Separate names with a comma.
Discussion in 'Mapping & Navigation' started by AugustFalcon, May 18, 2011.
Apparently not anyone.
Mine bounces a little bit when I come to a stop and I have track up. It's never been a big deal to me, but I don't stop much. Ha
I can't duplicate that behaviour. My Montana is as stable as can be.
Sent using strings and tin cans and Tapatalk.
Seriously? Trolling? For what? I have no stake in Garmin. I eval any/all GPS units that I think are worthy of use in the Adventure Riding activity.
I accept that you are technically proficient although your reluctance to be a good consumer is baffling. As for my knowledge of GPS technology, my involvement in GPS started with my work on NavStar in 1972. I have degrees in Electrical and Mechanical Engineering and a Wildlife Ecology degree (for personal interests).
I would agree there is little discussion left in this topic.
These are from 15:10 to 18:25UTC this morning...
(at same scale as floptana)
(scaled to fit)
(at the 76 scaled to fit scale)
Excellent charts, Steve.
Seeing how poor the Montana functions compared to the 76, you can see the new Oregon 6xx is even worse than the Montana here.
He seems to delight in showing how poorly his Montana works. Curse the darkness, kinda thing.
Sent using strings and tin cans and Tapatalk.
Floptana - 2.89
76 - 0.69
U-blox - 0.77
Floptana - 5.65m
76 - 1.4m
U-blox - 1.64m
Floptana - 8.83m
76 - 2.21m
U-blox - 2.64m
The floptana clearly has much greater position noise. I sort of suspected this from waypoints I created. The filter for determining GPS track when stopped in floptana clearly isn't up to the task of dealing with it. At least I understand/confirmed the nature of the problem.
I will do another with floptana on the roof overnight to maximize clear sky and minimize multipath even though that's not real world. I have large block walls and trees that will undoubtedly aggravate multipath noise and reduce sky coverage. The roof should have none of those issue and since the only horizontal thing up there is the metal AC, they'll even a get a rudimentary ground plane.
These are very easy to produce and don't take more than a few minutes. Anyone with Excel can do it. I encourage others to go ahead and we can compare.
Just setup track logging to Record Method "Time" and Interval to "00:00:01" (this is where mine pretty much always is).
Let it log for a couple of hours. If you have archiving on (necessary to grab more than 10000 points - a bit less than 3 hours) then you will need to append the archived tracks. I just copy the file(s) directly from the memory and use mapsource to append but basecamp does the job too. Save as a .txt or .csv file and import the data into excel. Throw out (or ignore) everything but lat/lon. Convert lat/lon to meters:
Lat(m) = Lat(deg) / 180 x 6.371 x10^6 (earth mean radius) x PI
Lon(m) = Lon(deg) / 180 x cos(Lat(deg)) x 6.371 x10^6 x PI
I make two columns that subtract lat/lon in meters from the first used sample to "normalize" it.
Graph the (normalized) lat/lon in meters by slecting both columns and inserting a scatter chart....
The new Oregon has GLONASS too? Wow. The U-blox has GLONASS and is Galileo capable but I only have GPS turned on.
I suspect they all may do better on the roof but that's ideal conditions, not real world.
I've always been very happy with the 76 GNSS performance. I think the qudrifiler helix antenna helps but I don't think that would explain the huge difference. I've done scatter plots with the 76 over 24hrs where the RMS error was less than 1m (closer to 70cm).
The antenna I have on the Ublox is not very good and has zero ground plane. The PCB is barely larger than the patch antenna and has the U-blox module mounted on the underside. It's advantages are that it is *very* small and runs at 5Hz. I use these in some machine control applications.
I've got some really sweet, extremely high performance units but they're a pain to setup and frankly I've already extracted data from them. 20Hz, RMS errors with WAAS of <70cm and RMS errors using one as a diff base <2cm.
Yes, the Oregon 6xx is GPS/GLONASS/Galileo/WAAS/EGNOS capable.
Don't be so hard on yourself.
Lack of floating point precision in Excel caused a small but not negligible error that added skew to the charts. They are updated as well as the standard deviations. They sort of look like flying spaghetti monster.
I'm still pretty impressed by both how good the 76 is and how bad the floptana is. I'll likely do the rooftop test in the next few days.
