On Nov 1, 3:21*pm, Alan Browne
wrote:

That wouldn't fly far - there are only so many viable gold codes -
though possibly many more on L5 with its longer code length.


There are actually around 500 balanced (roughly equal number of 0s and
1s) Gold codes in GPS, if you ignore the 2-tap mechanization shown in
the ICD. The 4-asterisk footnote was added to Tables 3-I and 3-II a
few years ago when the first list of expanded codes was published in
the ICD ( " **** The two-tap coder utilized here is only an example
implementation that generates a limited set of valid C/A codes.").

With zero Doppler difference between two PRNs, any pair of the 500 or
so balanced Gold codes would have the same peak cross-correlations.
The cross-correlation peak comes up a few dB at some Doppler
differences. Gold's 1967 papers showed that the zero-Doppler peak for
a 10-bit code is limited to 20*log(65/1023) of 20*log(63/1023), where
63 and 65 represent the excess of bit by bit agreements over
disagreements (or vice versa) between the two codes at a given time
offset. His papers also showed the probability of occurrence.

The log works out to about -24 dB. But user antenna gain, as well as
differences in satellite power, can bring the peak up more.

The bigger issue is that the broadcast ephemeris doesn't include the
PRN number. It's just assumed by the receiver that it's tracking the
one it wanted. But that's not a deficiency of the Gold codes, nor
really a deficiency of the original signal design, which is quite
elegant -- an awful lot of information packed into very few bits.

All just a quibble -- you're right that the newer signals with longer
codes will work better.