"Infineon" raceway POI set

[Nician87]

I was driving around Sonoma raceway (aka Sears Point or Infineon. See http://www.autoblog.com/2012/06/03/infi ... mporarily/) yesterday and the "certified" track set of POI seems a bit off.

The GPS data says I have a good fix and the points look excellent in the map view, but I have frequent problems with the turn 4 (T4) and turn 7 (T7) triggers not firing. Fortunately, the start/finish trigger does seem to consistently fire so I got good lap data during each of my sessions on the track.

Because T4 and T7 aren't firing, the sector times are wrong for most laps.

I tried editing the T4 and T7 points. T7 should be an angle of 355 deg not 45 and i moved T4 (N38 09.850', W122 27.802') and T7 (N38 09.968', W122 27.755').

That puts them spot on the grey line when I "test" the POI set, but I'm still not getting the triggers to fire when I recalculate. Even using "wide" detection corridors which are set to the default 25m/18deg settings. Looking at the map view of a lap where neither T4 nor T7 triggered, it appears that the triggers are much closer than that given that the GPS fixes are all HDOP=0.8 and ACCURACY_M=1.4.

What position fixes are used for the "test" view of a POI set? The "REF" set in the lap list? The first one in the list? Something downloaded?

I'm attaching two laps that should help illustrate the issue.

[Harry]

Yes, the named split triggers are not well defined for Infineon. Changed T4 to 5° and T7 to 345° and uploaded it to the server. Recalculating the above laps, results in 3 splits each (using Wide, which is the default for triggers with direction anyway).

- Harry

[Horstair]

There's something seriously wrong with Infineon/Sonoma/Sears... I just posted in "Using Laptimer" about the loss of sectors (two of them combine...) sporadically (all you get is sector one and then the rest). It also once in a while does not trigger right at start/finish. I always wondered why there wasn't a "Harry" LOAD of this track but some dude's name (only one in available tracks). Why does Harry's name show up on many West Coast tracks like Laguna, Thunderhill, Buttonwillows etc. BUT NOT Infineon? Does it even matter?
In any case, can we get Infineon fixed somehow? Horst.

[Harry]

Please load the updated track set (see my last post) from "Tracks" and check if recalculating your split time works now. Recalculation is done in the respective lap's details: Lap List -> any lap. Press the "Recalculate" button and select "Splits" to rerun the split detection using the updated track set. At least the two recordings Nician87 sent, work fine using the new track set.

The name mentioned actually does not make a difference. It is simply to honor the person who initially submitted the track set. For some tracks I do it myself using Google Earth, in this case "Harry" appears as the author. Please note that neither is good or bad, Google Earth is often not up to date, and users sometime add some weak information too. The only track sets you can expect extra quality from are those I have driven using LapTimer myself. They are marked with a certificate (check e.g. Nürburgring).

In case start / finish is missed, this is most probably due to a GPS accuracy problem. Please export any typical lap you have recorded using the .kml format and I will have a look into it. Please use my email address Harry@gps-laptimer.de

- Harry

[Nician87]

Yes, the named split triggers are not well defined for Infineon. Changed T4 to 5° and T7 to 345° and uploaded it to the server. Recalculating the above laps, results in 3 splits each (using Wide, which is the default for triggers with direction anyway).

- Harry

Thanks, I will try that if I can save my current tweaked POI points. I chose some different angles and moved some of the points to try and better locate them at the apex of the turns. I e-mailed you my version.

I had tried playing with the triggers, but couldn't get all 40 laps that were recorded in my recent lapping session to detect correctly even with Wide settings.

So I tried to develop my own approach and split algorithm. It turned out to be much harder than I thought to take into account all the different scaling (longitude compression related to latitude), direction, etc. However, I have something that works on all triggers of my 40 laps with no false positives on reverse approaches to directional triggers. Further, all the math is done with adds and multiplies and no need for atan(), sin(), cos() or other transcendental functions during lapping (sin/cos do need to be pre-calculated for PIO Triggers). I'm happy to share if you're interested.

[Harry]

Sure, I can certainly have a look into it. It is indeed complex to build an algorithm working reliantly in all situations / configurations / tracks. Detecting is the one thing, not detecting wrongly the other. I'm case I would have an order on POIs (i.e. know which one is next), it would be a lot easier. That would require a design change however.

Harry