If anyone would like to send me track logs, I'll be more than happy to create charts and give them the spreadsheet. Might be nice to get something more than a "sample size of one". It only takes me a couple of minutes to do them.
Rooftop results... were extremely interesting... at least to me.
Drum roll please!
Four hours beginning at 06:00UTC today...
Table of results:
More flying spaghetti monster
So that's the objective information.
Floptana in the nearly ideal conditions on the roof produces *much* better results than in the less than ideal conditions of the yard.
My yard has clear view above about 15-20 degrees but is surrounded by block walls and a house that is also block constructed. There are two trees about 20' tall, one to the east and one to the west. I chose a point in the middle of all this about 4' above the ground around nothing conductive except the wiring to power the units.
The test spot on the roof was at the highest point on top of my heat pump along the apex line. Completely clear view to the horizon and with the exception of the downward sloping roof no nearby sources for multipath.
Floptana is still significantly worse than the other two which always perform pretty close. The 76 consistently wins and I expect that is due mostly to the quadrifiler helix antenna.
However when we take the floptana to less than ideal, its positioning performance goes to hell. My yard is actually pretty good conditions for a GPS, certainly far better than in an enclosed vehicle and better than a lot of real world. Maybe it's very sensitive to multipath? It always seems to have satellites although the logs don't give me useful stuff like DOP, SNR or number of satellites so I don't really know whether the satellites are dropping or it's just noise/multipath.
Regardless, this may explain why some folks don't always see the same problems with the map orientation being too sensitive to GNSS position noise, assuming they have enough.... errr, I won't go there... to understand and repeat the same conditions that create the map orientation problems so many users complain of.
It might be interesting to do some more comparisons of the 76 (seems a pretty good baseline) and floptana in various conditions. These take very little time to setup and document. The ublox is more of a pain anyway.....
These are very interesting tests, Steve. I assume you were using the most current firmware available for the Montana?
I would be very interested to see the results from identical testing performed with a Montana loaded with Firmware 3.80, or older.
I think you might be surprised also.
Dang.... good point. I have to admit to being pretty complacent about updating. I usually only do it when I'm doing a bunch of other stuff with the unit (e.g. backing up, moving all the archives to storage, cleaning up all the .gpx's, etc.), not just sucking tracks from it. It's running 5.1.
I'll update it sometime in the next few days (prolly the weekend) and repeat the run in the yard with it and the 76. Very easy to do. The power strip with the usb power cords is sitting on the table and I hadn't put the extension cord away yet...
I'll be surprised if it's much different. Be nice if it was. If it is a multipath problem there are usually more important things than firmware although it has a role in it.
I ought to take a look at the solar flux and geomagnetic data to see if anything's been up too.
Firmware is *the* factor for rejecting the GNSS position noise. Unfortunately that's probably much harder to test objectively. They may stuff the filtered track direction in the NMEA stream. Maybe in the garmin binary stream? Even so, I'd have to log that to a PC and parse it out and that's more time than I'm willing to spend. These tests I've been doing take about a minute to setup/takedown (the majority of the time is coiling up the extension cord!). Post processing's biggest timesuck is appending and stripping out the log data but that only takes a couple of minutes....
I said 3.80 or older.
(make a complete archival copy of your Montana before loading older firmware)
The results will be very different, and probably not in the direction you might expect.
The preceding makes no difference to me. There may be things which are not perfect about the Montana, but is there a better unit you can buy for a reasonable price? No.
Don't even think of suggesting a 76. Been there, done that, had enough of that limited kind of gps. It may track slightly better but I'm not using my gps for Surveying, so I don't care. Its limitations in every other way are far too frustrating to me.
The Montana is still the best gps you can get at the moment, IMHO.
No need to reply SteveAZ - you're welcome to your opinions as I am to mine.
Wouldn't matter to me either. Two or three meters one way or the other is of no consequence. If a couple of meters of uncertainty over a short period of time is too much, then GPS is not the right tool for the job.
Also, is the 76 filtering the data in order to improve its stationary accuracy? IF so, then that might make it LESS accurate when in motion. The Montana may actually be giving the most accurate data that the GPS system has to offer, and only appears worse because it's not hiding the true results through filtering. A proper test for moto use would be a test while in motion, including errors due to time delays introduced in filtering.