V3 Tricopter Flip of Death

[x0054]

Hi all,

Yesterday I was tuning my V3 Tri from David from RCExplorer. I am using the integrated NAZE board. The rest of my setups includes 30A ESCs (http://www.rctimer.com/product-1323.html) from RCTimer running SimonK and Cobra 2217 motors with 1045 props. I got it tuned out really well on Acro, and I was just flying around in Stabilized mode. No problems what so ever. Then I decided to come in for a landing, I was just hovering it about 3 feet off the ground, just coming in for a landing when all of the sudden, without any warning, the tricopter flipped to the left and drove it’s self hard into the ground. It beat up one of the props a bit, and broke a few zip ties, no big deal. However, after fixing things up, now the tricopter flies really uneasy, I can’t really pinpoint what it is, but it just feels wrong somehow.

I did have some issues with sync on those ESCs. Any one had good experience with them? I ended up re flashing them with SimonK, which solved the sync issues. In any case, the Sync Issues would only crop up at full throttle. In this case I was coming in for a landing.

My failsafe is setup to cut off power on signal loss, so I don't think it was a radio issue either, plus I was only about 10 feet from the tricopter at the time.

I am going to reflash those ESCs again, other then that, I am not sure what else to try.

Any ideas what it could be, and why it flipped?

– Bogdan

 

[cranialrectosis]

To me this reeks of vibration issues.

I would put the Naze on a PC and run each motor up individually and watch the accelerometers under load.

Obviously, be careful with the props.

To me this sounds like something shifted or snapped and is now out of balance so that when you are in a stabilized mode and increasing throttle it jiggles the accelerometers to the point where the flight controller thinks up is down and then makes it so.

I would look at grub screws and shaft tighness, look for loose magnets a bent bell or a cracked boom. Check the mounting for the Naze too for anything loose or the Naze touching anything hard that is connected to the frame.

 

[x0054]

That sounds like it could be just that. I am going to swap to an external NAZE board on some foam tape and see what happens. On David's tricopter design the NAZE board is directly integrated into the frame, so it's literally bolted to the arms. The motors and props are balanced, but I guess it could still cause problems due to wind gusts or any number of things.

- Bogdan

 

[Cereal_Killer]

Dont do that, the way cleanflight operates now days you want the gyro reading EVERYTHING. Use a SW gyro filter if you feel vibrations are an issue. In fact most flight controllers now days need to be hard mounted so they can feel EVERYTHING that's happening on board. What level of low-pass gyro filtering are you running? Balanced your motor / prop combos?


I'm currently fly 5 of the Naze32 tri's (well 3 tri's, one is in a quad cause I didnt have a spare normal Naze and the other is a Y4) and I assisted in the development of it. My first recommendation is to ensure you have your board defines all correct. Did you copy / paste David's CLI dump or try to set it up yourself? Are you using the Naze plate as the upper or lower body plate?

 

[Ocean]

Dont do that, the way cleanflight operates now days you want the gyro reading EVERYTHING.

Even with running Boris' Betaflight V2 with g-tune. I certainly wouldn't want my Naze to pick up oscillations created by props of a 400-500 size Multirotor.

 

[Cereal_Killer]

Even with running Boris' Betaflight V2 with g-tune. I certainly wouldn't want my Naze to pick up oscillations created by props of a 400-500 size Multirotor.

Yes, especially with betaflight running low looptimes!

 

[x0054]

Hi all,

Well now I a little bit confused about what I should try. I am using the Naze board as the top half of the frame, with the PCB with the stock voltage regulator on the bottom. I used David's config file and firmware fork from RCExplorer's site, but I obviously changed the PIDs a little, to accommodate different electronics and bigger motors and props.

I double checked that the board is orientated properly in the Cleanflight configurator, it all checks out. Before the crash, I put 2 batteries through the tricopter, while tuning the PIDs, so, if the board was set incorrectly, I would have noticed I first tuned it in Acro mode, and when I was happy with that I flew it around in Horizon and Stabilized for about 5 min with no problems at all. It was doing flips and rolls like a champ.

In Stabilized mode it did tend to drift to the right, but I assumed it was due to bad accelerometer calibration. I recalibrate it before tuning, but the surface I placed the tricopter on wasn't perfectly level. But it was a minor drift, I wouldn't bring it up if it wasn't for the crash. So any way, after about 7 minutes of flight on the second battery, with plenty of power to spare, I decided to come in for a landing. The tricopter was very come and collected, just hovering in stabilized mode when all of the sudden it just jetted away to the left and boon, straight into the ground up side down.

