3DR Y6 RTL toilet bowl crash

Only one broken prop because it spiraled very slowly so I let it run its course rather than take a chance with taking over manually.

I posted here viewtopic.php?f=25&t=8542 with a note about discovering a missing nut on the tail assembly that maybe caused vibration issues with the FC/GPS/Mag as they are all connected together with three screws, and with the tail nut off it allowed the base plate to be easily flexed.

Previous to this event RTL worked fine for multiple flights.

Attached is the .bin file from the SD card. I was unaware MP 3.1.5 did not download all the files using Mavlink.

Doesn’t look like vibrations or compass or any of the like - roll and pitch diverge from demand, so I would suspect a motor, ESC or prop failure.

It could definitely be tuned better - running autotune would give it more ability to recover from a failure.

[quote=“jschall”]Doesn’t look like vibrations or compass or any of the like - roll and pitch diverge from demand, so I would suspect a motor, ESC or prop failure.

It could definitely be tuned better - running autotune would give it more ability to recover from a failure.[/quote]

Thanks for your response. I’m not sure about an ESC or motor (did have one defective in the kit…3DR replaced), but the props are secure and I replaced 3 that mowed some grass for good measure.

I flew today several times after checking all the nuts on the assembly; very little wind. RTL did not work properly again, as if it was lost and couldn’t find the GPS coordinates after reaching altitude. It started slowly circling like before. Alt-h gradually gains altitude after flying around a bit with more throttle, but Loiter works fine (holds darn steady), and so does Land. I was in a large open area this time so took back control after it was obvious RTL wasn’t going to cooperate.

Being a newcomer to multicopters (old timer RC flier still using 25 year old Futaba tx LOL), I’m not yet proficient in programming the Taranis tx; an old dog learning new tricks I guess. To use AT, channel 7 or 8 is recommended IIRC. I copied the setup by DIYDrones video youtube.com/watch?v=KBISRi4NEO8 which didn’t include setting up for AT. I have a X6R rx using SBus.

Since channel 7 is assigned to SF as being the ‘flip out’ to RTL as an overwrite, how do I get AT to work while in Alt-h?

Sorry for the noob question. I probably should have used trial and error to learn how to program the Taranis first, but didn’t want to chance wrecking my newly built Y6, so played it safe. I’ve had about 20 flights without a serious crash by taking baby steps. The RTL issue has me bothered though.

So I decided to start from scratch and reload the default parameters in MP for the Y6B as recommended by 3DR, then calibrate the ESC’s. I then went through the other calibrations except for compassmot which wasn’t a problem when checked originally. Last night it flew ok except it needed trim adjustments for roll issues; ran out of daylight so used Land mode and it worked fine.

Tonight I took it to the field, was messing with the trim while in stab/alt-h, put it Loiter, seemed ok, then in alt-h again it gained altitude instead of holding atl and was drifting a bit further out to my liking so decided to land…put it in Land mode but it didn’t land. Instead it slowly drifted away while losing altitude (similar to RTL issue but a linear descent). It was too far away for me to chance bringing it back in stab so I just let it go hoping it would land without damage. Even still I think the Geofence should have engaged as it was pretty far out there.

I could see it was going to be close to some trees and sure enough it flew into a pine tree fortunately only about 4-6 feet off the ground so I immediately disarmed. Only one broken prop and it landed against a fence tail up. My luck is going to run out at some point.

Unfortunately I don’t have the .bin file because when I overwrote the Y6B parameters it sets the MP ‘Log bitmask’ blank and I neglected to enable it. The dates are off on the SD card too; another issue.

Attached are the tlog files. The take off isn’t until about 70% in as I was going over the settings to make sure everything looked ok. The telemetry apparently lost contact because it flew considerably further than the logs show, all in Land mode; I didn’t try to switch back to another mode.

Is it possible the GPS and/or compass is defective?

Don’t use trim on your radio the controller will keep feeding it in. All trims should be at zero.

MP has 'Trim save" mode. It used to have ‘Auto Trim’ but I think it was removed. Adjusting trim is in the tutorials. After adjusting trim while in stab mode, land the copter, turn on ‘Trim Save’ switch (plus whatever other steps is required) to save the trim adjustments, then move radio trim back to zero.

I don’t see what my Y6 did last night has anything to do with trim settings. I used trim the night before and it landed fine several times in ‘Land mode’.

At this point, unless I’m completely missing something, it’s looking more like a hardware failure of sorts, but that’s why I posted the tlogs. It may very well be operator error, but I want to get it sorted out. Just when I thought things were improving, last night happened.

First off, LOG_BITMASK in your tlog is 26622.000000: INAV (deprecated), COMPASS, RCOU, CURRENT, CMD, IMU, RCIN, NTUN, CTUN, PM, GPS, ATT_MED. You should have a bin log with all that in there somewhere.

You’re way good on GPS satellites (sometimes you pick up 14!!!) Inertial navigation is functioning perfectly, which means compass and vibration are fine, but you were drifting at ~3 m/s in LAND mode. LAND mode does re-positioning via the LOITER roll-pitch input mode. I see that you had trim in on your radio during LAND mode - this may be your issue. Your nav roll is consistent with the transmitter input, which to me is the smoking gun.

It does not look like you moved the trims back to center after saving trim. The save trim functionality changes AHRS_TRIM_X and AHRS_TRIM_Y, not RC1_TRIM/RC2_TRIM - the radio input calibration remains the same and therefore you must return your trims to normal.

(LAND and RTL doing re-positioning is controversial. I used to be neutral in this argument, but seeing it causing people’s crashes is seriously getting on my nerves - I will send this thread to the dev group to bring the discussion back up.)

Here’s LAND mode extracted from your log, plus a graph showing the roll being consistent with the bad trim.

I was very wrong about there being nothing wrong with your compass in your first log. Something was causing it very big offsets that caused the problem.

On your RTL crash, I’m 99% certain that COMPASS_ORIENT got screwed up right before you armed for the crash flight. Your Z magnetometer starts behaving like it is one of your horizontal magnetometers during flight that is obviously level, which is clearly wrong.