An interesting day at the field today...
Took off after preflgiht checks, and eveything was looking ok. Hovering at around 40%, looked slightly twitchy - probably because the PID's were at default, and I was about to Autotune them into oblivion, so no biggie. Gave it a bit of gas to test motor aligment, and woooahhh - massive lurch to motor 2. Caught it, but now motor 2 was permanently dragging the quad to the left left and backwards. Got it on the ground eventually thank god.
Checked it over, and...nothing. Could find nothing wrong. No play in motor 2, no vibes or grinding - it was running smooth. MP was saying "Autotune activated"...which is strange, because it wasn't configured. Ok, so an APM freakout? Should be fine after I reboot it. And...
It was like someone had gone bezerk with the PID's. Landed immediately. Back to MP...no problems. Everything looked ok!
Tune down the PID's from 0.09 to 0.07, and rebooted. That helped...but it was still oscillating. So I basically had to call it - I wasn't going to get a good flight today.
On the bench, I found it!
When removing motor 2's prop...I noticed a bit of play. I wiggled it...nothing. I *lifted* it...hey presto! Got a full 2mm of lift! Then it clicked...these were pancake motors (Tmotor 3508-29's)...a nice wide diameter, but a very low stack height. Which means any change in the magnet-coil overlap would be immediately noticable. And this is where I think the problems were for me - when you gas it, the thrust tried to pull the bell from the motor...and if it succeeds, the power drops off dramatically! The play had come in because the two tiny allen bolts clamping the bell to the shaft had come loose, and were bouncing up and down in the 'flat' of the shaft...good job the flat was there - it would have completely flown off otherwise! 2-3 turns with an allen key, and bingo - tight as a drum again, but I'd lost my flight window to the weather (5-7mm hailstones!!!)
Less of an issue for more vertical motors, but definitely a problem for pancakes.
So there you go. Top Tip for today - if you run pancakes make sure your "motor play" check includes the Z axis!
PS - yeah, I know my gimbal ain't right...it's a martinez on defaults and need power, P and I. Will be fixed next time.
Comments
@Jason: I started buying short shafts from RCTimer that allowed the collets to fit flush against my bells, but still fit all the way up the collet. Haven't had sliding/bending shafts since.
I should clarify I'm not using pancake motors; just regular ones.
@ Randy: do you mean all stabilisation of the gimbal via APM? Martinez or Alexmos not required?
Interesting. I've lost at least two copters (2212 850kv) to this, but not since I switched to the newer style prop adapters that bolt to the bell housing. The old motors had long shafts that you attached a prop collet to. If you flipped in a crash you could push the shaft into the motor. If you didn't notice you could fly around and the motor would separate in the air with hard throttle and your quad would flip. I got in the habit of pulling on my motors after that.
Interesting failure. thanks!
I thought your gimbal looked like it was doing a pretty good job. I hope to put up some instructions on how to interface a brushless gimbal with the APM/Pixhawk over the next couple of weeks.
Thanks!
I recently had a bell slide ~10mm up a shaft in a crash (cut throttle, landed upside down.)
Always check + Lock Tight those grub screws!
Interesting. Did those grub screws not have any loctite?
I'm currently using iPower MultiMate motors, which is a pretty good motor, but was shocked to find no loctite on those screws. I almost lost one. I also found that if it wasn't for the flat, things would have ended differently.