I took a look at this too after it was posted on the Wiki. As I understand the code, the attitude solution is calculated at 50Hz, It is just that it is polled from the module on the paparazzi side at 15Hz.
I tried to read it at 50Hz, but it didn't work (probably overloads the processor with I2c interrupts or has some other timing problem). But reading at 40Hz seemed to work quite OK. It was a noticable improvement in the control response in Auto1.
It flew my plane with the same control parameters I use for IR sensors. The only thing I noticed was a pitch-down on acceleration and pitch-up on deceleration, which is to be expected.
----- Original Message -----
Sent: Wednesday, December 01, 2010 9:25 PM
Subject: Re: [Paparazzi-devel] Reply to:"Re: ArduIMU and Tiny autopilotcontinued (David Conger)"
Does anyone know how fast the gyro integration is done in the arduIMU (Atmel). (DCM propagation)
15Hz attitude data is sufficient for stable flight (slow of fast) but the lower bandwidth sensor also means you need to reduce the gain compared to 60Hz attitude, especially when flying fast (extra phase gives smaller gain margin).
But 15Hz integration is very very low on anything as small and agile as a Funjet. If the weather is smooth enough I do believe this will work, but it's also extremely easy to make is not work anymore by adding vibrations, turbulence, etc...
Even with a very efficient fixed-point implementation, a 20MHz 8-bit micro is "a bit" weak...
Paparazzi-devel mailing listaddress@hiddenhttp://lists.nongnu.org/mailman/listinfo/paparazzi-devel