An Overview of GPS errors and variation for three GPS units.
Click maps for larger views. Back arrow <= to return to document.

This presentation is an overview of the accuracy and consistency of three gps units. The first is the older model PN-20 Delorme gps, the second is the newer PN-40 which incorporates a different chip set and dual processors. The third is the Bluetooth BT-335 gps data logger by the GlobalSat Technology Corporation. The PN-20 and PN-4 0 have the STMicroelectronics chip sets, and the PN-40's is identified as the Cartesio chip set. They do not identify the PN-20 specific chip set. The BT-355 uses the SiRF Star III chip set. In general the BT-335 acquires a signal (indicated only by a blinking blue led) very quickly. The PN-40 is a great improvement over the PN-20 in satellite acquisition and screen refresh, and it appears is a little better in signal multipath rejection. And of course, the BT-335 data logger has no screen so it can only log data. If you have a Bluetooth receiver in a laptop computer or other hand held device, and have gps mapping software installed, you can use the system as a mapping gps.

The following two trips were used while recording data on all three of the gps units. The first was on Dutch Island in Narragansett Bay, Rhode Island. The test was in the spring when the trees were free from leaves and on a sunny day. This test included walking into and through bunkers on the old fortifications. Some of the bunkers were deep concrete where no signal could be obtained.

The second trip was down a the Wood River in Rhode Island on a clear day in the Spring, again before the trees developed leaves.

Some of the larger maps may take some time if you use dial up internet access. The software used for this presentation is the Delorme Topo 7 which comes bundled with the PN-20 and PN-40. The BT-335 data was saved in gpx format and imported into the Delorme Topo 7 software. The maps show the scale in the lower right corner. The PN-20 tracks are in green, the PN-40 are in red, and the BT-335 tracks in blue.


Dutch Island, Narragansett Bay, Comparisons


Overview of the area hiked.

This map above shows the lower portion of Dutch Island where the hike occurred. Note that all three gps had drastic errors at some time. The PN-40 and BT-335 were carried in shoulder pouches on a backpack, but the PN-20 was carried in a fanny pack by a separate person. This may explain some of the variation in the green PN-20 tracks, with the separation between the two carriers not more than 20-30 ft. Also the body could block the signal somewhat in the forward direction, which could also explain some of the variations.


The bottom tip of the island which has the lighthouse.

The land form was quite open towards the bottom of the island walking to the lighthouse, and the tracks are the most consistent of all on the island.

This map above is in the area that has granite Civil War fortifications and underground powder bunkers. The green PN-20 has some drastic errors when walking into and through the magazine bunkers. The green PN-20 island occurred out in the open.


Tracks around the Odgen Fortifications at the lower center.

The Ogden fortifications face south and can block the satellites in that direction, accounting for some of the PN-20 track jumps. Note that the Delorme systems seem to average the signals, which accounts for the quick jump out and averaging back to the proper location. The BT-335 logger does not average and was set to record every 30 seconds, ignoring data if the position did not vary more than 10 ft. The cluster of tracks at the lower right were while standing around and talking to other hikers for a period of time.


Tracks passing up through a deer stand in open brush.

In this section of the hike, the PN-20 seems to divert a fair amount from the PN-40 and BT-335. This section was open brush, although dense at times, no leaves were present.


The tracks around Hale fortifications.

The concentrated tracks are in the area of the large Hale gun emplacement fortifications. Even in this area the PN-40 and the BT-335 had large jumps in accuracy. This section is confused because we walked on top and into the bunkers of the Hale fortifications. All three gps units had trouble, even the BT-335 which in general is a very stable recorder. Looking at the tracks of the BT-335, you notice that the signal points are not averaged but jump around randomly. The red track for the PN-40 are averaged after the initial errant signal.


The start and finish of the hike at the beach landing.

This section at the beach has some major deviations that are unexplained. In general the observation is that the PN-40 is more accurate and stable than the PN-20, and the BT-335 is pretty reliable. Obviously more comparisons will have to be made but with the 3 gps units in a very similar location.


GPS Comparisons on the Wood River, Rhode Island


The Wood River, between RT-165 and RT-3 in Wyoming.

This test was kayaking down the Wood River, Rhode Island, in the Spring before the trees and vegetation leafed out.


The start of river run at RT-165 at the Canoe-Fishing State launch area.

The gps units were started in the parking lot and installed in the kayak. The PN-20 and PN-40 were laying in the bottom of the day hatch compartment, and the BT-335 was in a lightweight waterproof polycarbonate case in the front bulkhead compartment of the kayak. In general, the tracking on this trip was better than the one on Dutch Island. Since the three units were fixed in position from each other, there is no possibility that variations are due to separation of the units as in the Dutch Island tests.


Down the river about ½ mile.

This seems like the most deviation on the open river for the whole trip, where the green PN-20 track jumps around a little.


Down the river at one of the bends.

The tracks in this photo are typical of the trip. The BT-335 was set to record every 15 seconds, while the Delorme units seem to be going about every 3 seconds.


Entering Frying Pan Pond.

This section shows very tight matching of the gps units over the open water.


Lunch and relaunch below the road at the Barberville Dam.

At lunch in the upper center of the map, the gps units were all left on in their compartments. The Delorme units appear to stop tracking when stationary, while the GT-335 is still taking data every 15 seconds, ignoring data when within a 10 ft radius. The kayak was then carried across the road and down an embankment. At this point both the PN-20 and the BT-335 had problems with accurate locations. Note that when the kayak was launched, the green PN-20 drifts high compared to the others.


Starting out on the second section of the river, below Barberville.

This shows the drift of the PN-20 for the first section of the river where there is a 10-15 ft rocky embankment on the north side of the river.


The section of the river at Skunk Rd.

This shows a slight inconsistency for the PN-20 and PN-40 when going under the concrete bridge structure. I assume because the BT-335 was taking data every 15 seconds, the signal was more stable.


The final take-out ramp at RT-3 in Wyoming.

The tracks coming down the pond show very tight tracking position over the open water.

Of the three units, the most consistent is the Delorme PN-40 although when there is an errant data point, the averaging does make the track look funny as it averages its way back to good data. So instead of just one bad point, there is one bad point and a string of bad points working back to the good location. If the target is moving, that means a moving average coming back from the errant point. I would assume that the averaging does make a nice smooth track though, giving the gps a better looking performance. The BT-335 is good that the data logging frequency is variable and that it does not average data. There are programs that will average discrete data if needed. The BT-335 unit overall is very accurate. The only problem is the very user unfriendly software that comes with the unit. Limied pdf directions on how to use it, and it does not operate exactly as stated. When the command “load” means get the data from the gps, and “load” also means put the data in a table, it is difficult to follow the proper sequence. The software was done in 2006 and only apparently upgraded for Google Earth lately. I am hoping that there is other software on the internet that can be used. So the unit is good, the software is bad.

Rev. April 21, 2009