Hi everyone, here's a noob question! First, some background: I have just completed a build of a RCExplorer style TriCopter, and flashed ArduCopter firmware 3.1.3 on my APM 2.5 flight controller. I ran the Wizard and calibrated compass, gyro sensor, RC radio, etc. Did the ESC calibration, and all seemed fine. On first attempt out in field I concluded I needed to reverse elevator channel on my Turnigy 9XR radio, and most importantly, I need to go into the Mission Planner terminal window and set REV_CH7 to -1, as the back motor servo was turning in the wrong direction.
After these fixes, I attempted to lift off again, after verifying a successful GPS fix before arming and taking off. It was a success! Very stable even with default ArduCopter 3.1.3 parameters. I tested Loiter mode, and it seemed to move fine with Super Simple mode enabled. I enabled RTL, and it went straight for home position, waited five seconds, and landed fine, handling the rather strong wind this day. I took another short flight, flying higher this time and using Loiter mode, did a 360 panorama over the site, then RTL back.
However, on my third flight attempt, I flew at a low altitude away, then enabled RTL. It raised up to, I presume, 15 meters, and began heading back. It then put in a Rudder input on my TX, turning it towards me to get the camera to record in the direction it was heading, as it was heading RTL in a backwards orientation. Immediately after I put in the rudder input (`well, it actually did turn around first), it tumbled down out of sky and crashed on the ground. So, my question is: Are you supposed to stay away from the sticks during RTL? Or is there another factor that I should look for, in order to explain what happened? Maybe I accidentally moved the throttle input as I moved the rudder control? Any ideas?
All in all, I am pleasantly surprised about the simplicity in getting this capable controller up and running with relatively little effort. After reading and searching on forums I was prepared to go through more problems than this to get it up in the air. Now I just have to test, test, test to understand how the controller actually works and behaves. There are many parameters to possibly adjust, and fail-safe behavior to configure and test out. I read about sync issues with SimonK firmware on the ESCs and the ArduCopter controllers. I am thinking about trying an auto calibration procedure, but I am not sure if I it would be risky to do so? My motors are 750kv DT750 motors and I am using HobbyKings F-20A ESCs flashed with the latest 2014-03 SimonK firmware. Anyone here using similar equipment?
Replies
-Do log ACC, at least while testing - with only 3 motors to calculate - there will not be any reduced performance,.
-You will most likely discover that your vibrations are over the limit, check here:
http://copter.ardupilot.com/wiki/vibration-damping/
- Let the cables have some slack, the way they are mounted, can conduct vibrations as well.
-I am not familiar with the the case you have, bu usually - there's a piece of foam inside.
All conectors(openings) point up, in case of rapid descent, the pressure inside would drop. I am not sure if the descent or vibrations did that, but you can reduce this effect by having a plate on top of APM too, (sandwich it) - be sure the top plate do not touch cables/case and conduct vibrations.
Start with testing vibrations. , and verify that min throttle is able to keep it steady.
Hi Andre and Oliver,
This morning I did a test flight using the Auto mode with planned waypoints. Yes, I understand trying out things a step at a time is the best way. Even though it was short, I did on my very first attempts Stabilize, then AltHold (well, I setup Sports mode instead, which is a combination of AltHold and rate-controlled stabilize I understand? Sounded good to me), and then Loiter, and then RTL. They seemed fine, although to be as a beginner, it's hard to know exactly the borderline between "perfect", "acceptable", or "less than stellar" performance. It picks up the heading correct on RTL, and Loiter moves where I point the stick, so compass and GPS seems to work fine. After I rebuild it, I will run the Auto-calibration of PID values for sure.
As for logging, I could not get it to recognize "ACC" or "RCOUT". I guess you meant to enable IMU and MOTORS? That I did using the terminal and "enable IMU" and "enable MOTORS". So for this mornings Auto mission, I have data. I attached them as "logs.zip". As far as I can see, the X-Y vibrations are within the +/-3 acceptable limits? Remember that my booms are not of carbon fiber or aluminum, they are of wood. Wood has the advantage of dampening vibrations somewhat, instead of transferring them straight to the center frame as a harder material boom would. Still, of course, less is better.
This mission ended in a crash, and I suspect that something mechanical happened. Maybe a propeller snapped, or a motor mount dislodged? It seems it leaned to one side as it went down, and I found it with one arm buried a decimeter (a couple of inches) in the ground, the motor was found in the vicinity with the propeller blade broken in three pieces. The center was still securely bolted to the motor.
I had the DroidPlanner2 running on my tablet with telemetry link, and thanks to it pinpointing the final location it went down, it was easy and fast to find it. A real time saver. The pattern showed that it went down in a wide 180 degree circle as it came down.
Here's a video of the event: https://www.youtube.com/watch?v=UUkuyFDuihA
I realized I need to go over the build when doing the repairs and make sure all screws etc are locked with loc-tite or similar to prevent vibrations from loosening things up. So this last crash is a good opportunity to get over the initial excitement of it actually flying at all, and make the build more final.
Anyways, the attached logs can be reviewed for your analysis of the vibrations if nothing else?
logs3.zip
Yes, I used wrong terms then I asked you to log ACC/RCOUT, you got it right.
You can see a manouver start at line ~9.5k
your altitude is about 34m
as it turns, barometric altitude error grows, line 10217 logs 72m (more than twice at where it wants to be!)
in response to that , line 10241 logs a dropped throut (only 281) the same point in time logs the biggest difference in desired and actual roll (Drol/Rol) but it recovers fine as throttle increase and the false altitude drops.
Then , for no apparent reason, at line 10400 Mot1 output increases to correct Roll, by line 10410 Mot1 continue to raise , and the Roll increases way beyond DRol.
It seems to me like Mot1 Output is to blame, it went very high and overshot DRol by far, then dropped very fast, (10442) and started the tumble.
- next 100 lines, Mot1 stays at high throttle even when the copter is inverted.
I may be mistaken in my understanding of Y3 or the direction of rotation logged, but to me it seems like this could be some sort of bug - (it's not often I suggest that)
I hope a Arducopter developer to look into this and correct/confirm.
There's one more option, that this is some bizzare effect of bad PID's , I think your Rate D term is very small, but maybe it's common for this size/type of machine. - so do not care much for my opinion on PIDs , it's not about opinion :) - it's just a shot in the dark because it looks small to what I am used to.
I would try to do autotune over a soft area - and see if the result is very different from what you have now.
I too was looking for the vibrations but they're not logged. If I understand correctly that you have no vibration damping at all between the APM and the airframe other than velcro, then this either has been a problem or is likely to be one at some point.
Reviewing the video, I concur with Andre's speculation that maybe the throttle went below usable minimum. You can hear it throttling way down. In your initial description you used the term "it tumbled down" which had me thinking individual motor/prop/esc failure but in the video it's apparent that the vehicle is actually upright through the descent, and only flips over after impact.It sounds like it's trying to correct as it "falls" but with not enough throttle.
A general suggestion: When working with a newly built ship, try out one thing at a time until you have that thing working well and only then start combining things. So for example first fly only in stabilize. Next try altitude hold. Then loiter. Then RTL. Then maybe the fence. And so on.
-
1
-
2
of 2 Next