Could I please get a second opinion on a set of logs?  If I'm wrong, and this is something I could have prevented, I would like to know that before I fly again.

This weekend I suffered my first crash which I can be certain was caused by the APM. As you'll see from the 2-minute video above, about 1 minute into the flight the plane started making uncommanded roll movements, which became more and more extreme until the plane was uncontrollable and went down.

This was not a configuration problem. I had just completed a 98-waypoint Auto flight (without restarting, which may have been my downfall). This was to be the last of dozens of flights over two days with no problems whatsoever.

Miraculously, the plane survived mostly intact, and there were no mechanical or wiring problems whatsoever. Definitely nothing that could explain the crash.

In researching the attached logs, I found that the APM was rolling because it thought it was getting roll commands on ch1in. Channel 1 on my transmitter is only capable of 998-2014ms at the maximum limits (according to the same APM and my own exhaustive testing), but the log shows values well outside of that range. The PPM encoder was intermittently getting values for channel 1 that weren't coming from my transmitter. Trying to recover by switching into Manual mode made the problem worse because it just mixed more error in.

There are many sources of R/C problems, but only a few which will insert bad data on only one channel: the transmitter hardware, the receiver PPM decoder, or the PPM encoder. The transmitter software has hard limits at 1000-2000ms on each channel, so even if one of the sticks was intermittently bad it would never send the values seen in the logs. Any receiver problem would likely affect all the channels, not just one. That leaves the APM and its PPM encoder.

I believe the problem is the PPM encoder, but I can't completely discount ArduPlane, either, especially considering its odd behavior after the crash; alternating between Manual and Circle modes (I don't even have a switch setting for Circle) then going to full throttle the moment I find it and pick it up.

I'm flying:

• Bixler 2
• ArduPlane 2.74b
• ArduPPM v2.3.16
• er9x r798 NHT

The tlog for the final flight starts at 96%, and the dataflash log starts at about line 5815. I had to compress them to fit the attachment size limit, sorry.

Thanks!

2013-08-03 14-55-18.7z