While hovering I noticed all motors shut down few times and started again after about 1sec. I power AP board, Reciver and Roll/Pitch servos separately from different ESC BEC (with common GND only). There shouldn't appear brown outs. So what's the reason? [APM CPU 2560, soft v2.5]. Could it have something to do with bad Yaw and Throttle PID settings?
Might be a bad connection, check telemetry logs to see if something interesting there.
Look at my logs for 3rc channel and manual boost. There are signal drops with no reason. Connections are reliable. I use Futaba 10C transmiter and 14FS receiver and they work perfectly. All raw data from 6 sensors work great.
The most interesting is I don't see any problem without propellers. Might it be a problem with high current? I have Lipo directly under the main board.
Can you graph your barometric pressure altitude along with rc3 servo out?
You really need to double check the wires from the radio. This will certainly cause the situation. I've personally seen motors act this way because of a broken RC cable. The cable will break right at the back of the crimp connector where all the stress is. The plastic will hold the wires together, but the vibrations will cause the drop outs.
low battery ESC failsafe cutoff? Has happened to me. Basically, in a plane the engine would start cutting out, making ti obvious from the ground that you're about to run out of battery. WIth a plane that glides, that's not a bad idea... with a helicopter, not so good. So the ESC has a voltage cutoff perhaps. if it is programmable you can also set whether it cuts engines at high/medium/low voltage to make it more or less sensitive.
So, possibly just need to charge your LiPo?
thanks for suggestion, I'll check wires again.
I soldered the cables from servo 1mm connectors and quite good quality wires. All signal cables were connected with epoxy glue additionally.
It's definitely not Lipo fault. It happend many times and by full charged battery. I set all ESC to low cut off mode to protect from sudden crash. . I can clearly feel in the flight the lipo is nearly empty and have some time to land (similary to my helicopters).
Are you using Xbees or 3DR radios? We've seen interference from Xbees cause throttle cutouts before (solved by moving the Xbee further away)
It is weird that the throttle in doesn't match with the rc3_out. It's possible that the short downward spike from the receiver's channel 3 was missed in the logging because it simply didn't catch it.
Do you have an xbee? If "yes" could you try disconnecting it to see if that resoves the problem? I'm not saying that's a permanent solution but if it does it narrows down the problem a lot.
Can you post your logs as well?
Also if you have an atmel programmer, could you try upgrading the ppm encoder software according to this wiki page? I fly using the "ArduPPM_v0.9.87_Arducopter-APMv1.4.hex" version found in the downloads area.
your motors used hall sensor?or other?
maybe the problem is your motor driver.