The 4-ft long fiberglas antenna on the right is the 300Khz DGPS antenna. The short stuby post on the left is the support for the satellite antenna.
One question I get a lot is how do I ground the whip antenna? I was worried about that too. I currently have the ground strap of the antenna coupler attached to the aluminum backpack frame with some star-washers and a 6-32 bolt that goes through an unused hole in the frame. I was fully expecting to have to increase the frames coupling to my back (so that I could be part of the ground circuit) by putting some aluminum window screening under the nylon back-support. That would certainly give plenty of capacitive coupling of the frame to my back. It wasn't needed, at least not where I normally hike within 60-70km of a differential beacon.
This is the Y-cable pinout that I use for grabbing the NMEA signal from the GPS while the gps is plugged into the radio.
On May 1, 2000 the intentional degradation of GPS accuracy (called SA) was terminated. With this change a GPS without a DGPS unit attached became almost as accurate as a GPS with a DGPS attached. Here is the result of 2 separate 24 hour tests, one with DGPS and one without.
The above statistics were gathered with a home grown program I call salog. Feel free to grab the above distribution and run your own statistics. It should compile up with little difficulty under netbsd, freebsd, openbsd and linux.
If you've always wanted to see how stable the reading on your gps can be without the government-induced noise signal, but didn't want to spend the money for a DGPS radio, here is your chance. DGPS over the Internet page
This is a picture of the satellite tracks across the sky. The idea here was to use the GPS satellites for background illumination of a skyline picture in the 1.5 Ghz band.
Here is a 1024 sample point test of a stationary Garmin GPS-12xl with GBR-21 differential radio. The differential radio was tuned to the Coast Guard transmitter at Pigeon, CA (287 khz) at a distance of 57km. The USGS station mark that was used as the reference was "DRISCOLL 2", located in Fremont, CA. This mark was surveyed with a resolution of 0.3mm . (I'm not sure what the claimed accuracy is.) The sample points were taken with the tracklog set to "1 second" sample time. In reality this lead to an approximately 2 second sample rate over the span of half an hour.
A postscript version of the histogram.
It's interesting to note that the 12xl will transfer tracklogs with a resolution of approximately 9.3 mm during the transfer. (eg. It uses a number system with 2^32 == 360 degrees). Some other mechanism in the 12xl causes it to only output one out of every 64 of these 9.3 mm units, yielding an observed granularity of:
0.595 meters latitude
0.471 meters longitude (37.5 deg.)
This test was only done once and over a relatively short time so I'm not sure how meaninful it is as a statement of what is "normal".
I haven't figured out a way to get sub-meter data out of this unit without using the tracklog. The tracklog is only 1024 samples long. Reading out this tracklog requires playing with the GPS's buttons and turning off the differential correction. I can't easily automate this for a long-term test.
Long term averaging of the NMEA output shows a similar curve-shape and width. Two separate runs of 40,000 data points over a time span of 80,000 seconds each show an east-west standard deviation of 1.46 meters and a north-south standard deviation of 1.80 meters. These tests were done with using the DGPS over the Internet dgpsip server and client.
Here are some elisp snippets for converting lat/lon to distance and converting to from to/from the base-60 system (hh mm ss.s) to a saner base-10 system. (Did ancient Babylonians get invaded by a species with 60 fingers?) gps.el
This data was taken with the same differential setup as above. The positional accuracty should be +/- 2 meters or better 50% of the time.
The raw Mission Peak tracklog.
The Mission Peak waypoint list. Feel free to load this into your GPS. The data is in gpstrans format. Description of the waypoints.
The Mission Peak tracks and waypoint plotted on a postscript overlay for a 24k map. This can be printed out on transparency film and overlayed on top of the USGS "NILES California" 7.5 minute (1:24000 scale) quad. Note: the grid reference is relative to WGS-84. There are little NAD27 ticks to help in aligning it to the older NAD27 grids on the quad.
Alison's Mission Peak page with stories and pictures.
Here is a copy of my hacked version of gpstrans called gpstrans-1_1-wsr. Notable improvements over the originals sources are: 1) working checksums on receive 2) fixes some big-endian-only problems 3) compiles and runs on SunOS-4.1.4, BSDI-2.1/x86 and NetBSD-1.2/x86.
Marco Hyman has a much cleaner Garmin downloader for BSD/Linux. The format is a bit different from gpstrans, but if you don't have many old data files to stay compatible with this is a much better program to use.
Watch this space for a brand new, totally homegrown program that I call glink. (Hopefully that name isn't already taken. I hate it when that happens.) Unlike gpstrans this is a select()-based state-machine program that can easily handle asynchronous events sent from the garmin. There is no need for the program to be in a special "slave" or "master" mode. If the Garmin GPS sends it a packet, it will always be ready to record that packet. Glink is also more reliable than gpstrans on a slow cpus (like 386/486 portables) because it never does single char reads or writes. It tends to get far fewer receiver overruns.
Unfortunately Garmin hasn't yet come around to realizing that releasing some more details of their Garmin/Garmin protocol would be good for them also. This leaves my program (like all the other freely available ones) logging lots of uninterpreted data.
Punch this number into your GPS and and hit the goto-button. This is the coordinate of the parking space right in front of the USGS map store in Menlo Park, CA.
USGS 37.4548490 -122.1700144
A very strange differential GPS bug.
Here I use a common automotive vacuum pump connected up in pressure mode to blow a bit of air into the 12xl. The left battery tube is the one that has the pressure relief hole. I put a rubber stopper into it and gently pressurized the tube. Notice the air bubbles comming from 3 different spots. There is a very big leak at the top right of the window. There is a smaller leak on the top left. There is a medium size leak at the antenna connector.
Note the 12xl water-proof rating on my unit should really be called a water-resistant rating! It will stop splashes, but water sitting in the window ridge will bleed into the unit. Other 12xl owners might be well advised to test their units similarly before subjecting them to submersion in water.
One other thing worth noting is that the 12xl has a powered external antenna jack. This jack must be covered with waterproof material if the unit is to be used in a salt-water environment. The 5 volts on the center pin will cause so much electro-(un)plating in 5hrs of contact with salt water that the connector hole will be filled with a solid plug of green crumbly gunk. That gunk was formerly part of the connector.
Unless otherwise noted, all data presented here is in WGS-84 datum, usually in decimalized degree format (eg. hh.hhhhh).
Linux and BSD users: Coming soon: GPS mapping tools for BSD and Linux. www.gnomad-mapping.com
.
![[ Powered by Fedora Core ]](/icons/powered_by_fedora.png){kind=icon}
wolfgang.rupprecht+web@gmail.com
(Wolfgang S. Rupprecht)
WSRCC Home Page || Up One
Level
..