My latest email exchange with Garmin regarding "drift"

[jmundinger]

I have been trading emails with Garmin tech support regarding the "drift" issue with the Vista HCx. The conversation has been cordial and my sense is they are genuinely interested in resolving the issue.

 

I had sent them a track log - not just a screen shot, but the *.gdb file that was saved in MapSource - from a hike that contained a very obvious "drift". I also included some description of the hike that corresponded with various points in the log. I also brought to their attention a couple of the threads on this board where we have discussed the issue.

 

"drift", as we have used the term here, apparently is not part of their vocabulary. They did mention an issue with "multipath" but, as they described the issue, that issue does not seem to account for the fact that the Vista HCx has difficulty "finding" itself again after the "drift" has occurred.

 

Here is the last response that I received: " We have contacted the Engineer's with your last information and they have stated that they are working on a software update that should help resolve this issue. They believe this is partly due to this unit being a high sensitive GPS receiver.

 

Imho, as end-users, be are, in effect, beta-testers. That is not intended to be a criticism of Garmin but the reality of being consumers of all of these interesting electronic toys. Thus, it seems to me, that, if we really want these toys to perform to our expectations, we have a responsibility to make it easy for Garmin (or the other manufacturers) to fix these issues. I would encourage others to share track logs that have an obvious "drift" with Garmin. Note, that the "drift" issue is a track log with an obvious aberrant track leg that significantly deviates from the actual route and then has the appearance of returning to normal.

[+Timpat]

Thank you for this info with good points and with a great suggestion.

[jmundinger]

Just to be clear, here is a visual of the issue that we are talking about. Note the highlighted leg - somehow I managed to walk a distance of a half-mile in 24 seconds, at a pace of 78 mph. And, even though the log after that point is back on the trail that I actually walked, the gps was actually lost by about a quarter of a mile. I couldn't specifically from the log except that, at the end (right hand side) I was near a cache that the gps said was still another 0.25 miles away. After turning the unit off/on the unit said that I was about 100 ft. from it. The general location of the cache was near an obvious topographic feature, i.e. the knob that shows on the right hand side of the screen shot.

 

Edited September 18, 2008 by jmundinger

[MtnHermit]

Do I understand correctly?

 

Starting from Tenmile CG, the yellow track is correct. Then the blue leg jumps a 1/4-mile. Then the RH yellow track is "correct" but offset by the distance of the blue leg?

 

My low sensitivity Venture Cx will sometimes (low battery) jump to extreme positions, but never do an offset track parallel to the actual path. Wonder why high sensitivity is the issue here? My Nuvi 205W has a high sensitivity chip, never seen this happen.

[jmundinger]

Do I understand correctly?

 

Starting from Tenmile CG, the yellow track is correct. Then the blue leg jumps a 1/4-mile. Then the RH yellow track is "correct" but offset by the distance of the blue leg?

 

Sort of, but not quite. The actual hike started at the Lazyman Gulch trailhead, i.e. where the yellow track started. The hike followed the road to the hairpin and then followed the trail east from there to a point on the sw side of the know on the righthand margin of the image. So, based on the path that I actually followed, the unit started deviating shortly after I started hiking. Then, the unit regard a major aberration. Then, it sort of caught up with itself and started recording a more normal looking track again. But, I think most of that track log is in error. After, I turned off the unit and restarted it, the unit showed I was in a spot that actually corresponded to the terrain and it led me right to the cache that I was looking for.

 

I didn't post the log here, but after finding that first cache, I hiked another mile to the south to find another cache - through steeper terrain than the first leg and portions of it also were more heavily timbered. Then I backtracked to the trailhead. Throughout, the unit performed normally and laid down a track that was well within the standard error for the unit.

 

There are a couple of other recent threads regarding the recent software updates for the Vista HCx that also discuss this issue. There is at least one post of a track log that shows an aberration similar to what I posted here. There also have been a couple of posts in which people have talked about a track that plots on the map/photo differently than the path that they actually walked. That may be an issue, but it's probably something different from what others have described as "drift".

[SiliconFiend]

Do I understand correctly?

 

Starting from Tenmile CG, the yellow track is correct. Then the blue leg jumps a 1/4-mile. Then the RH yellow track is "correct" but offset by the distance of the blue leg?

 

Sort of, but not quite. The actual hike started at the Lazyman Gulch trailhead, i.e. where the yellow track started. The hike followed the road to the hairpin and then followed the trail east from there to a point on the sw side of the know on the righthand margin of the image. So, based on the path that I actually followed, the unit started deviating shortly after I started hiking. Then, the unit regard a major aberration. Then, it sort of caught up with itself and started recording a more normal looking track again. But, I think most of that track log is in error. After, I turned off the unit and restarted it, the unit showed I was in a spot that actually corresponded to the terrain and it led me right to the cache that I was looking for.

 

I didn't post the log here, but after finding that first cache, I hiked another mile to the south to find another cache - through steeper terrain than the first leg and portions of it also were more heavily timbered. Then I backtracked to the trailhead. Throughout, the unit performed normally and laid down a track that was well within the standard error for the unit.

 

There are a couple of other recent threads regarding the recent software updates for the Vista HCx that also discuss this issue. There is at least one post of a track log that shows an aberration similar to what I posted here. There also have been a couple of posts in which people have talked about a track that plots on the map/photo differently than the path that they actually walked. That may be an issue, but it's probably something different from what others have described as "drift".

Make sure you tell them that this problem manifested itself only after a GPS chipset firmware update. I believe it was version 2.6 that caused it, according to this thread (at the least, chipset version 2.3 does not exhibit this problem). That might help them narrow it down.

