GPS track question

[RickB.]

As discussed elsewhere, my backup Garmin eTrex Legend recently demonstrated inferior signal reception compared to my newer devices (76CXs/605). In addition, the eTrex seemed to record false tracks based on erroneous track points. Both the unit's weak reception (in top of pack, properly oriented) and a couple of these false tracks, one highlighted, can be seen in the below capture from a Liberty/Flume hike yesterday.
Doug Paul encouraged me to post this to discuss; hope it is useful and not an academic review of an obsolete piece of gear (but I know there are some users out there, still).

 

[DougPaul]

Several things can happen when a GPS is having difficulty getting good signals (either due to an inferior (compared to the latest and greatest) GPS or a poor skyview):
* It may record some inaccurate points. (Often the first and last couple points in a continuous section of track are less accurate than the others.)
* There may be gaps in the track (ie no points for sections of the route).

Background: The track log in my Garmin GPSes (presumably other brands are similar) consists of:
* One or more named tracks (<trk> in GPX)
* Each named track consists of one or more track segments (<trkseg> in GPX)
* Each track segment consists of some number of track points (<trkpt> in GPX)

Turning the GPS on and off will generally begin and terminate a named track in the track log. A long period with no signal will break a named track into multiple segments, but might also break it into multiple named tracks.

When I get a good signal for the entire trip (ie use my 60CSx... ), I often get the entire trip in a single track segment in a single named track.

If I have sections where there was no signal (ie went through a long tunnel), then a named track is likely to be broken into multiple track segments (or possibly multiple named tracks).

The easiest way to examine the structure of a track might be to either download it as a GPX file or convert it to a GPX file (MapSource or GPSBabel can do either: www.gpsbabel.org) and examine the resulting text file. The format is very human readable and I've mentioned some of the tags above. (You can also edit it in this format as well.)


The next question is how a computer program (or your GPS) displays a track. In the ideal case (all points are accurate and in a single track segment), all sequential points are connected by straight lines in the display. (I think this is how most displays work.) The issues come up about what to do with separate track segments: Do you simply connect them as if they were parts of a single track or do you put gaps in the display? Presumably the software author makes some choice or in one of my programs allows the user to select a policy. If you draw straight lines between the ends of track segments or named tracks, the lines can be wildly inaccurate with respect to the path that you actually took.

It looks to me like your image is from Garmin MapSource. IIRC, MapSource does not connect between track segments. It looks to me as if your track has a number of inaccurate points, has gaps, and is broken up into many segments (or many named tracks). (It would have been much cleaner, prettier, and more useful if you had used your 76CSx. You likely could have avoided the bivy...)

In Mapsource:
If you click on the "tracks" tab, you will see a listing of the track segments and if you double click on a track segment (in the tab) you will see a listing of the individual track points. You can also see the same thing by examining the corresponding GPX file.


In the GPS Bakeoff we recorded simultaneous tracks of 7 GPS/antenna combinations. The tracks are available for download (from the thread) if you want to view them. The difference bewteen the eTrex Vista and the 60CSx (electrically identical to the 76CSx) is dramatic... http://www.vftt.org/forums/showthread.php?t=14406. In fact, I bought my 60CSx as a result of this experiment.

Doug

 

[RickB.]

When I get a good signal for the entire trip (ie use my 60CSx... ), I often get the entire trip in a single track segment in a single named track.
Doug

I know how to analyze the tracks/data and I use the tools you mention. My question was focused more on how such wildly inaccurate points get recorded. They are far outside an explanation of the connecting of dots or an algorithm projecting a likely track.
I almost always get a continuous, single track for an entire trip with my 76CSx.

 

[DougPaul]

I know how to analyze the tracks/data and I use the tools you mention. My question was focused more on how such wildly inaccurate points get recorded. They are far outside an explanation of the connecting of dots or an algorithm projecting a likely track.
I almost always get a continuous, single track for an entire trip with my 76CSx.

I have no idea how much you may or may not know about how GPSes work--they are complicated devices and many if not most users do not understand many of the details required to understand some of their behaviors. Answering such questions requires a significant amount of guesswork on the part of answerer.


Marginal reception often gives inaccurate points. (Your track suggests rather poor and intermittent reception.)

GPSes use smoothing across several points to increase the accuracy. (The "raw" GPS output is position, velocity, and time. The position and velocity estimates for the recent past are combined with the current position and velocity estimates (using a Kalman filter) to increase the accuracy of the current position and velocity estimates.) Poor reception plus no history to use for smoothing often makes the first few points inaccurate. As the GPS is losing the signal, the last few points before a gap are often from poor signals and may even be totally from dead reckoning.

After loss of signal, a GPS may continue to output dead reckoning points without informing the user that they are dead reckoning. (A certain amount of dead reckoning is useful to fill in short gaps.) Some brands/models of GPS give up dead reckoning quickly giving the user fewer but more accurate points, certain other brands/models continue dead reckoning much longer thus creating the illusion of a meaningful track, but actually outputting junk.

GPSes are also designed to continue to operate when fewer that 4 satellites are received by making assumptions. For instance, when only 3 satellites are being received, Garmin GPSes will assume the altitude remains constant to compute a location. If your altitude is changing, this will lead to inaccurate positions. The front panel indicates a 2D solution rather than a 3D solution, but this information is not saved in the track log.

Learning how to distinguish the junk from the good points is one of the skills required for the best use of a GPS. As we have noted earlier, some of the modern GPSes have greatly reduced this problem, but they are not foolproof.

Doug

 

[Remix]

My conclusion is that Garmin is happy to log and display big (but occasional) position errors when the the GPS satellite gets weak or when the unit starts using different satellites to calculate a position. This has been my experience, and its in your data.

I have found that Garmin GPS tracks have much more noise in their tracks than my Magellen Meridian Platinum. I have used a Garmin 305 and a Garmin Colorado with National Geographic Explorer-- and I have seen big jumps in position on the map screen while the Garmin units were not moving.

Too bad the Magellen is obsolete and can no longer receive WAAS signals.

It would be nice if Garmin gave advanced users the option of rejecting weak signals at the expense of getting a Loss of GPS Alarm so cruddy tracks dont show up.

 

[RickB.]

While I'm frustrated with many things relating to Garmin products, the results above are from an early generation device and, as Doug has made clear, the newer, high-end Garmins are unlikely to yield the wildly inaccurate tracks depicted due to their improved sensitivity.
So, as I feared, this discussion will be (marginally) useful only to those still using the older devices. I'm no longer in that group.

 

[DougPaul]

My conclusion is that Garmin is happy to log and display big (but occasional) position errors when the the GPS satellite gets weak or when the unit starts using different satellites to calculate a position. This has been my experience, and its in your data.

I have found that Garmin GPS tracks have much more noise in their tracks than my Magellen Meridian Platinum. I have used a Garmin 305 and a Garmin Colorado with National Geographic Explorer-- and I have seen big jumps in position on the map screen while the Garmin units were not moving.

Too bad the Magellen is obsolete and can no longer receive WAAS signals.

It would be nice if Garmin gave advanced users the option of rejecting weak signals at the expense of getting a Loss of GPS Alarm so cruddy tracks dont show up.

You may have the comparison of the devices reversed.

There were knowledgeable reports that Garmins tended to terminate the track when signals got weak and Magellans were much more likely to report a fictional dead-reckoning track when signals got weak. (A dead-reckoning track can look smooth and reasonable if you don't know where it should actually go.) This led a number of users to the erroneous conclusion that Magellans were more sensitive than Garmins. When to terminate dead-reckoning is a judgment call--there is no hard and fast rule for the "best" threshold for termination. (These reports referred to an earlier generation of GPSes, not to the modern high-sensitivity models.)

There is another trade-off in how much smoothing to apply. A heavily smoothed track may look better after-the-fact, but may actually be less accurate. For instance, if you walk a trail that is straight, then suddenly turns 90 degrees and remains straight, a heavily (predictively) smoothed track will tend to look smooth on the straights but overshoot the corner while a lightly smoothed track will tend to look a bit noisier on the straights but follow the corner better.

The WAAS signals have not changed (although some of the WAAS satellites were moved a few years ago, generally to better locations for the ConUS). If your GPS used to be able to use WAAS, it should still be able to do so. The best WAAS satellite for the NE is low above the horizon toward the SW (in a geosynchronous orbit) and may be difficult to receive with a restricted skyview or in the trees. IMO, WAAS isn't of much value for hiking and I normally keep it disabled. (WAAS was designed for use by aircraft and is easy for airborne users to receive.)


I used to see occasional points off by large distances with my eTrex Vista (usually at the beginning or ending of a track segment), but it hasn't been a problem with my 60CSx.

FWIW, many users don't know the best orientation for and place to carry their GPSes... The best for an eTrex Vista is flat with the display facing upward (it has a patch antenna under the Garmin logo) and the best for a 60CSx is vertical (it has a quadrifilar helix antenna in the obvious "antenna" at the top). (The 76 series is the same as the 60 series except the case hides the location of the quad helix.) The eTrex Vista is much more sensitive to orientation than is the 60CSx (because the GPS chipset is less sensitive, not because one antenna is better than the other). The GPS will also work better if it is carried in a location where the user's body does not block any of the sky (ie on top of the user's head or on a pole above the user's head). (An external antenna in one's hat is one way of putting the antenna is a clear location while keeping the body of the GPS in a more practical location.) Again, this is a bigger issue with the Vista than the 60CSx for the same reason.

Doug

 

[DougPaul]

So, as I feared, this discussion will be (marginally) useful only to those still using the older devices. I'm no longer in that group.

You just have to ask the right question...

The performance of the modern high-sensitivity GPSes with regard to receiving poor signals under typical hiking conditions is so much better than the older models that I disrecommend using the older models for hiking.

If you must use an older-generation GPS for hiking, pay particular attention to where you carry it and and its orientation (see my previous post). Do not use battery saver mode (makes it easier to lose lock). Stop and check the GPS occasionally--if it has lost lock, stop and hold the GPS above your head in the proper orientation until it reacquires lock. (It is easier to maintain a lock than it is to acquire one and it is harder to acquire a lock while moving than when stationary.) If need be, go to a spot with a clearer skyview to reacquire lock. Learn to recognize when the indicated locations are likely to be inaccurate.

I highly recommend that one "practice" using the GPS on well marked trails where it is not needed for navigation to learn how to use it and how well it works under a variety of conditions.

FWIW, I started with an eTrex Vista and had to play the games described above. I upgraded to a 60CS (non-high sensitivity) and, while it was more sensitive, I still had to be somewhat careful (but it was pretty good with an external antenna in my hat). Now with the 60CSx (one of the first high sensitivity models), I can just turn it on, acquire lock, throw it in the top pocket of my pack, and generally get a good-looking track with few to no gaps. If I am intentionally mapping a trail I still use the external antenna in my hat for maximum accuracy, but I don't bother for normal hiking.

The Vista was usually able to maintain an adequate lock with the unit in the top pocket of my pack or in a pouch on my pack shoulder strap if I had a good skyview. However, it would usually lose lock under heavy tree cover no matter what I did. As noted above, the 60CSx is much better.

Any of the three above models work well for driving (mounted on the dashboard to the right of the driver) when away from tall city buildings. And any of them work well above timberline or anyplace with a good skyview.

Doug

 

[Remix]

I just want to offer a different opinion in case anyone hits on this using a search engine. It seems minds are made up.

My observations are based on using my Garmin Colorado. Its a newer device. It has displayed the same behavior as your older device when used on steep slopes when GPS satellites have been obscured. In fact the track data is quiet noisy when the unit is motionless.

To convince oneself that the Magellen does not do dead reckoning as DougPaul has suggested, a simple experiment can be performed. All you do is obscure the antenna from the sky and see how long it takes the Loss of GPS signal alarm to activate. I have done it, and do not need to rely on reports on the internet. DougPaul, you might want to try it.

And yes, it is true, for anyone else reading this, that most Magellen Platinum's no longer receive good WAAS data.
-The constellation was changed
-Any Magellen left on during the change will know about the change and continue to receive WAAS data.
-But if you ever reset it, or it loses its RAM contents, it will never receive WAAS signals again.
-There is a unofficial firmware update procedure on the net to fix this issue which basically works by editing the machine code and data stored in the ROM memory of the device with a hex editor.

 

[Remix]

---I deleted this duplicate post- Remix

 

[RickB.]

You just have to ask the right question...

Or the right person, perhaps..

 

[DougPaul]

My observations are based on using my Garmin Colorado. Its a newer device. It has displayed the same behavior as your older device when used on steep slopes when GPS satellites have been obscured. In fact the track data is quiet noisy when the unit is motionless.

The Colorado has a different GPS chipset (Garmin Type-M) than does the 60CSx (SiRF StarIII). I've seen reports suggesting that they have similar performance. There were some problems with the early software releases for the Colorado--is yours updated to the most recent release?

To convince oneself that the Magellen does not do dead reckoning as DougPaul has suggested, a simple experiment can be performed. All you do is obscure the antenna from the sky and see how long it takes the Loss of GPS signal alarm to activate. I have done it, and do not need to rely on reports on the internet. DougPaul, you might want to try it.

This experiment misses many conditions. The general conditions go from 1) >4 strong signals, 2) 4 strong signals, 3) 4 signals not all of which are strong, 4) 3 which may or may not be strong, 5) 2 which or may not be strong, 6) 1 which may or may not be strong, and 7) no signals. Each condition between 3 and 6 (inclusive) contains partial information which can be used for partial dead reckoning.

The reports are from people who IIRC are knowledgeable based upon a history of knowledgeable reports. I believe they are based upon observations of field performance which would include the intermediate conditions, although the immediate condition may not have been known to the observers. (One would need to log the NMEA data to determine which signal condition applies. The signal condition could also change second-to-second.) As I recall the reports compared Garmin vs Magellan but I don't recall if they identified specific models. The reports are at least several years old so they may not apply to recent models.

