how accurate is GPS in a phone

[Ollossm]

I appreciate to a degree all GPS could be deemed as accurate to a degree but it is that degree I'm wondering about.

I know the speedo on a 125 can only be deemed as accurate up to X speed and then is best used as an indicated figure rather than an actual figure.

As I've been wondering how accurate (or not) the speedo is on the CBF I took myself out today with the phone on the tank and GPS turned on.

0-25 speedo and GPS tied in
at 30mph on speedo the GPS was giving between 26 & 29
at 35 on speedo GPS again appeared top tie in
At 40 on speedo GPS was reporting 35/36.

With the 40 on speedo and 35 on GPS I thought I'd increase speed by 5mph - which in theory would have brought speedo to 45 and GPS to 40....when the bike hit 45 on the clock the GPS also tied in at 45

At these speeds it probably doesn't matter as it is being under-reported on the clock, but it is puzzling.

At one point going through a village the clock was just on 30 and the GPS was on 28 but the 30mph speed reduction warning sign flashed on.

Are the GPS units in phone less accurate than a GPS unit?
____________________
_______________________________________________________________
current bike: 2012 Honda CBF 125
past bikes: 2004 Suzuki VanVan 125 - 1988 MZ/Simson 50cc

[bikertomm]

I think they probably run off the same if not similar than what tomtom / garmin ect use..

I'm sure someone will be along soon to tell you for sure

But yes, generally your speedo will over-read.
____________________
07' Honda Hornet now full powaah! My guide on performing an oil change!

[janner_10]

As Tom said really, I assume it works same as a tomtom

MY sat nav in my old works van used to agree with the speedo until about 40 then the speedo would over read by about 10%.

I've never measured it accurately just a quick cursory glance.

Never repeated the process on the bike though cant say I'm bothered.
____________________
Yamaha FZS600 (Now gone to heaven) > CBR600F4i (SOLD) > '99 YZF-R1

[Derivative]

GPS software can't feasibly have a systematic error when recording speed unless there's a software bug.

All it's doing is plotting the points you're located at and doing a bit of 'speed = distance/time'.

The worst case scenario is when the software tries to find your average speed over too short a time period, because then the small inaccuracy in GPS location (up to about 10 metres I believe) will throw it off.

For example, at point A it could think you were 3m south of where you actually were, and at point B 4m north, adding 7 erroneous metres to the distance travelled.

However, that would only really be relevant for small time intervals.

If it gets the speed over the last 30 seconds on an open road, it should be pretty much bang on.

[Kickstart]

Hi

Can get inaccurate due to position errors, but longer term these should sort themselves out. Can lead to interesting maximum speeds though (apparently my YSR50 can do over 60mph, even if only doing under 50 the second before or the second afterwards).

Also accuracy will vary with going up hill or down hill.

All the best

Keith
____________________
Traxpics, track day and racing photographs - Bimota Forum - Bike performance / thrust graphs for choosing gearing

[Ollossm]

Thanks for the replies. Most helpful

I'd been on the road for an hour continuous riding when looking at GPS / speedo, although on occasions I was under tree cover which has caused issues in the past even with the Garmin.

However when the GPS was showing 28mph and the warning sign triggered I'd been on open road for about 5 minutes. Although I suppose the speed warning sign could be set within margins to trigger it.

It's not important just one of those curiousity things.

I shall try again at some point and try and pick a straighter and definitely less wind gusting route.
____________________
_______________________________________________________________
current bike: 2012 Honda CBF 125
past bikes: 2004 Suzuki VanVan 125 - 1988 MZ/Simson 50cc

[Derivative]

It may be useful to look at different software, if that is available on your phone.

Particularly if you can find an application with a variable sampling time.

Average speed over last 30 seconds will be much more accurate (provided you hold the same speed) than over the last 1 second.

The downside is that it won't update as quickly if you vary your speed, so most people prefer to have the quicker updates.

[Ollossm]

Tis an android phone so can change the app.

I was using Ulysses Speedometer, I'll try and find an app that has variable sampling as you suggest

cheers and thanks
Marc
____________________
_______________________________________________________________
current bike: 2012 Honda CBF 125
past bikes: 2004 Suzuki VanVan 125 - 1988 MZ/Simson 50cc

[Frost]

I've seen a huge difference between phones. I have a galaxy s2and the gps is pretty darn good for a single channel. A friend has an HTC sensation with isn't so good, but still not bad. Another friend has a galaxy Ace tat's all over the place! The position read will tell you you're 20 meters from where you actually are!
So long as the phone is reading lower than the speedo i'd say that was being acurate. Position skip on the phone would cause an over estimate of speed, going round corners will cause an underestimate as the software plays join the dots with the position fixes

[ocatoro]

[MarJay]

I was told that all GPS's are far better at measuring speed than just plotting points on the earths surface. Elevation changes etc would cause severe inaccuracies.

What they actually use is doppler shift which is incredibly accurate. I'd take a guess that your phone is way more accurate than your speedo.
____________________
British beauty: Triumph Street Triple R; Loony stroker: KR1S; Track fun: GSXR750 L1; Commuter Missile: GSX-S1000F; Cheap project: CBR900RR FireBlade
Remember kids, bikes aren't like lego. You can't easily take a part from one bike and then fit it to another.

[daemonoid]

[tahrey]

Their -precision- can be pretty variable, however the -accuracy- is generally quite good.

IE it may only locate you with a certainty of +/- 5 metres (which would be enough to cause notable errors at low speed), which may include a degree of drift, but if you're moving, it's less a random error and more that you're a sort of fuzzy dot. If you plot how far the dot as a whole moves, even though you may not actually be central within it, the speed and direction is fairly close to reality.

Of course there are spikes and other artefacts, and the smaller and cheaper the receiver, the worse they get. My crappy £40 satnav locks on much faster and more reliably than my phone in identical conditions, locates me in a smaller area of certainty, and displays a much steadier speed when the signal is suboptimal.

But you can work around that if your idea is to see how accurate your speedo is, and work out what, for example, it shows you for a real-world speed of 32, 43, 54mph (so you can keep within the gatso trigger limits even if otherwise pushing on a bit). Just set it up and drive at as steady a speed as you can over a reasonably flat and straight course. Either take the average measured over that section, or go with what speed it displayed most often and for longest. Rinse and repeat for the other speeds. Memorise.

And if you have one of those light-up signs that tells you how fast you're going nearby, make good use of it. They operate off the same radar or laser detector hardware as a speed camera, so can be quite trustworthy, so long as they're aimed up properly and there isn't much other traffic about. I've even had them pick me up with pretty good accuracy whilst I was on my pushbike. Even though most of them top out at 45mph, it's enough to help you calibrate your speedo for 20/30/40 limits, and acts as a "second opinion" to check your GPS against (which you can then use for 50/60/70).

According to two different phones I've recorded tracks on, thrashing a CG down a slight hill, with a following wind and a bit of slipstream, an indicated 80 is actually "only" 75 (meaning the estimated just-under-80 Vmax thanks to the bizarrely slightly-too-low(?) 14/45 sprockets is almost spot on; it seems to hit the limiter at an indicated 84 on a serious downgrade). There was a lot of noise on those tracks, when they were downloaded and plonked into a GPS telemetry graphing / mapping thing, including some crazy things like alternating between 0 and 200mph whilst the signal faded in and out on a bit of the M5 with a lot of bridges, but the data was still clear enough to shove a regular trendline (and a heavily averaged second one) through and pick out the actual story.

[daemonoid]

[Timmeh]

[tahrey]

I must have uncannily accurate ones round my way then, as they tend to agree with the GPS unit, except those that are quite clearly out of whack (reporting 15... 17... 43... 19... when you're doing low 30s on the clock).

And the GPS thing as above is much as it was described to me by someone who seemed to know their stuff quite well; if you stay still, a 10m accuracy wouldn't be enough to fire a satellite-mounted assassination laser at you, but it would be able to work out your speed and heading with remarkable accuracy if you started walking. The laser would still miss, but its aim would be off by about the same amount within a handful of inches each time. Something like that.

(If all it gets is your signal as a 1D dot that bounces around at random, how does it know what the accuracy is in order to report that as a variable-radius shaded circle on the map?)

Therefore I bow out, as it's going to be me reporting the remembered words of one person of your experience and position in opposition to, well, yourself.

[Timmeh]

Somewhat OT, but I wouldn't worry about being sliced and diced by some big laser beam whilst riding, although some think tank has suggested equipping gatsos with missile launchers.

In any case, it would be using optical guidance rather than GPS.

Guided bombs that do use GPS have massive warheads to counteract any inaccuracy in their navigation, and even then they'd be using the encrypted P-channel signals which can get your fix down to 10cm, even on a bad day.

've used military GPS whilst out playing with the TA, and that was reporting an error of some 0.4m whilst trekking though woodland, so it just goes to show how amazing it can be when using the right kit.

5m accuracy using civvy receivers is nothing to be sniffed at, in the early days of GPS the US government used to wobble the signal using something called Selective Availability (SA) which used to knock the accuracy down 50-250m in most places. In some countries like Libya, Iran and other terrorist-harbouring countries GPS was practically useless, with errors as high as 1Km.

As use of marine and air nav-based GPS became more widespread, people like Garmin invested huge sums of money into the Differential add-on system. Basically a radio transmitter with an exact known position would get a 3D fix from the satellites using the wobbled signal. It would then work out what the difference was and then transmit the information over radio. A GPS with a differential receiver could then use the extra information to undo SA and give a wobble-free fix, usually down to 10m or less.

SA was disabled on 1st January 2000 after they decided that it wasn't worth the extra hassle. The new Block III satellites currently being put into service have no SA capability at all. Instead, however, they can just turn the civilian signals on or off, at any time, for any satellite. This means that they could, in theory, deny GPS reception over any area of the planet of their choosing, which is far more effective than SA as it knocks out DGPS as well.


____________________
GSXR400 x2 | '94 RVF400 | '93 TZR125 4DL (again)
20:22:30 BLUEX5: i would love to be forced to undergo a**l plugging with different sizes
20:48:18 Temeluchus: comp you hunk of smouldering homos3x you

[Jim Mc]

It depends on the number of channels your gps supports simultaneously and the sample rate.

You don't need more than 3 samples to accurately triangulate but due to the fact that satellites are hurtling through space pretty fast, the more clock readings you get the more reliable your geometry calculations will be. The greater the difference in latitude between satellite samples helps here too, this is why GPS has problems above/below certain latitudes, less so these days due to the volume and orbits of satellites but it's still a possibility.

Then your sample rate, most commercial gps's have a 1hz update cycle which is 'good-enough' but the measurements are usually calculated linearly, which is again 'good-enough', until you go over a hill or around a corner which introduces error. The faster you go, the greater the distance covered between samples and the greater the error.

There are other factors too, but these are probably the most relevant.

[JP7]

Our cars at work have calibrated speedos, and the speed indicated on the GPS is bang on every time.

I had been told that GPS was very accurate, and this confirmed it for me.

[Dave70]

[Ollossm]

Day off tomorrow so I shall attempt to try it again.

A quick little trip today using another app but only in kph and it did seem to tie in with speedo rather more. But this was only a short little run.

Reckon a run down to (not so) Great Yarmouth tomorrow on the A47 should give me a better reading and therefore let me know how far out the speedo is.

cheers for all the replies, all helpful to me.
____________________
_______________________________________________________________
current bike: 2012 Honda CBF 125
past bikes: 2004 Suzuki VanVan 125 - 1988 MZ/Simson 50cc

[tahrey]

[angryjonny]

[Kickstart]

Hi

The speed warning signs seem to flash up around here at the speed limit at most, normally a couple of mph under the true speed.

I tend to presume that they are calibrated with a car driving towards them, but use any random car rather than one with an accurate speedo.

All the best

Keith
____________________
Traxpics, track day and racing photographs - Bimota Forum - Bike performance / thrust graphs for choosing gearing

[BG5067]

Ask apple
____________________
Current. Kawasaki ZX9R.

Theory passed 17/01/2014. Mod 1 Passed 03/02/2014 (0 faults) Mod 2 10/02/2014 Passed (0 faults)