Suunto Ambit3 Peak Trail Test

Original Review date: August 2015. Tested with Suunto Smart HR Strap and 1.5 – 2.4.1 firmware (updated August 2017)

While data collection pretty much finished up in June 2016, we’ve updated the feature set with more recent firmware and re-analysed the data. The comparison watch group has also been updated to include the Spartan Ultra.

Summary

The Ambit3 Peak, referred to as the Ambit3 from here, sets the benchmark for accuracy and reliability and endurance (with around 10,000km on the wrist so far). Rather similar to the Ambit2 only better, as it adds memory, mobile uploads, running metrics, and recovery tests, and activity tracking. The Ambit3 came out on top of the ‘high accuracy’ battery exhaustion trail run test at 15.5hrs with 19% battery remaining. The Movescount mobile app is stable these days, hopefully the past pain is nothing but history.

A current (ie. now dated) model round-up can be seen here, and the testing methodology is here

What kind of trail runner is it good for?

Any trail runner who values accuracy, reliability, and endurance above all else. Though like the previous Ambit2, runners need keen eyesight and hearing to accommodate the limited display and alert functions. Running mappers wanting to enhance OSM or other map data may also be better served elsewhere as you can’t manage or export your waypoints. Those needing 400m precision splits may also want to look elsewhere as lapping is a bit laggy.

What should you expect as a long term running companion?

A decent marriage, though maybe not quite running through till retirement as the Ambit3 is so good generally that it magnifies the few obvious issues it does have. Mostly the list is a carry-over of issues from the Ambit2 (limited display legibility, charging clip, no vibrate, waypoint management, no distance remaining/ETA when following a route). Oh, and the buttons aren’t so responsive, so give up any notion of accurate 400m splits. All that said the watch itself will be faithful and predictable on the trail, only showing any real accuracy issues in the worst of conditions (after others have failed).

Good Stuff

If not already obvious from the summary, the Ambit3 stands alone in terms of it’s positional, distance, altimeter accuracy, and endurance in our testing to date. Like the Ambit2 (you’ll be seeing that a lot in this review), it never had a day when it should have stayed at home, data recording was consistently good across all testing. Similarly, the concept of freezes, corruption, and resets during running activities is foreign to the Ambit3. Breadcrumb navigation allows waypoint proximity alerts keep you on track when brain fade sets in. The running metrics provided do an excellent job at feeding back performance and overtraining.

Bad Stuff

Lack of vibrate alert, data/charging-clip design, tiny/illegible secondary display fields, spongy and laggy buttons,  and lack of proper waypoint management top the list (yes, all unchanged since the Ambit2). Locked-down Movescount Suunto Apps is still an annoyance, be nice to see Suunto open up a little to get a more vibrant niche developer community generally. Historically both the Movescount webservice and Moveslink2 have been a little unreliable, though 2015-2017 has been solid so maybe that’s all behind us. Navigation also a la Ambit2, still very basic, though still better than the rest. Distance and ETA on waypoints that respect plotted course distance seems a pretty obvious omission.

Plea to Suunto

As per Ambit2 – vibrate alert please, a new clip design, and make the display options a little more flexible so your trail runners can actually read it while ripping up the trail. Give the buttons a bit of feel and kill the button-press lag factor. A GPS accuracy setting between Good and OK to give a 50hr endurance would be stellar. Movescount needs better tools to manage waypoints. More encouragement for 3rd party developers to move things along in terms of direct watch interfacing (mobile or desktop) and on watch apps especially given Suunto’s ongoing attempts at Android. Now that the Android Movescount app is actually working for some at least, how about adding lap splits to it? Navigation could also be rather more helpful, eg. distance/eta along plotted course.

Formal Testing

Accuracy

So far Ambit3 formal trail testing has covered over 560km on 66 separate days on the surveyed courses. A number of the activities were also specifically planned to assess the impact of poor satellite coverage (via tools that give Geometric Dilution of Precision). Potential GDoP was logged for every activity to better make comparisons across models.

The GPS pool against which the Ambit3  is assessed includes the following models: Garmin’s fēnix2, fēnix 3, and FR910XT, Suunto’s Ambit2 and Spartan Ultra, Polar’s V800, and a Sony Xperia Z5c running the SportsTracker app.

During testing multiple GPS devices are carried as a means of validation, if all units have a bad GPS day then a tin foil hat is donned. Though some days with poor GDoP potential saw all units a little degraded, no testing days saw GPS chaos across all models. Post FW upgrade outings we often a bit of a write off and were discarded.

Distance

The Ambit3 is leads the pack in accuracy. Overall  it registered 98.7% of the total distance, meaning it’d short you around 1.3% in your long-term training log if you run in conditions similar to ours. On a run by run basis in easy to mixed GPS conditions its performance is excellent with the best scores to date coming in at 99.2% and 98.9%  respectively. By comparison the test group averages were 98.5% and 98.2%.

Under difficult conditions the Ambit3 drops to second place at 96.5% to the (somewhat dubious) top performer on 96.6%, a little above the group average of 95.9% (yes the margins are tight). Putting it together as an overall accuracy score (accounting for differences in time in the various conditions) the Ambit3 came out top at 98.2%

Ambit3.abs.mean.accuracy

At this point you might be thinking those margins are pretty small, practically the same even. But there’s another way to look at accuracy where we call each recorded distance within 1% of the truth as a success. Think of it this way – you’ve just run a 10km PB and you want your glory on Strava or the online service of your choice. But your watch shows 9.9km at the finish, so no digital PB, and no bragging rights for you. So even a 1% error can be a bummer, let alone 4%.

Our ‘within 1% success test’ is much more sensitive in identifying how reliable each recorded distance run is. The differences between models now becomes a lot more pronounced. Here the Ambit3 recorded within 1% of the true distance 82.6% of the time in easy conditions compared with 66.7% for the second place getter and 47.3% for the group average.

This plummets for all models under difficult conditions to 10% for the Ambit3, 25.7% for the best performer (again this is dubious – we’ll get into why in a later roundup), and 5.6% for the group. In all, at 53.1% across combined conditions the Ambit3 comes out on top, second place is back at 35.8% with the group average at 33.4%.

Ambit3.99.PC

In challenging GPS condition outings (ie. typical NZ bush conditions) the Ambit3 comes in around the expected distances, though likely drops a couple of percent – hard to know till we properly survey the course. Though the true course distances aren’t not known, tracks largely follow contours and features, though off-course shadowing is seen especially against hard vertical faces.

Distance Accuracy Note: aside from the initial overall figure, which tells us if will run long or short on long term averages, we report distance accuracy using absolute percentages. In other words 5% wrong is 5% wrong, it doesn’t matter if it’s long or short, they don’t cancel each other out during your run.

What gives

When we plug all the factors we record into a model to show us which are associated with distance accuracy we can see the Ambit3 is strongly effected tree cover, but resilient to satellite availability and track bendiness. On the other hand the raw GPX feed accuracy and median track offset is predictive of good distance performance. So while accuracy will decline as the forest and canyons close in, the watch performance is rather more resilient to conditions than others.

Positional

The positional accuracy of the Ambit3 was pretty impressive across all conditions encountered. In unobstructed GPS conditions the median offset was 1.3 metres (mean 1.6). The Ambit3 also had the best performance in limited survey conditions with a median offset of 1.7m (mean 2.2m).

Ambit3.gpx.offset
Lower is better

Note: positional accuracy is measured as GPS trackpoint to the closest point of a surveyed track (or repeat track where course is un-surveyed), so figures reported here will be a lot better than the actual positional accuracy.

Satellite vs Watch Distance

And for the nerdier types among you, here’s how the watch recorded distance lines up with it’s raw satellite distance. The charts also show how each and every recorded lap distance pans out over the testing.

The recorded distance you see on the watch is in purple whereas the yellow is from the unfiltered gpx track. The vertical lines are the averages of the two. Our general performance expectation is for the filtered recorded distance to be a tighter distribution that pulls the average distance closer to 100%. And indeed this is what happens with the best performing watch we’ve tested.

We can see Suunto’s filtering/smoothing work to pull up the raw satellite data towards the true distance. Also notable are the generally conservative and tight distance distributions – though things are starting to go south in difficult conditions.

ambit3_dist

Ambit3 distribution of accuracy against unsurveyed courses

The Ambit3’s performance in limited conditions is further demonstrated in truly difficult conditions courses (heavy tree cover, steep terrain, gorges) where it is the best performer so far. Repeats a the 8km circuit had a median relative offset of 3.75m (mean 4.7m) with 19% of points greater than 10m.

Variance in accuracy over surveyed testing overall was solid meaning that not only was the average error low, but also stable. There were no instances of really bad GPS days so accuracy appears pretty predictable.

Track Appearance

The Ambit3 ‘best’ GPS setting appears to record at  pretty similar rate to the Garmin 1 second sampling rate. Visible random scatter was pretty contained. Conspicuous track shadowing was minor or absent in all but the most challenging GPS conditions, and as already noted, most commonly seen where a course would run parallel to hard steep/vertical surfaces.

The trackpoint cloud image below shows the density of GPS trackpoints from all runs against part of surveyed course. Both the limited (track on left) and relatively open sky (tracks on right) conditions on the right show nice, dense, symmetrical distributions of points near the track.

Ambit3_cloud_2017

Overall, tracks were well detailed and responsive to rapid changes in direction. In short, trail behaviour was excellent. Though again edged out by the Ambit2 in really difficult conditions.

Elevation

Elevation was tested with the FusedAlti setting on, this reduces the amount of barometric drift by quite a lot. Sure, it’s not perfect, but it’s enough not to be jarring when browsing the watch’s elevation plot. Regardless of FusedAlti it’s always best to manually calibrate elevation at the start of a run which is easy do.

Elevation range, when calibrated at start, is good. Total climb on the surveyed course was reasonable, coming in at an overall average of around 93%. Variation was also pretty good, on the worst day it came in at 79% and maxed out at 101%. While elevation is not quite the level of positional or distance accuracy it still is the best of the models tested so far.

Note: The survey course probably is a bit tough on GPS watches with respect to elevation with lots of rapid 10-30m climbs/descents. Watches would probably perform better on longer more pronounced climbs/descents.

Battery Saving Data Recording

Virtually identical to the Ambit2 if you’ve read that review you can skip this – the Ambit3 GPS can be set as ‘Good’ or ‘Ok’ reducing GPS sampling to 5 or 60 seconds respectively. This increases battery endurance from a Suunto stated 20hrs to 30hr or 200hrs(!). You can also set the recording interval for non-GPS data to 1 or 10 seconds. Without a footpod the distance obviously will come in short on the Good/Ok settings. How much short depends on how fast you are moving and how twisty the course is.

On a 161km mountain race (Northburn100) the distance came in about 10% short when set to the 60sec GPS setting, 10sec data recording and no footpod. On the up side it had about 72% battery remaining after 32hrs. And yes I was going that slow. Maybe I should have selected the 5sec sample and gone a bit faster to go inside of 30hrs :). An intermediate setting between ‘Good’ and ‘Ok’ would be nice here.

However, as the Ambit3 will update pace and distance from a footpod, its possible for the watch recorded distance to be comparable to the most accurate GPS setting. A calibrated footpod is said to be 99% accurate in ideal (road) conditions. The Ambit3 has shown itself to be 99% accurate distance-wise over mixed, but not overly difficult, trail conditions. Both footpod and GPS accuracy degrade under tougher terrain conditions, so until we see some data or do some mixed terrain footpod testing ourselves, it’s moot and we’ll call it a draw.

While the watch recorded distance may not be impacted by lower GPS sampling, a few things are. First, Strava works from the GPX data (ie. GPS positions) and it doesn’t respect your footpod, so it’s going to stiff you in terms of distance. Second your GPX track isn’t going to be as detailed.

Third, and most critically, the Ambit3 navigation doesn’t let you use reduced GPS sampling. If you choose an activity mode that uses reduced sampling and then turn on navigation to follow a course, the Ambit3 reverts back to high battery burn GPS mode. You can turn on navigation on only in times of need, but it’s a bit of a pain (you have to remember to turn it off after – and if you are on a 24hr mission your memory is not always the best).

Functional Tests

Data Syncing and Sharing

The Android app (version 1.2.6) was finally proving reliable for syncing on an Android 5.1.1 handset (Xperia Z5c), though older handsets may not be so lucky. That was a long, painful 6 months of public beta development. Lets not do that again. Apparently Apple stuff works well enough. [NASTY] Update version 1.2.7 of the Android app regressed and no longer synced. Version 1.2.8 needs daily repairing for it to work correctly. We are now on version 1.3.1 with watch FW2.2.16 and it now appears reliable for us.

On the PC side (again can’t speak for apple), once the Moveslink2 computer software pulls the data off your Ambit3 it sits in a hidden folder in the SML format which is a flavour of XML (like GPX, TCX, or KML). This leaves it able to be picked up by other desktop apps (eg. SportTracks) that recognise the format. The Ambit3 itself is invisible to the computer and other applications (ie. it doesn’t appear as a mounted device). Conversely if you use the mobile app to sync, you’ll have to manually download it from the Movescount site if you want to see it locally on the PC.

As well as activities, the Moveslink2 app also syncs watch settings back from the website to your Ambit3 (more on this in the general usability section). Syncing can be set to update a number of other sites like Strava and Training Peaks and you can also export individual activities in various formats including GPX, FIT, and TCX.

And a third party tool exists (MXActivityMover) that let you upload Garmin Connect activities (and TCX) files to Movescount for a reasonable one off license fee. Nice if you want to keep your stuff together. Haven’t seen any officially endorsed service that will suck your Movescount activities out in bulk, though it appears to be possible. And if you sync from a single computer, all the data is there in SML files).

Oh, and the OpenAmbit software that the Linux enthusiasts have cooked up for the Ambit1/Ambit2 doesn’t work with the Ambit3 series.

Interestingly we are now seeing a glonass.7d file in addition to the sgee.7d file in the hidden Moveslink directory, suggesting GLONASS satellites may now be in play. The GPS chip is capable, but it certainly doesn’t appear on the Suunto spec sheet. So it’s probably a file generated for the Suunto Traverse and other capable models only.

Navigation Functions

Again identical to the Ambit2 skip this section if you want – creating a breadcrumb course is easily done via the Movescount site using either Google Maps or OpenStreetMap. Both have an auto-follow roads function when drawing your course, and with the OSM version this includes trails. You can also import GPX files to Movescount if you’ve already got the course (though not any waypoints associated with the course).

And regarding waypoints (points of interest) there are two types which are managed and used separately – waypoints on a course and standalone. Course waypoints must be part of the course and only visible when navigating the course. Standalone waypoints are just as they sound.

You can create waypoints via Movescount or on the watch itself (if you know the coordinates). Thankfully Suunto have switched from Google to OpenStreetMap for waypoint creation as Google was hopeless for trail detail. But you still can’t export the waypoints you’ve collected on the watch at all, which makes OpenStreetMap editing or waypoint sharing painful.

You navigate to a selected standalone waypoint by line of sight bearings and distance, then manually select the next waypoint to navigate to if required. Unfortunately, when navigating a course there is no functionality to give distance to go or ETA against the course itself.

New for 2.4.1 firmware is the inclusion of elevation profiles into the planned routes, so we can see what climbs are coming up. Also included (new?) are alerts where we are heading off course which are most welcome.

Oh and turning on navigation stops any GPS power saving so you are back to a 20hr battery life (max). If we use a map and compass we don’t check it every second, so not sure why navigation forces 1sec GPS updates where I’ve set the activity to update less frequently. Though an update to current position via a button press might be good when we are seriously disorientated.

Pacing Functions

Ala Ambit2 – again skip if required. The Ambit3 doesn’t actually doesn’t have any pacing function as standard, but plenty are to be found as watch apps via the Movescount site (see the General Usability section for Apps rundown). The two types of pacing we’ve used are constant pace, and race time. Constant pace is what it sounds like, it measures your time or distance ahead/behind of where you’d be running at that set pace.

Alternatively race time is great for an actual race, these apps basically forecast your finish time on the basis of distance to finish and some kind of weighted pace function. You start running and it’ll give you an estimated race time if you keep it up – saves trying to do pace arithmetic during a marathon.

The problem with both of these types of pacing is that neither respect the changes in terrain you are likely to face on the trail. If your finish has a big hill at the finish, you’d better be banking some time and these kind of pace apps won’t help you with that. As far as we know there are no Suunto apps that allow for variable pace, either synthetic, or past performance which would be much more useful on the trail.

Battery Run Down

In testing the 14hr standard run down the Ambit3 came out on top of the test with 19% battery remaining after 15.5hr of trail running. At that rate of battery drain it looks like the Ambit3 would have lasted just over 19hrs (ie.  a little short of the 20hr on the box rating).

GPS setting was set to best, HRM was on, and smart notifications off. Though unlike the other watches tested, the Ambit3 was responsible for navigation duties – so had navigation on and was frequently referenced (was a tricky course). The Ambit3 used was about 1 year old at time of testing.

As stated in an earlier section, set to the ‘OK’ GPS accuracy setting and 10sec data recording the Ambit3 went 32hrs and ended with 72% battery remaining. Though it did drop ~10% of the supposed event distance. We’re guessing the ‘Good’ GPS accuracy setting ought to be good for 24hrs then (and will update when we get a chance).

Standard Feature Set

General Trail Running

  • GPS accuracy under canopy : Excellent
  • Consistent GPS performance : Yes
  • Rapid GPS Acquisition: Excellent
  • HRM : Yes
  • Cadence option: Yes
  • Battery 8hr (with HRM) : Yes
  • Barometer (for altitude) : Yes
  • Basic breadcrumb and waypoint navigation : Yes
  • Vibration alerts : No
  • Trail legible display: Poor
  • Open data access : Yes (except waypoints)
  • Sensor Standards Compliant : Yes

Ultra Feature Set (as per trail running plus)

  • Battery 14hr+ with HRM and high accuracy recording : Yes
  • Battery 24hr+ with HRM and down-sampling : Yes
  • Electronic compass : Yes

Nice to Have Features

  • Mobile uploads (Android/Apple) : Yes (Android reliable with 5.1.1)
  • Cadence (without footpod) : Yes
  • HRV (R-R) recording with recovery estimate/test : Yes
  • Footpod GPS override option : Yes (always overrides when footpod present)
  • Basic interval workout ability : Yes
  • Pacing function : Yes (via apps)
  • Position/waypoint autolapping : No
  • Feed/drink or run/walk timing reminders : Yes (via apps)
  • Everyday watch : Yes
    • Activity Tracking : Yes
    • Mobile Notifications : Yes

Non-Tested Core Features

While functions like heart rate and cadence are core, we’ve got no means of objectively testing them.

Nothing out of the ordinary seen on either heart rate or cadence behaviour. Running with a footpod showed no obvious difference to that of running with the Ambit3 alone regarding cadence. You pole runners should note though the wrist based accelerometer gets a bit confused when you use running poles (footpod works fine though). The footpod can and should be manually calibrated for distance. As was mentioned in the battery saving section, the performance of the footpod in trail condition is a bit of an unknown and should be tested in various conditions at some point.

Activity tracking is still watch only, no health type features on the Movescount website and no uploading activity/steps/sleep to other services. The performance level/difference features appear a pretty handy measure of long term running efficiency and current recovery/performance. These provide feedback for long term improvement and in-run performance in an intuitive way not seen in other GPS models.

General Usability

See if you can spot the difference from the Ambit2 review: Ambit3 audible alerts are pretty clear, though there are plenty of situations where we don’t want or can’t hear audible alerts (wind noise, bunch runs with everyone beeping, winter jackets, old age). The lack of a vibrate alert is puzzling, even if it does impact battery life.

The screen is reasonably sized but washed out compared to some, though the font is a litter crisper on the Ambit3 than the Ambit2. We found that while the dark display looks snazzier, the light display is more legible in daylight (unless you accidentally hit the backlight button). Conversely the dark display works better at night with the backlight.

The layout only allows for one large (ie. readable) metric on display at any one time. All of us have trouble reading the little figures at the top and bottom of the watch when running race pace, trail, or after 12hrs on the feet. It would be nice if layout was a bit more flexible to allow larger font sizes.

Current pace is rounded to nearest 5 seconds which may be an annoyance to roadies who want a more precise number to follow. We quite like the smoothed pace figure as it stops rabbit chasing. If you are want to run precise pace then average pace is shown to the nearest second, and lapping to capture intervals enables precision running. There are also custom apps which create nice smoothed current pace.

