[ahrs] Integer rate integration fix

[EwoudSmeur]

Fixed the integer rate integration. The conversion from int32_t to int64_t went wrong. Fast rates gave false quaternion updates.

[gautierhattenberger]

It looks good to me, but I can't test it.
If I didn't made a mistake, you can have issues at around 900 deg/s. So we clearly need 64bits since our sensors saturate at 2000deg/s at full range.

[fvantienen]

We already tested this on the ARdrone and seems to work as expected. Previously we had problems with fast turn rates, that's how we found the problem.
We debugged every step with a fast onboard debugger and noticed that you for sure need that 64bits it was supposed to be using but didn't.

[flixr]

Nice find! Seems this has been wrong and not noticed since Antoine added the implementation in 2011...
while at it, it would make sense to also convert this macro to a static inline function (no reason for this to be a macro, and functions are nicer to read/debug).

[agressiva]

HUmmm, in my last flights with quadrocotper, when i turn the YAW very fast the attitude start to work very bad.

When i say fast ... is very fast... much more than 500deg/s (spin at max yaw )
I will try the same next week.

Date: Thu, 31 Jul 2014 14:39:22 -0700
From: notifications@github.com
To: paparazzi@noreply.github.com
Subject: Re: [paparazzi] [ahrs] Integer rate integration fix (#782)

It looks good to me, but I can't test it.

If I didn't made a mistake, you can have issues at around 900 deg/s. So we clearly need 64bits since our sensors saturate at 2000deg/s at full range.

—
Reply to this email directly or view it on GitHub.