Unequal pseudoranges

[JohnNatHaysys]

I generated a file using a brdc file which was patched to make 4 of the satellites have exactly the same entries. This creates 4 satellites in the same place. Feeding the resulting file into GNSS-SDR should give exactly the same pseudoranges for these satellites. It does not. The diference is between 3m and 20m. I get the same difference if I feed the file into a HackRF and examine the Pseudoranges with a UBLOX F9T. There is also a slight difference in the position as calculated compared witht the position input. e,g, 51.000000 becomes 51.00003

Anyone seen this problem? Rounding errors in the calculations, but always in one direction? Since the only difference is the satellite SVN and the PRN code generated is there an error here?