Delorme PN-40, Oregon 450, iphone 4 accuracy

[+sloflyfisher]

I've had a gps for some years now, a Garmin etrex summit. Used it for backpacking. Decided to upgrade this year to a DeLorme PN-40 (for backpacking). Then I heard about geocaching. Good news, the PN-40 could do that. I also found out my iphone 4 could do that too. So I loaded up the Geosphere and Geocaching programs. Then, REI had this sale on the Oregon 450. So now I have three gps.

 

Now to my question. On each device there is a screen with gps accuracy stated in + or - feet (or meters). If the PN-40 says 7 feet, the iphone usually says 16 feet (iphone doesn't say how many satellites it's tracking), and the Oregon says 24 or 25 feet. I would expect this difference between the gps units and the iphone but not between gps units. Garmin says 24 or 25 feet is good enough and PN-40 is probably not accurate. This is with both units tracking 8 satellites, the same satellites, same strength as well. But, the ratio stays the same when PN-40 goes down. So PN-40 says 20 feet, now Oregon 450 says 80 feet and iphone is about 32 to 35 feet.

 

So, is what Garmin says true. If the PN-40 is overstating the accuracy, psychologically it make me feel better . PN-40 seems pretty good at locating caches. The Oregon 450 is a better interface. One thing about the iphone. If I look at satellite imagery map, I am usually within 8' to 10' of the location it pins me at.

 

Has anyone else experienced this wide range of accuracy on these units?

 

Thanks

[+RRLover]

On each device there is a screen with gps accuracy stated in + or - feet (or meters). If the PN-40 says 7 feet, the iphone usually says 16 feet (iphone doesn't say how many satellites it's tracking), and the Oregon says 24 or 25 feet. I would expect this difference between the gps units and the iphone but not between gps units. Garmin says 24 or 25 feet is good enough and PN-40 is probably not accurate. This is with both units tracking 8 satellites, the same satellites, same strength as well. But, the ratio stays the same when PN-40 goes down. So PN-40 says 20 feet, now Oregon 450 says 80 feet and iphone is about 32 to 35 feet.

 

So, is what Garmin says true. If the PN-40 is overstating the accuracy, psychologically it make me feel better . PN-40 seems pretty good at locating caches. The Oregon 450 is a better interface. One thing about the iphone. If I look at satellite imagery map, I am usually within 8' to 10' of the location it pins me at.

 

Has anyone else experienced this wide range of accuracy on these units?

 

Thanks

All of the above use proprietary algorithms to state what they consider to be their best guess

as to how close the device is to the reported coordinates. These formula are not shared w/us,

or each other. It's just a way to say you have a nebulas percentage of being within this

"guestimate" of feet to the stated position. The proof is in the pudding, take them out and check

them against a known corrected 'benchmark' and weigh for yourself the accuracy of each device.

Do this on different days and at different times, as the satellite constellation plays a major role,

. . . and is in a constant state of flux.

 

Norm

[+dfx]

what he said. the "accuracy" is neither a guarantee nor a promise, it's just an estimate, a statistical value. it means that there's a certain chance that the coordinates it gives you are actually within X feet/meters of your real location. that chance can be 99%, 95%, 50% or anything in between.

 

if you "only" see 25 feet accuracy on the oregon, make sure you have WAAS enabled. having WAAS reception brings your accuracy number down to half.

[+sloflyfisher]

Yes, WAAS is enabled. No I haven't taken it out and tested it against a known benchmark, which I will now do, but probably not before the long weekend. They may all be spot on, I don't know. Maybe it would be better not to include an accuracy figure. That way you're forced to check for yourself. Thanks for the replies.

[+dfx]

oh, and the oregon's WAAS performance pretty much sucks with the stock firmware. get the beta firmware and try again if you don't have it.

[+Team CowboyPapa]

Here is how to check your PN-40 at home:

1. Use the $40 trial certificate that came in the box to download the color aerial photo imagery for your house and its surroundings.

2. Cut and save a file with Zoom levels 13 - 17 and then transfer the file to you PN-40.

3. With the imagery up and active on the screen of your -40, stand in the middle of your driveway.

4. When it gets a stable fix, use the current position indicator to estimate the accuracy at Zoom Level 16 or 17.

5. Outside of that, you are just guessing.

[+user13371]

Here is how to check any GPS:

 

1) Go to a known, fixed point.

2) Set a waypoint.

3) Repeat steps 1&2 several times at the same place over the next few days.

 

Plot the differences between each point. The smaller the pattern it makes, the better the GPS is.

 

Some folks will tell you if your chosen spot isn't an established benchmark, you're only checking repeatability (not accuracy). That's a whole 'nother argument -- but for purposes of a quick check the end of your driveway is as good a "fixed" location as any other.

Edited September 2, 2010 by lee_rimar

[+TotemLake]

Also, you might want to maintain the devices 12" from each other. The RF they emit can cause enough interference to make your results questionable. Particularly the RF from the cell phone.

[+sloflyfisher]

dfx, my software version is 3.60. Noticed WAAS performance improved from 3.50 to 3.51. Don't know if updating would help, but I'll try. I'll try your suggestion Papa. I actually subscribed to the year map download. Seemed like a good price. Liked the idea of testing with an known benchmark that RRLover had. Thanks for everybody's help.

[+embra]

The known benchmark (i.e., one with adjusted, not scaled, coordinates) is the gold standard. Lee and TCP's suggestions are sometimes more convenient and certainly of value.

Edited September 2, 2010 by embra

[+user13371]

Max, I did say that was a whole 'nother discussion

 

Repeated readings over time are a requirement, and if you want to be very detail oreiented you could use Trimble Planning to note the conditions for each reading.

 

But in practical use, I've never been able to figure out how or why an established benchmark is better than any fixed location you might choose. Maybe on the first visit to make sure you have the datum set correctly. After that, what you're really looking for is the size of the scatter pattern for multiple readings -- and if you're going back to the same spot repeatedly, it doesn't matter if the USGS was there before you.

 

Does it?

[+embra]

I guess we're in that whole 'nuther discussion--or at least at the boundaries between the current discussion and that one.

 

Obviously, a single reading compared to a known location will immediately reveal the amount of error in the reading. And even here, we would ideally want a series of readings to see how consistent the amount of error is.

 

Without the adjusted coords benchmark, we can mark an identifiable location and repeat the exercise a number of times. On the second reading, it will likely differ from the first. At this juncture we don't know if the first is right and the second wrong, vice-versa, or (most likely) both are off. With successive markings we get a scatterplot which we would presume, I think, to be randomly distributed about the hypothetical "true" coordinates. The more readings, the less the error is likely to be...but we never have the degree of confidence in the amount of error as when we compare to the benchmark that is believed to be accurate to a matter of centimeters.

 

One problem, perhaps, is that it's not clear to me how many measurements we would want to record. In "practical use"--the key phrase in your argument--I think I agree that even without a benchmark, taking as many readings in as many conditions as possible as one has the patience for would yield a scatterplot whose tightness of spread would tell you all you need to know.

 

Concur?

Edited September 2, 2010 by embra

[+user13371]

I guess we're in that whole 'nuther discussion--or at least at the boundaries between the current discussion and that one.

I think I can continue AND stay relevant to the original question....

 

Do the test I described with each of the subject GPS devices. From first hand experience with two of them (and with a close relative of the third), I'll guess that the PN-40 will consistently show the smallest pattern, the Oregon 450 next (not off by much), and the iPhone 4 the worst (by a wide margin). And all of them will still be good enough to find geocaches.

 

And "good enough" is a key concept here.

 

If we were surveying a property line with a high-end unit and post-processing, I'd surely want some reference point with the kind of precision you're talking about. But when you're comparing consumer grade, handheld units only expected to be accurate within a couple of metres anyhow? At that point, knowing if the centre of that pattern is dead-on, "accurate to a matter of centimeters" of the exact coords is unimportant.

 

Re your other question (and still relevant to the OPs question): How many readings? My guess is at least six, evenly spaced over a 24 hour period. Fewer than that will not be enough data; and more would be better. I didn't completely pull that number out of thin air. The satellites being in 12 hour orbits. Spreading out readings by a few hours over the course of a full day should give a good sampling of satellite geometry.

Edited September 3, 2010 by lee_rimar