Help diagnosing errant behavior / crash for octocopter

I’m currently trying to diagnose a crash we had with our modified s1000+ craft, carrying a pixhawk running arducopter 3.6.5. We were having some takeoff issues all day due to wind and some inherent issues with the esc calibration (you cant do a traditional ESC calibration on the DJI ESCs included with the frame - though I hope MOT_SPIN_MIN / _ARM will help with that in future setups). On the last flight, upon entering auto mode (using a position modified mission we had used repeatedly, successfully throughout that day), the craft leaned to the side, moved horizontally, and never stopped, until finally descending / crashing.

I have colleagues remarking how it was probably a gps issue. While we were having some positioning issues earlier, I maintained 10+ satellites throughout all the flights, and a low HDOP. Any insight into what might have been the issue would be greatly appreciated

00000014.BIN (562.7 KB)

Your craft is woefully underpowered/overweight and barely flies. When you began the mission, the copter was pegged at 100% throttle and the battery was getting pulled below 18v and lost the ability to fly. You also have high vibrations especially at 100% throttle.

1 Like

I’m going to have to disagree. The payload we have for this platform is well within the max load For this configuration (though definitely out of CG - not something we can improve due to sensor requirements unfortunately) - we have been flying it successfully up to 3000’ agl within optimal power draw / voltages for approaching a year. We’re using custom li-ion packs which should be fine performing down to 18v (and were, earlier that day).

We were having some strong rolls (and corrections) on launch from takeoff that day, for reasons already described. My normal workflow with this craft is to get it airborne before beginning a mission (which all entirely consist of vertical profiling of the area, never a lateral motion). I’m just trying to figure out why this particular flight failed to attain level flight during the auto mode portion. In light of what you’ve brought up, do you think it may have been mechanical (and esc / motor failure on that side?)

What I stated was not an opinion. 100% throttle is 100% throttle.

You did not have a motor failure.

Based on your second post, I would look at the health of your batteries. They are sagging below 20v just hovering with a fresh charge. Was a battery not plugged in for this flight?

Is your mission supposed to accelerate to 30+ mph? I see that WPNAV_SPEED is set to only 110. Did the craft actually speed up to 30 mph?

Its a single battery. Those current spikes on the data log are definitely outside of the normal loads though - we are usually aiming for 60ish or less under hover / 1m/s ascend

No. Again, the mission entails only ascending and descending at around 1 m/s. It definitely was moving faster than that during the failure.

you are killing your battery.
120A draw 140A spikes. battery voltage drops down to 15,4V at one point.
you are using li-ion packs, what configuration? most cells are rated for 10 or 20A “max” rating, in reality maybe 5/10A. You battery resistance is to high -> voltage drops -> more power -> more amps -> more drop -> vicious circle.

your throttle out was saturated, no more power left. too heavy too less power.
try to have at least 1:2 mtow/thrust, better 1:3 to be save in winds.

out of CG payload means more load on specific motors -> that also is very bad for your setup.
cant you balance the cg for example with the battery?

i am flying 3 s1000 with pixhawk around 12-13kg mtow, and i did not find a 6s suitable li-ion battery that was up to a 60-70A current draw safely. also be aware that you have less power than with lipos!

Battery goes really low just before a GPS glitch, i’d say too low, around 15,5 volts.

so it’s a 1p battery and you’ve been drawing it this way for a year (flying daily???) and it’s liion… no way that battery is healthy unless it’s internally 6p…

just fyi, a copter has no issues dealing with wind if designed properly

if you want good help, you gotta give details such as when you say optimal power draw/voltages, give the numbers too

your craft purposely moved laterally, but why is not so clear to me… first thing i would do is correct the vibrations and resulting clips… all clipping was lateral in that log which falls in line with your gps/ekf errors and flight symptom… an additional and unrelated malfunction is your craft could not cope with the lateral maneuvers because it’s underpowered and this caused the crash

Its a 6s9p pack (titan power 31.5ah). While we’d like to be (and generally aim for) 50a, our average loads tend to be 40-70, depending on flight conditions (higher winds like friday or flying in high altitude situations like colorado being on the worse end). No, We’re not doing daily flights, especially not with the winter season on us. Yes, our li-ion batteries are stored and maintained properly, and still have relatively good IR readings compared to when they came in. I can get safe battery levels even at a 25 minute flight if needed (our flights were 17 minutes that day, and came in at above 18v when landing.) The 22ah lipo I brought faired a bit better obviously, and landed at around 21v.

Please trust me when I say I’ve done my research into our power needs. This is literally the first time I’ve seen it spike like that. Hopefully when I get it repaired, I can figure out how improve takeoff behavior (with mot_spin_min and mot_spin_arm), because it sounds like the issues cascaded based on what happened there.

Battery : 2708g Constant: 66A (form my experiance, in reality that is not safe to fly a heavy s1000)
what is the take off weight?
oakland alameda coliseum 09.03.2019 -> 48f/9° temp, thats very cold.
how long and to what temperatur did you preheat your battery?

From my experience, that was completely safe to fly the s1000 for our slow, vertical profiles, which is the only thing we do with this craft

~1.8kg sensor payload (normally 1.5, we has two mini anemometers strapped to the sides today), 2.7kg battery, and if the DJI spec sheet is correct, ~4.4kg frame weight. So, under 9kg?

It was in the low 50s that day on the ground during our work. Not optimal, but “very cold” is hyperbolic. They were preheated in the car prior to the operation. Battery Temps after flight were fine as well.

9 kg would be fairly light, and result in an amp draw about 40-45A @ 6s lipo so maybe max 50A @ 6s lion.
i would really measure the weight, things add up from my experiance.

i think we have to find out why the amp draw was more than dubble, and maxed out the trottle.

And I’ve definitely observed a similar average amp draw as you’ve described with this configuration in the past (at ~1000’ MSL at least)

I’ve been discussing and looking through the logs with a colleague as well, and (partially) agree on the power / underactuation issues that @Iketh raised. While within good operating weights (I measured all my additions @aamadeuss, and I strongly believe we wouldn’t need to attribute for more than another 100g or so from my initial estimates) the payload configuration does make for an inverted pendulum effect. It has been fine in our normal operating conditions so far, but with the amount of roll it had to compensate for on this launch, combined with the excessive amp draw, and room for improvement with the tuning (kentucky winter is highly unconducive to tuning outdoors, and this particular project only has been in progress since january), we ended up in a situation where it could not level itself.

I’m unsure if I can convince my supervisors to invest in a more powerful platform right now, but hopefully we can get this one running again. I’ll be dropping that 300g we added and have a few ideas to shave some more elsewhere.

Maybe i am wrong but i wouldn’t fly a 6S machine in wich battery sags to 15 volts.

you could use your battery to counterbalance your weight.
and i agree with the poster above. you lose a lot of thrust.

Again, this is the only flight where it dropped to that particular voltage. The worst case we had that day, prior to that was an incredibly brief sag to 2.7 / 2.8v per cell on takeoff. Do you just not believe me, or has this not been made clear at this point? I don’t know if there is a particular issue with that pack (I’ll have them all back in the lab next monday) but this is not a previously observed level for that pack before.

True, though to give more context on this - its a young 81000 ultrasonic anemometer we have mounted on top. 56 cm tall, and at least 25 cm off the deck of the s1000 (and if we could, we’d raise it higher). If I were to do it over, I’d definitely simplify the landing gear / battery tray area to allow for lower mounting. We currently leave the landing gear down in order to compensate some small amount of that axis.

if you post a log of a flight that it behaved nominally, it can help point to changes in the craft… preferably a log from a few months ago