ADUK Archive

I did a quick search and couldn't find the answer, so here is my question:

At a steady 60mph indicated speed on the speedo in my car, my satnav tells me I am doing 55mph.

Which should I have more trust in? Are satnavs more accurate than normal car speedometers?

Yes.

The speedo's are set to deliberately over read to protect the motorist. In days gone by this was partly a factor of the technology used - losses in the mechanical cables driving them, but nowadays there's no real reason for it other than to provide some protection from prosecution.

I believe in the past, if you were pulled for speeding and you could demonstrate to the officer that your speedo was under reading, you might have got away with it, once. Nowadays with cameras, no chance.

GPS is usually fairly accurate. It depends on how good a signal it's got, but for most purposes once locked on, should be better than most speedo's.

Mr Cholmondeley-Warner wrote:GPS is usually fairly accurate. It depends on how good a signal it's got, but for most purposes once locked on, should be better than most speedo's.

IIRC GPS is very accurate horizonatally, less so vertically. Therefore it will be most accurate on level ground, rather than driving up a hill. Also, due to the lag between samples it will approximate a straight line between points - so if you're driving up Zig-zag hill, it may be rather inaccurate. The lag will also lead to inaccuracies if you are accelerating or braking significantly. OTOH, on a motorway at steady speed it is likely to be a better guide than the speedo.

Mr Cholmondeley-Warner wrote:GPS is usually fairly accurate. It depends on how good a signal it's got, but for most purposes once locked on, should be better than most speedo's.

GPS speed will generally lag behind real speed. I think (and I hope I'm correct in this) that GPS speed is derived from successive GPS locations, each of which will have different degrees of inaccuracy. As a result of this, the software in GPS receivers will average the results of a series of speed calculations. Assuming my model of how this is implemented is correct, GPS speed will tend to be more accurate when driving in a straight line than when cornering. Finally in a moving car some of the successive positions will be missing due to changing reception conditions.

This will all lead to instantaneous inaccuracy, although I'm unable to put a figure on it. What is interesting, though, is the lack of detailed information from the GPS navigation vendors, so it suggests they'd prefer not to be liable for someone relying on their systems for any kind of accuracy.

I made my comment rather general for the reasons highlighted by Jon and Gareth. Steady speed in a straight line would be best for testing, and it's probably best to do that at a number of speeds, and then remember the speedo error, rather than trying to use the satnav as a "live" speedo.

The GPS needs more satellites to get a vertical fix, which may be a factor in the lower vertical accuracy.

If you've taken a handheld GPS out walking, you'll know it updates every few metres, but does sometimes lag a little. At 70mph on a straight road, the error is likely to be very small, though, and now that the random error has been switched off, the positional accuracy is going to be pretty good in a clear area with good coverage, too - maybe 5 metres.

One thing to rember is that GPS averages speed over short distances.

Whereas most enforcement systems (except VASCAR and SPECS) take an almost instantanous snapshot.

It is perfectly possible for your average speed to be say 70mph, even if you are doing considerably more than that at the instant you trigger that GATSO.

I find this webpage to be most informative:
http://www.aprs.net/vm/gps_cs.htm

In summary GPS speed is NOT calculated by taking two positions and the time to travel between them. It is calculated (in modern GPS receivers) by measuring the doppler shift of the GPS signal from each satellite; a much more precise measurement. The author states that velocity error is rarely more than 0.07 miles per hour.

Also note that Selective Availability, which the US Department of Defense used to degrade the quality of the GPS signal, has been turned off for several years, and hence you can ignore that part of the article.

Hope this helps.

Gareth wrote:GPS speed will generally lag behind real speed. I think (and I hope I'm correct in this) that GPS speed is derived from successive GPS locations, each of which will have different degrees of inaccuracy. As a result of this, the software in GPS receivers will average the results of a series of speed calculations. Assuming my model of how this is implemented is correct, GPS speed will tend to be more accurate when driving in a straight line than when cornering. Finally in a moving car some of the successive positions will be missing due to changing reception conditions.

Further to fishter's info, I found this:
http://www.gpsinformation.net/main/gpsspeed.htm
I recall reading somewhere when I was keen on GPS that the speed calculation was not simply position readings(relatively inaccurate)/time.

I'd agree that flat straight line speed readings would give a better "steady state" value, and this is what I use to calibrate my speedos.

Another thing to consider re car speedo accuracy is that there will be an approx 1.5% to 2% diference in your actual speed at the same speedo reading depending on whether your driven tyres are new, or down at 2mm tread.

Jimmy wrote:I did a quick search and couldn't find the answer, so here is my question:

At a steady 60mph indicated speed on the speedo in my car, my satnav tells me I am doing 55mph.

Which should I have more trust in? Are satnavs more accurate than normal car speedometers?

Find a measured mile close to you and run it and time it to test how accurate your speedo is. 60mph one minute 30 mph two minutes

Most measured miles are maintained by your local authority, give them a ring to find one close to you.