GPS errors

[ashley]

I've only got one GPS source at the moment, the trusty garmin 128, but will probably be adding a USB mouse to use with the laptop, as mentioned by someone else a few topics ago.
It's been cited by some that this would increase accuracy.

Could this be the case or is I am only benefiting from the increased equipment redundancy.
Surely both GPS units would show identical signals unless the satellites were being tested/faulty or being tampered with, in which case no amount of GPSs would help

<hr width=100% size=1>Live life to the fullest... think of all the people on the Titanic who passed up chocolate dessert.

 

[Talbot]

Whoever said that this would increase acuracy was in dreamland. Using WAAS, EGNOS, or differential GPS will increase accuracy. But at the accuracy you already have, why bother. The only real advantages to using a second GPS machine onboard is the redundancy and increased flexibility onboard.

<hr width=100% size=1>

 

[oldharry]

I dont really see how having 2 GPS sets aboard 'increases accuracy'. In my experience when they disagree then it simply increases the 'area of uncertainty'. Which one IS right? OR are they both wrong.... But as the error is nearly always well within visual range, it doesnt matter anyway.

After all, those of us still alive who learned to navigate pre GPS, had to live with much big areas of uncertainty, and survived it....

<hr width=100% size=1>

 

[Its_Only_Money]

The good thing about the Garmin 128 is that it gives you a meaningful EPE (est position error), mine frequently gives me a reassuring 13ft - on a 23ft boat that is more than enough /forums/images/icons/wink.gif As far as I know most GPS chip engines work with a max of 12 satellites (as does the 128), and I have never had all 12 in view despite a good antennae position...so even if the laptop s/w claims a higher number of theoretical satellites tracked I'd be cautious about the advantage in practice.

As I have no requirement to drop any kind of munition through my forward cabin deck-hatch without touching the sides I don't intend to bother with any of the accuracy-increasing/cash-consuming add-ons in the latest units either.....unless anyone can tell me of some must-have feature I'll be missing????

In assessing relative accuracy you'll have to take into account any differences in the view of the sky of the USB mouse and the 128's external antennae, it would be quite simple to have the USB unit have a significantly worse view and hence poorer accuracy due to its tracking fewer satellites.

<hr width=100% size=1>Rgds

Simon
Its Only Money
Fairline Sprint
Solent-based

 

[Koeketiene]

In my experience you'l just be benefiting from the increased equipment redundancy. A sensible precaution.

Doesn't add to accuracy at all. After all: a man with a watch knows the time; a man with two watches is never sure.

<hr width=100% size=1>Experience is a good teacher, but she sends in terrific bills.

 

[tome]

The Garmin 128 gives an EPE based on geometry only, which isn't very useful. All the same, it's a good receiver (I have one on board).

There are 24 active satellites in the GPS constellation and at least half of them will be round the other side of the earth so unusable, hence 12 channel receivers. In practice, you will rarely see more than 10 or less than 6 if you have a clear view of the sky. Some of them will be low elevation and therefore excluded from the fix. For a 3D fix you need 4 sats.

Adding a second receiver won't improve accuracy unless it's SBAS enabled (WAAS,EGNOS, or MSAS). SBAS provides 2 advantages:

1) Improved accuracy via differential corrections
2) Improved integrity monitoring including good error estimates

EGNOS (the european solution) is not yet operational, but is doing test transmissions. WAAS in the states is fully operational. We've observed EGNOS giving bigger errors than uncorrected GPS here in the UK, but this will change when it goes live.

Agree with your comments about antenna position.

<hr width=100% size=1>

 

[Its_Only_Money]

" The Garmin 128 gives an EPE based on geometry only, which isn't very useful"

What other error sources are there now that SA has been turned off?

<hr width=100% size=1>Rgds

Simon
Its Only Money
Fairline Sprint
Solent-based

 

[tome]

The other principle errors are multipath, ionospheric, and ephemeris. SBAS detects these also and gives you a useful quality indicator based on ground monitoring. It's designed for critical applications (including flight) andd will give integrity warnings within 6 seconds.

There was an incident earlier this year when a satellite clock went out of spec and wasn't detected by the ground segment for some hours, causing receivers to 'run away' at about 15 knots. SBAS would have prevented this.

<hr width=100% size=1>

 

[Its_Only_Money]

Ta - learn something every day....

<hr width=100% size=1>Rgds

Simon
Its Only Money
Fairline Sprint
Solent-based

 

[anabel]

But remember, the man with a stopped watch has the correct time twice a day !

<hr width=100% size=1>

 

[fireball]

and if he has 2 watches that stop at different times - thats 4 times a day!! Much more acurate!!

If your having to navigate to within 50' via GPS then I'd worry anyway!

<hr width=100% size=1>

 

[Oldhand]

<<For a 3D fix you need 4 sats>>

I thought you needed 4 sats for a reasonable 2D fix, the 4th Sat is needed to work out the error in the receiver clock and correct for it. Thus you really need 5 Sats for a 3D fix, don't you?

<hr width=100% size=1>

 

[tome]

The number of unknowns equals the number of observations required to solve the equation. Being an oldhand, remember you needed 2 RDF bearings to get a 2D fix? A third bearing gave you what is known as an over-determined solution, ie the cocked hat. This then gives you a quality indication as to how good your fix is.

With GPS there is an extra error - Time. Because the signals travel at 186,000 miles per second from 11,000 miles away, a tiny clock error in the receiver equals a big position error. This has to be corrected also hence an extra measurement required. So, for a 2D solution you need 3 sats but this doesn't give you a cocked hat whereas 4 does.

It follows that you need 5 sats for an over-determined 3D solution. Knowing that we are at sea level helps on boats, but not when climbing mountains.

Hope this makes sense!

<hr width=100% size=1>

 

[Gunfleet]

could you explain a bit further

For those of us who haven't worked on survey ships.
<<Because the signals travel at 186,000 miles per second from 11,000 miles away, a tiny clock error in the receiver equals a big position error.>>
How does that work Tome? 186,000 miles per second from 11,000 miles away means the signal takes abt 16th of a second to get here (if I've been counting my fingers right). If I'm doing 4 knots and the satellites are geostationary why doesn't it require a gross time error before I notice?


<hr width=100% size=1>

 

[tome]

Re: could you explain a bit further

Blimey, I might have cornered myself here. It's a difficult subject to explain in text on a forum without any diagrams.

You're quite correct- it takes about 1/16th of a second to travel 11,000 miles. GPS works by measuring very accurately the delay time from the GPS signals ex satellites to the receiver. An error in the receiver clock of a mere 1 millionth of a second will add an equivalent error of 344 metres to the range of that satellite.

GPS is currently running at about 10m total error budget, so that magnitude of error is unacceptable. We could put more accurate clocks into the receivers (eg rubidium or gallium) but that would add an extra £20,000 to the cost of yer Garmin.

So the answer is to make time another unknown, and to solve that also as part of the GPS solution.

Does that make sense, or have I muddied the waters further? BTW, the satellites are not geo-stationary, but travelling in inclined polar orbits.



<hr width=100% size=1>

 

[Gunfleet]

Re: could you explain a bit further

Great reply. it has given me so many things to think about I can't reply!

<hr width=100% size=1>

 

[BrendanS]

Re: could you explain a bit further

I can't compete with Tom in terms of knowledge, but this is one of my favourite subjects, and I've built up a veritable wealth of bookmarks on interesting GPS sites.

The effect Tom talks about in which you ignore the time measurement, and calculate the position independent of time measurements is explained very well here:

<A target="_blank" HREF=http://www.forestry.umt.edu/academics/courses/For351/GPS.htm>http://www.forestry.umt.edu/academics/courses/For351/GPS.htm</A>

Assume 2-dimensional space and think of distance in terms of time

With perfect clocks - we are 4 sec from satellite A and 6 sec from sat B (in 2-d this is enough to locate us at a point)

If we used a receiver that was a second fast - it thinks we are 5 sec from satellite A and 7 sec from satellite B therefore our calculated position would be miles off

Add another satellite, with perfect clocks, satellite C is 8 sec away
With the one second receiver error, all three incorrect measurements will not intersect

The computer in the receiver trims off time from all measurements until it finds an answer that lets all ranges go through one point

In 3-d we need to make 4 measurements to cancel out any clock error therefore need 4 satellites above the horizon for accurate positioning




<hr width=100% size=1>Me transmitte sursum, caledoni

 

[Gunfleet]

Re: could you explain a bit further

Tome and Brendan
I think I've got it. If you're using a sort of psuedo time derived from the signals any inaccuracies at the satellite clock will lead to very large discrepancies in the ranging calculations and therefore inaccurate cocked hat. Yes?

<hr width=100% size=1>

 

[tcm]

Re: equipment redundancy

i note that some people have more than one gps, which is a start, but not that good. The idea is that you'll always have a spare. But what if the first one goes pop as you go out of the marina? That wd mean you are on just one again, so you need three. There again ,this is so sensible that three gps's on board will probably become the norm, perhaps even a regulation so four is a better bet, all with separate power sources. Plus another one allowing for the fact that with four machines you've multiplied up the chances of a failure, i suppose. Five gps's should do. i think....

<hr width=100% size=1>

 

[BrendanS]

Re: could you explain a bit further

The times at the satellite clocks are extremely accurate, and each satellite carries 4 atomic clocks!. Take those as given (typically!)

The discrepancies more commonly occur at the receiver (your gps) for a whole variety of reasons.

But yes, if hugely wild cocked hat occurs, or if the cocked hat is way off where the gps thinks it should be (hugely simplified), the computer in your gps can ignore the time signals, and calculate an accurate cocked hat by working out where the intersection should actually be if it ignores the time signals


For a techie delve into what can be accomplished with commercial receivers, rather than what we typically have on leisure boats, this is a longish but worthwhile read
<A target="_blank" HREF=http://www.gpsworld.com/gpsworld/article/articleDetail.jsp?id=132983&pageID=1&sk=&date=>http://www.gpsworld.com/gpsworld/article/articleDetail.jsp?id=132983&pageID=1&sk=&date=</A>

<hr width=100% size=1>Me transmitte sursum, caledoni