Dual XGPS 160 accuracy issues

Discussion related to external and internal GPS / GLONASS / Galileo / BeiDou sensors
Tripper
Fewer than 10 Posts
Fewer than 10 Posts
Posts: 5
Joined: Wed Jan 16, 2019 10:07 pm

Dual XGPS 160 accuracy issues

Post by Tripper » Wed Jan 23, 2019 7:13 pm

I've bought a XGPS 160 and am using it with HLT. It works fine but when I export laps to view in track attack or circuit tools two laps from different sessions on the same day have an offset of 10m+.

The XGPS sits on the dash, has an unfettered view of the sky, and reports 'excellent' positional accuracy.

Do I have a faulty unit? Anyone else seeing this problem?

Incidentally the battery no longer holds a charge after 3 months of ownership and I have to have it plugged in. I doubt that's relevant but thought I'd mention it in case.
User avatar
Harry
Site Admin
Site Admin
Posts: 9402
Joined: Sun Sep 12, 2010 10:32 am
Location: Siegum, Germany
Contact:

Re: Dual XGPS 160 accuracy issues

Post by Harry » Fri Jan 25, 2019 9:13 am

LapTimer stores GPS data as it is coming in when defaults are used. At least for Circuit Tools, I know they use their own set of triggers (s/f) positions instead of using the ones provided by LapTimer (part of VBO export). So any gap in location of start / stop is a result of this different definitions of reality. This usually does not influence lap times beyond GPS achievable accuracy. Early start and early stop compensate each other.

Harry
Image Image Image Image
Tripper
Fewer than 10 Posts
Fewer than 10 Posts
Posts: 5
Joined: Wed Jan 16, 2019 10:07 pm

Re: Dual XGPS 160 accuracy issues

Post by Tripper » Fri Jan 25, 2019 3:42 pm

Here's a screenshot of what I mean (from trackattack):

Image

That's an enormous variation. The minimum number of satellites is 12. The start points are more or less identical. I was just wondering if this was normal for an XGPS 160 or if I have a unusually inaccurate unit.
User avatar
Harry
Site Admin
Site Admin
Posts: 9402
Joined: Sun Sep 12, 2010 10:32 am
Location: Siegum, Germany
Contact:

Re: Dual XGPS 160 accuracy issues

Post by Harry » Fri Jan 25, 2019 5:18 pm

What is the source for the red and the green. What is the time gap between both?

- Harry
Image Image Image Image
Tripper
Fewer than 10 Posts
Fewer than 10 Posts
Posts: 5
Joined: Wed Jan 16, 2019 10:07 pm

Re: Dual XGPS 160 accuracy issues

Post by Tripper » Fri Jan 25, 2019 5:52 pm

The red lap was at 16.05 GMT on 21st Jan, the green lap was at 13.45 on the same day. Both laps are from the same XGPS 160 on the same dash of the same car. The red lap was a 1.01 and the green lap a 1.06.
User avatar
Harry
Site Admin
Site Admin
Posts: 9402
Joined: Sun Sep 12, 2010 10:32 am
Location: Siegum, Germany
Contact:

Re: Dual XGPS 160 accuracy issues

Post by Harry » Fri Jan 25, 2019 6:31 pm

Thanks for the clarification. What you see is the so called GPS/position drift. Be prepared for an in depth answer. ;-)

When talking about position accuracy for GPS (and GLONASS / Galileo etc), we need to talk about absolute accuracy and relative accuracy.

Let me define absolute accuracy as the difference between your actual position and the position measured by the GPS. This is what matters e.g. for navigation, architecture, geo services. Relative accuracy is the difference between two measurements sampled for the same actual position within a short timeframe. They will differ although you are in the same spot - we have noise in our measurement. To measure lap times from positions, relative accuracy is what matters. It is tightly related to the accuracy stated by the sensor and what you see as accuracy in LapTimer's GPS View.

Absolute accuracy for GPSes can be quite big. Within a short time frame (e. g. a lap), it will be a kind of a shift following a certain direction and distance. You can see it in your diagram. The green line is shifted roughly North-East at a roughly constant distance. You can assume that both the red and the green line are shifted by a certain direction and distance. This direction and distances drift the whole day - slowly and random.

GPS and other systems can be augmented by systems to compensate this drift. The XGPS and most modern devices feature SBAS (Satellite based augmentation system). To make it work, a satellite with correction information needs to be locked. This is not always the case. You can see this correctional satellites in LapTimer's Satellites View, it has a special color (need to check which one it is). SBAS is far from perfect, the drift is not only time but location dependent. So even the XGPS with a correctional sat locked will have a drift.

