Was doing a long steady hover for current calibration and just about to land. At about 1m the copter flipped and crashed. Not too much damage. Mechanically I cant see anything that may have caused it and I cant see anything in the logs.
Obviously the crash was the last second or two. Any ideas?
This is a tough one from my perspective. It seems to be behaving very nicely for what I presume is among the first flights of a copter. You could probably do with a compass re-calibration (or MagFit, once a bit more flight time can be logged). Otherwise vibration is low, and motors don’t indicate frame twist/misalignment to any significant degree.
The log just ends abruptly as if power was suddenly lost. Check all wiring carefully.
Also, check your battery voltage monitor calibration and make sure it’s not misreporting battery status. It would be helpful to know the actual, measured post-crash battery voltage. However, this isn’t even much of a smoking gun, as the values logged appear to coincide with expected reality.
There are a couple of “IMU0 fast fifo reset” messages present, but they aren’t occurring at a high rate, and I don’t see any crazy logging or filtering parameters. It looks like you have logging enabled to analyze and configure notch filtering, which is a good next step (after crash analysis). Some discussion of that message is in this thread.
Will be curious to see analysis by some of the other folks who typically chime in on these topics.
1 Like
Agree with @Yuri_Rage regarding a power loss problem. When the log ends there is no Pitch or Roll to speak of so the power loss came 1st.
Unrelated but you should have addressed the Output oscillation present before now. It looks like you made some rather random changes to The Rate Pitch PID’s and these parameters are still at default:
PSC_ACCZ_I,1 (change to 2X mot_thst_hover)
PSC_ACCZ_P,0.5 (change to mot_thst_hover)
These are at default too:
MOT_SPIN_ARM,0.1
MOT_SPIN_MIN,0.15
And you can set these as a start for the Notch Filter:
1 Like
This is a new setup (on a 10 year old frame) and I was just trying to get things right one by one. This flight (and the previous one) were with a 4s4p li-ion pack and it was obvious from the previous flight that the current calibration was way off. I had an interdependent inline Mah logger onboard to confirm my suspicion that the calibration was wrong. The previous flight was hovering for 52mins!
As you say, the logging appears to just stop just before the malfunction so I guess there was some sort of power issue, all though when I approached the quad it was certainly powered with all FC lights blinking, but it may have been a quick short or something.
I did enable logging for notch filter because the first flights had terribly bad vibration but I later isolated the FC and the vibration reduced well within acceptable levels. I forgot to turn off notch logging. Anyway Im not sure when filters need to be used. I think in my setup they are not required but please advise?
I also had some PID tuning assigned to a RC channel which I also forgot about but Im sure neither of these omissions were the cause of the crash. Its defiantly not a good tune at the moment as a gust can cause some roll oscillation.
I think (hope) the problem may be due the the inline current monitor stuffed into a confined space with the other electronics.
@dkemxr Where do you see an output oscillation? What is that printout for notch? I dont think I saw it presented quite like that. Difficult subject I need to spend more time on.
Always. A better/tighter tune can generally be achieved.
Here. Its not terrible but it can be improved.
Lower the Rate PID’s for Pitch and Roll (Try 0.1 for P&I) and you have the D-term for Pitch way too high I think at 0.01. Set that back to default. And set the PSC parameters I mentioned.
1 Like
was that table you posted an automatic suggestion from some software or you filled it in manually?
It’s from the Filter Review Tool Filter Review Tool
You can configure the filter and test it using “Estimated Post Filter”. But it needs to be real world tested also and it might need adjustment after dealing with the output oscillation. But I will guess it will be fine as is.
Don’t miss the loop rate increase.
1 Like
@dkemxr .
Tried all your settings and did an autotune. Unfortunately had another mystery crash. Logs just stop as before. Afterwards all the electrics were still connected and FC lights flashing in normal way. Very frustrating. Quite a lot of damage this time. Im suspecting a bad FC - Matek H743 v1 which sometimes fails to boot but apparently its a known problem. Both crashes happened after a long time airborne.
