Galileo ephemeris selection and TGD correction

[derekng201222]

I am creating my own program for single-point positioning and I want to cross-check my results with RTKLIB. However, I have encountered problems regarding Galileo ephemeris selection and TGD correction. When selecting the ephemeris for satellite clock error calculation, I think it skips the F/NAV ephemerides (only the ephemeris with bit 9 set in the data source field and with the smallest time difference to the observation time is considered). Am I correct in this ephemeris selection procedure?

Next, there is the correction of the observed pseudorange. The code seems to get the TGD from the last field of broadcast orbit 6 of the first ephemeris in the nav file. I obtained different results here.

I am using the broadcast ephemeris data product from NASA, so it has both I/NAV and F/NAV for each Galileo satellite in the same file. Each message type may contain multiple entries at the same epoch. Because of that, the F/NAV message is always selected if the first ephemeris is used, regardless of the message type. Should I restrict the selection criteria to only the I/NAV type?

I would also like to know whether the TGD value always comes from the third value of broadcast orbit 6.