Minor update

I’ve added the gravity complementary filter and a few other bits to hopefully work together with it:

  • dlpf is now set to 2 – 94Hz with 3ms lag (from 3 – 44Hz and 4.9ms lag)
  • motion processing frequency of 71Hz (from 43Hz)

The aim of the first change is to ensure we get to see the acceleration spikes, so the filter can take them out of gravity.  We can’t let the dlpf take them out as the motion processing needs to see them to calculate net acceleration with gravity removed.

The aim of the second change is to increase the resolution of the motion processing cycles so that fewer cycles are impacted by acceleration spikes.

Both of the above changes are only possible due to the recent increase of speed in the code from ∼450Hz to ∼900Hz.

Net result though wasn’t significant in the first test flight – same climb well above what the flight plan defined.  Not a problem, I realized it would need tuning.

What was a problem is the battery bank I use to power Phoebe.  It appears they don’t like impacts.  I’ve gone through 4 now – it seems as though the USB B socket gets detached; the bank charges fine, but there’s no power from the USB B socket.  Ironically, they are beautifully made, and I can’t find a way to open them up to fix the problem.

I could do with a super low profile USB B plug, but I’ve never seen such a thing.

Instead another tenner to Amazon for yet another battery bank.

2 thoughts on “Minor update

    • All filters are low pass; dlpf and my integration averaging takes out high frequency vibrations from the props / motors; the gravity CF operates nearer the 0.5Hz level (still needs testin / tuning), filtering out real acceleration spikes (i.e. when the flight velocity target changes from hover to ascend) and allowing through the drift in gravity measurement by the sensors. The aim is then that the (accel – gravity) * time calculation of velocity does not drift – this happens primarily in the Z axis preventing hovers.

      Still haven’t had the chance to test this properly yet due to various breakages. Hopefully I’ll have a better idea if this works later today.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.