WSIB: accuracy of consumer grade models?

[Bow Lake]

Hi All,

I'm looking for a new GPS primarily for hiking and geocaching. After reading different opinions and reviews, it looked like either the PN-40 or the Oregon 300, but both appear to have issues, however I am reluctant to buy older technology units. I also noted that there appears to be very little tech info behind the accuracy claims. I have two mapping/survey grade Trimble GPS pocket PC units (not really suitable for hiking) where I can get 1.5 foot accuracy with post processing to 95% confidence level. In not being very happy with Trimble, I did look at the Magellan mapping grade GPS unit and it turned out that their reported accuracy is only at the 65% confidence level.

 

Has anyone looked at true horizontal and vertical accuracy of handhelds? (put the GPS on a known benchmark position) Certainly GPS position repeatablity (as in geocaching) is typically much better than true position accuracy.

 

Also, any experiences loading free public maps and air photos (such as from state GIS web sites) onto handhelds? Do the manufacturer's block you from doing this?

[+twolpert]

If your standard is mapping/survey grade units, you're going to be disappointed with any recreational handheld. Not sure if you're doing your post-processing on the pocket PC or elsewhere (not in the field). If you are not doing it in the field, you should be looking at your current results [/i]without postprocessing, though.

 

I have not seen anything posted on systematic tests of either of these models against adjusted benchmarks. My own (entirely non-systematic anecdotal) experience with the PN-40 and single benchmark is that you get about what you would expect -- anywhere from right on top of it to 10 or 15 feet, depending on the constellation.

 

AFAIK, neither manufacturer publishes anything about their EPE formulas. So you have no idea what the confidence level is. My personal take is that it can't be very high on the PN-40 (can't speak to the Oregon) because the EPE figures seem fairly optimistic. By "not very high" I mean somewhere in the 50% ballpark. But it's just a guess. YMMV.

 

Makes no difference for caching anyhow. Even if your unit is right on, you're at the mercy of the cache owner as to the accuracy of the cache coordinates.

[+dakboy]

I have not seen anything posted on systematic tests of either of these models against adjusted benchmarks. My own (entirely non-systematic anecdotal) experience with the PN-40 and single benchmark is that you get about what you would expect -- anywhere from right on top of it to 10 or 15 feet, depending on the constellation.

I really wish I could find the huge thread over on DeLorme's forums where people ran through a number of accuracy & repeatability tests with their units. Including using adjusted benchmarks.

[+twolpert]

I have not seen anything posted on systematic tests of either of these models against adjusted benchmarks. My own (entirely non-systematic anecdotal) experience with the PN-40 and single benchmark is that you get about what you would expect -- anywhere from right on top of it to 10 or 15 feet, depending on the constellation.

I really wish I could find the huge thread over on DeLorme's forums where people ran through a number of accuracy & repeatability tests with their units. Including using adjusted benchmarks.

If you mean the one that Chip Noble started, I thought about it. But my recollection is that it degenerated pretty quickly into random discussion with very little hard data. And seems to me that what little there was revolved around repeatability -- out on the trail, back on the trail, do my error circles overlap? That made sense since they were looking for validation of their EPE formula, but doesn't say much about objective accuracy.

 

Edited to add: Here you go, but prepared for glazed eyeballs.

http://forum.delorme.com/viewtopic.php?f=1...t=shirt#p104879

Edited November 20, 2009 by twolpert

[Grasscatcher]

What I have come to expect over the last several years from my Garmin 76CSx looking for "adjusted coordinate" Benchmarks is within 1-2 meters.

I use UTM coordinates .

 

Very very seldom is the error greater than that to "known" coordinates, and that is under "real world" conditions.