I have seen the same behavior, which resulted in a nasty crash over the weekend:
Was flying FPV and at some point my RX went in to fail safe (action: RTL). It then came back to ALT-HOLD upon which I decided to engage RTL, just to make sure.
The Quad pointed towards home and went zipping towards home. WP_SPEED was set to 10m/s (36km/h, or 22 mph) (I just upped it from 5m/s/) . I noticed it descending rapidly and to my surprise the speed went way over the 10m/s. As a matter of fact it went well over 50km/h and it had descended so fast (and was not out of sight) that it crashed in to my neighbors roof at 70km/h (video went blank), skidded of and slammed in to the side of another neighbors house.
Result, not bad: broken props (carbon), broken arm and broken motor mount (I replaced the props with cheap plastic ones and she now flies much better using less power(!). However at 50% instead of 60%.
Cause: I suspect that the software prioritizes speed over altitude. When it increases speed, it simply does not have enough power left to maintain altitude. It does not explain though why the speed went well over the set speed of 10m/s. I never had the issue when the speed was set to 5m/s. The OSD and telemetry were showing a decrease in altitude (in other words: the APM was aware that it was descending).
Am I correct in my assumption that the software favors WP_speed over ALT_HOLD?
how did u tuned alt hold parameters...iwe started to notice same problem on my y6...
thank you...when you were talking about your problems i was not aware i have the same bcs i didnt fly for a while....also tought problems are more disaster...flying forward descend and flying backward ascend yes?mine is like that...
2 things to check if you don't mind
just a quick question have you changed the 'RAW'in logs download to 'INV',thank you,Marty
sorry IMU is that the new raw,Marty.
Getting to know this release :-)
I've been tuning my TBS like quad the past days.
The alt hold and loiter were already working (taken into account that I use an APM1 with Mediatek on this frame).
The only issue I had was that, when in RTL, the quad started off like a Porsche towards the home.
I searched a little in this forums and turned my quad into a Beetle, now it is going home more slowly.
Also tried a guided flight with the Droidplanner om my tablet and must say it was nice to see that it was working quite well.
This was my test frame, so now moving to the real TBS with the APM2.5 and Ublox.
My initial idea was, if I can get it tuned on an APM1 it must certainly work with an APM2.5 :-)
My testing was done using trunk. I also have my baro in a container, and very little vibration.
If the weather holds I'll do some test runs and post my logs later. I'll try to explore the variables, like inav-tc and the amount of ventilation my container has.
Yes, that's correct what was RAW is now called IMU in the latest versions.
Thank you very much Dave.
I just had a crash in this version. The Multirotor started behaving realy strange. Can somone help me understand what happened. First the GPS behaved strange taking forever to lock. But I don't use it so I did not wait for more then 6stelites then started flying.
Then I was flying around in Acro in average wind gusts. A little unstable but ok for just flying around.
Then suddenly it would not obey me I have flown this multirotor many hours total,with no such problems. And I Got some random controol back, when trying to make it come back to me it responded random. Luckily it ended up in a bush with only some broken props. and not a broken nex5r camera. I don't know what to look for in the logs, because those logs to me are some alien language :) I supply mission planner logs and flash logs.
the flying starts at ~76% off the tlog file don't kow if its the same with the flash log because I dont understand how to read those files. was leting it sitt outside for a while before flyig to let the gps get new locks since it was not even connected to the multi rotor since friday.
//Edit spelling and stuff