Related

Auto calibration of altimeter during activities (details from Garmin and suggestions for improvement)

bellostom
over 2 years ago

Hi all and excuse my lengthy post

There are several threads regarding how auto calibration is working in Fenix6/7. Coming from suunto and coros, both brands have the auto calibration algorithm very well implemented.

My old Fenix 5 did not have auto calibration, which was one of the main reasons I sold it. Since then I read that Garmin had implemented auto calibration using also DEM and I believed that it would work well....

So having purchased the Fenix 7 and doing some activities along with my Apex Pro, it was evident that Garmin's implementation of auto calibration, was falling short in terms of auto calibration.

I then did a classic test where you manually calibrate the altimeter to a value off the actual one, start an activity and hopefully auto calibration will adjust the altimeter. Coros worked well. Fenix 7 after 3:30 hours of the watch sitting in the window, with the GPS altimeter showing the correct altitude, never adjusted the barometric altimeter.

So I contacted Garmin regarding auto calibration, describing the issue and test I did. I must say that the responses I got from Garmin UK customer support were exceptional. We exchanged several emails and the guys forwarded several of my questions to the engineering team. Below is the response to the test I did

Hi Thomas,

So I have had an update from our engineering team, so this device appears to be working as expected,

If you calibrate manually, even to an incorrect elevation, it takes priority over other calibration methods. DEM calibration takes place during an activity if auto calibration is on, and there is a swing in barometric pressure or drift in the altimeter.

The barometer does not gauge elevation. It gauges pressure changes. When you calibrate the altimeter, it sets an new standard of normal pressure for the barometer, a base level. From there, pressure changes corollate with increases or decreases in elevation based on the starting elevation (the calibrated elevation). 

Better yet, Your test demonstrates that their altimeter/barometer is working well because it did not experience enough drift to recalibrate to DEM or falsely read pressure changes. 

This can also hold true during an actual hike outdoors. If you calibrate to the wrong elevation at the beginning and there are no major swings in pressure over a short time or drift of the altimeter, it would not recalibrate because expected pressure changes would be occurring.

This feature is designed to deal with unexpected changes in pressure that could lead to false altitudes corelated to base pressure. I hope that helps

Main takeaway from the above email

1. Manually calibrating the altimeter before an activity, takes precedence over all other methods. Even if you enter a wrong value.

2. The auto calibration is not as aggressive/smart as Suunto/Coros. It will only trigger if it detects an exceptional pressure change. And this despite the fact that it is the only one which can also use DEM data. 

My next question was regarding the pressure drift that would trigger auto calibration

The thing however that is not clear, is the below
" DEM calibration takes place during an activity if auto calibration is on, and there is a swing in barometric pressure or drift in the altimeter"
If you are hiking uphill for instance there is constant shift in both altimeter and barometric pressure (gaining altitude decreases pressure)
How does the calibration distinguish between pressure change happening due to weather and not due to gaining altitude?

Again they came back with an answer from their engineering team

Hi Thomas,

We spoke yesterday, and I advised you I'd come back to you with an answer to your question. Here is what my engineering team have said:

"The speed at which the change takes place would be a major factor. Pace over horizontal distance, compared to elevation gained, and what is realistic versus what is not. When a storm is moving in, pressure drops quickly and causes a spike in elevation that looks a bit different from a steady hike up the side of a mountain. Think of it like rate of vertical ascent/descent over horizontal movement speed."

Please let me know if you have any additional questions.

Although I was very satisfied by Garmin's input on how the algorithm works, I still feel they can improve it to be on par with other brands, especially since the watch also have DEM. Coros/Suunto only have GPS to work with.

I have submitted a request for improvement here

http://www8.garmin.com/contactUs/ideas/

Anyone else that feels that this is something worth having as well, please post a similar request

Regards

  • 0 over 2 years ago
    bellostom said:
    Anyone else that feels that this is something worth having as well

    Any improvements are always welcome. 

    bellostom said:
    I then did a classic test where you manually calibrate the altimeter to a value off the actual one, start an activity and hopefully auto calibration will adjust the altimeter. Coros worked well. Fenix 7 after 3:30 hours of the watch sitting in the window, with the GPS altimeter showing the correct altitude, never adjusted the barometric altimeter

    I'm confused by your test as it seems incomplete.  If you manually calibrate elevation to an incorrect level and leave the watch still, I would expect it not to change because you "told" the watch this was the correct elevation for that location and you're not moving.  Garmin even states GPS elevation can have an error of + or - 400 feet.  This is the reason manual calibration is available.  What happens once you move?  Auto cal using DEM should then kick in and correct itself. Have you tried this?  What support told you is absolutely correct.  That being said; could auto cal be tweaked to better determine actual elevation change from atmospheric pressure changes? Absolutely!  It worked fairly well in the past, but a recent change in Garmin's algorithm has made things worse in my opinion.  If Suunto and Coros are changing the elevation using the same test, then they are overriding manual calibration.  I'm not sure I'd like that.  It's fine if the elevation is correct, but the sole point of manually calibrating is to correct errors that exist with GPS and even DEM. 

  • 0 over 2 years ago

    Wow, i am surprised by the great answers you got from Garmin Support. German Garmin support only copies info from the Garmin Q&A articles that I can find in a 2 sec google search as well. So next time i will contact Garmin UK. 

