Map / Data issue

[architect]

I get random and consistent spikes through my sessions after I recalculate Acceleration using GPS data.

Also exit or enter speeds are at the wrong points or just not shown at all.

Thanks

[Harry]

There is no such thing as an entry or exit speed. The speeds shown are the max speeds (on straights) and min speeds (in corners).

To remove the spikes, select a stronger smoothing when recalculating acceleration (slider to right).

- Harry

[architect]

Thanks. Can you do a quick explanation of how the spikes occur why hard smoothing is not recommended? I had to set mine 3/4 into the right in order to get it reasonable.

EDIT: Just found your other response:

The GPS (either external or smartphone) should be positioned on the motorbike's vertical axis. In case it is positioned to the left or right, there will be an impact - but I doubt it is too critical (but haven't done the math). Spikes are from GPS inaccuracies / noise. When calculating lean / lateral from GPS, you get a dialog that allows you to smooth the calculation - to work around this noise effects. Just repeat the process, you don't lose anything the second time.

- Harald

[Harry]

When smoothing too much, you will smooth away max values. The level of smoothing required depends a lot on GPS update rates. While 1 Hz data needs almost no change, higher update rates need smoothing.

- Harry

[architect]

Is the smoothing just an average of the max and mins? To a point where I can actually get it looking rather smooth, the Max has been chopped by almost 50%. That's why I'm not sure what the best approach in "reading" my data is ...

What causes the noise or inaccuracies? Should I trust the Max or Min more? Just trying to get a best practice approach going forward.

[Harry]

Calculation of lateral acceleration is done by deriving G from speed and the radius driven for consecutive GPS positions. This mean any inaccuracy of GPS positions goes into this calculation. The shorter the distance between two fixes are (higher update rates produce shorter distances), the bigger is the effect of inaccuracies (which are typically the same the same for high and low update rate GPSes). Smoothing is done using cubic splines using some "smallest error" algorithm - complex stuff.

Both spikes from noise and too strong smoothing produce wrong values. The "correct" value cannot be derived. But what you get is a consistent picture. You can recognise which corner you go near a limit and which ones not. So it is all about relative comparisons, not absolute values.

My guidance is to smooth until you have eliminated all spikes going to the wrong direction of the corner and stop here. This makes values reasonable and allows you to analyse your relative corner performance with values not smoothed more than necessary.

- Harry

[architect]

Thanks for your patience and explanation.

Does the lean angle/map first uses the gyro inside the phone before I smooth using GPS data? Because the lean angle issue I have is still there, even though I've ensured I've mounted my phone correctly and ensured the G forces are within 0.05 at neutral. There are no longer only positive or only negative values, but they are still spikey and low in value. It's not until I recalculate Acceleration and smooth until I get something that appears more realistic.

I find I'm pretty much have to be smoothing to the strongest right now and I'm not sure that's right/normal. This is at 5hz.