62/78/Montana Track Point Repeatabililty

Discussion in 'Mapping & Navigation' started by Countdown, Nov 7, 2013.

  1. Countdown

    Countdown Long timer

    Joined:
    Oct 20, 2003
    Oddometer:
    6,078
    Location:
    Carson City/Ridgecrest
    A couple of people in the New Montana thread have pointed out what they think there is a problem with Track Point recording in the new generation Garmin units. Some of the Track recording issues noted are:

    1) Active log records only one continuous track per day even if unit is moved while turned off producing meaningless long straight lines.

    2) Backup Active Log is stored in main memory not on Data Card so it is not actually a hardware backup.

    2) Active log records numerous different points while unit is stationary and record mode in "Auto".

    3) Active log "jumping" while unit is stationary causes map to rotate.

    These issues have nothing to do with Accuracy of position only repeatability of calculated position. The difference between new generation and old generation (60/76) units appears to be in the algorithm that the unit uses to determine if the unit actually moved.

    I would like to know other user's experience.
    #1
  2. Countdown

    Countdown Long timer

    Joined:
    Oct 20, 2003
    Oddometer:
    6,078
    Location:
    Carson City/Ridgecrest
    I have only used my 78 a couple of times but both times were very frustrating.

    I was scouting a way for the PCQ to go around several locked gates on Sierra Pacific land. I had downloaded 120 tracks of possible roads to scout.

    1) While riding, every time I stopped at an intersection to look at GPS for possible alternatives, the map kept rotating and I couldn't tell which way to try. I now realize that the solution is to use map in North up not Track up. My old 76 never had that problem unless the bike rolled backwards slightly.

    When I got home and uploaded the tracks for the new path, I noticed that every time I stopped for 20-50 seconds, the 78 recorded 5-10 useless track points. My 76 did not do this. Since I record tracks to pass on on the Internet or to give riders at my Dual Sport rides, I would have to carefully edit the track to delete all these useless points before I post or give out.

    Attached is a file with tracks of several of my stops.

    Attached Files:

    #2
  3. mcnut

    mcnut Long timer

    Joined:
    Apr 26, 2006
    Oddometer:
    1,391
    Location:
    Bakersfield CA & Sammamish WA
    Thanks for starting this thread Jerry, I was resisting cluttering up the Montana thread with questions/comments which may not be exclusively Montana related. You have raised several issues, my focus will mostly relate to; location/track point deviation, the spinning map and what roll the compass (S models) plays. While I don't own any of the units in question my trusty old 60CSx will likely be replaced with one of them in the not to distant future. I also find it very curious that we are seeing some GPS behavior which is reminiscent of the Selective Availability days of the distant past which baffles me.

    I think the first thing I would suggest is that feedback from owners of the new version Oregon 6X0 units might also be enlightening. These are the most recently released Garmin units.

    There has been a tendency for GPS units to exhibit this type of behavior to varying degrees all along but for some reason and in spite of steady improvements to the satellite constellation (no SA, WAAS), receiver chip sets and 3 axis compass it would seem the latest offerings from Garmin might not perform as expected in this regard.

    Are others experiencing these behaviors?
    Why is this happening?
    Are there settings which help?
    Is it firmware or hardware related?

    I hope SteveAZ will offer his input and help with data collection as he has worked out a great method to graphically display the results and doesn't seem to mind crunching the numbers. I'm going to play with my old 60CSx, intensionally change a few setting to see if I can get it to act up.

    Bruce
    #3
  4. mcnut

    mcnut Long timer

    Joined:
    Apr 26, 2006
    Oddometer:
    1,391
    Location:
    Bakersfield CA & Sammamish WA
    Jerry, I looked at each of your tracks about the only way I easily can by using the 80' scale in MapSource and I have to ask......

    Are you saying that all the track points in this file were recorded while stopped? It does appear that for the first 2-3 points and last 2-3 points of each track the GPS thought it was moving a 5+ mph and 100s of feet. Also some look like the motion is along an existing mapped route, not random.

    Attached is your file which I edited by removing the track legs with a velocity of >5mph.

    Given we are probably looking for error on the order of 10s of feet I'm thinking that any deviation while stopped will be difficult to measure via MapSource or BaseCamp at lowest their scale of 80'.

    and

    May well need some basic testing/data collection protocols to insure results are somewhat repeatable.

    Bruce

    Attached Files:

    #4
  5. Countdown

    Countdown Long timer

    Joined:
    Oct 20, 2003
    Oddometer:
    6,078
    Location:
    Carson City/Ridgecrest
    Yes each track shows me slowing down to a stop, then stoped for 10-50 seconds, then starting up again.

    Actual error is not the issue, only repeatability when unit is not moving. The fact that you can not see the scatter when stopped unless zoomed in is also not the issue. The issue is that the extra useless track points cause the map to rotate because the GPS has determined that the unit is now moving in another direction. The extra points also make filtering the track down to 500 track points for a "Saved Track" to distribute to users more difficult but not a real big deal.

    Digital measurenent systems all have some finite linear error +/- 1/2 LSB of digital resolution so even the slightest change can toggle the LSB causing two different answers for esentially the same input. This problem is accentuated when the resolution is much greater (finer) than the accuracy (the case of all GPS units). In this case the digital reading can be all over the place yet meaningless because they are all within the accuracy of the true measurement. This is what is bothering the two other gentlemen.

    Most systems do something to eliminate this toggle at the human interface because it is very distracting and has no real information.
    #5
  6. Countdown

    Countdown Long timer

    Joined:
    Oct 20, 2003
    Oddometer:
    6,078
    Location:
    Carson City/Ridgecrest
    The first question I have about the firmware is how often does the unit make a position calculation?

    Aslo, does it make a seperate Lon and Lat calculation? Off the top of my head, seems like there is no way to calculate both of them with one calculation.
    #6
  7. wbbnm

    wbbnm Long timer

    Joined:
    Dec 18, 2007
    Oddometer:
    2,440
    Location:
    Abq NM
    Now that you convinced me to set my Montana to automatically create track points, I have that problem too when stopped.

    I never had it when I used a fixed distance.

    I don't like the jitter points because it makes it much harder to find where I stopped for any length of time. I used to just scroll thru the track points and look for a big time gap. Now I have to watch the track cursor as I scroll thru and see where it seems to stop.

    I might go back to fixed distance with very small distances but above the jitter like maybe 100 ft and then filter down later.

    I think the lat, long, elevation calculation is a closed-form solution of simultaneous equations using the time delay information in the GPS signal (as opposed to requiring an iterative solution). So all three would computed at essentially the same time. (At a few points in my life I had to actually do this for a work projects, but it was so long ago I don't remember much.)

    I suspect the update rate depends on how often the unit receives a complete set of timing data.

    I think you could figure out the fastest rate by selecting track point updates based on time and looking for the fastest time in the menu.

    I'm too lazy to get my unit out and try these things myself right now.
    #7
  8. NorthernTraveler

    NorthernTraveler Long time Adventurer

    Joined:
    Jan 1, 2009
    Oddometer:
    495
    Location:
    The Big Mitten
    Some of the newer Garmin units with both Russian & US satellite receiver chips talk about having a faster sample rate of 10hz and the older units having a 1hz sample refresh rate.

    So it might be safe to assume that the units we are talking about have a 1hz refresh rate. Of course, the older units are probably slower.

    One way to check is to see what is the most frequent sample time rate you can set in a given model.
    #8
  9. mcnut

    mcnut Long timer

    Joined:
    Apr 26, 2006
    Oddometer:
    1,391
    Location:
    Bakersfield CA & Sammamish WA
    For those of us who think in terms of minutes/seconds;

    I believe older units (My 60CSx) default and fastest setting is a sample rate of 1 per second (1HZ) and if set to Battery Saver mode this slows to 1 position fix every 5 seconds (and WAAS is automatically disabled).

    Latest Garmin units??????? I think SteveAZ has been referencing 1 HZ for the Montana.

    Bruce
    #9
  10. wbbnm

    wbbnm Long timer

    Joined:
    Dec 18, 2007
    Oddometer:
    2,440
    Location:
    Abq NM
    I broke down and checked my Montana. The fastest available was 1 sec between track points.

    While I was at it, I reset the track point creation mode back to Distance with a setting of .02 mile. Will try this for a while.
    #10
  11. NorthernTraveler

    NorthernTraveler Long time Adventurer

    Joined:
    Jan 1, 2009
    Oddometer:
    495
    Location:
    The Big Mitten
    The EchoMap 50 is touted as having 10 position readings a second.

    The new Oregon models also have the GLONASS receiver but their spec sheet doesn't specify how often it gets a position reading.

    There are other new units out with the updated chipset.
    #11
  12. Countdown

    Countdown Long timer

    Joined:
    Oct 20, 2003
    Oddometer:
    6,078
    Location:
    Carson City/Ridgecrest
    I agree with the Jitter flaw in the 78, distance setting is the obvious solution.

    Since I filter to 500 points per about 100 Mi, 0.2 is the starting point for distance. I would think you need at least 10x increased resolution for tight stuff which would be 0.02 which would record 5,000 points per 100 Mi.

    Even at .01 Mi (52 ft), I think that is greater than the Jitter so it should no record more than one point when stopped.
    #12
  13. Countdown

    Countdown Long timer

    Joined:
    Oct 20, 2003
    Oddometer:
    6,078
    Location:
    Carson City/Ridgecrest
    My gut feel is that position calculation is free running at some fixed rate, independant of track point recording since it is also used for navagation which is independant of track recording.

    My 1st question is what is used to determine heading for Track Up map orientation? The last vector calculated, the last vector based on some filtering, the last vector based on the last recorded track point? Based on my ONE use, it appeared to rotate about as many times as I noticed multipul recorded track points when stopped. Again wierd if you were navagating a route not following tracks on map?

    I have no idea what all the sources of error are but if there is any chance of something like "single event upset" I would continually run a filter that discards/outweighs any outlier calculations. A difference in this filter between 76 & 78 would explain the new Jitter flaw. In any event there is a difference in some algorithum.
    #13
  14. Countdown

    Countdown Long timer

    Joined:
    Oct 20, 2003
    Oddometer:
    6,078
    Location:
    Carson City/Ridgecrest
    Yes fasest for 76Cx, 78, & Montana are all 1 Hz. However the user interface is H:M:S which is wierd since that constant must be stored in only one unit (seconds or mS or something).

    Again is that setting, the Min setting, or some other faster fixed rate the actual Pos calculation rate?
    #14
  15. Countdown

    Countdown Long timer

    Joined:
    Oct 20, 2003
    Oddometer:
    6,078
    Location:
    Carson City/Ridgecrest
    Again I have a strong feeling that various units have different filters of pos calulation before user interface.
    #15
  16. mcnut

    mcnut Long timer

    Joined:
    Apr 26, 2006
    Oddometer:
    1,391
    Location:
    Bakersfield CA & Sammamish WA
    I would think the pos solution refresh rate available is dependent on intended use. Battery life is often an issue with handhelds so Garmin offers a Battery Saver mode which reduces the sampling rate resulting in reduced power draw by reducing the load on the processor and display redraws.

    For hiking a fix every 5 sec. is probably overkill and works fine for my MC use and I would think marine use. That said a pos fix every 5 sec. or even every second might seem an eternity at 550 mph for aircraft navigation. Guessing the EchoMap 50 (a marine unit) platform, fix rate of 10/sec, may be shared with some of Garmin's aviation products.

    Bruce
    #16
  17. Countdown

    Countdown Long timer

    Joined:
    Oct 20, 2003
    Oddometer:
    6,078
    Location:
    Carson City/Ridgecrest
    Very good point. I wonder what really uses power? It seems to be associated with pos calculation, I use 76 (non-x) at computer and batteries last forever if you "use with GPS off" on satellites page.
    #17
  18. UrsaMobili

    UrsaMobili Canoodian Expat

    Joined:
    Sep 17, 2010
    Oddometer:
    75
    Location:
    Back on the left coast, on island with no ferry
    Has anybody considered that the satellites are not "actually" stationary? Their "floating" motion can explain the randomness and directionality of the track points.:norton
    #18
  19. 250senuf

    250senuf Long timer

    Joined:
    Feb 17, 2008
    Oddometer:
    2,487
    Location:
    Nelson area, Kootenai, BC, Canada
    Yup, and they go from being at the horizon to overhead which probably causes all sorts of variables.
    #19
  20. wbbnm

    wbbnm Long timer

    Joined:
    Dec 18, 2007
    Oddometer:
    2,440
    Location:
    Abq NM
    I believe these satellites are in Low Earth Orbits as opposed to high stationary orbits. As such they are zipping right along and not exactly "floating".
    #20