I didn't have a chance to fly it today to see what happens, the winds have been literally 50+MPH all day long. I am concerned for our solar panels flying away like a kite. If the weather comes down, I'll try flying it again tomorrow.

- Bogdan

 

[cranialrectosis]

What we don't know from Cereal_Killer is the size of his copters. I have 8 with one in production (Sparky II board). I have one old MW Pro (yes it still flies), one KK2, and 6 Naze32 boards. The largest copter I am running is an Anycopter with 1000kv motors and 10" props on the MW Pro. That MW board is mounted on two layers of hobby foam gently compressed with zip ties.

I use layers of foam for 8" or larger rotors. The minis just use little O rings on the standoffs. These are NOT necessary for the maiden, but after a few crashes, this mitigation keeps me in the air with less hassle. I learned this trick from Soma (inventor of the WarpQuad). I note that my latest mini-kit from Soma (Alien 6") includes O rings for vibration mitigation on the flight controller.

Low frequency vibrations can easily confuse the acceleromerters on a board. High frequency vibe is not nearly as bad.

The longer your rotors, the lower the frequency of vibe and the more powerful that vibe is. On a mini-quad, vibe is no problem. 5" and 6" rotors running at high speed require little or no mitigation to use autolevel. However, once you go on up to about a 10" prop, you are in a radically different world and you MUST mitigate or low freq vibe is gonna dump you in the snow.

Smaller copters are simpler to build, and maintain. This is one of the reasons why.

 

[jipp]

im planing on a 680 size Ap project for spring. so i will want to be sure to isolate teh FC from the frame good then.. as im assuming that will use 12" props. i know with minis you do not have to worry to much about prop balance.. but yeah larger id imagien you have too be very anal about it.

one thing larger copters have over smaller.. space.. which for you pros may not be a big deal but man.. it is nice to have lots of space to work in.

chris.

 

[PHugger]

I hate to say this, but I've experienced some similar strange behavior with Naze32 on Tricopters - Particularly after hard landings.
Does your 3D Model respond correctly in Cleanflight?

I have had to flip 180 degrees and then later have to flip it back (without changing the board orientation).
I would re-flash and reconfigure - re-calibrate the accelerometers too.
See if that helps.

Something is wacky with Cleanflight on a Naze32 (even David's custom version).




Best regards,
PCH

 

[Cereal_Killer]

The longer your rotors, the lower the frequency of vibe and the more powerful that vibe is. On a mini-quad, vibe is no problem. 5" and 6" rotors running at high speed require little or no mitigation to use autolevel. However, once you go on up to about a 10" prop, you are in a radically different world and you MUST mitigate or low freq vibe is gonna dump you in the snow.

I know I'm new here but I'm NOT new to RC, I've been flying aircraft for 18 years, I've been flying multirotors since we were attaching heli tail gyro's sideways to the booms and flying then without flight controllers, I'm also VERY active on RCGroups and cleangflight IRC...
I'm not trying to argue but please dont call me out cause you think I dont know what I'm saying since I have a low post count here, I guarantee you I do. I am also a senior student research adviser at Ohio University's robotics lab. Again I'm not looking to pick an argument, especially cause I do recognize I'm new here, all I'm saying it I really don't appreciate being called out like I'm an idiot cause you dont recognize me.



Now back to the issue at hand: If you're having to soft mount your Naze with larger props you are still having a physical problem. My most used RCExplorer build is a 650 with 1138 HQ multirotor carbon props (originally I built it with 12's but that was a little mucj for the 2820's). Again ALL of my builds are with the FC hard mounted (tho note, my largest, an 800, runs PIXHAWK so while it's hard mounted the IMU has internal dampening inside the pixhawk's case). With the current algorithms in cleanflight (especially in LuxFloat and the lowe the looptime the more it matters) you really have to get that ACC sensing EVERY LITTLE vibration that happens. If you're having vibration issues from your motor / prop you need to balance them as a unit.


jipp: Are you tuning ONLY your PID's or are you tuning your gyro filtering settings too? Do you have blackbox you can post some logs? I'd be happy to review them for you.

