The next installment in my Nexus 7 tests has been a proper field test, collecting more meaningful data. For this test, as before, using a Qstarz GPS818X bluetooth GPS receiver running through the Bluetooth GPS Provider, points were recorded on the tablet using the Droid2CAD app. The same features were recorded with a Garmin Geko GPS in WAAS enabled mode. The bluetooth GPSs was able to acquire a differential lock and the accuracy was under 3m, getting as low as 1.8m. The Geko also acquired a differential lock. The accuracy of each point was not recorded but was between 6 and 2m.
The drawing above shows the Qstarz points as triangles and the Geko points as circles. The results are close but not exactly the same. One reason is that the feature recorded, an old coal shaft, was covered in heather which made it difficult to judge where the edge was. The difference is between 1 and 4m but could be simply put down operator error, not walking the same route around the shaft mound. In the center is a roughly square area of different vegetation which is interpreted as the position of the shaft. This should have been very similar, however it is not. The shape and proportions are similar but there is a slight difference in size and, as can be seen above, the bluetooth results are offset by approximately 3m. Plotted against Google Earth satellite imagery the geko points are a closer fit than the Qstarz points. In order to explore this difference further and get a better understanding of the inherent errors my next test will be on a site where I can more accurately record the same detail using different GPS hardware.
This took place on a local sports field recording points around a rugby pitch.
The points plotted above over a 5m grid represent tests on two different days following the same methodology. On both days the points were recoded with the Nexus 7 and Garmin Geko. Both device achieved a differential lock. The pale blue triangles are the Geko points and the diamonds were recorded with the Nexus. These represent points from the Qstarz GPS818X bluetooth GPS receiver with a set, coloured red, coming from the on board GPS. The latter without a differential lock. All the points have a similar clustering, with occasional outliers. The Geko reported accuracy from 6m to 2m. The Nexus with Bluetooth GPS quoted between 3.1 and 2.1m. The internal GPS gave an accuracy of 6-12m.
My analysis based purely on observation of the plot is that there is little difference between the receivers, with perhaps a slightly tighter clustering from the Gecko. This despite the differences in the reported accuracy.