Thread: About 37xx converted to 34xx satellites reception accuracy

Continuing our interesting discussion about precision of 3490, converted or not, in this post I present another test, this time only with the original 3490, but in a different environment than the one I did earlier (post #368). At first I used the same 3490 real and a 3790 mod with Kunix # 1 fw (3790 boot.ini 3490 fw_all), but in an urban environment with many high buildings and trees in almost all the way. Both gps had a very similar result, apparently weak, as can be seen in the tracklogs off-road and lock-on-road, on the satellite photos from Google Earth. The tracks showed that the off Road were irregular and very imprecise, with considerable deviations of the roads. The tracks lock-on-road sometimes had some crazy moments deviations, such as jumps to other routes and other aberrations. Compared to mine, the tracklogs off road of Neil’s test (post #346) were much more regular in its trajectory, both for nuvis that were more accurate as for the inaccurate. So, which the question: after all, the difference to the tracklogs of my test (post # 368) was due to quality of gps units tested or what? The answer, I believe, can be from the analysis of tracks that now present. Despite having been made only with the original 3490, the tracklogs off-road (pedestrian mode) and lock-on-road (trip log enable) were quite regular and accurate, staying on the real road almost all the way, as seen in the pictures attached.

After all so what can explain such differences? Answer: the environment test. My test of the post #386 was made in a densely populated area with many buildings, trees and other obstacles. Besides the decrease of signal strength and number of satellites tuned has another very important complicating factor in these environments: the reflection of satellite signals that occurs in environments with many buildings, greatly increasing the inaccuracy in the recording of GPS positioning. Interesting is that the reflection signal can trick the gps in your calculations intensifying sometimes falsely signal strength of satellites, base time to calculate the position of the same and more, distorting the accuracy shown on screen (more or less) and in recording of tracklogs. How to know that? Just check the tracklogs off-road and see the distortions: in those recordeds in environments free of these obstacles and reflexes, the tracklogs are accurate and precise and the screen accuracy is almost constant. In my test was 3 meter, practically 100% of the path, that was not an absolute value but walked over to of the reality most of the time.. In tracklogs recorded in environments with many buildings were the opposite: imprecision and irregularity of tracklogs.

So, if any of us says not having problem with the accuracy of the 3790 converted, it seems that may not necessarily be a mere matter of a few different hardware versions, but the environment in which the measurements and navigation usually occurs. For example, let's take the Neil’s case (sorry, lol ..) in post 346: we observed by satellite photos that he seems roaming through open environments, residential areas without tall buildings, near the coast, in short, environments for privileged people, lol .. As a result of the Garmin lock-on-road system, that fixes almost all of the inaccuracies in these environments, the result of the 3790->3490 fw does not show abnormality. If anyone has any doubt about these statements can check by doing your test, even if only with his one gps, recording tracks simultaneously in lock and off road mode in both environments suggested: places with many buildings and open spaces, and check on Google earth, pretty easy..

Conclusion, now I am more satisfied with the level of accuracy of my 3490 (which even I thought about selling, lol) and best of all, I believe that I now understand a little more why occurs some distortions and aberrations in relation to apparent (on screen) and real (tracklog off road) accuracy. Lack doing this test with 3790 converted with Kunix fw # 1 (in the open environment) and with 3790->3490 fw full (in adverse environment, since Neil has done this test in open environment with this fw), what I will do as the unit back to my hands.. Thus, I believe that it is possible to reach consistent conclusions about the accuracy of GPS road in urban environments with a single device and different firmwares, because most of us do not have shop or collection of gps. Having multiple units is a privilege of a few but compare various units, is undoubtedly the ideal condition…Luckily we have a expert and “collector” that is always ready to assist us


Legend tracks: (of the same gps, simultaneously recorded)

- 3490 off road: red
- 3490 lock-on-road:yellow



[Only registered and activated users can see links. ] One of the rare moments of imprecision, justifiable by the presence of buildings and trees



[Only registered and activated users can see links. ] Lock-on-road misplaced and off-road track on the path indicates error in the gps map (where the track was fixed lock-on-road)


Below, another rare error in path, with measured deviation of 28 meters (buildings and trees). Sounds like a clear case where the GPS uses satellite signals reflected in edifications ..