Searching tips to maintain steady altitude during fast flights with quad

I am using latest arducopter stable version on my rpi3+pxfmini quad.
I am searching for tips in order to maintain a steady altitude when i move the quad in its fatest speeds.
The effect i have observed is a loss of altitude when i go too fast, that altitude being automatically recovered when i slow down. The only solution i found for the moment is to limit the max. Inclination angle in the params to 30 deg max instead of 45. But that only partially solves the issue. Probably 20 or 25deg would allow a proper flight
I have this issue flying Althold or Poshold but obviously not with loiter or rtl or auto modes.
My quad has gone through autotune an is flying well for all kind of modes
Thanks for any tips possible! :slight_smile:

If it pops back up, then it is probably because air movement through the frame is causing an air pressure drop around the barometer. This makes it think it is climbing when it shouldn’t be, so it descends. When you slow down, it sees the correct pressure again, realizes it is low, and climbs back up. Solving this can be a pain. Depends how you have the pixhawk mounted.

1 Like

Thanks for the advice
My flight controller is mounted below a top frame cover. It is strange to me to imagine a big depression as air seem to freely circulate on all sides, but yes it definitely souns like a possibility.
Couldn’t it as well be that my motors are not powerful enough to maintain altitude and in the same time move at top speeds? Something like “not enough rpm” on the props.
Fyi in stabilize mode i get a hover at around 50% power

Look at the logs
Agree with Matt your profile creates depressurisation

Not in this case. When that happens and altitude begins to decay, the target altitude will decay along with it. And since the target altitude decays with the actual altitude, when you slow down, the copter will maintain the present altitude rather than popping back up to where it was before.

Thanks again.
Do we agree i should see constant altitude (vs actual altitude variating depending on speed) in the logs if the “depressurisation” effect is confirmed?
I will check as soon as i am back from holidays