Autosport Labs

So I've just been looking at the data that was sent to race-capture during the shakedown I did yesterday. Mostly it seems pretty reasonable, but I'm surprised at the accuracy (or lack thereof) of the GPS data.

This is the event:

http://www.race-capture.com/events/7965 ... her-s-jedi

Some of the laps look pretty good. Lap 28, for instance:

http://www.race-capture.com/events/7965 ... ?laps[]=27

But others less so, lap 27, for instance:

http://www.race-capture.com/events/7965 ... ?laps[]=26

(I really didn't straight-line the chicane to that extent!)

And others are wildly off, for example lap 24:

http://www.race-capture.com/events/7965 ... ?laps[]=23

Is this indicative of something being wrong in the installation or configuration?

it looks like for some of laps where you might've been 'teleported' the telemetry data just hiccuped and there was a break in the data. How are you doing telemetry- via the phone app or internal telemetry module?

In other cases, there might be a drop off in GPS satellite data. What is the terrain like around the track? Also, where do you have the GPS antenna mounted relative to the engine and other radio transmitters? You do want the GPS antenna as far away from other RF transmitters as possible. If you're rear/mid engined then nose of the car is a good point. Also make sure the GPS connection is a little on the tight side of snug.

Here's your summary data where you can see the # of satellites dropping off.
http://www.race-capture.com/events/7965 ... di/summary

In the 2.8.0 firmware we're developing we're adding some more GPS diagnostics channels which will report the reported Fix quality as well as the GPS "Dilution of Precision" indicator which can show you how good the field of GPS satellites look.

Also, if you carefully mouse over the charts where it looks like you've been 'teleported' you'll see that there are missing samples between the jump in the points:

http://www.race-capture.com/events/7965 ... ?laps[]=26

Thanks Brent.

I'm using the internal telemetry. I'm not sure what the signal was like at the circuit (the signal on my 'phone was a bit patchy, but I use a different network on my phone from the one I'm using for the RCP). Is there any way to log signal strength?

The car is a mid-engined spaceframe with GRP bodywork. I have the RCP mounted just ahead of the dashboard (just above my right knee) so it's about as far away from the engine as it can be. I'm using the supplied GSM aerial (i.e. it's inside the bodywork) but I was assuming that that should give pretty good reception given that the bodywork is GRP and shouldn't block the signal much?

Does the telemetry not support acknowledgement and retransmission? Quite a few UK circuits have patchy connectivity—I'll be amazed if there's connectivity at every point of Cadwell Park (very hilly terrain and heavily wooded in places), for example.

The GPS aerial is mounted very close to the RCP, also inside the bodywork. It would be easy enough to move it so that it's further forwards, and outside the bodywork (down near the front wheel) if you think that that's likely to make a difference? I'll check the connection of the GPS aerial

Thanks!

We don't have a technique for automatic backfilling, yet. working on that!

It looks like the GPS data was pretty good overall judging by the lap times despite the missing data. The charts in the summary where satellites go to zero - is that when you came in after a session?

brentp wrote:We don't have a technique for automatic backfilling, yet. working on that!

Cool

It looks like the GPS data was pretty good overall judging by the lap times despite the missing data. The charts in the summary where satellites go to zero - is that when you came in after a session?

Ah! Quite possibly, yes (something that would make this *much* easier to determine would be if there were at timestamp on the data?). So this is probably an artefact of the RCP being power-cycled?

So that suggests that the only problem is telemetry backfilling? Which isn't something I need to worry about immediately.

Thanks!

Correct, we have a UTC timestamp (in millis) on every sample - you'll see that in the logfile on the SD card, and is also sent via telemetry.

-Brent

brentp wrote:Correct, we have a UTC timestamp (in millis) on every sample - you'll see that in the logfile on the SD card, and is also sent via telemetry.

Cool. Is there any way to see it on race-capture.com?

Not currently - We actually recently hid that channel b/c it was cluttering the interface (number was so long!)