[time-nuts] Carrier phase tracking

Magnus Danielson cfmd at bredband.net
Mon Feb 19 16:25:57 EST 2007


From: Didier Juges <didier at cox.net>
Subject: Re: [time-nuts] Carrier phase tracking
Date: Mon, 19 Feb 2007 11:19:28 -0600
Message-ID: <45D9DC20.8060807 at cox.net>

> Layman explanation, be nice to me please...
> 
> The notion of carrier with a spread spectrum system is theoretical. 
> There is no "carrier" signal being sent continuously and modulation 
> sidebands that contain the information, as you would with an AM signal. 
> This is more like an FM signal, where the carrier is not always present 
> in the transmitted signal, depending on the modulation index, or like 
> single sideband, where the carrier is purposely removed. If you looked 
> at the GPS signal with a spectrum analyzer, it would look just like 
> noise. You need the right correlator to see something.
> 
> The "carrier" information is reconstituted in the receiver by software 
> algorithms, which essentially remove the modulation to compute what the 
> carrier should be like.

Well, if you use traditional tracking methods, the carrier is actually
detected in the digital frontend after it has been decoded by multiplying the
signal with the code replica (on mark) and integrated separated from that of
the code tracking integration using the code replica (early-late). This is
traditionally done in hardware, where as some moderns receivers does this
front-end processing in software. Personally I don't think it is a very wise
choise, where as others raise it to the skyes. Sigh. Ah well. You can't argue
with those blinded in their beleif. Being able to do it in software is cool
thought.

> Didier KO4BB
> 
> PS: it used to be early cheap GPS receivers could only decode one 
> satellite at a time, so they had to train on the signal from one 
> satellite, then decode it, then switch to the next and so on in 
> sequence. This delayed the availability of a fix by a lot, and tracking 
> while moving, well, sucked. Then parallel receivers appeared, where the 
> signal processor was powerful enough to decode 4, then 6, then 8 then 
> all 12 signals at the same time (in parallel).

These signal processor is actually dedicated ASICs.

> You may remember when parallel receivers became popular, all advertisements
> would prominently display that feature. Now, you take it for granted. There
> are at most 12 visible satellites at the same time, so there is no need for
> more than 12 channels (at least for a "single frequency" L1 receiver)

Poul-Henning touched on this, and I agree. But when you have 12 channels, you
should be pretty well covered. But time-nuts would not be time-nuts if they
didn't go that extra distance for ultra-performance. :)

Cheers,
Magnus



More information about the time-nuts mailing list