We tried using an APM 2.5 as a flight data recorder in a Cessna 208B. It was built into a 3D printed box with own power supply and recording done on a mobile phone.
There is one problem though: When the aircraft is in a steady bank the APM HUD indicates horizontal attitude as the resultant g-load is downwards relative to the APM. The difference between APM indicated attitude and aircraft artificial horizon is shown below. (Aircraft AH indicates about 30deg LH bank in top right corner)
UPDATE: THE PROBLEM SEEMS TO BE THAT ARDUCOPTER CODE WAS LOADED AND IT WAS FLOWN AS DISARMED, INSTEAD OF USING ARDUPLANE - THANKS DEVELOPERS!
Here are two link for logs (one from Mission Planner, second from APM) from my flight in December 2013:
Both from the same flight:
airplane - ultralight VL-3 - http://www.vl-3.com
APM 2.5 with GPS installed. Main usage was a "distance trigger" for ortophoto.
3 cameras (Cannon SX280HS, Samsung NX1000 and Fuji E-X1 was used - triggered by servos (all from the same channel - ch6).
All time as manual flight, but now.... just received two servos for MGL Avionics Autipilot and my plan is to install its into VL-3 for 2-axes controll (without rudder) and to use APM as autopilot for ortophoto flights (lawnmover pattern). Those servos have RS-232 and CAN interfaces, so it will be necessary to make converter from PWM to RS-232 or CAN and MGL Avionics Servo software protocol (description available on their website).
I am in the process of using an APM for a data recorder on a UAS. We have an Israeli UAS and we put the APM in side the airframe to record and transmit the data back to custom software that takes samples from the UAS telemetry and the APM telemetry. The software timestamp the data and produces a report with the sampled data. The report will show the error between the sensors in the autopilot system. We plan to use this as a proof of concept project so that we can test other systems and create a database of Flight control system (FCS). We plan on flying next week and we will have more info on how well it worked.
Thanks! I'll update the blog.
yes, that would change things a lot. The Copter code disables centrifugal corrections when disarmed (to reduce the number of complaints made by copter users that their HUD moves about when the copter is on the bench)
Try again with the plane code, and make sure you use a good GPS (eg. a uBlox) and that it has clear view of the sky.
I see droneshare identifies the vehicle as Arducopter. Maybe I had Arducopter code loaded - could that be the problem?
I just did my first upload to droneshare. Hope this link is sufficient for you to view the log file:
Could the high airspeed (around 100 knots) be the problem?
Yes I have logs although it is quite a large file as the flight was about 2hrs. GPS velocity was correct I think, as it plots accurately in Google maps and I use a Ublox GPS. I can upload to droneshare if you are interested.
What wasn't accurate is the heading. It was only accurate until we did a 180 deg turn on the runway before take-off. Maybe I should have calibrated the compass again before flight. Still, I checked the portions during flight when the heading was correct and the roll was still incorrect.
Do you have logs or at least the parameters you used?
If the GPS velocity information is correct then it should get correct attitude.
Hi Murray ,
i doing same test 1 years ago , the main problem found was the wrong roll info when plane doing turn in air ... centripetal force i thought that could produce that problem ... Now with new gps correction could be nice to understand if problem solved .
Right, as Pieter is implying, if you have a GPS connected it should use the change in velocity from the GPS to correct for centripetal forces like you're seeing here.
What you've done is a great test though! We tend to dry the APM/Pixhawk around in cars to test rather than Cessnas!