UltraTrac on 945 VERY, incredibly inaccurate (cycling) for activities?

UltraTrac accumulates distance by toggling between GPS and WDR (accelerometer) as it progresses through the satellite active cycle.  While GPS is 'on' it will use that as the source for calculating distance traveled.  While it is 'off' it will SWITCH to the cadence of your arm swing as the source.  It will NOT connect the position between GPS reads and try to determine distance.  It's a simple toggle.  Visualize running on a treadmill while holding the handrails.  You won't accumulate accurate distance.  It's not that the watch assumes you're not moving, it's that your arm swing isn't providing the feedback of movement.  Same on a bike.  Or driving in a car.  Or any motion where you're not simulating the typical arm swing that it needs to accumulate distance.

As such, I can't imagine an implementation that would be useful while cycling.  Probably not the intended vision for UltraTrac.  I've found it to be quite accurate when running and use it frequently.  Distance accumulation is dependent on the calibration of the accelerometer and so that's somewhat out of your control.  For me that implies while I consider it very trustworthy for running, my results would likely be suspect while hiking.  I just find it extremely reliable for my use case - trail running.

You could build a compelling case that distance accumulation should operate differently while cycling since it leverages additional sensors (compass, gyroscope) to simulate GPS position while creating your 'track'.  This is really only used for course navigation in UltraTrac mode however and will not impact distance.  Again, that's by design and isn't really a 'bug'.  If the design would change for cycling then that would potentially 'break' running as you would lose the advantage of using WDR for distance accumulation.

I guess the most viable solution for extended cycling tours is a portable power bank.  That seems like a fairly logical solution to address capturing activities that span a longer duration that what you're able to record on a single charge.

I should add that I use UltraTrac with the Fenix 3 and not the 945 but the basic execution of distance accumulation is standard across all devices.  I wish I had a 945 to assist in additional testing but it seems like your results (and Flip's) speak for themselves so hopefully this is helpful in explaining how it works in practice.

Has anyone tried what happens if you have a speed sensor on your bike?  Not sure if it disables Ant+ sensors but they should be extremely low energy compared to GPS, so it shouldn't.  Similarly, I'm guessing running with a footpod would give you more accurate distance while using ultra-trac mode than just the in watch sensors.

Why don't they clarify this then?
The way it's "billed" is simply as a "bit less accurate GPS tracking", for an activity, which would imply almost ANY activity, right?

I'm starting to see the picture (and debating on returning my 945, going back to my VA3/Ambit3 combo, until something better exists).

Yeah, I could carry an extended power bank, I suppose, but the whole idea of UltraTrac (in my mind anyway) was usability like my Ambit3Peak, where I could get 120 hours of pretty-darn-accurate data, from an activity such as cycling (Suunto seems to have figured out the correct algorithm for this some years ago).  It's one less thing to deal with, after a long day in the saddle, setting up camp, cooking dinner, etc, and then to have remember to charge my watch. 

It'd be about every-day or every-other-day right now, I'm getting around 3-4%/hour drain, on regular GPS (just GPS, on Galileo or Glonass) with the built-in cycling activity, and around 5-6%, in "UltraTrac - "battery-saving mode".

So yeah, I'm a bit disappointed, having moved from a VA3, thinking I'd ditch my old Ambit3, for the 945, and be able to use it for at least a 5-7 days, when touring, not just local rides...

I think Suunto should be MUCH more forthcoming, about UltraTrac; the way it's sold is that it "simply saves battery on (ALL) GPS activities", and doesn't say something like "*only GPS activities <= 4mph, or something along those lines...). 
Yeah, it would kind of dumb, but at least people wouldn't be buying it, expecting to use it as such, or needing to read reviews of the 935 and UltraTrac, to "figure out" that it's not going to work for most GPS scenarios.

Yep, this is the secondary issue with UltraTrac, currently, on the 945, I'm pretty sure (I don't have a big enough sample-size to be definitive, at this point) it's using MORE battery, quite a bit more, than just a regular (non-UltraTrac) GPS activity, wihch seems COMPLETELY counter to how it's described/sold, right?

Thanks for collaborating on this, I'm wondering if there's a reasonable way to "file a bug", for at least the battery-life issue?

I've done some searches, but I'm not finding a mechanism, other than to call CS, which seems like it'll mostly be a "big fat waste of everyone's time" (mine and theirs)?

At the LEAST, I'd sure like to see them fix the battery issue, and perhaps make it work better, or close to the 935 functionality anyway.
I have a friend with a 935, an Ironwoman competitor, so she takes her tracking pretty seriously, whom I asked about this, before starting a ride Thursday.  She'd used Ultratrac on her 935, and some "some inaccuracy" similar to the 20% or so others have reported, but nothing like the 945, where it appears to be about 25-35% accurate, if moving >4mph or so...

Yeah, I thought about this, a bit, but I'd have to get a "bike pod", for every bike I wanted to use UltraTrac on, which seems a bit inane (I have a few bikes, some might never be a candidate), at 40/pop.

If Garmin meant that "if you want to use UltraTrac at >3-4mph, or so, you should buy a xxx-Pod", they should say that, so one could take that into consideration, when buying an already crazy-expensive watch, that now you need to invest in a bunch of "accessories", in order to use the (seemingly independent) GPS mode (UltraTrac),

My Ambit3Peak does this, with zero issues, in either the 50/120 modes, it still tracks my distances, VERY close to as if I were in regular GPS mode (I'd say the 50 is maybe 2-3% off, and the 120 is 3-5%, if that).
I'm okay with Garmin not being at that level, I guess, but they make it sound like it is, in the "sale statement", IMHO: 

"The UltraTrac feature is a GPS setting that records track points and sensor data less frequently. Enabling the UltraTrac feature increases battery life but decreases the quality of recorded activities. You should use the UltraTrac feature for activities that demand longer battery life and for which frequent sensor data updates are less important."

This statement does not say to me "you need more sensors for UltraTrac", or "you will get 20-30% accuracy using UltraTrac, if moving faster than a walk"...

Not to mention the fact that currently, UltraTrac uses MORE battery, than just standard GPS mode.

I fail to see how Garmin measured this watch out at 36/60hr for GPS use, any way you slice it, but that's an "adjunct" issue, one that's hopefully going to get better with GPS/firmware updates, I suppose.  I bet I couldn't get 20 hours, currently, of just-GPS cycling, before having to recharge (this is based on two weeks of riding with the 945, so far, and the drain-rate while riding).

For curiosity sake - it would be interesting to see what a service like My GPS Files would come up with for distance if you dropped one of your test rides out there.  My guess is that it would probably be pretty close since it would use track points for the source.  That's just a guess though.

Yep, I tested this today (for some reason I don't seem to get email-notifications, from this group, despite being subscribed to the thread, visually).

It works out to 19.5mi, or within about 2-2.5mi of the actual distance.

I admit that I expected this, too, given that the GPS tracking really isn't "that bad", in UltraTrac mode, it just looks like the algorithm for "in-between GPS points" seems to think you're not moving, at all, almost like you "magically transported between them" ;-]

So yeah, that's about the size of it.  I found some 935 posts, where people had pretty good success with UltraTrac, not great, but maybe 10-20% off, instead of 70-80%.

Does anyone know if there's a method for filing a bug, officially?  I could provide logs, etc, the whole bit.  Mostly, I think they just need to go review the 935 code (seems like this should've been mostly inherited, anyway...), and fix the algorithm, accordingly.  Even though I write very little code anymore, I bet I could make this change, pretty quick, if had both sides of the source ;-]

Has anybody actually used Ultratrac for its intended purpose  - ie undertaking a long, very long GPS activity - rather than speculate what it might do? 

I mean to say, how many people will actually need Ultratrac to record a bike ride?

FlipStone said:

I the end it still used about 3.5%/hr which is quite a lot imo for ultratac though but ran for just 30 minutes and usage calculation is very unstable then still. Also, as it was a training where I needed to run in a certain zone I was gettnig warnings (vibrations) very often when I was just under and in the zone so that prolly also has an impact on battery life

This for example is just not an example of what Ultratrac is for and completely misleading to make assumptions about battery life. Go do an activity of 15hrs plus with Ultratrac (on foot) and then report back.

Why wouldn't UltraTrac be "for" an activity that isn't "continuous", necessarily?  Is this actually the documented purpose?

I would argue that my use of UltraTrac (not having to charge for say a week, at a time, bike touring, or backpacking, or ...).

Even though I'm not doing a 30hr effort (I did a 300mi race/event once, and only once double-centuries are SO much more fun, IMHO), I still want to be able to record say 45 hours of GPS-and-distanced-tracked data, as part of a "larger activity", say a 2-3 week bike tour, encompassing say 120 hours of actual pedaling.  This would involve say 15 days of pedaling 8-ish hours/day.
Even outside of that, what if you "pause" during an Ultra event, of some sort (I took 10 minutes out of the saddle, on my 300mi deal, to eat a quick, but "actual meal", which probably resulted in a faster finish, overall.  Would that imply that my event was therefore "invalid", because was split into say 200mi/100mi?  

No disrespect, truly, but is there a definition that says that it has to be used in this fashion, to be "valid"?

Either way, if the "GPS bug", or the algorithm that interprets the data is assuming "zero speed", between GPS points, the data is going to be "unusable", regardless of the length of the activity (unless it was an incredibly short activity, say 1-2 minutes, then it might actually be more accurate, currently).
The other type of activity that might work "better", possibly, is when you're moving really slow, perhaps alpine climbing or similar.

Just to be absolutely clear, I'm not saying the GPS is incorrectly tracking me, in UltraTrac; it actually does a pretty fair job of the route followed (see my activity above, or others' too).
It's the part where it averages-out the time in-between GPS points, to zero-speed, that's making "a mess" of the activity roll-up.

I'm confident this is s/w fix-able, if I can figure out how to roll it up, as an actual bug.

If it's never going to work, or it doesn't work, Garmin should change how they "sell" this part of the system, because the description sure makes it sound like it could work for all of these "activity types".
My Ambit3, from some years ago, does a great job of this exact thing (with some obvious shortcomings of its' age), I can get 100+ hours, of VERY accurate activity tracking, in the lowest GPS setting, while riding.
I wanted to add HR, hence my move to the 945 (and my assumption about UltraTrac, from the marketing specs/description).