New Prius in the mountains

When you consider the scale of the signals being received and the basis on how it works, it is amazing it works at all!

I know what you mean of the lateral jumps, usually because of a poor internal crystal if it happens even with 5 locked satellites. Where do you mount your GPS antenna, or do you rely on the internal antenna inside the unit inside the car?

 

At the times I saw this and could check the constellation, there were not 5 satellites visible. Mountains can obstruct a large fraction of the sky.

GPS mounted on the dashboard with standard antenna, as most non-car-integrated users use them.

 

I live in the Rockies, I know mountains.

Only time I ever get less than 5 is when I am actually covered like a parking garage or the Eisenhower tunnel. I do have an externally mounted antenna though, actually the whole receiver is external. I have found that glass in newer cars really obscures GPS signals. The same receiver inside the car on the dash facing the sky, looses signal strength by greater than 50% on received signals, causing many satellites to be dropped or so weak that they cannot be locked correctly.

If you like to track your progress, I highly suggest a unit with an external antenna port so it gives you that option if you so desire.

 

Dang, I forget to prepare for another roadtrip opportunity. So yesterday I made do with the equipment on hand. Results:

Prius odometer: 400.9 miles (reference)
Garmin Nuvi 755: 402.9 miles (+0.5%)
Garmin 60CSx: 424 miles (+6%)
Mileposts: 125 miles (+0.2%, from 124.8 miles on odo)

The Nuvi was plugged in and mounted on the driver's side opera window. The 60CSx had several handicaps: Battery Saver mode (power cord not available) cuts tracking accuracy, passenger seat placement hurts reception, and it doesn't lock to a mapped road. It is loaded with a hiking topo map, not a driving map.

Interval results were taken at six stops. The Nuvi results were very linear, never varying from the overall ratio by more than a single increment.

The 60CSx interval results varied wildly, from less than 3% to more than 11% error. At various times I saw it accumulate 2 miles of error in 10 miles of driving, 6 miles in 80, then no more error in the next 80. And in under 8 miles through the Mercer Island lids, Mt. Baker tunnel, I-90/I-5 interchange, and Convenction Center underpass, it gained over 4 error miles. The image below shows some of this error. The recorded track, in mostly 5 second increments, incorrectly goes down the 4th street ramp towards the stadiums, backs up to the interchange at 139 mph (blue segment), charges again down the same wrong ramp at 285 mph, then returns to I-5 at 291 mph. Under the Convention Center, the track flies into downtown Seattle at 166 and 194 mph, shifts into reverse at 216 mph, then forward again at 142 mph before returning to I-5.

Before GPS I used mileposts for odometer calibration. Last year a 101 mile stretch of I-90 read +0.2% compared to the odo before a discontinuity ended the measurement.

Yesterday on another eastern WA road, Highway 26, I found essentially the same result. The first portion was actually running 0.3%, then the difference seemed to slightly shrink twice for possible road reallignments. I had expected 15k miles of tire wear to show a greater shift. Further refinement of this method requires that I be a passenger recording much more data, not a driver just keeping mental track.

Without knowing how GPS units accumulate trip distances, I cannot automatically presume them to be more accurate than markers on surveyed roads over long distances. For example, if the second-by-second segments are computed to the nearest foot but rounding/truncation matters are not handled correctly, that error alone could easily accumulate to 0.5%. I have just enough experience with road surveys to have some idea what errors to expect, but no similar experience with GPS internals.

With this post I now admit that GPS can be a very good trip odometer. But please take this as a demonstration that it can also be very bad, if the circumdstances are not set up well.

A future trip will compare 3 GPS units, with the 60CSx promoted to plug-in power and the other opera window, while an old Garmin III gets the handicapped conditions.



PS For odometer comparisons, tires are Bridgestone Ecopia, 18k miles, 43 psi in front.