regarding SNR unit I used
https://github.com/swift-nav/piksi_firmware/pull/385 to piksi_firmware and build firmware and update,
the effect was such that, SNR rose +31 (under "Tracking" in the "pixie console" "dB-Hz" found signals were tracked over those that scan for)
[note graph here was more transparent if possible it would be excluded from the these PRN chart which "pixie" actually does not track]
S1 value of observation in RTKLIB increased but the problem remained, solutions RTKLIB as if they jump quickly from single (spp) to float and fixed Get around
the problem is probably in the fact that the acquisition to specific PRN is not continuous
eg. in RTKLIB disappears signal to a satellite which has the highest C / N0 in the "pixie console"
and so, eg. in the "pixie console" I have a constant amount of satellites np.7 and RTKLIB after conversion (SBP> RTCM) the number of satellites jumps between 3 and 6.
(view in attach RTKLIB-Nsat-plot.jpg) and fixed solutions are jumping about 50m (view in attach RTKLIB-Position-plot.jpg and RTKLIB-GndTrk-plot.jpg)
Color in plot:
Red= singiel(spp) solution
Orange= float solution
Green= Fixed solution
P.S. I think it would be best to implement the format "sbp" in RTKLIB