OK, here's my latest GPS query ...

[George]

Yesterday, I went out on the tandem with Roger Green.

If I look at the history files on the device itself, or if I upload them to Garmin Connect, it appears that our return route was 36.88km. If I upload the files to Ride with GPS, that site says we did 37.1 coming back. And BikeHike says we did 37.09.

The difference between the last two is negligible, and may be just due to a different approach to the rounding off of figures for presentation. But I'm surprised that they are so different from the device's own figure: more than 0.2km out. Can anyone explain the difference? For example, could it be because the on-line applications are making assumptions about the route we took through junctions, etc, which aren't correct? And is it safe to assume that the Garmin's own figure is more accurate than the on-line apps' figures? Or does the Garmin itself make certain calculations/assumptions, which might be incorrect? If we pass under a motorway bridge and it briefly loses satellite reception, does that affect anything?

[Andy Terry]

If you mail me the GPX file, I'll have a look.

[Andy Terry]

There are a few spurious trackpoints at the start of the file, indicative of not zeroing properly.

When these points are discarded, the total distance is 36.8km. There is a break in the track outside the Jinney Ring, although you did not nip in for a quick cuppa. I calculated the distance by adding up all the point-to-point distances. There are about 1000 points in your track. If you import the 'corrected' track into RWGPS, the distance is also 36.8 which confirms my maths.

Try this procedure:

1. switch on GPS
2. move to a place where you can get a view of the open sky and hence a good strong signal i.e. somewhere in the open, away from buildings
3. give the unit a minute or two to lock on to the satellites and establish its position (on mine there's a circle that gets smaller, some units have an 'accuracy' reading in metres which you want in single figures)
4. now, and only now, zero the mileage
4a. If you are obsessive about ascent, this is the time to calibrate and zero your altimeter (if available)
5. ride
6. at the end of the ride, switch off as soon as you stop i.e. before carrying the GPS into the house

[George]

Thanks, Andy.

I'm slightly unsure what you mean by 'zeroing'. My unit has a 'Reset' button and a 'Start/Stop' button. I've always assumed that the 'Reset' process was just a file management process (the existing 'Current' file converted to a History file, and a new 'Current' file started, but no trackpoints recorded) and not therefore dependent on satellite reception. Is that assumption false? Is it maybe the case that when you press 'Reset' it records a 'starting point', even if you haven't actually pressed 'Start'?

Or, when you talk about 'zeroing', do you mean pressing the 'Start/Stop [recording]' button? I did do that in the open air, albeit while still standing close to the café, and possibly therefore in its satellite signal shadow. I must confess that I don't usually watch the unit while it locks on to the satellites, since that takes several minutes, and I know it'll warn me if the process has failed. I didn't get a warning, so I assumed it had succeeded, and carried on. Between pressing Start/Stop and actually starting to ride, the tandem was wheeled about 10 metres.

We did have to give way to traffic by the Jinney Ring, so the Auto Pause feature will have kicked in then.

I'm intrigued to know how you quickly and easily pick up trackpoints as 'spurious' and make point-to-point distance calculations. Do you open the file in a spreadsheet, for example?

[George]

PS, do you think tree cover can affect the accuracy of the satellite lock-on, leading to similar spurious trackpoints at the start of a ride? Because I've noticed similar minor discrepancies with rides starting from home, where I always start the unit outside, but under trees.

[Andy Terry]

Yes, I meant 'reset'. And mine doesn't have start/stop - when it's switched on it's always recording.

So, to sum up:

- get into the open
- power on
- wait for satellite lock-on (don't move)
- reset
- start
- ride
- stop
- power off

Dense tree cover can affect satellite reception, yes.

Can you turn off the auto-pause feature? What's the point of it?

I'm intrigued to know how you quickly and easily pick up trackpoints as 'spurious' and make point-to-point distance calculations. Do you open the file in a spreadsheet, for example?

I use a piece of software called GPS Utility which lets you display individual trackpoints, but I think MapSource can do the same. Also you can see there's something funny going on if you zoom in with RWGPS

http://ridewithgps.com/routes/209651

It looks like you started your ride in the middle of a field.

Point to point distances you can do in Excel with a bit of Pythagoras.

[George]

Yes, I meant 'reset'. And mine doesn't have start/stop - when it's switched on it's always recording.

Well, I think the 605/705 are a bit different in that respect. When the unit is switched on, it isn't necessarily recording; it only starts recording when you press Start/Stop. So it simply looks as though I had poor reception when I started recording (although there must have been some reception, or it would have said 'Can't find satellites').

So, to sum up:

- get into the open
- power on
- wait for satellite lock-on (don't move)
- reset
- start
- ride
- stop
- power off

I'm sure you're right, Andy ... but going through the whole procedure at the café can be a bit of a fag, especially if you are in company and you want to use the 'Navigate' feature (it takes several minutes to work out the directions). In this instance, it would have meant the two of us walking out to the middle of the car park, and then spending at least 5 minutes standing around waiting for the GPS to lock on and calculate its route before pushing off.

Can you turn off the auto-pause feature? What's the point of it?

Yes, I can disable it, but I've always kept it enabled because that's what I'm used to on an 'ordinary' computer: it means your average riding speed really is just that, rather than a door-to-door average that reflects other variables, such as whether you got caught at traffic lights.

I use a piece of software called GPS Utility which lets you display individual trackpoints, but I think MapSource can do the same.

Thanks, that's handy to know; I'll investigate.

Also you can see there's something funny going on if you zoom in with RWGPS. It looks like you started your ride in the middle of a field.

As it happens, I had independently discovered this myself since my last posting. Not only does it appear that we started in the middle of the field, but the route has us riding off the car park through the fence, then back on again, before exiting a second time. Presumably all on account of it needing some time to find the satellites. I wonder whether it is slower to lock on if it is doing other things at the same time, e.g. navigating? (A bit like a PC that is running several capacity-intensive processes being slow to execute a command.)

[Andy Terry]

Remember how this thread started. You were questioning a slight difference in distance figures produced by RWGPS, Garmin etc. If you want spot-on accuracy, follow my suggestions. If you're not that bothered, carry on as before. The GPS I use for running (the 305) takes forever to 'acquire' satellites and I have got some very strange looks while I wait, motionless, arm outstretched, for it to 'lock on'.

And this is probably why I use a 'traditional' computer for distance and average speed. The GPS is just for navigation.

[George]

Well, the daft thing is that I'm not actually all that bothered about the 0.2km discrepancy itself. We are only talking about a casual, low-intensity ride, after all, and it makes no practical difference whether it was 36.9km or 37.1.

Really, I was just curious as to why there was a difference. And, in fact, even though I have learnt a lot in the course of this exchange, I am still curious as to why there was a difference. After all, both Garmin Connect and RwGPS were reading the same data, complete with spurious trackpoints at the start. Yet Garmin Connect seems to have 'known' that the jumping in and out of the field could be ignored, while RwGPS counted it as part of the trip.

[Andy Terry]

Really, I was just curious as to why there was a difference.

Without knowing how the thing calculates the 'distance' on its display it's difficult to say.

Of course, you and I both know that a properly calibrated computer is by far the most accurate way of measuring distance.

[Andy Terry]

After all, both Garmin Connect and RwGPS were reading the same data, complete with spurious trackpoints at the start.

Possibly not - GC might just the displayed distance without re-calculating from scratch like RWGPS does.

[Andy Terry]

After all, both Garmin Connect and RwGPS were reading the same data, complete with spurious trackpoints at the start.

Possibly not - GC might just the displayed distance without re-calculating from scratch like RWGPS does.