MPH on Satnav

steve-L

redux wrote: »

Even if that's right, then as I said the error due to the horizontal projection distance is trivial almost all of the time.

The cosine of 2.5 degrees is 0.9990, which means on the steepest UK motorway the effect is about 0.1%, or about 0.07 mph at 70 mph

True, the real reason for inaccuracy is the way geodetic data work.
WGS 84 uses the earth's centre of mass/gravity to create a approximate model of the earth. This oblate spheroid is just a mathematical model of the earth not taking into account local asperities. Military GPS systems use a 3D model both above and on the earth, car GPS systems do not. They also receive full resolution data.

Tyre tread depth varies from 8 mm new to 1.6 mm when worn to the legal limit. That 6.4 mm variation is approximately 2% of the radius of most wheel and tyre combinations, for example 195 65 15 has 317 mm radius or 205 60 15 has 314 mm.

So tyre wear between new and replacement accounts for approx 1.4 mph variation at 70 mph. It is certainly not accurate to anything remotely approaching 0.1 mph

That is not what I said, for a given tyre circumference once calibrated it is certainly accurate to 1mph and there is no reason why not 0.1 mph. If it is calibrated on new tyres correctly inflated then it will only ever indicate faster than actual....

And the speedometer instrument error is often more than that, such as the 2 to 5 mph in the original post, and not necessarily a linear proportion of speed.

The speedometer instrument error is consistent for a given tyre circumference.

The point is that the SatNav/GPS speed will unless you are in a perfectly straight line always indicate SLOWER than actual.
The algorithms are more complex however as they try and fix your estimated position of +/- 10m to a direction and previous speed and also a known road.

This is one reason why when you overlay say google maps and Sat photo's the road's are sometimes a few metres away from their 'actual' position.

The GPS algorithm uses reduced resolution data with random errors +/- 10m added by the military system. By taking subsequent readings the actual can be found by triangulation for a stationary object. (For a missile flying in real time it can't be triangulated quickly enough).

So your car GPS corrects for this by assuming you are on a known road with a known previous velocity and vector.

Hence your error in GPS is not constant (like the speedo) but depends on your velocity, sample rate and availability of GPS.
From time to time it will calculate your velocity as say +1,000 mph or -2000 mph.... but it uses logic to remove these spikes.... knowing they are impossible..... the same happens when you don't take the slip road the GPS tells you to and you see yourself continue before it has the 'confidence' to decide you missed the exit and reroute.

You can play about and see this using google maps on a smart phone. Often your location may be very large ... hundreds of meters of diameter.... before a new fix updates and reduces the error, wait long enough and it will get increasingly accurate.

Hence GPS is inaccurate over both longer distances (that are not perfect straight lines) and shorter distances due to sample intervals and the built in random inaccuracy of civilian GPS.

Despite the deliberate calibration errors (to ensure it never tells you you're going slower than you are) in the car speedo the speedo is constant until pressure changes or tyre wear occurs. The amount of tyre wear on the average journey is minimal.... so relying on GPS speed is not really a good option.

It would be hard enough getting clocked at 45 in a 40 mph zone and showing a manufacturing defect as the reason and then getting the points taken away.... trying to argue you were using the GPS speed would be pretty pointless as they would simply ask why didn't you use the instrument provided with the car!

VSynth

If I put my car on cruise control so that my speedometer reads 70mph and run a few miles, the cars on-board computer says I averaged 67mph and the satnav says 65mph.

I think that speedometers are designed to be biased to display a higher speed than you are actually travelling at.

I usually put my cruise control on 73mph to compensate.

steve-L

VSynth wrote: »

If I put my car on cruise control so that my speedometer reads 70mph and run a few miles, the cars on-board computer says I averaged 67mph and the satnav says 65mph.

I think that speedometers are designed to be biased to display a higher speed than you are actually travelling at.

I usually put my cruise control on 73mph to compensate.

All of the above.... but also when on cruise control the car will be allowed to slow a little before it applies more gas... hence the average speed will be slightly lower.

The bias is for the outer circumference of the tyre that changes according to tyre wear and pressure.

You can test this if you find an area of motorway marked out for the helicopter speed traps with the white marks. These are a known distance apart so you can measure it (using the mile-ometer) and then round it to the nearest.

Then you know how much latitude you have.
Unless your actually worried about breaking the law rather than getting a ticket most cameras are set to 10%+1 hence 78 mph so you'd be extremely unlucky to set the cruise to 78 and get a ticket as your probably doing 74-75.

smudger1946

just believe the speedo as we did before PRATNAVS!

Smudger

forgotmyname

Or you could goto a place that calibrate tacho's and borrow their rollers.