Mavic Mini Fly RTH set, triggered by loss of signal, but lands quickly instead

[slup]

...It is possible that there is an untold logic in the firmware of the flight controller ...

I've felt this for a long time now ... perhaps it's a try to remedy "something" but it only moves the problem elsewhere --> descents.

First: all these decent have occurred in a tight cluster around the recent FW update.

Second: we saw a massive increase in blow away's before this

Third: it's not the first time I see the fronts decrease revs before the descent ... like can't rev up rear more, let's try lower fronts to get more negative pitch ... strategy (which leads to decreased lift & unintentional... or intentional descent)

There's more here than meets the eye ...

 

[PeP]

Thanks a lot for thoughts and analyses
- I need to digest it in the coming day(s) and see how various hypotheses can be substantiated / falsified, possibly through som targeted further test flights.
Will revert then.

 

[PeP]

Yeah ... seeing the failsafe action change to Hover also. This can happen for the Mini if the GPS signal becomes weak according to the manual ... but this doesn't seem to be the case here with 17-18 satellites.

View attachment 93387

But both from looking at the video clip & reviewing the graphs from the log this COULD again be that "uncommanded descent" we have seen previously ...

At RTH the AC turns, ascending & start to pick up heading speed ... then suddenly it pitches up to +15 degrees (typical for the uncommanding descents) & then the Z downward speed picks up. Just before the AC shows pitch/roll instability... it drifts slightly in the wind direction, which isn't strong, to the right & descending, once below the houses the side movement stops & it only descend.

I think that the DAT log can give us a bit more to look at --> Flight log retrieval and analysis guide (under point 3.)

View attachment 93388

slup, from your attached diagram, I am puzzled by the red "hspeed" graph. Between T=528 and 534 (6 sec.s) the hSpeed changes 40 times, oscillating between 0 and approx. 1 - 2 m/sec (20 peaks). Can this be as per design? - @new2mavic (post 20) suspects a faulty motor or rear propeller, and this hSpeed behaviour makes me think, he is right, that there is a motor problem.

@new2mavic , My rear propellers inspected, no dents, cracks or bends can be felt or observed. On right front propeller, there are two dents along one edge, not visible to the naked eye, but can be felt with a running a soft finger along (less than 0.1 mm, my guess). A drone repair tech. has earlier inspected and did not think this to be cause for concern.
If I end up changing any way, I am going to make before/after flights as identical as possible, and check flight logs.

 

[new2mavic]

My rear propellers inspected, no dents, cracks or bends can be felt or observed. On right front propeller, there are two dents along one edge, not visible to the naked eye, but can be felt with a running a soft finger along (less than 0.1 mm, my guess). A drone repair tech. has earlier inspected and did not think this to be cause for concern.
If I end up changing any way, I am going to make before/after flights as identical as possible, and check flight logs.

It just take a small change in the pitch of the prop to produce that effect. The change cannot be detected by simple visual inspection. You may take a look at this thread on how a prop change has fixed the problem : Strange messages during normal flight

 

[slup]

slup, from your attached diagram, I am puzzled by the red "hspeed" graph. Between T=528 and 534 (6 sec.s) the hSpeed changes 40 times, oscillating between 0 and approx. 1 - 2 m/sec (20 peaks). Can this be as per design?

No, you can't draw any conclusions from that... it's caused by the sampling & averaging of the data. The telemetry in the TXT & DAT logs is transmitted from the aircraft to the RC at 10 Hz & then displayed a bit different depending on what signal you chose to look at.

Heading speed from the TXT in 2 ways:



Velocity H from the DAT:

 

[sar104]

No, you can't draw any conclusions from that... it's caused by the sampling & averaging of the data. The telemetry in the TXT & DAT logs is transmitted from the aircraft to the RC at 10 Hz & then displayed a bit different depending on what signal you chose to look at.

Heading speed from the TXT in 2 ways:

View attachment 93574

Velocity H from the DAT:

View attachment 93575

Note that CALC_hSpeed is not averaged - it's just spurious data, and has never been a useful derived field. The correct horizontal velocity is CUSTOM_hSpeed, computed as

CUSTOM_hSpeed = √(OSD_xSpeed ² + OSD_ySpeed ²)​

 

[slup]

Note that CALC_hSpeed is not averaged...

No of course you're right, slipped in my non native language there ... in hindsight I should have used " not smoothened out" or similar perhaps.

 

[sar104]

No of course you're right, slipped in my non native language there ... in hindsight I should have used " not smoothened out" or similar perhaps.

No problem. But it's worse than that - even smoothing won't fix it. Those data are interspersed with spurious zeros and other incorrect values: