Different or Same???

[+RB_Nielsen]

I start this out by stating that I am not well versed in the particular electronics of GPSr technology.

 

That being said, it seems probable that the tracking electronics in any/all GPSrs are going to be basically the same. There are only so many ways of electronically locking in and tracking a satellite signal. It appears there is just one way of doing it well, and by now it is no secret among all manufacturers of how this is done. Furthermore the cost for the electronics to do it well is not significant, and so the very finest tracking technology is commonplace in every GPSr unit made today.

 

If I am right about this (and I suspect that I am), then the difference between GPSr manufacturers and all their GPSr models is negligible insofar as the basic tracking capability is concerned. So all the hoopla amongst users about which GPSr is "best" (at least as far as tracking is concerned) is no more than that - just hoopla. Now each manufacturer would want their clientele to believe differently, I am sure.

 

The salient difference, then, between models and between manufacturers is not in the basic satellite tracking per se, but in the various other "goodies" that are packaged into each model. Some models have better screen resolution, others have better maps, some do a credible job providing turn-by-turn directions that get you to your destination, yet others have a built-in compass, etc, etc. It is these additional functions for which we lay out the cash and have our fun.

 

It seems worth noting that the sizzle and desirability of owning a particular model GPSr is really in this additional functionality. One fact that makes this assertion credible is the absence of claims by the manufacturers that their products are superior in terms of their tracking accuracy. Various users of the products certainly make such claims, but those who produce the GPSr units are, to my eye, rather silent on this point.

 

I'll end where I started - I am no maven in this arena of GPSr expertise and my naivety may be showing rather brightly, but my sense of it is that there is some truth here.

 

Comments from those more learned would be most enlightening.

[+leatherman]

 

Preparation, the first law to survival.

Mokita!

[+apersson850]

The option for the hardware manufacturer to select different types of antennas, for example, certainly does imply a difference in performance between different models, even if they are of the same heritage. Garmin manufacturs both units with patch and quad helix antennas, which don't have the same characteristics, as far as satellite tracking is concerned.

 

Still, most units can track satellites good enough to tell you where you've been, where you are and where you are going (provided you've informed your toy about your plans!). So, I agree with you that a lot of the fuzz about which unit to use, is what additional features you want. Some may be jumping for joy over the Jumping Master in the eTrex Vista, just to give one example.

 

The general user interface may also be important. Such a simple thing as if the unit can talk your language could be important, for us who knows English just by rumor, not by lifelong practice...

 

Anders

[+Alan2]

I agree that once you get past antennas and where you're operating the GPS, the accuracy between units seem to be the same in getting the coordinates. But besides raw coordinates there's other kinds of accuracy such as in the mapping. Maps that more accurately reflect the geode of the earth and actual location will better place the arrow of where you're located. This is especially important while you're navigating.

 

For example, if a topo map in your GPS is shown off by 75 feet, the map could place you on the wrong side of a stream or gully even though the coordinates read accurately. Having to hike two miles to get to a bridge to get to the other side because of this 75 foot error can be serious.

 

I'm sure there are other examples.

 

Alan

[dave and jaime]

i agree that the only difference between different units ought be the fit and finish, user interface, and add ons. since all units share a common background there should be little or no difference in practial use if all things are kept equal. having said that antenna type and other add on will cause some minor differences in performance.

 

a good way to test this theory would be to take several different units and place them in an area of even cover and give the units a couple of minutes to get a good fix. all units should display the same coordinates within the error tolerance of the unit, if when accounting for individual error tolerances the coordinates are the same then you have shown that in that specific location the gps units are created equal.

 

i would suspect that this test would confirm that all gps are created equal.

 

however it has been said that the gps recieves coordinates, i believe this is wrong. i believe what happens is the gps receives time coded signals from the satalittes and then uses a computer algorithm to compute a position. difference in coded firmware could, in theory, cause discrepancies.

 

what does all this mean, i would suspect that any differences in bare bones performance would be well within the error tolerance of the gps and when you spend more money you are buying bells and whistles not better performance.

[+apersson850]

Alan is right, but the problem for the potential buyer of a GPS unit in this case is: "How to know?"

 

As far as I can understand, there is only one way to know if the GPS you are going to buy actually has a map that's accurate enough for your needs. Buy it and try. Well, unless you have some nice friend, that can loan you a GPS for some undetermined amount of time, to let you try out all possible locations you like to go to.

 

Besides, I wouldn't be too surprised if it turns out that, on the average, all GPS units are about as good, or bad, when it comes to mapping accuracy. One may be better here, another better there. At least here in Europe, it seems that they all use maps from NavTech. If I compare the coverage of my MG Europe with the navigation system you can buy (for an awful amount of money) if you get a new Volvo today, they turn out to be almost identical. No big prize for anyone who can guess who supplies maps to Volvo.

Although I understand that the user of the maps (in this case I mean GPS manufacturer by "user") order them with different data contents, like with and without address lookup capability, I doubt that a company like NavTech makes more or less accurate maps for Garmin and Magellan, to mention two of the more well known manufacturers.

 

So, at the end it comes out to what features you want, and how you like the user interface of a particular unit. The readability and size of the display may be very important to some that's 50+ of age, but less to somebody who's 20. Perhaps.

 

Another reply came in while I was writing. You're right about that the GPS can only measure is't distance from the satellites. It then computes the position of itself. Here comes some rather complex software into the scene. There may very well be differences in different manufacturers way of handling difficult signal environments (multipath rejection, reception on different channels comes and goes etc.) There may also be differences between units of different age. At least providing not all new research goes into firmware upgrades of the old ones.

 

If I check the firmware upgrades for my unit, I can find things like "Improved WAAS reception in shaded and reflecting environments". Since they do nothing to the hardware itself, that must be improved algorithms for trying to determine what's real signals and what's just noise, coming down from the sky.

 

Anders

[+leatherman]

If you take several digital watches of the same model, you will undoubtedly have a couple that will not hold the same time. That's just one function. Not several satellite tracking and timing functions at the same time.

Like Anders said the computation of the information being received by each GPSr is processed differently. If you don't have the latest firmware, even two of the same unit will not process the signals the same.

 

So even the same isn't the same.

 

Preparation, the first law to survival.

Mokita!

[Kerry.]

The term GPSr is used to cover a fairly wide description of device capabilities.

 

In terms of accuracy, all handheld GPS receivers are created equal, they have the same basic parts and use the same basic signals. It is what they do with what they get that makes the difference (to a point).

 

The "which is best" type question appears to be more directed to the "general" user type interface, which is really only taking a continuous feed of positions from a "GPS engine" and turning those positions in something that are useful and meaningful to the user.

 

This GPS engine has its own built in firmware/software (apart from the main "user" software/firmware), which can also have some user control from user system but in many cases especially in the cheaper units (like many handhelds) configurations are hard-wired.

 

Even though one receiver might use all satellites down to the horizon another with the same engine might be hard-wired to a mask out satellites below a specific cut-off angle. In practice there might not be a lot in it but technically the masked out lower satellite model is the better option but not many users would be aware of what models have what masking cut-offs and why. Most are too tied up in the the 12 channel myth/more satellites the better syndrome.

 

Obviously there are other "refinements" in the engine function (and also the antenna) that does/might/could make some difference, which might not be immediately obvious to most users.

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[Kerry.]

The term GPSr is used to cover a fairly wide description of device capabilities.

 

In terms of accuracy, all handheld GPS receivers are created equal, they have the same basic parts and use the same basic signals. It is what they do with what they get that makes the difference (to a point).

 

The "which is best" type question appears to be more directed to the "general" user type interface, which is really only taking a continuous feed of positions from a "GPS engine" and turning those positions in something that are useful and meaningful to the user.

 

This GPS engine has its own built in firmware/software (apart from the main "user" software/firmware), which can also have some user control from user system but in many cases especially in the cheaper units (like many handhelds) configurations are hard-wired.

 

Even though one receiver might use all satellites down to the horizon another with the same engine might be hard-wired to a mask out satellites below a specific cut-off angle. In practice there might not be a lot in it but technically the masked out lower satellite model is the better option but not many users would be aware of what models have what masking cut-offs and why. Most are too tied up in the the 12 channel myth/more satellites the better syndrome.

 

Obviously there are other "refinements" in the engine function (and also the antenna) that does/might/could make some difference, which might not be immediately obvious to most users.

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[+apersson850]

For a typical consumer grade product, like my eTrex, the cut off angle isn't even stated in the specifications, let alone adjustable by the user. So there is no way for the user to know if it discriminates everything below 5° or what.

 

Anders

[Kerry.]

Anders, that's right, they don't but then not many do, so I suppose the next question is WHY not.

 

I believe there's a lot of "little" things that users of consumer grade receivers aren't told.

 

Some manufacturers won't confirm some of the things that are "known" but then they don't/won't deny them either.

 

I've a fair idea what your etrex mask is and I see no reason why it's not mentioned in the specs?

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[Kerry.]

Anders, that's right, they don't but then not many do, so I suppose the next question is WHY not.

 

I believe there's a lot of "little" things that users of consumer grade receivers aren't told.

 

Some manufacturers won't confirm some of the things that are "known" but then they don't/won't deny them either.

 

I've a fair idea what your etrex mask is and I see no reason why it's not mentioned in the specs?

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[+apersson850]

Soooo, what's your "fair idea" then, if we come down to some good old numbers? 5°? 10°?

 

Anders

[+GOT GPS?]

Would be interesting if there was a beginner course in GPS technology, and an advanced one for electronics people.

 

------------------------

My home page about GPS units and information

[Kerry.]

quote:

---

Originally posted by Anders:

Soooo, what's your "fair idea" then, if we come down to some good old numbers? 5°? 10°?

---

 

I'm not aware of any hard-wired greater than 10 degrees and some are around 5 and then 0 degrees is fairly common (actually very common).

 

For serious GPS observations the minimum is normally 15 degrees and even as high as 20 degrees for some applications but these receivers are configurable.

 

Catch 22 with WAAS is the higher the mask cut-off the less chance of receiving these at low horizon latitudes. I'm not that sure if WAAS transmits corrections for all satellites in view (even if both base and roving receiever were seeing the same sats).

 

With WAAS the base receievers have an elvation mask set then that will also dictate what satellites the roving receiver will use as it can't use satellites that coorections aren't transmitted for.

 

Similar with dGPS systems as the reference receiever also has masking set and besides they only transmit corrections for a maximum of 9 satellites in any case.

 

Augmentation systems really induce a masking cut-off angle by default.

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[Kerry.]

quote:

---

Originally posted by Anders:

Soooo, what's your "fair idea" then, if we come down to some good old numbers? 5°? 10°?

---

 

I'm not aware of any hard-wired greater than 10 degrees and some are around 5 and then 0 degrees is fairly common (actually very common).

 

For serious GPS observations the minimum is normally 15 degrees and even as high as 20 degrees for some applications but these receivers are configurable.

 

Catch 22 with WAAS is the higher the mask cut-off the less chance of receiving these at low horizon latitudes. I'm not that sure if WAAS transmits corrections for all satellites in view (even if both base and roving receiever were seeing the same sats).

 

With WAAS the base receievers have an elvation mask set then that will also dictate what satellites the roving receiver will use as it can't use satellites that coorections aren't transmitted for.

 

Similar with dGPS systems as the reference receiever also has masking set and besides they only transmit corrections for a maximum of 9 satellites in any case.

 

Augmentation systems really induce a masking cut-off angle by default.

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[dave and jaime]

just a thought : it is known the AMD cpus have issues with operating heat and potential difference when compared to say a celeron. is this a potential consideration when buying a gps, the chip manufacturer?

[+Alan2]

GPS Accuracy in a real geocache test

 

Here’s the result in real geocaching GPS accuracy. The set-up was that each cache finder to a NGS benchmark, cache http://www.geocaching.com/seek/cache_details.asp?ID=10814 would report their GPS coordinates at the benchmark to the cache web page. The benchmark center was on a 7 foot wide, 2 ½ foot high granite slab. You would have to reach to set the GPS in the center and to read it. Benchmark is in an open area although there are a couple of small trees nearby. The actual NGS coordinates are 40 44' 34.46640"(N) 073 50' 42.31409"(W) which places it in Flushing Meadow Park, Queens, NYC which is in Waas range.

 

The following GPS’s were reported used including my own Vista when I set up the cache. Readings were taken with 13 different GPSr’s on 12 different days between 11/25/01 and 3/24/02 at different times of the day. Weather conditions were not reported.

 

Garmin Vista (4)

Garmin etrex (2)

Garmin Summit

Eagle Expedition II

Magellan Map 330

Magellan 300

Unknown

“Cheap” Garmin

Garmin Street Pilot

 

Summary of results. Difference calculated from NGS coordinates. WGS84 coordinates in DD MM.MMM were converted to UTM using Mapsource program. Distances from benchmark calculated taking the square root of the sum of the squares of the difference between the UTM latitudes and UTM longitudes in meters and converting to feet.

 

12 readings under 12 feet

5 readings between 13-25 feet

1 reading at 40 feet (note that this was the Magellan 300 which is an older unit and reads to 2 decimal places only - about 50 feet accurate I’ve been told).

 

All were single readings except for Eagle (13 readings = 3.3 feet avg.) and one of the etrexs (7 readings = 25 feet avg.). There was one WAAS reading reported (my Vista at 3.3 feet).

 

I think the results speak mountains as to the accuracy of the equipment we are using and the standard algorithms they probably are all using. Of course, in heavy cover, or reflective circumstances, these results could be quite different but at least the "normal" range seems similar.

 

Since there was only one Waas measurement at one meter from the benchmark, one cannot state a general determination reagrding Waas.

 

Alan

[dave and jaime]

alan, i think what you have shown is that these gps are essentially equal in accuracy. the issue of tree cover and weather would test antenna sensitivity. i guess then that user interface ought be your major consideration when making a purchase.

[+Alan2]

dave and jaime. I agree that's why I posted the "test". I believe what's really amazing is not only the similar results regardless of manufacturer, but just how close a consumer grade GPS will get you; at least in a clear, non-obstructed location where you can stabilized the unit to acquire satellites.

 

It would be interesting to do a test under more adverse conditions to do a comparison. It would be pretty difficult to set up I would think.

 

One of the more interesting topics I would like to see is a comparison of non-hardaware features between manufacturers and models. Features such as support and service, mapping improvements, responses to upgrades requested by users, operational features, the little things that make using a particular unit more desireable. MOst "arguments" in the forum I've seen over the last year have been either around antenna, size, weight, display, etc but not software or "soft" feature type things. Maybe you guys could start that topic.

 

Alan

[Kerry.]

Accuracy is a funny thing and limited single point in time positions really doesn't give an "overall" indication of the system or receiver accuracy, they are just that a single position in time.

 

It could be dangerous (in some applications) to assume that GPS is such and such based on very limited data.

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[+Alan2]

The "test" involved a dozen and a half readings with over a dozen different units, more than a half dozen different models by different manufacturers, and equally as important taken over a period of 4 months. While the place was the same spot, it certainlky was NOT the same position in time. I think that "test" really does show the accuracy of consumer grade GPS's. I'm not aware of any other test involving so many different units checking accurancies over such a long period of time. MY hat is off to Garmin, Magellan, etc. and the US Military.

 

Alan

[+trippy1976]

We did a test of GPS accuracy in relation to another person's measure coordinates at the first MiGO gathering. The experiment was organized by Rusty and the outcome was... I dunno, interesting. Nothing conclusive.

 

[edit: Oops. Meant to post the link to the write up about the event and experiment. http://www.mi-geocaching.org/event_stories.php?itemid=20 ]

 

As far as value of a GPSr, I also suspect the radios are at least similar - the differentiator will be form factor, durability, and the software. I haven't touched anything but my Venture since I bought it. I prefer it for its size and value. Value's another good selling point. My unit does an admirable amount of thing, though nowhere near what the higher end Garmin units (Or Magellan, or...) do. But it was pretty affordable at $140 at a boat show and it came with the data cable.

 

--------

trippy1976 - Team KKF2A

[+Alan2]

Tripppy: Your 6-8 feet for the majority is consistant with the results I was getting. This could be the reason the mfr's have changed (lowered) the "circle" of error shown in GPS's as they are consistantly more accurate than was figured.

 

ALan

[Kerry.]

Alan, I realize what your saying but running a handfull of receivers at the same point regardless of the time frame is not within the meaning of GPS accuracy.

 

Just because there was some consistency and that's what it is consistency doesn't mean that if one was to take all those receivers somewhere else that the results would be the same.

 

That's the thing with GPS and the specifications (including accuracy) it is a global system, the worlds first global utility and that's what the system has to cover (the entire world) including accuracy specs. That entire world also includes a few hundred km's into space.

 

The system is currently running with all spares and things are fairly good but drop back to the design 24 and then loose any 1 or 2, which is a requirement (and possibility) of the specs then the world wide capability is on a minimum footing.

 

There are several reasons why the system is consistently more accurate than figured but why manufacturers continually want to "massage" error circles (EPE's etc) is really a bit of a con.

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[+Alan2]

quote:

---

Just because there was some consistency and that's what it is consistency doesn't mean that if one was to take all those receivers somewhere else that the results would be the same.

---

 

How do you it won't be as accurate? MY test and trippy's test showed a majority of accuracy readings under 12 feet in two different places. What tests have been done to show this kind of accuracy is not consistant in other areas?

 

Alan

[Forester II]

I think what Kerry is trying to get at is that under nearly ideal conditions, such as where your test was, is that the units could have consistent results. If the conditions were less than ideal then your results could be wildly different.

[Kerry.]

quote:

---

Originally posted by Alan2:

How do you it won't be as accurate? MY test and trippy's test showed a majority of accuracy readings under 12 feet in two different places. What tests have been done to show this kind of accuracy is not consistant in other areas?

Alan

---

 

Without hindsight you really won't know what the difference in accuracy will be. That's one of the reasons why the 95% accuracy figure is spec's the way it is and don't forget there's another unknown 5% .

 

As for tests enough. It's actually really quite easy to show there is no consistency in accuracy in different areas.

 

Just for the sake of this particular exercise the following is some random accuracy examples for 3 locations on October 7, 2002 (Doy 280). These figures are the accuracy in metres RMS for the 24 hours based on C/A L1. The receiver types are unimportant as it really doesn't matter.

 

Kingston, Jamaca - 1.9m RMS

Casey (Antartic) - 10.6m RMS

Manama (Bahrain) - 23.4m RMS

 

One can be assured that there's certainly no consistency between any of them and there certainly would be a greater "inconsistency" in the period by period (minute by minute) accuracy throughout the day at Manama compared to say Casey and to a much lesser extent Kingston.

 

The dynamics of the system simply can't be expressed that accuracy will be such and such at all places at any time, it simply don't work like that, hence the way accuracy is defined.

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[+Alan2]

Notwithstanding what ForesterII suggested that you were talking about non-ideal conditions, I believe that you, Kerry, meant from what you wrote that even under ideal conditions, there's no consistancy. I totally disagree.

 

I don't know what your figures prove Kerry. You don't explain how those figures relate to how a GPS is operating. It seems like you're trying to present some data that you got somewhere about one satellite being off a certain amount of meters depending where on the earth you're measuring it. You pull some data out of the blue and ipso facto "prove" that GPS do not give consistant accuracies like was shown on my example or Tripppys. How do these figures prove what the GPS is calculating? You're just assuming the GPS will therefore be wrong but you offer no proof. What do your figures mean relative to the coordinates the GPS is displaying? The GPS will be sampling many satellites, maybe averaging them, I don't know all the algorithms used in these units, but I'll bet you $100 that we can do the same test anywheres in the world and get the same accuracy results as long as the test is set up the same way - cLear sky and lay the GPS down so it stabilizes. Even if it's 95% of the time, that's pretty good. That means that 19 times out of 20 a dozen GPS's will be within a dozen feet of the actual coordinates.

 

That's my point about accuracy and unless someone actually does a real world test to prove inconsistancies, statements about inconsistancies are opinions not fact. Actually the test is really simple. Find a nearby benchmark with a clear sky and lay it down and read the coordinates. Get your friends to do that too. Then see how far off it is. The $100 bet is still open.

 

Alan

[+Searching_ut]

There are several groups that monitor the accuracy of GPS on a daily basis. Check out Sam Wormleys sites for some good information on the subject:

 

http://www.edu-observatory.org/gps/gps_accuracy.html

 

For information on accuracy at many sites thruout the USA, and the latest FAA report from the ground sites that provide WAAS corrections:

 

http://www.nstb.tc.faa.gov/reports/pan_last.pdf

 

David Wilson has also been tracking accuracy for years:

 

http://users.erols.com/dlwilson/gps.htm

 

If you have a Garmin unit, and can set it for track logging at timed intervals, set it stationary, log a track for a few hours, then view it on something like expertGPS. Even with WAAS turned on, I find my wander and drift at least 25 feet or more most of the time. My Magellan averages a position so I can't do the same with it.

 

For what it's worth

 

Jeff

[+trippy1976]

I only know what I've seen. In that one instance, we all ended up within 6-12 feet of each other.

 

What was interesting was that Rusty baselined the 'spot' with a GPS that does waypoint averaging. He let it average out for an hour and took the reading.

 

We all ended up within 6-12 feet of each other... but we were a good 45 feet from where Rusty thought we should be.

 

All I'm saying here is that I think GPS accuracy is reasonably consistent, but yeah - I can concede that any number of conditions may change the results. Afterward, Rusty re-averaged his point and ended up in the same spot we did. So to us it was pretty obvious that you can expect up to or at least 45 feet of variation.

 

It would be interesting to see more tests done like this at various locations at various times.

 

--------

trippy1976 - Team KKF2A

[Kerry.]

No Alan I'm talking about every day conditions on any day, in the real (and the rest of the) world.

 

If you've forgotten the figures were in response to your following comments as to why the accuracy won't be the same elsewhere. If you think GPS accuracy is 6-12 feet "everywhere" then your blindly mistaken.

 

quote:

---

quote:

--------------------------------------------------------------------------------

Just because there was some consistency and that's what it is consistency doesn't mean that if one was to take all those receivers somewhere else that the results would be the same.

--------------------------------------------------------------------------------

 

How do you it won't be as accurate? MY test and trippy's test showed a majority of accuracy readings under 12 feet in two different places. What tests have been done to show this kind of accuracy is not consistant in other areas? Alan

---

 

Those figures relate to the real world and I'm not all that sure if there's many ways I can say that, the real world on any day, they are real data recorded just as if you were there.

 

I'm not really sure if you understand what accuracy is all about and how accuracy will be different at different times, in different locations for many and varied reasons.

 

Lets try and explain that those accuracy figures are based on real data at those specific locations on Oct 7, 2002 just as if you'd been there yourself with your trusty handheld. On that day at those locations that's the sort of accuracy you would have experienced, recorded, observed or whatever you want to refer to it as.

 

Those figures are not opinions they are real world data and one simply can't get any more factual than that.

 

Those "facts" say that on that day in Kingston one probably would have noted a "consistent" accuracy as the grouping for that location over the time period was very tight (1.9m RMS) but that is entirely different if the same group were standing in Manama as there's simply no way the accuracy would have been the same.

 

There's about 500+ stations around the world that have been/are recording "factual" everyday GPS data 24 hours a day / 7 days a week that would eat your $100 bet without even blinking.

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[+apersson850]

I think VisualGPS is a good tool for that. Can be downloaded for free, in a simple version.

 

Just connect it to the GPS, and let it sit there, transmitting its position, for a day or so. You'll quickly find out what accuracy is all about. I agree with Kerry, but it's perhaps not so easy to visualize for yourself, unless you see it in a graph.

 

The thing is, that even if several GPS units will show about the same position (provided they are at the same place) most of the time, all of them will, at more or less random occasions, drift away significantly. That's why the accuracy is defined at a certain percentage of time.

 

Anders

[Kerry.]

a graphic now there's an idea (and a quick one at that). The following is the plot of position distances every 30 seconds (over 24 hours) relative to the average position for that location on that day. The distances are metres (not feet) and based on actual recorded real world data (not opinions)

 

. Same day, same times through the day (every 30 seconds), same GPS system, different locations AND diffferent accuracy (very).

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[+Alan2]

First let me apologize especially to Kerry for my strident comments. They weren't called for and I should have respected your point as well as others.

 

So now I'll stop pushing my belief and start asking questions so I can learn something.

 

I accept that the signal "drifts" so that if you were to measure and run a graph over a 24 hour period you will see a lot of drift and inaccurancies or inconsistancies. But isn't it also true that the GPS displays the error circle in feet or meters. So if I was trying to get an "accurate" reading I would wait as I do now for the error circle to be minimum and then note the coordinates. It is at this point that I say that we do get accuracy within feet. I would discard results where the circle is off more than 15-18 feet. Does the 24 hour graph take that into consideration or are we looking at raw data?

 

An analogy to this would be if I was to step on a scale to take my weight and graphi the scales results. The graph would be all over the place except I wouldn't consider the results until it stabilized or in the case of GPS when the error circle is the smallest. The rest of the data would be discarded. Do the graphs allow for human interpolation?

 

Alan

[+Alan2]

So let me restate my original point that accuracies are consistantly wihin feet on a clear view of the sky with the GPS sitting stable if you take the reading when the error circle is at its minimum.

 

Does this now make sense?

 

Alan

[Kerry.]

No, not quite as accuracies might/can/will vary from location to location and to some extent vary during different times of the day.

 

For the same time of day different locations will have different conditions, have different satellites in view, different satellite geometry, will have different atmospheric conditions and a few other differences (that can make a difference).

 

Not all the satellites are equal either and as more of the older sats are replaced with better technology and timing capability things are getting better and better with every new launch.

Some of the existing sats are well past their designed life but still going strong but with some being more than 10 years old that is old in satellite technology terms.

 

The 24 hour type graphs are simply based on the computed positions same as what any handheld is producing and using to compute the things it computes.

 

One thing with error cirles, EPE's or whatever is that they are just that an "estimate" as the unit has nothing to compare the positions it's computing to anything absolute. They are an estimate and probably more effective in showing the "possibility" of the position improving relatively over a period of time and not in an absolute sense.

 

Without knowing how error estimates are exactly calculated (and no manufacturer is prepared to outline the methods used for propriety reasons) makes it rather difficult to know what the figures "exactly" mean. However since error estimate routines obviously get changed from time to time it's really appears more a case of massaging the output to suit the current system. If there was any real science in the error estimates then they wouldn't need to change them and there would be a direct comparison regardless of the software version or manufacturer, but there's not.

 

Confidence in an accurate reading (or the best possible accuracy at the time) requires other considerations than simply the estimated error the unit displays.

 

When Selective Availability was set to zero just about all manufacturers were caught out fiddling the error estimates as without SA the real position improved by a general factor of 7 and yet the estimated error only showed reduction of half so obviously somebody was being conned back then.

 

Generally this assumption that accuracies are consistently within feet is a bit of a myth. If you take the position when the error estimate is at a minimum (if one looks at it for long enough) then one is probably best to assume that will be the best position available (at the time). Still no guarantee just how good the position is (in an absolute sense) but "should" be the best at the time. For sure there will be times when the "actual" position is with a few feet but that % will be relatively low, similar with 12 channel receivers, the % of the time there is actually 12 satellites to use is also reasonably low compared to the general average of around 7 to 9 the "majority" of the time.

 

On a world wide average system accuracy is stated (guaranteed ) to be less than 13m 95% of the time and less than 36m 95% of the time on a worse case scenario. Those figures are also Signal-In-Space as the custodians of the system don't have any control over what users do on the ground. They spec what they have direct control over.

 

Even averaging doesn't do all that much (especially these days) as in some cases averaging can effectively decrease the accuracy, not improve it over different time periods.

 

The issue with the scales anology is that with GPS SPS positions one really doesn't know what to throw out, what is right and what is wrong as there's nothing to absolutely compare the data to. All data must be taken into account, which is what the estimated error is based on anyway, if it goes up the position is presumeably getting worse and visa versa. It will fluctuate more so than actually settle. Human interpolation works well in hindsight but at the time the human doesn't really know.

 

I consider EPE's etc more of a relative indication "tool" more so than any absolute indication of accuracy.

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

 

[This message was edited by Kerry on October 15, 2002 at 11:12 PM.]

[+Mopar]

Hey Alan, lemme translate that into english for ya! (great job though, Kerry!)

Since the GPSr has no way of knowing its sitting on a benchmark reading 40 44' 34.46640"(N) 073 50' 42.31409"(W), it doesnt actually know what readings are accurate and what ones are 75ft off. If you use your scale analogy, its like the pointer never stops swinging. The GPSr makes an educated guess based on propriatary algorithms and averaging. In various places and times, these algorithms do a better job then others, and I would bet your $100 that they are "massaged" to produce better results in areas their sales are highest (USA). Even still, since the GPSr has no way of calculating where it is other then the GPS signals it receives, it's actual accuracy is always a guess, at best.

 

Illegitimus non carborundum!

[+apersson850]

There is another figure, usually abbreviated DOP, which a unit like an eTrex doesn't reveal (unless you are listening to the NMEA strings it can transmit).

 

As far as I've understood, DOP is a "fact", a value that can be calculated by known formulas. It describes how good (or bad) the satellite geometry is at a certain moment. A DOP of two means that the conditions are twice as bad as with a DOP of one.

 

So, if one can see the DOP value, at least you can look at something, that changes during the day, and where you can know for sure that conditions are at least more favorable at this time, rather than at that time.

 

Or am I wrong now, Kerry??

 

Anders

[Kerry.]

Anders, at least the DOP's have some underlying common standard that actually relates to the system. Now if every manufacturer displayed some of the DOP values (just the main one or two as there's quite a few of them) then at least those numbers are comparable between different units and also at least the formula's don't change with every second software release.

 

Really the DOP means more than the number of satellites as a low DOP with say only 4 sats is generally a stronger solution than a high DOP from say 8 sats.

 

DOP values are certainly incorporated as "part" of the error estimates.

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[+Alan2]

With due respect to everyone's viewpoint, I have a real world test over a 4 month period that consistantly gives readings to within a few feet using many different units. Your viewpoints are speculation unproved by any testing. Don't you think that I should be given the benefit of the doubt since my test and trippys gave the only real-world results?

 

You're not allowing operator descision making. You're using raw data based on 30 second sampling over a 24 hour period to back up your statement. But the GPS operator waits for stability before marking the coordinates. Knocking the mfr doesn't prove anything. That's an ad hominen attack to "prove" your point.

 

Prove your statement. In scientific circles that's proving or disproving a hypothesis by repeating the test to get the same or different results. Wouldn't it be interesting to see what happens? Prove I'm wrong. Otherwise your positions are not based on any experimental data but on opinion. Very unscientific don't you think.

 

If you could set up a number of these in different parts of the globe, the experiment would be even more meaningful. Find a benchmark, easily acceptable with a clear view of the sky and create a virtual cache asking the finders to post the coordinate reading of the GPS to the cache page.

 

Let's do it.

 

Alan

[Kerry.]

Alan, everybody is welcome to their viewpoint but if you want "tests" then it's tests you'll get. Honestly don't have enough web space to have all on line but this should be a start. If you want more "tests" then not a problem.

 

Ok the 5 channel v 12 channel myth based post and pre Selective Availability. Note in the pre SA accuracy figures your claimed "consistent few feet" is also a myth. Oh by the way ALL the following are supposedly "1 second" data and I say "supposedly 1 second" data because even though just about every manufacturer spec's 1 second output that tends to be a bit of a myth in many cases. That's easily "proved" and is fact not fiction.

 

http://www.cqnet.com.au/~user/aitken/gps/gps_5v12.htm

 

Some "facts" on averaging but more importantly in this case ALL 1 second data and don't see any accuracy that "consistently a few feet"

 

http://www.cqnet.com.au/~user/aitken/gps/gps_avg.htm

 

http://www.cqnet.com.au/~user/aitken/gps/gps_obs.htm

 

http://www.cqnet.com.au/~user/aitken/gps/avg_pii.htm

 

Some "real world" data at the time Selective Availability was set to zero and don't let anybody tell you it's "switched off" because technically it's not, just set to zero.

Oh all "1 second" data and very "real world" and still don't see accuracy consistent "to a few feet", do you?

 

By the way the 95% accuracy in this case prior to SA being flicked was 47.2m (~155 feet) display EPE was 14m (~46 feet). With SA off 95% accuracy 6.3m (~21 feet) and EPE 7m (~23). Who was kidding who?

 

http://www.cqnet.com.au/~user/aitken/gps/sa_off.htm

 

Another "real world" parallel example of handheld positions compared to the "real position". A bit over 5 metres (17.1 feet) but notice the relative consistency in the "accuracy error".

 

http://www.cqnet.com.au/~user/aitken/gps/1085_1.jpg

 

Another "real world" example highlighting how even the antenna orientation can affect accuracy. More importantly ALL "1 second" data and again very "real world". What's even more interesting with this one is this handheld can actually be recorded to 0.0001' where as most still can't get past the 0.001' barrier which does make the "few feet" rather meaningless with some units since the best a 0.001' unit can technically display with condifence is about 2.6m (~9 feet).

 

http://www.cqnet.com.au/~user/mattk/gps/1104_ant.htm

 

Even more accuracy comparisons on sequential days. Oh the position accuracy scale is 0-20 metres (0 - 65 feet), each horizontal line spans 5 metres (~16 feet). One might notice the importance of the dates and the interest in "actual" accuracy on this day.

 

http://www.cqnet.com.au/~user/mattk/gps/pa_1091.htm

 

Well Alan I don't know about your viewpoints but your starting to sound like a salesperson and some of them are very unscientific.

 

Oh by the way when dealing with accuracy there isn't a great deal of "scientific" difference between supposedly 1 second updates and 30 second updates espeially with SA off.

 

Well Alan it's your call, is that enough "real world tests" or do you need more . More is not a problem and if it takes more to kick this "consistent few feet accuracy" garbage (and garbage is exactly what it is) then so be it.

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[+Alan2]

Kerry: We’re talking past one another. All your links to tests are basically giving the same kind of results but is different than the consistency I’m talking about. Your consistency is looking a raw data over a period of time that shows drift of the accuracy due to obstructions, signal strength, etc. Or just plane inaccuracies built into the system. I do not disagree with that data.

 

What I have been stating is that a person using a GPS with a clear view of the sky can mark accurately within a few meters consistently by allowing the GPS to settle and wait until he know he is getting a signal that is reliable using the error circle. That’s what my test appears to have shown.

 

Let me give you another analogy. You didn’t like my last one about weight and scales. Let’s say I’m target shooting. You have a motion detector tracking my aim point. If you look at the graph it produces, I’ll be all over the target. You than draw the conclusion there’s no way I can consistently hit the bulls-eye. The graph “proves” it. So I shoot off ten rounds and amazingly 7 are in the bulls-eye, two are in the next ring and one is off target. You say that can’t be. But I waited until I knew I was on target before squeezing off the shot.

 

Same thing with the GPS. You’re looking at the graph over a period of time. You’re not considering human intelligence or action in this case waiting for the signals to “settle” just like my squeeze point with the rifle.

 

You have a lot of knowledge about this stuff. Why don’t you see if you can come up with an answer to why I got such results; and trippy too? Why do you think it’s happening? Why else did these results happen that a dozen people over 4 months in over a dozen readings get such accuracy? Luck? Are I and eleven other people lying about the results we got? Coincidence?

 

A real test is not hooking up a GPS to a recorder and tracking the coordinates over time as the links your presented show. A real test would be to set a similar type test that I did at a benchmark or other known position. Then test the results and see what you get. The $100 bet is still on the table. If you’re interested, we’ll select a third party to set it up and run it. Game?

 

Alan

[Kerry.]

No Aln I'm talking and your not listening (too anybody).

 

Yeah, it's funny that all tests are the same, doesn't that spell something out to you or are you just going to ignore that fact. They are generally all the same 'cause that's simply the wau it is.

 

And I'll say it again a person without hindsight DOES NOT KNOW what the real reliability of the signal/position is becasue they have no idea where they really are i the first place.

 

The error circle are simply ESTIMATES nothing else, nothing absolute.

 

Lets hope you don't start confusing "Accuracy" with "Precision" but I'm afraid I fell some further confusion (and mis-understanding) coming on. Totally totally and utterly total different to your target shooting.

 

Some users will convince themselves of anything but when one understands what it's all about and even the simple "facts" that there can be 3m (~10 feet) error due to clock stability, 4 odd metres (~13 feet) in the Ephemeris prediction error, between 5 and 7 metres (~16 to 23 feet) due to atmospheric (ionospheric delays etc) plus a few other minor errors then to come up with a statement that GPS accuracy is a "few feet" consistently is totally and utterly rubbish.

 

The answers are all above Alan, the "facts" are as above so you will obviously choose to believe what ever you like.

 

Your obviously going to convince yourself of anything but then so did a few users (now deceased) who also ignored what the real situation is/was all about.

 

However you keep up this "few feet" myth garbage and I'll cut you off at every turn because your giving users a totally wrong and very inapropraite (and to some extent dangerous ) impression of accuracy.

 

You've made some rather ridiculous unsubstaniated claims/statements and raved on about every thing lacking "scientific tests" (how convienient).

 

Time for you to substantiate your claims with some of your own so called "scientific proof" .

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go

[+Alan2]

So is it a bet or are you going to chicken out?

 

alan

[Kerry.]

No bluffing tactics here, just up front with your facts, it's that easy.

 

Ok lets explain what this is all about (again ).

 

October 4, 2002 (the blue lines), October 5, 2002 (the green lines), same site, same receiver, 24 hours 1 second data. 95% accuracy 7.7m (25') and 9.4m (31') respectively. There's an 11% and 14% chance respectively of at some time throughout the day seeing a position (for real not imagined) within a "few feet". That translates into approx 2.5 hours on the 4th and 3.4 hours on the 5th out of 24 hours.

 

Would you call that "consistently" less than a few feet and lets try and be truthful about this (for a change).

 

everybody keeps telling me where to go

[+Alan2]

Your last post is very interesting.

 

So, is our bet on or off?

 

Alan

[Kerry.]

Seeing you have this fascination with estimated error being the ultimate accuracy indicator maybe you'll see some rather obvious things in this plot. Won't bother with any meaningful numbers as you don't (don't want to ) appear to understand what they mean or imply anyway.

 

So the blue line is the "actual" error relative to the known point and believe me (but that's been a problem so far) this point is extremely well tied down .

 

The Green line now that's the "actual" Error position (maybe read this point again Alan, the "actual" error position) as gets displayed on the screen, you know the one you accept so blindly as being the indicator of when a position is good or not so good. Notice I've left your "less than a few feet line" (the red one) for your reference

 

Now remembering this is "actual" data both the actual position error AND the Estimated error would you like to explain why between 0400 and 0500 as the position error improves the estimated error (the green one remember) goes up . Also a bit strange (don't you think ) that when the "actual" position uncertainly (error) does actually increase the estimated doesn't follow it accordingly

 

everybody keeps telling me where to go

[+Alan2]

Good morning Kerry!

 

Your last post is most persausive. Are you now ready to back up your belief with a small experiment and a small wager? If $100 is too rich for you, how about fifty?

 

Alan

[Kerry.]

Alan, actually it's the night after the (your) morning before so your just getting into Thursday and here Thursday has just about gone.

 

Things can get more persausive but am I getting any where that's the pertinent question.

 

There's this one thing your missing I don't work in "belief's" or magic or smoke or mirrors and until you have a way of getting the background about some the magic then that's going to make any conclusion to you most difficult.

 

since you've been willing to part with your cash why not find a local charity, I'm sure they'll appreciate it (you can tell them I sent you if you like )

 

Goodnight.

 

Cheers, Kerry.

 

I never get lost everybody keeps telling me where to go