 

[jhitesma]

If it helps I recognize you Cereal and appreciate you stopping in here at FT I can't remember just where I remember you from (I'm guessing RCGroups but want to say you were around the MW forums too? Seem to remember it was somewhere more technical) and I'm very interested in hearing your input.

I still haven't got my naze32 tri to fly - though after the last round of debugging I'm pretty sure it's an ESC acting up...just haven't had time to try swapping bits around to confirm my suspicion. Though I admit the flood of "Just put a KK2 on it" comments left me wanting to take some time off from both the tri and this forum: http://forum.flitetest.com/showthread.php?22211-Time-to-try-a-tri

Soma is a good guy and makes some killer frames. But he's also more than glad to admit how much he still has to learn on the electronics side (though he's learned a lot more than most for sure!) It's worth noting that moto who works with Soma closely and makes the Tornado board and the motowii boards for the warp designed the motowii board as an integrated solid part of the frame. And having flown one I have to say it was one of the most enjoyable MW boards I've ever experienced. I was doubtful at first but after experiencing that I became less and less convinced that vibrations are as big of an issue as everyone makes them out to be. That isn't saying they can't be an issue...but it is saying a hard mounted board can be a good thing.

Oh and Cereal - didn't realize you were at OU. I did my time there from 93 to 99 in VisCom. Dying to go back for a Millers superchicken one of these times I'm back in Ohio Can't believe the Union burned down though We still hang out with Lou (the old owner) fairly frequently since he's just a few hours away in San Diego now.

 

[cranialrectosis]

Cereal_Killer, I do not intend to call you out. I am not conflicting with you on the issue only the solution.

My argument is that the larger your rotors, the more time you will have to spend mitigating (foam, O rings, and/or motor/rotor balancing) to fly well particularly using an autoleveler. I believe you are stating the same thing and recommending balancing hardware.

Running perfectly balanced hardware always works well. For those who don't balance after every crash, O rings and foam provide a bit of leeway but are never a total solution either.

Hard mounted boards are fine when everything is balanced. It is simpler to keep things balanced on a mini. The motowii is a great board for a WarpQuad but I wouldn't want to maintain it on an 800mm hex. I am sure it can be done and the results would be impressive, but I would rather fly than balance motors. This is partly why I fly minis anyway.

Welcome to FliteTest Cereal_Killer. I for one, care nothing about anyone's post count. We can use your expertise here particularly where it comes to tuning and troubleshooting.

 

[x0054]

About Hard Mounting FC

Cereal_Killer, sorry you got the impression that people were dismissing your input, I think it's pretty interesting take. It goes against conventional wisdom, but that doesn't make it wrong.

My personal experience, I have one quadcopter running 1045 props, and they are very beat up from many crashes. It's running Sparky 1 with Tau Labs on it, and it runs like a champ. I intend to get new props for it, and balance them properly, but as it is right now, with 5mm of foam under the FC it runs like a champ.

My other quadcopter is a custom 250mm X frame running Sparky 1 with latest Cleanflight. The frame has a clean cage, with battery, FC, and all of the FPV gear mounted on a rubber isolated frame. Here is a shot of it at work:



That's my absolute favorite quad to fly. It's snappy, super fast, and rock solid steady. But I will admit that I am far from the best pilot, and my experience is obviously much more limited. Am I missing out on not hard mounting my FC? I mean, what are the real benefits of hard mounting the FC? Is it more snappy, more responsive, more stable? I would love to learn more.

About the Tricopter Flip

As for the flip with my Tricopter, today I connected Cleanflight to it, and was very surprised to find out that the 3D model was on it's side. I think there might be something wrong with the ACC. After clicking the Calibrate ACC button, the model became flat again. It responds exactly as it should, and all of the settings are as they were before, basically David's settings with slightly different PIDs. Then I moved on to test the motors, to test them for vibration. With props on I cranked the motors up to about 1/3 power (just before takeoff) and looked at the ACC values. The lines were mostly nice and flat, with a few small jitters. Basically, something I would consider to be just fine.

