I finally managed to do some tests in a better environment with the NEO-6T dongles and they do work great. Also… I purchased TW-2410 antennas through digikey with 5m cable and they work great. They’re possibly a bit overkill and more expensive ($75). Since NEO-6T doesn’t use Glonass, it’s likely that L1 GPS antennas are a better fit costwise. Anyway, with these I get 3-4 sats in green, 1-2 others > 40dB and the others lower. RTKLIB is frequently using 6-8 satellites to calculate a solution. Magnet antennas seem like a great way to fix this on cars to improve the groundplane, but when I was about to put this in my bag along with my computer and SATA HD I luckily had a very quick second thought :). So be careful how you transport them.
With static mode on my rover, I got a fix in 30 seconds through RTK. Amazing result. Although I must say the fix was established through fix/hold and this is very risky, because I noticed it got stuck in a specific integer solution and sometimes the wrong one.
I noticed that the position indicated as base station also makes a difference in how fast and how well the rover gets a fix. So a better indicated base position helps to achieve a better and faster fix and helps especially to maintain the kinematic fix.
Unfortunately, searching google for rtklib and “base station”, no links describe what’s the best process to determine the base station position if you do an independent setup in the field. Most links already assume some sort of stream that you have access to that can be used.
One link suggested to use “SINGLE” mode on the base station setup to do this, but this provides very little practical information. Obviously, as SINGLE doesn’t converge on screen, it seems it’s similar to having a normal GPS trying to determine its position. It keeps fluctuating and the ratio / numbers describe the variation of the signal in particular directions.
The way I interpret this and what I ask you to confirm is this:
- Set up base station + antenna and make this fixed, otherwise you have to start all over again.
- Run single mode for up to 6 hours and make sure to collect a log of the stream (not solution necessarily)
- Convert the log into RINEX file format with RTKCONV
- Post process the rinex files to obtain an average of the position.
- Use the average as the indicated base position.
Or is there a better way that provides better results?