I have had exactly 0 successful flights with Pixhawk so far. This quad flew fine on APM other than the fact that my clone APM had a bad barometer, which is why I switched to pixhawk.
My quad is about 1.5kg AUW, SimonK 30A esc's with Afro NFET V2014-01-19 firmware, Tiger MT2216-9 1100kv motors, Arducopter 3.1.2.
First crash was on a throttle punch, motor 1 just stopped and the quad crashed. Troubleshot it to an ESC sync issue (which didn't exist on APM...). To fix it I added the ground wire to the ESC servo connector. Tried autotune again, and although it didn't crash arm 1 would dip during the autotune. So then I updated the ESC firmware to the latest. Also soldered the motors direct to the ESC. Seems to have resolved the issue, at least when the quad is strapped down.
Went to try today. Before attempting autotune I did a couple of mild throttle punches to see if I had any issues, and sure enough the quad flipped and crashed inverted.
Now, even with default PID's that don't fly good, doing a mild throttle punch shouldn't result in the quad flipping over and crashing correct? I'm scared to get to autotune altitude knowing the quad has this problem, so far my crashes have only resulted in minor damage due to the low altitude.
I've attached the log and would be eternally grateful if someone can assist me in getting this $700 paperweight flying properly. I'm especially confused why I didn't have these problems with APM but do now with the Pixhawk.
Replies
See the videos in the thread linked above.
Hi, thanks, and sorry that was a little silly a question I asked. I just thought of it in the context of the PWM signals, and never heard of anything about sync there...
As far as I can see in the SimonK source , the Afro ESCs use the ICP interupt to capture the PWM signal. That is, an interupt is fired each time the beginning and each time the end of a pulse comes on the RC input. The interrupt handler does not do an awful lot of work, but it is also not super minimal. The commutation (sync) is done from a main loop.
It is not impossible that very frequent PWM pulses could have some effect on commutation phase, sometimes.
Regards
Soren
Well I think I've narrowed this down to a Pixhawk only issue so I've posted a new message with video in the Pixhawk forum:
http://diydrones.com/forum/topics/pixhawk-causing-motor-sync-issues...
Did some troubleshooting and it does appear that I have a sync issue still. So I am going to try some changes on the ESC firmware side of things to see if I can resolve it. At least now I can recreate the issue with the quad strapped down instead of finding the problem terra firma style.