Forerunner 935 GPS Accuracy/Performance

@FlipStone: not a reply to you as such, just some general observations. But I do believe people have unrealistic expectations when it comes to GPS watches in general

Tend to agree. My "benchmark" is does my track allow matching of all Strava segments (save for ones with terrible traces themselves) en route.

Very rare now to miss a segment...

If I am creating a segment I will review my runs/rides on it until find the "best" one but by and large with the 935 they have been good enough without delving back into my "back catalogue"!

Garmin Forerunner 630 w/HRM

I have been reading this thread with interest as I am having problems with my pace readout on my 630. So it would seem that the pace metric works better as speed goes up for a number of physics reasons.

Well, OK, I will buy that but it is very disappointing. One of the main reasons I bought the 630 was to see my pace. I am a casual user and do as much walking as I do running. I just got my 630 and only have 1 run and 2 walks so far, all at the same track and under good conditions weather wise. During the run, the pace (11 min/mi) was very stable during the whole run as indicated on the watch and also on the graph after the run. But the two walks were terrible in real time on the watch during the walks. It would jump up and down by 7 or 8 min/mi in the space of a few seconds while moving at a constant pace. This was everywhere on the track and over the entire course of 1.5 hours, all the time. After the walks, the avg pace (15:49/mi) on GC seemed about right.

But this doesn't do me much good. I usually have a slow warm up mile and a slow cool down mile and want to know my actual real time pace on the miles in between. I can't even rely on the average because the slow paces skew it. So it looks like for walking at least, I have no way to get my real time pace.

I understand that the GPS metrics might be effected by speed, but it looks like a difference in about 4 min/mi between walking and running makes real time pace perfectly useable during running but renders it more or less useless for walking. That seem like an awfully high sensitivity of speed to useable results.

I know! I will speed up my walking to 13 min/mi........................................

Not the worst GPS tracks I've had, but almost as bad as the F2 I had a few years ago. Total distance seems OK, but have seen a few strange deviations on tracks. We'll see how things go, but the F5X seems much more reliable to me, for now.

toddsdn - you might want to try looking at lap pace in combination with current pace to get a good idea of how fast you are going at walking speeds.

There is the also the question of where you are doing your activities. If near tall buildings or under trees then the fluctuations in current pace at slower speeds will probably be more noticeable.

Here is some data for people who want to use the watch for hiking.

Hiking last Saturday - this was a tour I guided for a friend:
Elevation data from barometer was very accurate. The difference in up/down is due to the fact that I stopped the watch at the peak and restarted it some meters below. Conditions were perfect for barometer as there was stable sunny weather.
GPS data shows some distortions, especially in the first part as the path is densily covered with trees and passes some close steep rock formations. With tree cover only (no rock walls) the watch did a good job, sometimes putting me some meters away from the actual path. Being a hiker, it doesn't really bother me. The upper part to the peak is clear open sky and the watch did a great job as expected.
Settings were GPS 1sec recording. Software 2.40

http://www.mygpsfiles.com/app/#utBH9pex

Climbing/Hiking last Sunday:
Again elevation data was very accurate. Same conditions as Saturday, stable weather, clear blue sky.
GPS puts me away from the path by a few meters in tree covered areas, however, it follows the many turns quite accurately. When coming closer to the wall (which is about 450m high, very steep) the GPS starts to go crazy quite a few times. I don't really care about the GPS in that condition as it behaved ok on the hiking passages and only struggled when climbing (and being very close to the wall).
The return from the peak seems to be quite ok, besides the canyon you need to pass. This is a 60-70m wide canyon with steep rocky walls on both sides (I would guess a couple of hundred meters high), so no wonder GPS went crazy here as well. Heart rate seemed to be off in the first hour, then it got better.
Settings were GPS+Glonass 1sec. Software 3.30

http://www.mygpsfiles.com/app/#ufhbHfVD

Summing up:
Barometer did very well. GPS altitude would have given exagerated numbers. I will have to see how it goes with changing weather conditions. In this case I might recalibrate on the peak if altitude is known or GPS signal is strong.
GPS did ok, not great, not bad either. It struggled were i would expect it to struggle. Fast turns don't seem to be a big issue, but the watch put me away by a few meters from the actual path when tree cover was dense. I used hiking sticks for the "walking" parts, so I am not sure how much that affects accuracy as the watch is pointing differently to the sky. In any case, I hope and expect Garmin to improve accuracy with future updates, right now I am quite happy with it. Still need to validate when running, will post those results as well.
Having owned the Fenix 3 when it came out and returned due to bad performance, I would say this watch does much better (only compared to initial F3 firmware, cannot say anything about later updates). I am also not sure whether software 3.3 did better than 2.4. In terms of heart rate accuracy I would think so.
Will also try out smart recording, maybe that provides better tracks for hiking, although I think they are not too wonky as a hiking path is rarely straight.

Will do some more hiking this weekend and also test the watch on my usual running track, for which I have some other tracks to compare it to.

toddsdn - you might want to try looking at lap pace in combination with current pace to get a good idea of how fast you are going at walking speeds.

There is the also the question of where you are doing your activities. If near tall buildings or under trees then the fluctuations in current pace at slower speeds will probably be more noticeable.

Thanks for the idea. Yes, I have found a definite degradation doing walks at the park track as far as the pace goes. Running is no problem with the pace. I have found that I can live with just checking the pace every few minutes while walking as it averages out good enough. I am just walking for exercise and want to maintain a constant and somewhat known pace. Hell, I am finding more and more that I rely on keeping in a heart rate zone anyway so it's not a big deal. I am glad to see though that the pace works nicely and smoothly when I run because then I really do want to know my pace accurately at all times.

Not the worst GPS tracks I've had, but almost as bad as the F2 I had a few years ago. Total distance seems OK, but have seen a few strange deviations on tracks. We'll see how things go, but the F5X seems much more reliable to me, for now.

Something must be wrong with your 935. I also have both watches and the 5X is far inferior from a GPS tracking perspective. In addition, the 935 locks on to GPS almost instantaneously, while the 5X takes much longer. Pace is also much better on the former, which goes without saying due to its accurate GPS. Of course, having a footpod alleviates that issue.

To me the iPhone track is more wonky than the 935 which ought to add false distance but are both on every second recording for a direct comparison?

I think discussions of "accurate" distance are interesting but I would more trust a professsionally measured course than an average of GPS traces whose accuracy, especially in these sort of environments, we are discussing in the first place. Bit like calories there always seems a tendency to think the higher value must be right because that is what we hope for.

To me the iPhone track is more wonky than the 935 which ought to add false distance but are both on every second recording for a direct comparison?

I think discussions of "accurate" distance are interesting but I would more trust a professsionally measured course than an average of GPS traces whose accuracy, especially in these sort of environments, we are discussing in the first place. Bit like calories there always seems a tendency to think the higher value must be right because that is what we hope for.

I cannot find anything in the Runtastic App settings which explains what time interval is used for recording. I don't know if the iPhone has a standard setting for that. If anyone knows where to look for it I am happy to do so.
FR 935 is on GPS+Glonass and 1 sec. I might also try with GPS only in the future and see if Glonass has helped with this challenging track.

I agree with you, I would just have guessed after >100 runs of this track the average distance captured by 3 different iphones would probably reflect the "real value". But you are right, I have no proof.
To the calories comparison, I would hope the 935 is right and not the other way around :o I'll see if I can borrow some other devices to compare.