[time-nuts] Thunderbolt - trouble locking with some types of antennas

[jimlux]

On 8/1/17 12:11 PM, Graham / KE9H wrote:
> Dider:
>
> This is a CDMA signal. (With a 'chip' rate that far exceeds the information
> rate.)
>
> If you put a different correlator on every multipath signal, which are each
> differently delayed in time, then they can be independently demodulated.
> (Or time shifted and added back together with some quality indicator for
> weighting.)
> So, in CDMA, multipath is used as a form of (time) diversity reception and
> will improve the signal to noise of the combined signal.
>
> By definition, the signal with the least time delay either is, or is
> closest to, the most direct path.
>
> You don't have to necessarily fully demodulate this early signal by itself,
> just know what its timing is. (And remember where it was, if fading in and
> out.)
>
> Since multi-path is a destructive mechanism in most narrow band radio
> systems, the above may not be intuitive to people not familiar with CDMA
> and "rake receiver" based systems.
>
> Multipath helps, not hurts, these systems, as long as the multipath delays
> are most of one chip apart, or more.
>

For GPS raw (off the air), when they post process at JPL, they use a 
fairly sophisticated correlation process, incorporating an estimator of 
the underlying time delay trajectory: you can use a later "big peak" to 
help recover the early "first peak", for instance.

Rake receivers (under the name "adaptive equalizers") are also why 
digital TV works fairly well in high multipath environments - with 
modern receivers that do this.