a STRYD lite.
Thank you!
What are paces? Plural of pace I guess? The av should be OK. Maybe the actual 1 is a little relative but who cares? And the max doesn’t serve any goal either
Pace is very 20th century by the way, from the days we didn’t have live speed. And we were running on tracks…. And were drawing graphs with a pencil. I’m from those days.
I think I had this conversation with someone else before? I do a 5k training plan (see menus above) by far the minority of people ask for speed over pace (maybe 60% pace per km and 20% pace per mile maybe then split equally between mph and kmh). Pretty much 100% of the faster people ask for pace/km (i do not offer a track option for time/lap/400m). Neither makes you faster or slower and both rely on the same inputs just expressed differently at the end of the calculation.
I love drawing graphs with pencil.
First, english is not my native language. Second, tracks are the ideal since you have access to that, which is not my case (and that’s the reality of many people in here). And yes, in the past I used to only use a stopwatch and measure my distance by counting blocks and then using google earth to take distance in a public computer because I didn’t have enough money to just buy a sport watch or even a smartphone.
“This could be a GPS firmware innovation and equally apply to other Garmin watches also on GPS firmware v3.X. If you have any feedback, below, I’m sure many will be interested to hear it.”
In the last days my FR945 beeps every km exactly in the same moments of my old FR610. So now it seems very accurate. But in my experience the FR945 is sometimes very accurate, then it comes a new gps firmware and things get bad again, and so on. I hope that now that accuracy is good Garmin doesn’t mess things again.