They’re not? Since when, dammit? Why was I not informed??!
Ah, well, screw it.
A neat Java applet showing lots of satellite positions (Satellite>Select lets you highlight one of the GPS satellites, for example).
That’s pretty sweet, thanks for sharing. Those GPS birds are really out there.
They sent out an e-mail - didn’t you get it?
The biggest reason GPS satellites aren’t in geosynchronous orbit is that if they were all in a single ring, it would be hard to estimate your position accurately. You want them to appear scattered broadly over the sky from your position.
GPS satellites know the time with stunning, spectacular accuracy. They send out a stream of signals that follow a long repeating pattern so that a receiver can recognize where in the pattern it is. The receiver guesses a correct time (it can’t get the time from the satellites directly because it’s using the same degree of freedom to figure out how far it is from the satellite, right?). It then figures out how far it would be from each satellite based on its stated position and on where in that satellite’s long repeating pattern it is. It only knows the distance, in other words, based on its guess about the time.
Then, it compares its distances from all the satellites, looking for a single point that would be all those distances from all those satellites. There won’t be such a point, because the time guess was wrong - but adjusting the guess one way will increase the scatter in point estimates whereas adjusting the guess the other way will reduce it. It keeps adjusting its guess about the time until the total system error estimate is minimized - then the position estimate is a good one.
In the process the GPS device has to develop an estimate of the correct time of day to within about a nanosecond for each foot of error it delivers. So if its position estimate is drifting within about 10 feet of the correct mean, its estimate of time of day is good to about a ten millionth of a second.
So if they’re so damn sophisticated, and they take relativistic effects into account as per a recent GQ thread, do they do something to account for continental shift and its effects on ground station location? Hey, you can’t sacrifice any precision here!
In an attempt to contribute something to this thread to show that I know a little about this, and I’m insecure:
Napier, you forgot to start off the explanation with a ‘geosynchronous orbits are, by necessity, right above (or very near) the equator - thus, in a single ring.’
Geosynchronous things have to be right above the equator, because they have to travel great circles (at the same speed as the earth spins), and all other great circles would make the earth wobble (from their point of view) underneath them.
I recently became interested in satellite tracking as it pertains to amateur radio. One of the things I was startled to learn was the tiny amount of information needed to calculate a satellite orbit. Below is what is called the TLE (two line element) information for one of the GPS satellites.
GPS BII-05 (PRN 17)
1 20361U 89097A 04351.92204424 -.00000051 00000-0 00000+0 0 2792
2 20361 55.3358 290.8750 0175418 205.7918 153.3326 2.00561084100585
That is all that is needed to calculate its orbit. So one can see that a receiver downloading the almanac for 24 satellites is not as difficult as it might sound.
Just a little terminology nitpick…
Geosynchronous is any orbit that has a period equal to the time it takes for one rotation of the Earth, regardless of where it puts the satellite over the Earth’s surface at any time. What you’re describing is better referred to as a geostationary orbit. All geostationary orbits are geosynchronous, but not the other way around.
And, as was mentioned above, most GPS satellites are in neither of these orbits. Except for two [one for the east coast, and one for the west coast] that are transmitting WAAS [Wide Area Augmentation System] data. Although maybe these aren’t the only two anymore.
Bup, yes, you’re quite right.
Servo, gee, yes, you’re right, Bup’s not quite right.
While we’re at it, would an orbit with a period equal to Earth’s rotation BUT IN THE OPPOSITE DIRECTION be geosynchronous? Yes, of course, it’d be difficult and perhaps pointless to put a satellite there, but I was just curious.
I’ve been very interested in the Garmin GPS 17HVS receiver. Less than $200 and yet it offers some pretty fancy things, like carrier phase information in binary format. Every second, the phase angle of the carrier for one satellite (which runs at over a billion cycles per second) is reported to one part in 2048, and the number of cycles since the last report is given as a long integer. Over half a minute or whatever this reporting system cycles through all the satellites (or all the ones it’s receiving, I guess). So the resolution of this system is smaller than half a trillionth of a second, in which the radio signal is going to travel something like the thickness of paper. I don’t know exactly how accurate the clocks on these satellites is, but believe it’s about this good over a day (of course my receiver’s clock is going to diminish this accuracy considerably for the number I read, to say nothing of the much bigger atmospheric and other errors).
Whoops! Whoa!
Driving along a little while ago, sifting and winnowing in my mind, it occurred to me what a brainfart I’d just documented.
GPS satellite clock rates are good to something like 1 in 10^12 over the course of a day, but that’s not 10^12 seconds per day. A day is almost 10^5 seconds long. The satellite clock could drift 100 ns in a day at that rate. I think they do other things to improve on that, but there’s nowhere near 10^-12 accuracy in the time of day.
What a Dope!