Yesterday I had several successful tests of Loiter mode, so today I decided to continue testing in windier conditions.
After confirming that I had a GPS 3D fix, I switched to Loiter mode. It was windy and gusty, but the quad seemed to be doing well. After a while, though, it looked like the deviations from the loiter position were getting larger, so I switched back to Stabilize mode to bring it back to base.
That’s when the problem started. The quad started to move away from me. And very rapidly too. So much so, that I found I was holding the pitch stick full towards me. Yet it continued to drift away. For a bit it recovered, but then it flew away even faster. Eventually I lowered the throttle and deliberately crashed it into the hillside.
I had to trek up the hill to retrieve it. When I reached it some 15 minutes later, one of the props was still spinning. One other prop was stalled and the motor was hot. Luckily it survived mostly intact. Electronics, ESC, motors are all OK. After I replace the booms and a couple of props, it should be good to go again.
I’ve attached the telemetry log. I am hoping someone can help me figure out what went wrong.
 I lost orientation and the control inputs were inadvertently directing the quad away from me rather than towards me. I know I had the pitch stick pulled all the way back. The logs show that the quad didn’t yaw around to face me (till the end). So I don’t think that’s what happened.
 The wind overpowered the quad. However, it was able to counter the wind in Loiter mode. And the quad has decent power with 4x160 watt Avroto motors turning 11x4.7 props.
 Lost Radio signal. But then I was able to lower the throttle to crash-land the quad ...
 Since it was fighting the wind in Loiter mode, it accumulated a large I-term (windup) that required a large control input offset. (I remember this from the early days of Arducopter 2.0 code last year).
Details: APM 2.0, Sonar, Xbee, 2.4.1
Have you got the .log file as well, as I couldn't get your .tlog to play.
Thanks for offering to take a look. I've attached the log file below.
Not sure why you can't view the tlog file -- I checked again by downloading from the forum and I was able to view it. But anyway, I hope you can view the log file.
The graph of Roll/Pitch & Roll_In/Pitch_In is below. While I'm not an expert by any stretch of the imagination, it does look like your quad was commanded the wrong way, the pink line is your input and the green line is what the quad was doing pitch-wise.
For a long period pitch (pink) was held in and the quad responded by pitching (green). Just before 4100 the pitch command was released briefly and the quad went level briefly, then pitch was added again. At 4300 there was a direction change (and what looks like a 360 deg yaw - not shown on this graph) and then just before 4600 the crash.
I would say disorientation (partly from the logs and partly because I've had it myself too and also crashed then). It seems the quad was going away and pitch was added to try bring it back but the quad was probably turned around and thus going backwards and so just went further.
Were you flying in simple mode by any chance?
Thanks. It does appear that I lost orientation. I'll try and correlate these logs with the tlog. There it seemed like it yawed only shortly before the crash, but then I can't explain the orientation issue.
It was definitely not in Simple mode. I only had Stabilize, Alt Hold, and Loiter configured.
I just saw the new thread, "Massive Drift Issue and accumulated error. Is it a PID issue?". Jason Short responded by saying there is a known issue and a fix is in the works.
My guess is that I encountered the same issue as well (Point  in my original post).
Sounds like leans problems.
I'm sorry, i advise all users to stay at the gound and wait the exit of 2.4.2/2.5.0 for this reason.
Thanks for confirming, Marco. I only wish I had read the entire "2.4.1" released thread before I took it out for a spin on Monday.
Sorry but this is NOT the 'leans', I have analysed many of my logs showing the leans very clearly and this log shows nothing of the sort!
The log shows pitch stick input and the quad responding in that direction. Please don't confuse this with the 'leans'.
I had a similar problem with my quad. I never went out of stabilize, though. It was night and I hovered in the front yard for a while, then went straight up. I pushed forward and everything seemed fine until I tried to pull back. It never stopped going forward. I am watching my bird fly away and I am in a residential area (no hill to crash into). At one point it was 1/4 mile away at 750 ft altitude. I got it turned toward me thanks to all the lights I have that indicate which way it was going and got it back down my street. It was moving fast and I had to hit a tree to stop. I thought this thing was going to kill somebody! Thankfully it crashed and fell onto the asphalt busting a few things up. Scared the crap out of me!
I was using 2.3.xx, not sure. I have lost trust in this. Too much experimentation for where I am in trying to fly. I like the idea and I am sure it will eventually be solid, but I had to go buy a commercial controller for now. Trust in the control electronics is fundamental for me.