ADVrider

Go Back   ADVrider > Riding > Layin' down tracks
User Name
Password
Register Inmates Photos Site Rules Search Today's Posts Mark Forums Read

Reply
 
Thread Tools Search this Thread Rating: Thread Rating: 68 votes, 4.60 average. Display Modes
Old 11-04-2013, 02:04 PM   #9916
DRTBYK
Long Haul Adventurer
 
DRTBYK's Avatar
 
Joined: Mar 2003
Location: North Central Washington (state)
Oddometer: 4,567
Quote:
Originally Posted by speedygee68 View Post
To all who shall see this message: Greetings!! Please send me your aluminum cans and recyclable bottles. I just ordered the Montana 650 bundle. I guess I could tell my sandwich maker that it's a safety item. Keeps me safe and from gettin' lost, right? Gulp, RIGHT?
You'll only need about 11,000 cans and bottles to pay for the Montana at the current deposit-refund rate in CA. As for keeping you safe - not really. That's your job. It will however let you know right where you are.

Have fun.
__________________
Cheers,

Dan
All Things GPS
Reviews at www.GlobeRiders.com
DRTBYK is offline   Reply With Quote
Old 11-04-2013, 06:31 PM   #9917
clint64
Adventurer
 
clint64's Avatar
 
Joined: Mar 2006
Location: Duluth, GA
Oddometer: 89
Quote:
Originally Posted by speedygee68 View Post
To all who shall see this message: Greetings!! Please send me your aluminum cans and recyclable bottles. I just ordered the Montana 650 bundle. I guess I could tell my sandwich maker that it's a safety item. Keeps me safe and from gettin' lost, right? Gulp, RIGHT?
I'm in the same situation. Just ordered the Montana 600 bundle myself.

clint64 screwed with this post 11-05-2013 at 02:52 PM
clint64 is offline   Reply With Quote
Old 11-05-2013, 02:51 PM   #9918
Timpo
NORTH WALES TRF.
 
Timpo's Avatar
 
Joined: Jan 2007
Location: Downtown Wombatislava, England.
Oddometer: 2,084
Quote:
Originally Posted by speedygee68 View Post
To all who shall see this message: Greetings!! Please send me your aluminum cans and recyclable bottles.

I'll drink to that Dude......
__________________
The Rave Generator......
Timpo is offline   Reply With Quote
Old 11-05-2013, 03:46 PM   #9919
Rocky TFS
Gnarly Adventurer
 
Rocky TFS's Avatar
 
Joined: Apr 2012
Location: Hailey, Idaho
Oddometer: 466
Quote:
Originally Posted by atlas cached View Post
Read Bug 07 again. If you duplicate the conditions presented in Bug 07, you will experience the behavior reported.
I did, automotive mount, compass off, and I see only a slight occasional jitter in the map. Guess mine doesn't have this problem. BTW, I tweaked my automotive profile long ago, setting compass to auto, which is why I could not see this before.

What I DID notice, with compass off, is that if I made a partial turn as I came to a stop, like turning into a street and stopping halfway thru the corner, is that the map did more of a turn than the car did. If I stopped straight ahead the map never moved. Go figure.
__________________
2010 Yamaha TW200
2011 BMW F650GS Twin (traded)
2014 BMW R1200 GS WC
Rocky TFS is offline   Reply With Quote
Old 11-05-2013, 04:13 PM   #9920
atlas cached
OX Ambassador
 
atlas cached's Avatar
 
Joined: Jun 2012
Oddometer: 685
Quote:
Originally Posted by Rocky TFS View Post
I tweaked my automotive profile long ago, setting compass to auto, which is why I could not see this before. Go figure.
When you place the Montana in the Powered Automotive Cradle, the Compass is automatically disabled, regardless of user settings.
atlas cached is offline   Reply With Quote
Old 11-05-2013, 04:50 PM   #9921
Rocky TFS
Gnarly Adventurer
 
Rocky TFS's Avatar
 
Joined: Apr 2012
Location: Hailey, Idaho
Oddometer: 466
More Horse Beating

Well, I did some plots ala SteveAZ, and here is the result:





So what I did here was to set up the Montana in the middle of my corral, hundreds of feet from any building, with the track set to record once per second, WAAS on, and let it record for 2.5 hours. Since I'm hopeless at math, I found a spreadsheet converter for Lat/Long to UTM, which has meters as its units. (Also to try a slightly different methodology to see if it produced the same results.)

