[Paparazzi-devel] Test flight results (Lisa/M, ETS, Pitch-loop, GPS)

[Marc Schwarzbach]

Hi,

did a long test flight yesterday with results probably interesting and related 
to some previous issues...

Setup: Easystar, Lisa/M 2, UBlox Lea-4, Eagletree Speed on I2C2."Old control"



1. Probably most interesting because related to recent discussions: Pitch loop. 
I started manually and commanded to home circle. Altitude hold was really bad, 
looking at the data, there is a phase offset from desired climb to desired 
pitch which led to huge oscillation around desired altitude. Looking at the 
code, I think the reason are the integrators of the auto pitch loop winding up 
and taking time to desaturate.. The values of the integrators (limits in the 
code, not gains) seem quite high. Is this dependant on control frequency?
1.a There is no setting file for airspeed in "old control" (like 
basic_airspeed). I wrote one, probably should be integrated?
Should "new control" perform better anyway? Then it should/could be made 
standard in the example airframes, which it is not at the moment...

2. The Airspeed sensor works on the I2C2 port with 3.3V (also written in the 
specs). But initialisation seems to be a problem. Only after 
(Luftboot)programming, the sensor delivers data. If I just plug the battery in, 
airspeed is constant 8.. Sensor display shows numbers like usual. Should there 
be some kind of pause in the ETS init? Or is this an I2C related problem?

3. The V_BATT voltage reading was wrong with the standard conversion value. I 
had to alter it from 0.00485*adc to 0.00459*adc. Were there different values of 
resistors in the production in the voltage divider?

4 GPS... The performance of the Lea-4 (navilock module) seems to be much worse 
with the configuration of the gps_ubx_ucenter.xml module compared to the 
offboard configuration with the conf/gps/ublox_conf.c. Especially altitude is 
drifting like crazy and z_dot looks bad (most times close to 0 and not fitting 
to the actual altitude changes). The differences seem to be Airborne 2G instead 
4G and EGNOS activated. I will check this in future flights but confirmation of 
this issue by others would be good.


The Easystar really showed its forgiving nature in this flight. It was the 
first flight with Airspeed and all the tuning could be done in auto2 without 
switching back (20 minutes) :-) Looked crazy in the air, but never unstable. 

Cheers

Marc