Naze32 Cleanflight Crashing Three Minutes Into Flight

luketheduke

New member
Hi All,
Thanks first of all for all you do to get us newbs up in the air! I am building my second quad from scratch this time. I used the mini racing quad power pack for electronics with a Naze32 Acro and a homebrew H frame. I built a tether that supplies 12.6V/30A to an XT60 plugged into the quad. Besides a little issue with two of the escs(That's another thread...), it has been all good. For testing I have it hooked up to two ropes so it can only go up and down. I've told myself that I won't take it off till it will fly for 10 minutes with no gimmicks. I recorded this video on one of my flights and this repeatedly happens: https://www.youtube.com/watch?v=xP-ijypL81I The quad will take off normally in Angle mode then about three minutes into flight, the quad will rapidly start to wobble and lose altitude. I have tried to increase the throttle but that doesn't seem to do anything. Any ideas? I'm flying the latest version of Cleanflight.
Thanks,
luketheduke
 

newpilot01

New member
hi Luketheduke

I'd try flying the quad off the ropes.

The rope idea is cool cause it can keep the quad safe from unexpected flips/flyaways HOWEVER maybe small oscillations your quad experiences due to turbulence etc is being picked up by the board and its trying to correct the wobbling however the ropes are making the craft behave in a manner the board is not programmed to handle? This ultimately results in a positive feedback loop which eventually brings your quad down. I'm not familiar with the Naze32 however after seeing the motion of the quad on the ropes it looks like small oscillations are becoming big ones rather quickly
 

Bricks

Master member
Landing straight down like that you are encountering your own prop wash which contributes to the oscillations. Take it off the tether and see if it still oscillates if it does then tuning the PID`s are necessary.
 

Invid

Member
Looks like a combination of your P gains being a bit too high and the quad getting hung up descending on the ropes. You can't tune it properly tethered like that.

If you are able to fly somewhat, I'd try dropping your Roll P by 10% and attempt an untethered hover outdoors. Pitch P might be high also, but I can't tell because the tethers are masking motion.
 

Snarls

Gravity Tester
Mentor
I wonder if your ESCs are overheating causing power to drop to your motors and the rest of the motors freaking out to keep flying. Especially if it keeps happening at around the same time into your flight. You're running your motors at 12.6v constant, but in real flight it will start at 12.6 and decrease from there. So the electronics may be overheating from extended run time at 12.6v. If you want to keep it tethered to power you may want to run it around 11.7v. Also from the video it looks like your ESCs are hidden from prop wash. If you are hovering in place for a long time without the air from the props running over the ESCs, they will have a higher temperature as opposed to if you are flying around and getting at least air flow from lateral movement over the ESCs. Anyway, this is all just a guess. Feel your ESCs after the crash happens and see if they are too hot to hold for more than 2-3 secs.

If all else fails just go ahead and fly without the ropes. The naze is a very solid board and if it flies well for 2 minutes it should fly well for the rest of your battery and session. You look like you have it under control so go ahead and let it fly free! :)

P.S. If you are wanting to tune, tune acro before you tune angle.
 

luketheduke

New member
Just curious: I don't exactly like A)flying in acro and B) flying off the ropes... Is there a FT group in the Detroit, MI area that could help me with the tuning?
 

crlock

Senior Member
hi, did you calibrated your radio and receiber with cleanflight? i found out that my resolution was too poor and thus my copter was getting a really bad input which translated into bad oscilations in the tail (i fly a tricopter)
cheers.
 

Craftydan

Hostage Taker of Quads
Staff member
Moderator
Mentor
a) nobody does at first. keeping the craft level is one more thing to mentally juggle, while you're balancing a marble on top of a ball . . . and if it's poorly tuned, a ball that keeps randomly moving on it's own :p Eventually, you will get comfortable with it, and then you'll find the auto-level is actually countering your command -- fighting you -- and then you'll want it off . . . until then, you're not going to like it . . . but you can only tune the gyros in acro.

b) The ropes are great for debugging, but if there is ANY tension on the rope, the craft is not truly free to fly itself -- you don't really know what's going on as it fights against it's restraints. the debugging efforts using ropes were always completely free to spin in one axis, and that axis alone was debugged. this is great for making the software work right on functional hardware, but not good for testing hardware to verify function.

As for groups in Detroit . . . not that I know of, but could be . . .

As for tuning yourself . . . from your comments, keep in mind the controller has layers of control. At the heart is the PID (or PI) loops for the gyros only. It looks at where it wants to be in angular rate and where it is, and adjusts each motor to get where it wants to be. your control commands enter by changing the "where it wants to be". Turn on a self-level loop, and it creates another loop between your control commands and gyro loops -- another P(ID) controller that re-interperets your commands, mixes the error with the accelerometers and adjusts the "where it wants to be" for the gyro loop.

the take-away of all of this: The gyro loop is never off. It has it's own PIDs, and if it is wrong, no amount of tuning in the other loops will help . . . they generally only make it worse.

I hate to say it, but I agree with others . . . take her off the ropes, and gently fly her in acro. Don't go high -- just enough to get out of the ground effect -- and be quick on chopping the throttle off. If you've got intermittent behavior in the ESCs, settle that first -- if it happens on the bench, it WILL happen in the air. Once you've got comfortable control of her, slowly bump up P until it jitters, and back down, then slowly increase I until it can hold an attitude for a few seconds without drifting -- the goal is highest P w/o jitter, lowest I without drift. Just tuning P and I will help tremendiously. build up your skills more, then bump D up a touch and readjust P and I (until a change in D gives you no change in P and I).

*THEN* adjust the self level to taste.
 
Last edited:

luketheduke

New member
I flew the quad today in manual mode on the ropes and I got the same issue after 3 min. Please correct me if I'm wrong but this rules out a problem with the P loop in angle mode, correct?
 

luketheduke

New member
I flew the quad in angle without the ropes and it still occurred after three minutes. So the problem I'm having has nothing to do with the ropes... Any ideas folks?
 

Snarls

Gravity Tester
Mentor
Are you still flying off a tether? If so try flying off a battery. Perhaps the power supply can not sustain the power needed for more than 3 minutes.
 

luketheduke

New member
Are you still flying off a tether? If so try flying off a battery. Perhaps the power supply can not sustain the power needed for more than 3 minutes.

I checked for that already. I hooked up a multimeter to the output and the voltage was steady throughout the flight.
 

luketheduke

New member
I finally got gutsy enough to fly it in Acro/Manual without the ropes. I flew with the tether and just like clockwork, 3:22 into the flight, the throttle dropped way down, it wouldn't climb and the right two motors weren't responding as much as the left two. Could this be an issue with PIDs for roll and pitch? Today I will try flying with just P to see if that can fix it...
 

luketheduke

New member
I got the same thing in Horizon mode. I'm uploading the video to Youtube, it'll be a few hours. I tuned the PID in Manual a bit and now when I get to the 3 min mark, it doesn't wobble as much, just the throttle goes way down.
 

cranialrectosis

Faster than a speeding face plant!
Mentor
Most modern ESCs will detect low voltage or voltage sag and cut power to the motors. It looks exactly like what I am seeing in this video. Modern ESCs sense the voltage drop and they bring the copter down "gently".

When testing my new Alien I was using old 4S 30C lipos. They were voltage sagging in seconds as the copter drew more current than the old 30C lipos could provide. It looked EXACTLY like your video. After a minute or two the copter would just sink to the ground and no increase in throttle would make a difference. Testing the lipo after showed a solid 4v + per cell.

Under load the voltage was dropping close to 3v per cell but after the motors shut down the lipo showed 4v per cell. The ESCs were detecting low voltage under load and bringing the copter down.

I suspect your issue is one of available current. 20A is not much power. I think your ESCs are detecting voltage drop due to current restrictions and setting you down gently to save the lipo the ESCs think they are connected to.
 
Last edited: