savnat speed accuracy

If the refresh was faster the digital display would be hard to read.

My favourite ever Viz definition -

Stephen Fry: A stupid person's idea of what a clever person is like.

Ian

As you say, calculating the Doppler shift isn't difficult as you know the satellite speed. The receiver signal processor locks to the 1.023 Mb/s Coarse Acquisition code using a code locked loop and the 1575.42 MHz L1 carrier frequency, the latter using a phase locked loop. (L1 also carries the slow speed ephemeris data)

The Code Locked Loop is processed to produce pseudo range information which includes the error in the receiver's internal clock added to the real geometric range of the satellite. The PLL measures the carrier phase rate (apparent satellite frequency with respect to the receiver's local oscillator/timing clock) -- this is the Pseudo-Range Rate (PRR).

All the PR+PRR data from all the satellites in view goes into a Kalman filter to produce the combination of past & present PRs and PRRs from many satellites to generate the Position, Velocity & Time (PVT) estimate.

Assuming no satellite errors GPS signal carrier wavelength is about 20 cm. The PLL in the receiver can measure the carrier phase to about 1 cm (1/20th of a cycle). If you measure for 1 second you see ~1 cm/sec of velocity for a given satellite (0.023 miles/hour). Factor in the geometry and the fact that there are multiple satellites you multiply this by horizontal dilution of precision for horizontal speed. HDOP is rarely > 3 so the "system" velocity error is rarely

satellites are not perfect.

As you are using carrier phase rate to determine speed you can measuring the relative L-band carrier frequency (i.e. Doppler offset) to 1/20th of a Hz which isn't bad for a box you hold in your hand that cost about £100.

There are two reasons you don't get faster updates. One is the receiver black box usually only outputs one data set a second. The other is that the display side of the device may smooth the readings still further.

That is how GPS receivers measure speed, not by distance/time calculations.

It's sort of right, though.

The position estimates feed into a chain of filtering and smoothing algorithms and the result is used to guesstimate position and speed.

If you're going steadily in a straight flat line, then the estimate will be at it's best.

Ah, so they orbit the earth. Presumably a GPS system would need at least three satellites per hemisphere (N/S) to maintain 24/7 coverage?

Every GPS user needs a minimum of 3 satellites visible to get lat/long resolution, and 4 to resolve altitude. So world coverage needs lots of satellites. I understand the minimum required is 24, with about 31 currently in service.

John

Actually the GPS calculates your speed from the Doppler shift on the signal from the satellites. It's nothing to do with distance/time measurements.

Tony

the doppler effect is only accurate if your tragectory is either directly towards or away from the satalite , any position other than directly towards or away will give an inaccurate reading.

Time distance is far more accurate as each satalite transmits its signal at a given time , 3 or more signal receptions will give you a precise position anyware within reception range , as these figures change every few milliseconds its a simple task for the gps to calculate speed direction and position

If you used the doppler effect you would still need time and distance calculations from several other satalites unless of course you were travelling at right angles to the satalite which is near inpossible

And how does it correct for the Doppler caused by the satellite, which is moving significantly faster than the car?

By using the time+distance estimates, maybe?

No, by using information from the satellite which tells the receiver processor how fast the satellite is going. It then becomes relatively simple to calculate the offset. Speed is calculated from data which is already required so it isn't as if something extra is required.

On early receivers (such as the first generation Garmins) the slow processor put a limit on the mathemetics the device could do between updates so some had quite low maximum speeds, 99kts for the first handheld Garmin. This was commonly misrepresented as an artificial limit - it wasn't, the proceesor couldn't work fast enough to cope with faster speeds!

On at least some of the early Garmins it was indeed an artificial link. The Garmin 95 aviation GPS was exactly the same as one of the other, cheaper ones, but (a) would go up to 999kt instead of 99kt and (b) cost a lot more. The difference was a single wire link on the circuit board.

Ian

The GPS set knows the position and velocity of the satellites to very high precision.

Ian

That would be savellite natigation then?

Presumably it has to use that information from *several* satellites, to figure out the angles involved?

Do you have a good link for the algorithms and stuff the latest generation of receivers use? I know what they used to do when I was involved in it, but I guess they've moved on...

Surprisingly the SiRFstar datasheets and the like don't give that much detail..

The same number it needs to get a horizontal fix so a minimum of three.

Ta. That's my light reading covered for Christmas ;-)

Just slow down to about 10 mph under the limit like everyone else.

It's a complete nonsense with modern pulse counting systems. Which should be accurate to within 1% at worst - assuming the correct diameter tyre. And since tyre wear makes the speedo over-read which is the safe option there's no excuse for it being fast on new tyres.

That's a big issue though.