So after getting 9800 points into two columns as UTM Easting and Northing, I scatter plotted them to produce the above graph. I didn't "normalize" them because I don't know how, but since each gridline is one meter, the results are quite plain. My data agrees almost exactly with Steve's as to position scatter.

However, I then chose to put the (approximately) 95% confidence ellipse onto the graph. Then I added a few waypoints, of which the red ones were created with the unit stationary, WAAS on, walking away for a minute or so then back and logging the waypoint, and the yellow ones by picking up the unit, WAAS off, returning to the exact spot, setting the unit down in the same orientation and logging the waypoint. While not exhaustive, it does illustrate what to me is an important point:

What is the likelihood of a randomly created waypoint being significantly closer to the actual physical point than the scatter graph would indicate?
Well, my take is that it is quite high. I repeatedly walked away from the first waypoint, which was physically flagged, and back and never ended up more than a meter way, on several days. (Zoom set to max of 20 ft.)

Hmmmm. So I did some real world testing. First, I walked away from the point with tracking at 1 second, WAAS on, walked around the perimeter of the corral fence about 2 feet away, then saved that track, showed it on the map, and walked it backwards (not tracback). At max zoom, I could tell when I was three feet from hitting the corner posts, when I sidestepped three feet, and when I was three feet from the original waypoint.

I tried this with my old 96C and I could not even tell whether I was inside or outside the fence. Now my 96 may not be working real well anymore, since I could not get it to do differential with WAAS, and it never got better than 21 ft. accuracy on the satellite page.

OK, so being fascinated by why my Montana seems so accurate and Steve's so inaccurate, I did a second test. I went down to the river to find some heavy cover, and walked about 400 yds thru the woods with both GPS's held out in front of me, talking three perpendicular trips of 30 yds or so to plant a flag and log a waypoint on both units. WAAS was on for both units.

Hmmm, very interesting results. The Montana's accuracy degraded significantly, as expected, and the 96 became nearly unusable, as shown by its track which looked like spaghetti.

However, even though the Montana showed only 18 to 35 feet accuracy on the satellite page, the track showed every place I left the trail and it got me to within 10 feet of the flag all three times. The waypoints were not quite as accurate, being off by 15 to 20 feet. Go figure. Strangely enough, the 96's waypoints were closer to reality even though the track was much worse.

So here is my take on this controversy, which may influence readers of this thread and folks who are beginning to wonder if they bought a mediocre unit.

I simply can't agree with SteveAZ's categorization of the Montana as very poor in the accuracy department. My testing indicates that it is extremely accurate in track logging and track following. I could follow my track back in a dense fog and never get more than 5 to 8 feet off it.

And this is in spite of the fact that my scatter plot is nearly as bad as his. Logically, I would agree with him, but practically, it doesn't seem to translate into poor real world performance.

As for holding onto the sats, my 96 lost reception several times on this walk in the woods and the Montana never did, although it degraded to 35 feet at times.

Can we stop beating this poor horse now?
__________________
2010 Yamaha TW200
2011 BMW F650GS Twin (traded)
2014 BMW R1200 GS WC
Rocky TFS is offline   Reply With Quote
Old 11-05-2013, 05:36 PM   #9922
SteveAZ
Beastly Adventurer
 
SteveAZ's Avatar
 
Joined: Sep 2002
Location: AZ
Oddometer: 1,114
Good stuff. I like! Numbers are great, much better than subjective guessing when it comes to things that are easily measured like accuracy/precision. BTW, I've been pretty good about not beating this, it's others that keep bringing it up!

Yes, your numbers are very close to mine:

Rooftop: 3.52m 95%, 3.06m 90%, 2.02m RMS
Yard: 11.0m 95%, 8.83m 90%, 5.65m RMS

Your horizontal 95% confidence is hard to determine since you haven't come up with the horizontal error, only the n-s and e-w, but if we take (3^2+5^2)^0.5 that gives us about 5.8m, about splitting the difference. My guess has been that the problem is with multipath from nearby objects and the ground. Since you did your test without the nearby objects, but close to the ground and it came right between my rooftop which is very clean with regards to multipath and my yard which isn't, it lends at least some credibility to this suspicion.

For reference, the 76 numbers:

Rooftop: 2.81m 95%, 2.42m 90%, 1.68m RMS
Yard: 2.47m 95%, 2.21m 90%, 1.4m RMS

Since the results from the 76 were similar with the uBlox, I'm guessing that conditions (Ionosphere, constellation geometry, etc.) was better during the test in the yard.

BTW, 3m 90%, 2mRMS are pretty typical of most GPS chipsets with just a reasonable antenna. So I'm still firmly in the camp that it is not a very precise GNSS machine. Heck your data supports that when you compare it to the 76 or the uBlox. It clearly suffers quite a bit when conditions are less than ideal. It does seem to get a position fix very quickly (fast cold start time) and keep satellites pretty well, but it's no better than most present day GNSS receivers I've been using, including the uBlox with the weak antenna I was testing with.


If you want to send me a tracklog, I'll throw it into one of my spreadsheets and send the spreadsheet back to you so you can massage it as you like. It will probably give you better results than the UTM conversion and it does most of the work with the horizontal error instead of n-s, and e-w. They also make the plots I've been doing. It's work I've already done and have no issue sharing...


I don't know (remember?) anything about the 96. It certainly sounds as though it isn't performing at all well though and isn't a good unit to make baselines against. I used both a 76 and a uBlox LEA-6 and they produced very similar results - with the 76 winning a bit. I suspect the quadrilier helix antenna in the 76 is what is giving it the edge.

Again, my suspicion is that the Montana suffers a lot from multipath interference. I'm guessing it's mostly the antenna system but there could be something else going on.

Look, if you don't care about accurate positioning, the bugs and lack of routing features/functions then floptana is probably fine for you. I utilize reasonably surveyed waypoints frequently enough, something a flagship handheld would be expected to do well and the lack of key routing features/functions leave me wanting every time I use it. Frankly the bugs don't bother me that much but it still seems pretty crappy that a two year old, flagship handheld would still have them.

In the end, for me it was the wrong choice. Oh well.

YMMV...

SteveAZ screwed with this post 11-05-2013 at 05:41 PM
SteveAZ is offline   Reply With Quote
Old 11-05-2013, 07:25 PM   #9923
Rocky TFS
Gnarly Adventurer
 
Rocky TFS's Avatar
 
Joined: Apr 2012
Location: Hailey, Idaho
Oddometer: 466
Quote:
Originally Posted by SteveAZ View Post

I don't know (remember?) anything about the 96.

Look, if you don't care about accurate positioning....

YMMV...
The 96 should be nearly identical to the 76 except for the aviation database and features. Same processor, same antenna, same box. I know mine USED to get differential to approximately 10 feet, but it does not now. It IS ten years old, after all.

I DO care about accurate positioning, which is why I went to all this trouble to show that in spite of poor numbers in a scatter test when compared to the 76, it seems to do just fine in real world applications like geocaching. (which I have no explanation for, but which was pretty obvious from my track testing.)

The next test I want to conduct is to get the Lat/Longs of some nearby survey markers from the USGS website and see just how close the Montana can get me to them, just as a control. Presumably any recently re-surveyed markers will have accurate Lat/Longs. In the meantime, I'll keep going out to the corral with a blindfold and see if it still gets me to within 5 feet of my first waypoint every time.