However, when testing the rear motor / prop the model started to lift up a bit, I turned the power down, and the tail fall down, hitting the concrete floor. It wasn't a big fall, just 2-4 inches. It only lifted a little. However, right after the fall, this is what happen to the ACC:






The X Axis went all crazy, as you can see. It stayed like that for 40-60 seconds, I wanted to see how long it will go on. Eventually I went back to the Setup screen and clicked the calibrate the ACC button, and that fixed it. What is this? Is the ACC defective, or perhaps something is off in the settings? Any ideas would be helpful. I am pretty sure this odd behavior is responsible for the flip, but I have no idea what caused it or how to fix it. I think this time it was triggered by the shock of the rear arm hitting the concrete, but again, it was from only 2-4 inches, and it was a very minor shock. I recreated it by manually picking up and dropping the tail, and here is how it looks:



By recreated, I mean I recreated the fall, to show that it was a mostly minor impact. I was not able to recreate the ACC jitter again, even after picking up and dropping the tail, as well as the entire tricopter multiple times on the concrete, even from as much as 12 inches. So it looks to be one of those annoying intermittent faults. So..... thoughts?

 

[x0054]

On a side note, Jhitesma, next time you are planning on being in SD, drop me a line. Maybe I'll visit family in SD, and meet up with you for some flying. We can crash each other's tricoptres

 

[jipp]

jipp: Are you tuning ONLY your PID's or are you tuning your gyro filtering settings too? Do you have blackbox you can post some logs? I'd be happy to review them for you.

hi, iv not started the build im just collecting parts for the 680 size. but you bet ill take any advice/help i can get in the future with this project.

right now im just using CC3D with a zmr 250 which flys fine.. we will see how the icky versacopter likes cc3d as that is what im working on now even tho im ready to box it and start work on my 250FPV project. just not feeling any love for this versacopter at this point.. but once i get a new RX with strait pins we will see if i can get it air born.

also welcome to the forum. look forward to your input as well. it really is a small group of good guys. everyone gets along fine here.. its not like on rcgroups where everyone is out to cut someone throat so if you read something as negative here i would just think of it as a miss understanding unless someone comes strait out and says they are insulting.. seems that is why i like it here over RC Group that place is to cut throat for my noob self.

chris.

 

[PHugger]

x0054 - I'm glad to see you seem to have gotten sorted out.
I'm really suspicious of crashing the Naze on my Tricopter.
It always seems to start acting goofy afterwards.
Perhaps there is something about the that model (Tricopter) that makes it more susceptible.



Best regards,
PCH

 

[x0054]

Well, I think I figured out the immediate cause, but not the underlying reason for it. I wonder if it's software or hardware related. On one side, intermittent problems are usually hardware related, but on the other, pressing the ACC Calibration button fixed the issue, and that's a software procedure. I don't know enough about the software side of Cleanflight to be sure how exactly it does the calibration and what happens during the calibration procedure. Not being able to reliably reproduce the problem also is frustrating. I really don't what to strap my gimbal and gopro to this frame until I figure this out. Though, I guess I usually fly in Acro any way, and from what I understand, the ACC isn't really used for Acro. But I also like to take off and land in Stabilized mode.

- Bogdan

 

[x0054]

So this is a quick update. It looks like the problem might be with faulty hardware

The winds have come down, and I took the tri out to fly again today. With the integrated Naze, the Tricopter was jumping and shaking very sporadically. Even in Acro mode it would be hovering, and then all of the sudden it would jerk upwards or to the side with no provocation. So I decided to try another FC.

I switched to my spare AfroMini board and copy and pasted all of the settings. I had to change the orientation, of course. The AfroMini is mounted on 3 layers of 3M mounting tape. It's so light that I don't believe the 3M tape provides any meaningful vibration absorption. After switching to the AfroMini, the Tri flies very nice! It needs more tuning, for sure, and I am not very good with tuning so it's going to take some time. But it's reasonably locked in and very responsive, and does not exhibit any strange behavior like with the integrated Naze.

I am guessing that the MPU chip is defective, and it's feeding some strange readings to the integrated Naze. Now I am trying to figure out if I am going to replace the MPU chip with another one. It looks like a tough surface mount part to reflow though. Decisions, decisions.

- Bogdan

 

[jipp]

i know the naze is popular but iv read so many threads on bad naze boards.. i guess that is what happens when you have so many suppliers. and dealing with such little parts. I have no idea if the quality of each component has any effect or not but my guess would be yeah.. hence cheaper the part seems logic would say it has had less quality control tests on it, so cheaper to manufacture.

just glad you are able to get t flying.. i personally do like a clean build.. but sometimes like here maybe its not the best idea? not sure how you would replace it with out buying a whole new integrated naze plate which i would assume cost more than just a naze board?

chris.