L1 vs L1/L2 Post processing
This is a strange bug? maybe?
I downloaded GPS observations & I get a ~0.30m difference between L1 only processing & L1/L2 processing using GNSS Solutions.
In the field, I forgot to bring the dongle so it was processed using only L1 data. The vector came out as fixed. Problem was the vector distance disagreed with the TS distance by 0.30 m. Odd.
So when I got back to the office, I reprocessed using the L1/L2 with the USB dongle. The vector now agrees with the TS to within 0.02 m.
A different solution is expected, but that's quite a difference. How long is the vector between points?
The reason for multi-frequency combinations, in addition to solving for ionospheric effects, is to allow the combination of frequency data in order to accurately determine the integer unknowns. While your L1 solution may indicate it is fixed, the correct integers may not have been determined.
For long baselines frequency combinations are needed. For short baselines an L1 solution is ideally better as the frequency combination process introduces noise into the solution. What is a short baseline? Some say 5 others 10 kilometers.
When I did processing, we always computed an iono-free solution to assist in integer fixing and then used the solved integers in an L1-only.
For Short vectors L1 is sometimes better, but I have had few cases where I could not live with L1 only solutions. At times I have discarded the L1/L2 solution.
I do not think your big difference had to do with the fact that it was L1 only. I would suspect that you had better orbit data by the time you got back in the office. The biggest difference being between broadcast and post orbits, igu, igr or igs.
Paul in PA
If you shot it with a total station, I would suspect the baseline length was suitable for L1 only. Perhaps your observation time was too short? I've had my complaints with GNSS Solutions in the past, but the processor itself has always been impressive. Can you repeat the bad solution now?