And yes, I will send the tracklog, to see if your statistical analysis matches my eyeball guestimate from the graph. BTW, I chose to use unconnected dots because I think it gives a better visualization of where the denser area of points is.
__________________
2010 Yamaha TW200
2011 BMW F650GS Twin (traded)
2014 BMW R1200 GS WC

Rocky TFS screwed with this post 11-05-2013 at 07:39 PM
Rocky TFS is offline   Reply With Quote
Old 11-05-2013, 07:58 PM   #9924
SteveAZ
Beastly Adventurer
 
SteveAZ's Avatar
 
Joined: Sep 2002
Location: AZ
Oddometer: 1,114
Quote:
Originally Posted by Rocky TFS View Post
....I DO care about accurate positioning, which is why I went to all this trouble to show that in spite of poor numbers in a scatter test when compared to the 76, it seems to do just fine in real world applications like geocaching. (which I have no explanation for, but which was pretty obvious from my track testing.)
I was "surveying" points with the floptana and was noticing that they didn't seem nearly as repeatable as other units.... that's a big reason I got suspicious about a potential concern with error and did the plots. I wanted some confidence I was getting good positions. They just weren't as repeatable as I'm accustomed to. The plots merely back that subjective feeling up...

Quote:
Originally Posted by Rocky TFS View Post
The next test I want to conduct is to get the Lat/Longs of some nearby survey markers from the USGS website and see just how close the Montana can get me to them, just as a control. Presumably any recently re-surveyed markers will have accurate Lat/Longs. In the meantime, I'll keep going out to the corral with a blindfold and see if it still gets me to within 5 feet of my first waypoint every time.
I have access to many extremely accurately surveyed points (<1cm) that are used to test GNSS receivers... I should do the same nest time I get the chance.

Quote:
Originally Posted by Rocky TFS View Post
And yes, I will send the tracklog, to see if your statistical analysis matches my eyeball guestimate from the graph. BTW, I chose to use unconnected dots because I think it gives a better visualization of where the denser area of points is.
Good enough.... I'll send it back to you as dots. It's very easy to change....
SteveAZ is offline   Reply With Quote
Old 11-05-2013, 10:58 PM   #9925
atlas cached
OX Ambassador
 
atlas cached's Avatar
 
Joined: Jun 2012
Oddometer: 685
Quote:
Originally Posted by Rocky TFS View Post
I simply can't agree with SteveAZ's categorization of the Montana as very poor in the accuracy department. My testing indicates that it is extremely accurate in track logging and track following. I could follow my track back in a dense fog and never get more than 5 to 8 feet off it.
May want to google accuracy vs precision. What you are describing in the quote above relates directly to precision, not accuracy. The Montana can easily guide you back along the path just recorded with excellent precision, regardless of how accurate the positioning data was.
atlas cached is offline   Reply With Quote
Old 11-06-2013, 04:35 AM   #9926
SteveAZ
Beastly Adventurer
 
SteveAZ's Avatar
 
Joined: Sep 2002
Location: AZ
Oddometer: 1,114
Quote:
Originally Posted by atlas cached View Post
May want to google accuracy vs precision. What you are describing in the quote above relates directly to precision, not accuracy. The Montana can easily guide you back along the path just recorded with excellent precision, regardless of how accurate the positioning data was.