Professional GPS systems use GBAS (ground based augmentation systems). They measure the position at a well known location and can compare the well known coordinates with what is measured. The difference is the drift. A GPS used on track can be augmented with this information to almost neutralize the drift. We are at a completely different price point here, four digits at least.

What does this mean? Lap timing depends on relative accuracy only as we can assume that the drift at the start and the end of a lap are almost the same. So no problem with this. To compare two lines recorded at different days or a few hours apart are hard to compare with uncorrected measurements. As an alternative to GBAS, LapTimer could auto-counter-drift laps to make them comparable using some numeric optimization algorithms (minimize the overall difference in positions). It doesn't currently and I'm not sure any system does it.

I'm aware this answer is probably more than you wanted to hear. :-)

- Harry
Image Image Image Image
Tripper
Fewer than 10 Posts
Fewer than 10 Posts
Posts: 5
Joined: Wed Jan 16, 2019 10:07 pm

Re: Dual XGPS 160 accuracy issues

Post by Tripper » Tue Jan 29, 2019 12:10 pm

Thanks very much for this detailed explanation. Having checked the accuracy of timing by comparing with video I have no complaints in this regard. It's very close.

GPS drift is totally new to me. I don't really understand how 'noise' affects the calculation, it's just comparing times transmitted digitally. And also how it does so in way which are very similar to other calculations temporally close but in ways that differ considerably over longer periods.

Presumably the accuracy claim of a GPS unit is relevant to its 'absolute' accuracy rather than relative to the previous point. So an accuracy of +/- 2.5m should cause a maximum difference of 5m. I'm seeing at least 10m.

Harry, there's no reason you should answer these questions, it's hardly part of your support remit! Thanks very much.
User avatar
Harry
Site Admin
Site Admin
Posts: 9402
Joined: Sun Sep 12, 2010 10:32 am
Location: Siegum, Germany
Contact:

Re: Dual XGPS 160 accuracy issues

Post by Harry » Tue Jan 29, 2019 12:28 pm

No problem, I hope many readers are interested in stuff like this. :-)

Position calculation is not digital at all. GPS receivers receive signals from multiple satellites with a common reference time. The location is derived from signal run times (a satellite being further away has a longer signal run time than one nearer by) - kind of a super-triangulation. Run times are influenced by atmospheric disturbances, reflections etc. You can see this easily by standing still anywhere and watch the position reported in LapTimer's GPS View. It is changing all the time.

Accuracy delivered by receivers are "statistic values", nothing that is measured or precise. In fact, GPS chip manufacturers just commit that a certain percentage of positions delivered are within the accuracy reported. Something like 95%. And it is definitively relative accuracy.

So you see deriving lap times and assessing their accuracy is not a trivial exercise? ;-)

- Harry
Image Image Image Image
Tripper
Fewer than 10 Posts
Fewer than 10 Posts
Posts: 5
Joined: Wed Jan 16, 2019 10:07 pm

Re: Dual XGPS 160 accuracy issues

Post by Tripper » Tue Jan 29, 2019 2:42 pm

Harry wrote:
Tue Jan 29, 2019 12:28 pm
Position calculation is not digital at all. GPS receivers receive signals from multiple satellites
What I meant was that those signals are digital. They contain digitally encoded information including, presumably, the time of transmission of the signal and information that identifies the satellite, and (I'm guessing here) information about the position of the satellite.

If I understand you correctly you're saying that conditions affect the speed of the microwave signals causing the variation in the calculation result but changes in those conditions happen slowly enough that calculations that happen close in time are sufficiently accurate relative to each other.
User avatar
Harry
Site Admin
Site Admin
Posts: 9402
Joined: Sun Sep 12, 2010 10:32 am
Location: Siegum, Germany
Contact:

Re: Dual XGPS 160 accuracy issues

Post by Harry » Tue Jan 29, 2019 2:53 pm

Yes, the signal is digital, but position calculation is not as lined out in my last post. Measuring signal runtimes is what leads to "noise", i.e. small variations for each measurement. This noise has no special direction, but is chaotic. The drift discussed has its roots in slow changing variables like clock drift, ionospheric delay, or probably inaccuracy in the satellite positions going into the calculation.

- Harry
Image Image Image Image
Post Reply