If your interested in the logs and perhaps have suggestions -
EDIT - as it lost power crashed- can I retrieve the new PID values it found. Also there was some messages ‘bad gyro health’ (could loop rate 800 be too high?).
Haven’t reviewed the log, but I had a v1 slim that acted up a bit, and I replaced it with another H743 autopilot after a while and never looked back.
I don’t think there’s much value in increasing the 400Hz loop rate unless you have a very specific need.
Since new my H743 v1 wing frequently wouldn’t boot. Otherwise it seemed ok but I replaced it with a H743 V3 which has never given any problems. Anyway I wanted to get my old big quad flying so got this v1 from storage. Still often dosent boot but perhaps its a symptom of MCU freeze.
I will put the loop rate back. Was just following Dave’s suggestion.
Thinking out loud -
Wondering about the 4V5 rail. Its supposed to be good for 500mA. I have a RX and a GPS unit on there. The TDMX rx is supposed to be less than 200ma and I cant find info for the GPS - Foxeer GPS M10Q-250. I believe the M10 units are supposed to be low current. Unlikely to be the problem I guess.
You might be onto something there. The Matek M10Q module only draws 13mA, but I do recall other folks overpopulating the 4v5 circuit and running into issues.
A poor connection/solder joint could be responsible for power issues, also.
I power my current-hog F9P modules on a direct line to a Mauch PDU because they quickly overcome most autopilot voltage regulator circuits. Certainly a much different scenario, but worth noting, I thought.
1 Like
No problem running a higher loop rate with an H743, I do on all of mine.
2 Likes
I’ve always avoided doing so, but that was based on assumption rather than founded knowledge. Learned something!
During this last autotune flight (log above) there were several messages on Yaapu about bad gyro health. I cant find anything in the log to show this (unless its an IMU message). I thought it may have been because of a high loop rate as it didnt happen before when set to 400.
I cant imagine its anything other than a bad FC as the battery plugs directly to board and logs should continue. I suspect if battery did disconnect there may have been a bit of data showing Batt V plummeting.
Anyway I have ordered another H743, v3 this time. The v1 uses the ICM20602 for the second IMU and it has a lot of bad press in other forums but I doubt it could be blamed for FC freeze. Perhaps it could?
Actually it’s the 1st IMU, 2nd being the old standby MPU6000.
Look at your parameter file INS_GYR_ID,3080482 (primary being used)
If you look at Andy’s 7" LR build he used the Lumenier version which is the same as the V1 and set EK3_IMU_MASK to 2 to bypass the ICM20602. The V1 I have I have not done this and it’s been performing OK. I think you should get a new FC but did you perform the IMU Temperature Calibration?
No I didn’t perform a temp calibration, but just read about it now. Every day I learn something new about ardupilot.
I already have a H743 wing V3 ordered. I already have one in my Goose plane and its been fine (so far).
How does 3080482 tell you its a ICM20602? I’m realizing how little I understand.
Mission Planners Ctrl-F Menu , “Decode HWID’s” button. Or connect and Setup>HW ID it will tell you the device types.
1 Like
Happened again after rebuild and different FC. Once again, nothing to see in the logs, they just stop.
Yesterday evening I did a battery test hover for 53 minutes and all was perfect. Went out today and after a couple of minutes it just dropped. On reaching it, battery still connected and FC blinking lights look normal.
New FC - Matek H743 wing V3
New Battery.
Most of the wiring is changed.
I guess it can only be one of the peripherals, GPS/Compass, sik433 radio TDMX rx
The ESC is 4 in 1 but the 5v is not linked to the FC as the 743 is directly connected to the battery.
I guess the only way forward is to fly without the GPS and sik radio. Anyway its so intermittent it will be difficult to confirm its not happening … until it does.
The log in case anyone can figure it out.
EDIT - Last time I was suspicious of too much hanging on the 4v5 rail. I meant to put RX power in the 5v servo rail but didnt.
How’s your battery monitoring?
Have you verified that the reported voltage is accurate? You seem to be demanding a lot of endurance, so it’s critical that your monitor is well configured.
What’s the measured pack voltage at the end of one of those tests?