And yes, it is true, for anyone else reading this, that most Magellen Platinum's no longer receive good WAAS data.
-The constellation was changed
-Any Magellen left on during the change will know about the change and continue to receive WAAS data.
-But if you ever reset it, or it loses its RAM contents, it will never receive WAAS signals again.
-There is a unofficial firmware update procedure on the net to fix this issue which basically works by editing the machine code and data stored in the ROM memory of the device with a hex editor.

Wow--what a design flub! The almanac* is only valid for a limited time period (up to several months). Once that period is over, the GPS should search the entire sky for the satellites. A reset should clear the almanac and cause a fresh start searching for the satellites. (At least some models of Garmin GPSes will start a complete search if the satellite is not found after some amount of searching. This allows the GPS to escape a bad almanac.)

* The almanac contains coarse orbital info for all satellites. It is transmitted continuously by all satellites and is primarily used to speed up acquisition of the satellites. WAAS is an FAA add-on to the DoD GPS, so I'd guess that the regular GPS satellites do not transmit almanac info for the WAAS satellites.

Doug

 

[Remix]

Doug-

I just wanted to correct your assertion that the Magellen Platinum units should receive WAAS data. It is incorrect. Both sets of firmware are up to date, but since you are suggesting that the Garmin firmware may have a bug, we can only hope for a fix in the future.

All I can do is reiterate to you that the tracks recorded by the Magellen are broken, but not wildly off, and do not need subsequent editing with GPS babel.

There is no theory involved..just observations over a few years.

Rick-

The newer Colorado series apparently has the same behavior as your Legend.
Sorry if you dont want to hear that for some reason.

 

[RickB.]

Rick-

The newer Colorado series apparently has the same behavior as your Legend.
Sorry if you dont want to hear that for some reason.

No, Remix, your inputs have been helpful and I want to hear them. This forum clearly needs more than one GPS expert. But the 76CSx that I've been using for a couple of years definitely does not exhibit the wild tracks that I posted above.
I posted because the erroneous tracks/points make navigating with the eTrex impractical. If it displayed only points recorded, with effective smoothing/dead reckoning or not, it would be one thing but the errors are half a mile or more at times and the resultant large tracks obscure the real points in the display in the field. I wouldn't have been surprised by an input that my GPS were faulty.

 

[Remix]

Well, I don't consider myself an expert, but I have been frustrated by the erroneous track data, especially after I thought I was upgrading my Magellen to a newer model with better performance.

Its like the technology went sideways--more features but at the price of the essential function of the device.

(Putting display visibility issue aside, it seems to work much better on the kayak with many satellites visible and no weak signal conditions. For working with unknown and changing tides and wind effects in unfamiliar areas, I find it more valuable than for hiking. )

 

[DougPaul]

I just wanted to correct your assertion that the Magellen Platinum units should receive WAAS data. It is incorrect. Both sets of firmware are up to date, but since you are suggesting that the Garmin firmware may have a bug, we can only hope for a fix in the future.

Re the Magellan Platinum: the product description claims that it should receive WAAS data, but I'll take your word that they may not anymore. A flaw in the hardware/firmware, IMO. I see a first-available date of 2001 so it wouldn't surprise me if Magellan has ceased issuing firmware updates.

Re the Garmin Colorado firmware: The problems that I mentioned were early teething problems. Considering how long the Colorado has been out, I wouldn't expect too many more updates.

The newer Colorado series apparently has the same behavior as your Legend.
Sorry if you dont want to hear that for some reason.

Huh? I have a Vista, not a Legend. I don't have a Colorado either.

Doug

 

[RickB.]

Well, thanks, guys.
Doug is right, you have to ask the right question. I really wanted to know (and still do) if the wildly inaccurate track in my map capture is within the norm in the performance of the eTrex Legend, in forumites' experience.

 

[DougPaul]

I really wanted to know (and still do) if the wildly inaccurate track in my map capture is within the norm in the performance of the eTrex Legend, in forumites' experience.

I thought a bit more about your (GPS) problems on Moosilauke and went back and reread your original "trip report". You had a mile or so along the ridge and summit cone where the skyview is essentially unobstructed. I would have expected the Legend to reacquire the satellites in this section even if you didn't stop and hold it up to the sky. (The route that you show in the map at the beginning of this thread is all in the trees, so I'm not surprised that the GPS had difficulty there.)

It is possible that your Legend is a "low performing" unit due to manufacturing variation, age, or abuse by the previous owner. Or perhaps it shifted in your pack and was ultimately in a poor orientation.

In 20-20 hindsight, another strategy for dealing with not having maps loaded would have been to go back to the summit, set a waypoint, and execute a goto on that waypoint so that the GPS would tell you the distance and direction back to the summit wherever you went. You could then use that info to locate yourself on the map and hopefully find your way back to the carriage road or one of the other trails.

Anyhoo, all is well that ends (fortunately without permanent damage)...

Doug