Also, even with lower performance GPS's, they usually have much better short term precision. I.e. guiding along a track done within 15min will be much closer than one done much later. It's tough to put a time on it, but in the precision GNSS industry 45min is a number thrown out a lot. Satellite geometry (and less so, ephemeris) and ionosphere are primary factors.
SteveAZ is offline   Reply With Quote
Old 11-06-2013, 06:42 AM   #9927
Rocky TFS
Gnarly Adventurer
 
Rocky TFS's Avatar
 
Joined: Apr 2012
Location: Hailey, Idaho
Oddometer: 466
Quote:
Originally Posted by atlas cached View Post
May want to google accuracy vs precision. What you are describing in the quote above relates directly to precision, not accuracy. The Montana can easily guide you back along the path just recorded with excellent precision, regardless of how accurate the positioning data was.
Good point. So even though I was able to follow the track along the fence without bumping into it or getting more than 3 feet away from the actual physical track 2 feet inside the fenceline, since I don't know the fence's Lat/longs precisely I can't say that the Montana was accurate?

Next step is to go back out there when it warms up this afternoon (2 days later) and follow the track again to see if it is still in the same place relative to the fence when tracking it.
__________________
2010 Yamaha TW200
2011 BMW F650GS Twin (traded)
2014 BMW R1200 GS WC
Rocky TFS is offline   Reply With Quote
Old 11-06-2013, 06:52 AM   #9928
atlas cached
OX Ambassador
 
atlas cached's Avatar
 
Joined: Jun 2012
Oddometer: 685
Quote:
Originally Posted by Rocky TFS View Post
Good point. So even though I was able to follow the track along the fence without bumping into it or getting more than 3 feet away from the actual physical track 2 feet inside the fenceline, since I don't know the fence's Lat/longs precisely I can't say that the Montana was accurate?
Comparing the Montana reported position to actual known position is accuracy.

Comparing the Montana reported position to previous reported position is precision.

Quote:
Originally Posted by Rocky TFS View Post
Next step is to go back out there when it warms up this afternoon (2 days later) and follow the track again to see if it is still in the same place relative to the fence when tracking it.
I appreciate the tests conducted and find the results very interesting :)
atlas cached is offline   Reply With Quote
Old 11-06-2013, 07:25 AM   #9929
SteveAZ
Beastly Adventurer
 
SteveAZ's Avatar
 
Joined: Sep 2002
Location: AZ
Oddometer: 1,114
Quote:
Originally Posted by atlas cached View Post
....I appreciate the tests conducted and find the results very interesting :)
Me too. My sample size of one makes for lousy statistics!

And you accurately (hahaha) point out that we are only measuring precison, not accuracy. we need surveyed points for that.

I'm concerned with accuracy since I survey points with one GPS and then use those points in others... but like most users, precision is more important in general.

Rocky TFS's data in the 2.5hr (a "lite" sample) sample he had was much better than his original numbers suggested. Much closer to the 76/uBlox numbers I got. It would be interesting to see a longer sample and one done where there's nearby walls and maybe one where there's cover.

I would welcome others [again] to send me track logs (1pps). I'll send you an excel spreadsheet that generates the plots, the RMS and confidence numbers (and lots of other numbers)....
SteveAZ is offline   Reply With Quote
Old 11-06-2013, 09:06 AM   #9930
Countdown
Beastly Adventurer
 
Joined: Oct 2003
Location: Carson City/Ridgecrest
Oddometer: 4,709
Quote:
Originally Posted by SteveAZ View Post
And you accurately (hahaha) point out that we are only measuring precison, not accuracy. we need surveyed points for that.
Actually I think you all have been talking about repeatability (within the accuracy) not accuracy.

The repeatability may frustrate many users (me) with recording a bunch of track points while stopped and causing the map to rotate while you are studing it to see which way to go.

Accuracy of 5-10 meters is overkill for a navagation GPS. When routing, I can't believe how precise my Nuvi is at freeway interchanges.

Again why don't you guys or the moderator start a new thread titled "Using a cheap Navagation GPS for precise Survey work"?
__________________
Jerry Counts
Countdown is online now   Reply With Quote
Reply

Share

Thread Tools Search this Thread
Search this Thread:

.
Display Modes Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is On

Forum Jump


Times are GMT -7.   It's 05:29 PM.


Powered by vBulletin® Version 3.8.5
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright ADVrider 2011-2014