Accuracy Of Time Displayed On Gps

[+Delta-S]

Am I correct in understanding that the GPS receiver is getting its time information from the satellites? If so, then could this be considered a very accurate time source - like the US NIST clock?

 

If it matters, my GPSr is a Magellan 315.

[+jacques0]

I think I'm safe in saying that having a GPS receiver is like having your own personal atomic clock in the palm of your hand.

 

Anyone else agree?

[+WildGooseChase]

I just compared the US NIST clock on the web and my GPSr and they are within one half second of each other. Don't know what that means though.

[+listerine]

The time in the internal clock in the GPSr is very acurate but the time displayed isnt as acurate as the interal clock. The unite dosent devote alot of power to displaying time. But it would probably be one of the most acurate clocks you would find in a house.

[+Cardinal Red]

Delta-S Posted on May 25 2005, 12:12 PM

 

Am I correct in understanding that the GPS receiver is getting its time information from the satellites?

 

After a period of inactive storage, watch the displayed time the next time the GPSr is searching for satellite acquisition.

When you get a lock you will see the time jump when it syncs with the very precise atomic clocks on the satellites.

[+jacques0]

Here is an excellent article explaining GPS time.

 

In short, the answer to your question is yes.

[+Bob Blaylock]

  I have several timepieces that work by receiving and setting themselves by the WWVB broadcast from the NIST.  Whenever I've compared these with the time displayed by my eTrex, there's always a noticable difference (on the order of 1/4 to 1/2 of a second).  There's never a perceptable difference between any of the WWVB-based timepieces, if I compare them just after they've successfully used the WWVB signal.

 

  I've occcasionally wondered which is more accurate.  As has been pointed out, the GPS ought to contain an extremely accurate account of the time, in order to be able to do its thing; perhaps the display of the time is not entirely in synch with the internal representation of it.

 

  As for the WWVB timepieces, I am approximately 900 miles from the WWVB transmitter.  Using 2,000 miles as a basis for my calculation (to make a generous allowance for the signal bouncing off the upper atmosphere rather than reaching me straight-line), I estimate that the delay between the signal being transmitted and my receiving it should be on the order of 1/100 of a second — much too small to produce a perceptable discrepancy.

 

  I guess listerine's answer is quite plausible:

The time in the internal clock in the GPSr is very acurate but the time displayed isnt as acurate as the interal clock. The unit doesn't devote alot of power to displaying time.

 

  Though I know that the GPS must internally have the most accurate time represntation, the WWVB-based timepieces, being designed and optimized for use as timepieces, probably display the most accurate time.

 

  I wonder if a more accurate display of time would be a feature found in higher-end GPS units.  Then perhaps it's like one of my old rants about clocks in computers.  You can pay thousands of dollars for a computer, and you can be sure that the clock therein will drift off by several seconds in a day, while you can spend $20 (or less) for a digital watch at Wal*Mart, and you can be quite confident that this <$20 watch will keep time to within a second per week.

[+mikeatnight]

I work for an ISP and our switches connect to to the phone company switches. The switches require perfect time otherwise it will lead to more disconnects. We have 2 GPS receivers mounted to the top of our switch room for the timing.

[+blindleader]

Whenever I've compared these with the time displayed by my eTrex, there's always a noticable difference (on the order of 1/4 to 1/2 of a second).

 

While the GPSr has extremely accurate time internally, when it is synced up on the satellites, the display of the time for you to read is far down on the list of the processor's important things to do. So I wouldn't expect the time display to be much better than the nearest second most of the time. Remember, the display does not update every microsecond or even every tenth of a second. It's more likely once per second (consumer hand held units). Still, that should be good enough for anyone relying on a visual display for time ticks. If you need better, then you probably already have a radio controlled "atomic" wrist watch.

 

Cardinal Red noted that the displayed time takes a jump when the unit gets a lock. the internal clock in a GPSr doesn't have to be very stable because it can get synced up very nicely to an atomic clocks in orbit for the short time it takes to work a fix. It really doesn't really have to be any better than the piece of junk in your desktop PC. That means it's cheap, probably the same as those produced by the hundreds of millions for the cheapest wristwatches. In practice, though, manufacturing processes are always improving, and even the cheapest quartz clocks might be very accurate by now. I have a Seiko wrist watch that needs a monthly correction that's rarely over two seconds.

 

The link by jacques0 really is very good at explaining how a GPSr gets the correct time. Just a curious irrelevant side note: It doesn't seem to mention that the UTC based time displayed by the GPSr and the time actually used by the entire GPS system are different by currently about thirteen seconds (no leap seconds), but that's another story.

[+NightPilot]

There are several time systems, including GPS, LORAN-C, UTC, and TAI. They all have different time, but the differences are precisely known, so it's easy enough to display whichever you like. If you want to see all these at once, plus your local time, go to this site. It has more than you may have wanted to know about time and timekeeping.

[+park2]

One of the older Trimble portable units had a 1pps output pin that went active for a brief period (few mS?) and it was spec'd to within 1uS accuracy. As the other posters correctly pointed out, screens back then could have errors of over a second depending on processor loading, but this pin was designed to still give users high accuracy. To use it, you would have to 'read' (with a processor) the time on the other comm pins and that time was exactly valid when the 1pps pin went active