Accuracy of Satnav speed estimates

[alaric]

Hi all.

I was just looking at Gordon Lund's post re the rear diff on Rhubarb, which has some discussion on calibrating the speedo with a new diff ratio.

A satnav should be a reasonably accurate tool to assist with this. But that presumably depends on how the speed is calculated. If the reading is an average then the error should be pretty small - there will presumably be a lag determined by the length, or time constant, of the average.

In most cars that I've used the satnav in, I find that the speedo is reading about 5mph fast. The exception is my wife's golf, which is pretty much spot on - typical! When the V8 Esprit is doing 90 on the speedo, it's actually doing 85mph according to the satnav; that's probably saved us some speeding tickets.

Has anyone heard that the satnav speed reading may be wrong?

All the best.

Sean.

[berni29]

Hi

Apparently if you are going up or down a hill when you take the reading it will less accurate than on flat ground. Makes sense.....

Berni

[mopho]

I am sure there is some slight variation between GPS units, but they are usually more accurate than speedometers. I do work for a car magazine, the testing equipment used is all GPS based and very accurate.

Most speedometers intentionally read optimistically (some say so that you can't argue with the police that your speedometer was wrong when you are caught speeding). I've been told that it is a requirement in Germany for Speedometers to be 10% optimistic, so I am surprised to heat that your VW appears to be correct

Even the Ferrari I was driving a week ago was optomistic, it said we were going 165mph, when in fact we were doing 161mph

[alaric]

Crikey, only a 161; I'd take it back and complain!

Re the going up and down hills, that's a good point.

On the road speed will be mph_nav / cos(inclimation) I guess, where mph speed is the sat_nav displayed speed and inclination is the angle of the slope relative to horizontal. An inclination of 25 degrees will give you 10% error according to this, which is, if I've got this right, about 1 in 2 (a one in one would be 45 degrees so that sounds about right). For a 1 in 10 we're looking at a 5.7 degree inclination and speed error of 1%.

It's a bit worrying if the police won't accept that the speedo is in error - I assumed they would have something more accurate.

All the best.

Sean.

[hatman]

It is my understanding that, in this country at least, it is legally permissible for speedos to 'over' read by a small amount but illegal for them to 'under' read. The manufacturers, not surprisingly, take due heed of those requirements and in some cases (Italian scooter manufacturers come to mind) the latitude is regarded as a useful, no-cost top speed tweaking device. An easy way to produce 'fast' machinery that can be boasted about in the pub later!

[denicholls2]

Mopho wrote:

I am sure there is some slight variation between GPS units, but they are usually more accurate than speedometers

----------------------
A lot more accurate. Variation is introduced only by the GPS fix, which is always within about 100 feet and usually under 50. By definition, there can be no variation based on the same fix between units; the calculation is simple triangulation math that must be the same for all GPS devices in order for them to give a fix at all.

Speed math is simply delta between current and last fixes divided by time between same.

In practice, you will see a GPS vary from a fixed relationship (as noted, usually in the neighborhood of -5 due to intentional error in the speedo) only when the satellite connection is spotty enough that a position cannot be established with each poll. I've seen periods of around 5 seconds when the speed did not update due to this.

My Lotus (Smiths) speedo is the most accurate I've seen, usually within 2 MPH through the range I've checked it (say, 0-70). This is a Europa S2 with stock 336 transaxle.

If you have a consistent accuracy problem with the Smiths speedos, then your needle is probably in the wrong place. It's a simple task to correct it once you're inside the case.

[alaric]

Re the speedo needle position. That reminds me, for about a year I trundled around in a Fiesta XR2 which underead by about 10mph - well before satnav so I timed it against motorway mileposts. I'm having trouble remembering now, but I think it was the return stop that was at fault; the mileometer was spot on though - long distances always matched up to expectations. I never was stopped for speeding as it happened, but in the event planned to declare that as far as I was aware there was no problem. It was a constant offset vs speed though, so not a gearing related issue.

Sean.