Time to GPS fix is as quick (maybe quicker) as you’d expect for a unit that pre-caches satellite availability, when recently synced and in the same area as last used we talking a few seconds. The Ambit3 is a departure from previous Ambits in adopting the BLE (bluetooth) standard. While this means no more ANT+ compatibility, the BLE standard is fine for running sensors. We’ve used Polar and Adidas sensors without a problem. Unfortunately though, while bluetooth is capable of letting the watch know when sensor batteries are running low, it doesn’t appear that Suunto have implemented this function in the the Ambit3. Battery life of the bluetooth sensors seems to be a bit shorter.

What definitely is not good is the unchanged data/charging-clip design. The Ambit2 one had to be replaced after 8 months as the contact pins became permanently recessed. Evidently this seems to happen a bit, and Suunto did a warranty replacement for this. Not sure why a watch with great build quality is shipped with such an average charger design.

A feature of the Ambit series is the antennae ‘wing’ built into the strap, we’ll take a punt here and credit that as part of the reason it fares so well in terms of accuracy – so it can stay. Buttons are pretty logical, and are context dependent, doing different things on different screens and modes in a pretty direct way.

Button feel is a bit poor though, kind of squishy with no positive feedback. There is also a lag on the lap button which frequently results in frustrating double lapping when running fast intervals (is more likely an issue for roadies wanting precision splits). And if you are running very short intervals the split times are a bit hopeless.

Like the Ambit2 set-up and configuration is largely done via the Movescount site or App which is then synced back to the watch. This makes watch setup fast and easy compared with fiddling with the watch itself. Be great to leverage this a bit more on the computer at least, copying and moving display screens for example or sharing setup with other watches or users.

Those who’ve been using Ambit watches for a couple of years will know that historically both Moveslink2 (the computer syncing application) and the Movescount site have had a few issues with reliability. During formal testing we’ve had no issues but the Ambit2/3 watch owners fondly recalls the multiple day outage of Xmas 2014 (and the aforementioned issues with Moveslink2). Hopefully Suunto have upped their game on server resilience and app reliability.

Another aspect of the Ambit3 that causes a bit of pain, if only very infrequently is the fact that firmware updates wipe the watch history/recovery data. As the Ambit3 has only had three updates to date it wasn’t too disruptive. It’s also worth noting that the performance of early Ambit3 firmware releases were solid performers, the updates simply added new features. NOTE: the last FW2.2.16 update in July 2016 kept history/recovery data.

The core Ambit3 functionality is padded out via custom watch Apps available on the Movescount site. Many useful if not core functions are provided via app by Suunto and Joe public alike. Good examples being race pacing, current incline, and tides (for coastal running). Though annoyingly the Suunto developed apps are locked down and unable to be adapted by mere mortals. If you don’t live in a region covered by the Suunto tidal app, tough (unless you can figure out how to write it from scratch). The user apps by contrast generally have their code public so it can be adapted as required.

Some additional functionality like advanced workouts, and mobile display mirroring (and mobile uploads as mentioned previously) are available via the Movescount mobile app. We haven’t really played with them much, though they do work.

Like shoes and backpacks, strap comfort is a personal thing – try before you buy.

Long Term Verdict

After +10,000km of use the Ambit3 has been reliable+ during all running activities. Never had a freeze, reset, unexpected battery drain, or unusable data tracks or distance. We’ve done plenty of high intensity trail races, the odd full day hike, and a good year’s worth of run and bike commutes. We do occasionally see some obvious GPS shadowing, though this invariably is in sub-optimal conditions.

The one issue we did find during long term testing was the 99 waypoint bug. Collecting 100 or more waypoints prevented the Ambit3 from sycning until the waypoints were manually deleted from the watch. Update 2.0.45 appears to have fixed this.

Not sure what triggers it but we also experience intermittant loss of paired sensors. Have had to repair the HRM a few times now. Wierd.

Looks like we are finally through the worst of the pain on the Android Movescount app too, though version 1.2.27 and 1.2.28 have gone backwards a bit in connectivity. That was a pretty ugly ‘open-beta’ process but it seems Suunto did cough up enough resources to get over the line, for a time at least. Still not a rosy pic in Suunto-Android land.

And our watch strap has finally given out after ~10,000km on the wrist. Otherwise the Ambit3 is still solid.

Advertisements