    I also have the impression that DEM calibration is sometimes making elevation readings worse. I assume it is because of the quality of the DEM data that Garmin is using. 

  • 0 over 2 years ago in reply to who knows

    Hi. Why should it matter if you move or not? The issue with how auto calibrate is working right now, is that it seems to only kick in if the pressure changes drastically. I have not tested it but I do not think it will make any difference, based on support's reply. The test I conducted proves that although the GPS altimeter had settled, DEM was never queried in order to adjust the altimeter

    Furthermore, I do believe Suunto/Coros approach is correct, because once GPS altitude stabilizes (based on the vertical accuracy provided by the chipset), it detects the offset and adjusts the altitude. Being still should not matter.

  • 0 over 2 years ago in reply to #Matthias#

    Yes, they were excellent in their responses and I complimented them about it. 

  • 0 over 2 years ago in reply to bellostom
    bellostom said:
    Why should it matter if you move or not?

    Because you told the watch that the GPS and/or DEM elevation was wrong by changing it manually.  Once you move to a new location and back, it should adjust accordingly.  I just did your test.  My elevation is 13 feet.  I manually set it to 30 feet where it stayed for an hour.  I went on a run and when I returned the elevation read 15 feet.

    bellostom said:
    Furthermore, I do believe Suunto/Coros approach is correct, because once GPS altitude stabilizes (based on the vertical accuracy provided by the chipset), it detects the offset and adjusts the altitude. Being still should not matter.

    If Suunto, Coros, Garmin, etc. has an incorrect DEM elevation of 25 feet at my location when I know it's 13 feet and it eventually changes back to 25 feet without me moving after I manually put in the correct elevation, how is this the correct approach by returning to incorrect data? 

  • 0 over 2 years ago in reply to who knows

    Actually one cannot  manually affect GPS/DEM, only the barometric altimeter. It is nice to hear that your test was a success. I will also test it outdoors, since I agree that it is a better approach. Maybe indeed it does not kick in only when it detects major pressure changes. I really hope so. Btw, you also own the Fenix 7 or a previous model?

    Now regarding Suunto/Core approach. Your example I feel is a bit extreme. I am not saying that their approach is fail safe 100%, but my experience with both, especially in long running activities, is that they both are pretty accurate.

  • 0 over 2 years ago in reply to bellostom

    Yesterday I manually calibrated the altimeter from 520m to 560. After 5 hours I went for a 50min run. No adjustment was made whatsoever. This seems consistent with what garmin replied to me unfortunately. 

  • 0 over 2 years ago
    bellostom said:
    My old Fenix 5 did not have auto calibration, which was one of the main reasons I sold it.

    I am afraid  your basic statement relies on being misinformed. Not just Fenix 5 had auto cal see https://www8.garmin.com/manuals/webhelp/fenix5/EN-US/GUID-BC734846-01A7-4F33-86D4-DFBDBC06CDB4.html , but also earlier models like Fenix 3 see https://www8.garmin.com/manuals/webhelp/fenix3/EN-US/GUID-BC734846-01A7-4F33-86D4-DFBDBC06CDB4.html

    The only differences were that those used saved location data, barometer data and GPS data, but not DEM data, evidently because they did not have maps (except for 5X).

  • 0 over 1 year ago in reply to -

    I have a Fenix 3HR from 2016, Enduro from 2021 and the Fenix 7X SS from a few months back. When it comes to measuring Total Ascent, I still rely on the F3HR for consistency and accuracy. For a six year old watch that does not have all the fancy controls, my F3HR isn't affected by strong winds or sweat or heavy rain during an activity. The watch has gone through so many ultra trail races (distance and elevation gain matched race organizer's published data) and is also my bike computer mounted on the handle bar. Because of its reliability I use it as a reference when testing the Enduro and F7X.

    Compared to the F3HR and I think to some extend the F5 also, it looks like things went downhill with the F6, Enduro and F7, so disappointing that so many owners struggled with accurate elevation gains. Instead of enjoying the watch and its long battery life, every trail run and bike activity comes with a suspect Total Ascent numbers. I really shouldn't have to but being paranoid, I bring my F3HR along as long for comparison. I'm on my second Enduro, Garmin replaced my first piece when I consistently recorded off the charts Total Ascent (eg a 34km trail with 2600m of Total Ascent the Enduro recorded a ridiculous 3200m). 

    Maybe the F3HR had better software algorithm that match its simpler technology while the newer models just got way too complex. 

  • 0 over 1 year ago in reply to jan.shim

    I did not mention, Jan, that I had owned a F3HR before getting F5+.

    Although I had to go throught two replacement processes due to a malfunction of their altimeter, in the period when they worked they really worked well. And to be frank I had no problem with the altimeter of F5+ either, although I am really "choosey" when we speak about altitude measurement. 

    I have more problem with my F6X, but the whole thing more about a comparison of a pretty accurate F3HR and a relatively accurate F6X. So I cant say that the newer Fenixes are definitely worse and worse as regards altitude measurement. Or I just have become older and older and dont care so much. :-)

    As regards the calculated Total Ascent the situation is totally  different, I also have some suspicion, but I have always had this suspicion with all Fenixes. I dont say they are inaccurate, I just say that compared to distance covered I dont trust them at the same level.

    So absolute altitude is OK for me, aggregated changes of elevation is so-so.