[speede541]

 

…They believe this is partly due to this unit being a high sensitive GPS receiver.[/i]

 

 

Me thinks "high sensitivity" is Garmin-speak for low-precision, i.e. "cheap."

 

Looking at all these "drift tracks" various users have posted, this is my take:

 

The receiver calculates its location by being indiscriminate, receiving a high number of "junk" signal and averaging them to guess at a position. In doing so, it can perform under less-than-ideal locations such as thick tree cover, canyons, etc., where previous models performed weakly.

 

But in doing so, the software needs to keep the position from bouncing all of the the place. It appears to me that it is doing so by basing each position fix off the the previous one, especially if signal reception is weak. So if plot point #2 is off by just a bit, plot point #3 is being based on relative movement from plot point #2, and is off by just a little bit more.

 

Perhaps this is done to hide the receiver's weakness? Perhaps because it can't discriminate between a "good" signal and a "shoddy" signal?

 

But it seems evident that none of these logs are random. The ones I've looked at appear to follow the general shape of the trail, just that they're usually greatly compressed and sometimes slightly re-directed by a few degrees.

 

So hopefully this truly is just a software fix, and not some fatal flaw in the hardware. It'll just take some creative programming to address the problem, if indeed Garmin has accepted this error really does exist and is serious about correct it (given it's a two year old model and their priorities & energy might be needed elsewhere, dadgum us HCx owners).

[MtnHermit]

…They believe this is partly due to this unit being a high sensitive GPS receiver.

Me thinks "high sensitivity" is Garmin-speak for low-precision, i.e. "cheap."

Have to agree with the cheap comment. The primary reason for adopting the Microtek chip is cost not "High Sensitivity" My Nuvi's next-gen STM Cartesio chip is yet a still lower cost solution than the Microtek, albeit w/o the drift.

 

That suggest the location number comes from the chip and its firmware, not a product of Garmin's math magic. Perhaps STM is giving up some "precision" for stability.

 

A couple of weeks ago I walked an old railroad grade with my eTrex Venture Cx and Nuvi 205W, the eTrex had the more consistent track back and forth than the Nuvi, much to my surprise given its ancient Bravo chip.

[+Stu The Hiker UK]

QUOTE Me thinks "high sensitivity" is Garmin-speak for low-precision, i.e. "cheap."

 

Yep.....Have to agree on this one.

 

I have an etrex H and for me it does what I need, but I've noticed the "x-files drift"

 

A couple of times a week I walk the exact same route (exercise ) and mount the etrex H on the top of my pack for the entire journey.

 

My reasons are simple, I can get back and compare previous tracklogs, see how fast I walked certain legs, see how long I stopped for breaks, and basicly get a good idea of how my fitness levels are improving.

 

However it's also made me well aware of this "drift" thing, where some of the tracklogs are spot when overlayed in Memory Map and then other days they're a little bit off and then the extreme where they are all over the place. In some places it will miss part or all of a track and only jumps back on track if I get to a sharp deviation in the trail. It's also had my walking average as high as 62mph !

 

I've emailed tracklogs and screen shots to Garmin, along with my constant moan that I can only use six characters for naming, but as yet have not recieved a single reply.

 

Now I don't want to rubbish my little etrex H, because it's serving me well, but I do have to agree that it pulls in everything, good and bad and then makes a guess as to where I realy am on the map. Now althought that might not be so good it's far better than they days of having no signal when under dense tree cover. It seems we have to put up with the ups and downs and keep sending feedback to Garmin in the hope that these issues can be fixed in future software updates.

 

As for the etrex H, I will still give it 10/10, even though it sometimes gives me sketchy tracklogs

The way I use it with Memory Map serves me well.

Edited September 20, 2008 by stuthehiker

[MtnHermit]

However it's also made me well aware of this "drift" thing, where some of the tracklogs are spot when overlayed in Memory Map and then other days they're a little bit off and then the extreme where they are all over the place. In some places it will miss part or all of a track and only jumps back on track if I get to a sharp deviation in the trail. It's also had my walking average as high as 62mph !

Can you correlate the drift to battery charge? I've found low batteries increase the chance of "bad" points.

[+Stu The Hiker UK]

However it's also made me well aware of this "drift" thing, where some of the tracklogs are spot when overlayed in Memory Map and then other days they're a little bit off and then the extreme where they are all over the place. In some places it will miss part or all of a track and only jumps back on track if I get to a sharp deviation in the trail. It's also had my walking average as high as 62mph !

Can you correlate the drift to battery charge? I've found low batteries increase the chance of "bad" points.

 

The walk I do is only 3.5 miles long and I always start it on a fresh set of charged 2300 batteries, and at the end of the walk the battery level is still high.

 

So I think I can dismiss that as being my problem but not sure if that's the same with other units.

 

It has propmeted me to do the walk now a few times with batts that are almost out of charge and see if I get any marked changes in the data and will post my findings here.

[qwer993]

Just like to add that the colorado runs perfectly well on four alkaline cells in a separate battery pack connected to the usb interface (I use this under very cold conditions, batterypack under clothes). The initial potential is then 6,4volts. In my opinion the colorado generally should be running on a higher voltage than two AA cells supplies. The reason for this is that running on four cells allows for a complete drainage of energy from the four cells (final cell potential ~1.0volt), whereas running on two cells does not allow for a complete drainage (final cell potential ~1,3volt).

 

Maybe running on four cells drifts may never occur? I havent seen any drift using the external battery pack, but I havent used it much. Using two internal alkalines I have seen it once, a few days ago on 2.7 software.