M. Adam Davis wrote: > On 12/4/06, Alan B. Pearce wrote: > > This seems to be a benefit of having many more parallel channels > > in the receiver. More channels allows it to try so many more > > satellite combinations in parallel, so it finds them faster. > > I recently learned why some GPS receivers claim to be 16 channel, when > it's useless to listen to more than 12 satellites at once (There are > 24 satellites, a typical receiver cannot see more than 12 since the > earth blocks at least half of them). > > Some extra channels are reserved for WAAS augmentation. > > But most newer chipsets only have 8-10 correlators - these are the > computationally intensive bits of the receiver that decode the GPS > signal. The remaining channels are fairly dumb receivers that give > signal strength, and they are constantly changing frequency looking > for all available satellite signals and the associated signal > strength. When one is found that is better than one currently being > decoded, a correlator is assigned to it. Not quite. There is only one RF/IF path through the receiver, and *all* of the "channels" are correlator channels that process the data stream coming out of the A/D converter. Remember, the GPS signals from all of the satellites are spread-spectrum signals that all occupy the same space in the frequency spectrum. Furthermore, the signal strength over that space is about 10 to 20 dB *below* the thermal noise floor at the antenna. In other words, you can't even detect a GPS signal at all unless you have a correlator properly aligned with it. It's the "coding gain" of the spread-spectrum system that gets you a signal you can actually recognize, and the code-division multiplexing (just like with modern cellphones) is what allows all of the satellites to occupy the same frequency spectrum. You are correct in that a few correlator channels are usually being used to search for additional satellites as a background activity, and they can also be used to decode WAAS. However, it may be possible that a receiver uses more than one correlator to track a given satellite for the navigation solution. The technique is to keep one running a little "early" with respect to the satellite signal, and another one running a little "late". This can give you a bit more precision in picking the actual time delay of the signal, relative to finding the "hump" at the center with a single correlator. It also gives you more information about which way to track when the phase changes. > The reason a new GPS may take several minutes to boot up is that it > doesn't have a current almanac. That's only part of the story. The bigger issue is that on a cold-start, a receiver doesn't even know what the date/time is, or where it is in the world. It has to do a "blind" search through a 2-dimensional space (code phase and Doppler) just to find a single usable satellite signal. This takes time, and the more channels you can throw at it, the better. However, once it has the first satellite, a lot happens quickly. The receiver now knows what time it is, and since it has some basic information about satellite orbits programmed into it at the factory, it can now figure out which side of the Earth it is on, and which other satellites are likely to be visible and approximately what their Doppler shifts are going to be. This narrows the search space for additional satellite signals considerably. > I wonder how difficult it would be to make a simple receiver > that simply does an FFT on the GPS spectrum. Impossible. All you'd see is the thermal noise. > I'm still waiting for a one channel GPS receiver made only for > synchronizing clocks. They used to make one-channel receivers, before correlator hardware was reduced to a single chip. The tradeoff is that your local clock needs to be extremely stable, in order to provide a useful navigation solution while multiplexing the one channel among the available satellite signals. With multiple parallel channels, the long-term stability of your clock becomes a non-issue, since it drops out of the difference equations. There's no reason to build a modern receiver with only one channel, since you get 12 or 16 channels on a single chip anyway, and even in a time-only application, this has a huge advantage at start-up time. GPS-based time reference receivers *do* maintain their full timing specs even when only one satellite is available. -- Dave Tweed -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist