The VBOX VB20SL height accuracy at 95% CEP is ~6 meters.

http://www.racelogic.co.uk/_download...B20SL_DATA.pdf

 

Thanks for tracking that down, Dock! The 6 meter spec makes me wonder if we should rely on measured altitude changes in the 6 meter neighborhood. My 25 years in engineering have taught me that there are oodles of places for measurement error to creep in, but it's nonetheless still tempting to believe a number or graph displayed by a measurement tool. We all need to balance healthy skepticism with an open mind.

Jeff

 

One of the problems with GPS claimed accuracy versus actual accuracy is the impact of the 95% probability formula. The GPS systems I use in my work generally have a real time computed accuracy of 0.05 nm (300 feet), which includes the 95% calculations, but the real time global accuracy is much better than that.

 

I have enjoyed reading this soap opera for some time...just a little to add. My company specializes in applying statistical methods to understand measurement uncertainty (among other analytical statistics). I have a humorous quote from a high level executive from a fortune 100 company. We gathered enough data to suggest a certain measurement system being used was totally incapable of measuring what it was intended to measure. His quote was...At least it gives us a number. Talk about irrational...anyway..all of the measurement systems being used have error (uncertainty). And until you understand what it is and account for it, you could be using a random number generator. Jean's thoughts on multiple runs, etc. and using an average is part of the story (averaging can reduce the error if it is appropriate to average). One must understand the stability and precision of the measurement system (not the accuracy) if one wants to make comparisons (same car before/after mods or between different cars). Now back to your regularly scheduled debate.

 

Dock, that is not the same GPS engine in the Pbox. That is their uplevel Vbox tool that has a 20mhz gps engine whereas the Pbox is only 10mhz. The CEP of the Pbox for absolute postioning is 95% CEP of 2.5m. The 95% CEP of height is 10M. Here is the manual. All the specs for the GPS engine are on page 45

http://performancebox.co.uk/Download...al-English.pdf

 

If the height accuracy is 10m for the entire run, then the slope calculation would be accurate.

 

I understand why you say that, but I'd sure like to see an analysis supporting it. A ruler with zero in the wrong place will always give a wrong measurement, but the difference between 2 measurements may be spot on. I don't know if that's a good analogy for GPS altitude calculations. For all I know, a better GPS analogy may be a ruler with a zero point that randomly changes with each measurement. In that case, the change in elevation calculation would be as inaccurate (or worse!) as any individual measurement.

Jeff

 

If the starting point altitude is an actual 500 feet, and the end point is an actual 550 feet, the difference is 50 feet. If the GPS calculated the start point at 510 feet (the 10 foot vertical error) and maintains this same 10 foot error thought to the end point, the end point GPS calculation would be 560 feet..for the same 50 foot difference.

 

The supposition above that I enlarged and made bold (and fixed the typo ) is the one I'd like to have confirmed by a GPS expert. If it's true, then yes, an absolute error in altitude is a don't-care because we are only interested in the change. But if it's not true, then the error in a measured altitude change may be as large as the error in absolute altitude.

Jeff

 

Simple math actually...variances add. So there is a variance around the reading (mean) at the start point and a variance around the end point. The total error is the sum of the variances. Oh yeah, what is absolute altitude? What measurement system is used for this?

 

The one that uses sea level as zero? I use furlongs.

Jeff

 

What time of the day and at what location on Earth?

For when its cold in the winter

I know enough already...Notice how the fun has stopped since a statistician started talking...

 

GPS uses a base geoid of the earth's surface and measures altitude above that surface. This geoid is not often accurate in terms of the actual global geoid, but it is still a base that is used to measure deviations from. Will the calculated altitude match the altitudes on a topographical map? Probably not. Can the calculated altitudes be used to accurately determine the receiver's altitude above the earth? Probably not. But it would be possible, assuming a consistent deviation error (50 feet in my example), for it to measure the difference in altitude between two points over a relative short distance. Keep in mind that these two calculated altitudes don't need to be accurate in terms of absolute altitude, the system error just has to be constant for the length of the run.

GPS receivers are best used to measure speed and distance, not altitude. Although in theory the unit could determine the altitude difference between two points accurately enough for the 60-130 slope calculation, I certainly wouldn't accept it until running tests on the unit itself.

 

Dock, thanks for the info, much appreciated!

Statmeister, don't be so hard on yourself. I'll bet with a few drinks you become the life of the party. (kidding!)

Jeff

 

Where is TB's graph?