On 6/3/19 12:47 pm, John B. Slocomb wrote:
On Tue, 5 Mar 2019 16:44:03 -0800, "Mark J."
wrote:


I think it's the trigonometry of the computation; the GPS (as I
understand it) measures distance to a collection of satellites whose
positions are well known, then computes location from triangulating the
results. I'm guessing that since most of the satellites are usually not
directly overhead, but rather the line of sight to the satellite is
usually be much closer to tangential to the earth, then very small
errors in the distance-to-satellite computation turn into much larger
errors in the altitude computation.

I think this is why higher-end bike GPS's have a pressure-based
altimeter as well, to correct the fluctuations in the GPS-computed
altitude. I know my Garmin Edge's regularly solicit known altitude
input at the start of a course.


I'm not sure about how accurate GPS really is but back in the day, the
seismic folks had a large "Black Box" that they used to locate their
seismic lines on the chart that they said was accurate to within feet.


Mark is pretty close to the correct reason. Most GPS receivers
intentionally track satellites that are close to the horizon as opposed
to direct overhead. This is so that the X-Y part of the position
information is most accurate, at the expense of less accurate Z position.


The overall accuracy depends greatly on the GPS receiver quality. The
cheap receivers (say $50 ea) may be within a few meters, while expensive
receivers ($500) are 10 times better. If you pay more ($10,000) and
incorporate corrections for atmospheric conditions and such, accuracy
can be better still.

The difference between cheap and expensive is largely down to the
stability of the oscillator used to time signals. The antennas can also
be an expensive part and play a big role in accuracy and reflected
signal rejection.

But... Even cheap GPS receivers are relatively stable over a short
time. They usually produce a position, speed and heading once per
second. The previous position, speed and heading are combined with new
measurements in a special filter, that usually results in better
accuracy than if the measurements were used alone.

The only times I've noticed real problems is when you cycle relatively
fast around tight corners. The GPS position effectively cuts a little
off the corner each time, modelling it as a series of straight lines.
Hence your road speed appears to be slower than it really is and you
appear to accelerate again when the road straightens out.

More expensive GPS receivers can produce calculated position results
more frequently than 1 per second. More powerful processor. More power
consumption. Unlikely to be in a battery operated consumer grade bike
computer.

But I find that they are accurate enough not to miss the magnet and reed
switch.

--
JS