[Canberrauav] successful testing of Telemaster today

Andrew Tridgell tridge at samba.org
Sat Dec 10 20:37:11 EST 2011

Steve and I spent nearly 6 hours at CMAC testing the Telemaster
today. The repairs that Jack did worked very well!

We did a total of 41 minutes of AUTO flight, along with some LOITER and
of course some manual flight. For the AUTO flights we concentrated on a
750x200 meter rectangular circuit that Jack and I had been testing in
the simulator. The circuit is setup to loop indefinately, and provides a
good guide for tuning.

We didn't have any problems with holding altitude, so that problem seems
to be behind us now. What we are now concentrating on is getting the
parameters just right to keep airspeed, altitude, roll, pitch and
throttle all as smooth as possible, to maximise the effectiveness of the

We tried 5 different sets of tuning parameters, plotting the results
between runs on the laptop and trying to work out what we need to change
to get better results. After the time at CMAC, we then came back to my
place to look at the data some more.

There were a few key results:

 - the natural flying attitude is much more nose up than the "top of
   fuselage level" that we had been so carefully using. It needs to be
   about 6 degrees nose up from that level. 

 - we were able to significantly improve the tuning over those 5
   flights by looking at the graphs of the results and tweaking a few
   parameters each time. We should be able to repeat the changes with
   the mugin.

 - we found that the airspeed sensor data is very noisy, which leads to
   the throttle changing quite rapidly, which in turn leads to poor
   values for the other key variables. Steve and I worked on some code
   changes to better filter the airspeed sensor values. I'll test those
   some more before pushing.

 - we also found that the accelerometer data is very noisy, and seems to
   correlate with throttle. We suspect motor vibration, and that this
   leads to aliased noise in the accelerometer data which leads to a
   poor attitude solution (particularly pitch). The noise is quite
   severe and we really need to fix this. I'm setting up some code for
   bench testing, so I can try different vibration damping materials.

We also found one problem with the repaired airframe. Without the struts
the wing ends up rotating a big during flight (so the front of the wing
is about 1cm off center at the end of the flight). We may need a pin to
hold the wing centered.

Now I've got a whole pile of data to analyse and some vibration to

Cheers, Tridge

More information about the Canberrauav mailing list