Hexacopter wants to lean over before liftoff

ArduCopter Copter 3.5
1

Hi,

We have a 900 mm 7 kg ±hexacopter that wants to lean over to any side during rev up. This happens before liftoff. If we lifts off “violently” it leaves the ground and hovers.

Second problem is that is acts sluggish. Changed P (Roll and Pitch) from default 4.5 to 10, which made the hover a little bit more stable but not as good as we wants it. The lift off is still the same, it still leans and leans until it falls on its side.

Why dont it stabilize when it starts to lean?

It leans to one of the three sides which booms dont have a leg. It acts like it is stuck, but it shouldnt continue to lean tils it falls over. We start on plain cut grass.

The hexacopter is a Vulcan Mantis with three long legs, looks like this one:

2

We would need to see some .bin files from the FC of these takeoffs to be able to draw any conclusions.

3

You can’t be timid with the throttle. This is a very common beginner mistake, and the same thing happens with new flybarless heli pilots…

The second you arm the motors the PID controllers become active and they will try to get the airframe level. If you are timid with the throttle, the PID controllers get real stupid, real quick, and if you do get the aircraft to lift off, chances are very good the aircraft will tip over. BTDT.

I had been flying fly bar helis for a long time, and I thought I knew what I was doing, but the first time I tried to get my new FBL heil into the air it tipped over, and it kept tipping over until I found out that “chasing cyclic” to get the rotor disk “level” before lifting off is a big Bozo No-No.

The point is, if the aircraft is balanced in roll and pitch, if you did a proper 6 point calibration on the accelerometers, and if the radio is calibrated properly, the aircraft will take off straight up.

4

Boland:

Oldgazer, you are right, we are beginners with these big rigs (jump from 680 to 900). I have been flying around 20 multirotors since RCExplorer started with woodframe-tricopters. =) But none this big. We were too cautious. But when we forced it up more quickly it flies. But still, too soft or sluggish flying, we dont trust this one, yet. Havent tested anymore since I started this thread.

5

I can’t say it better than @oldgazer1
They will do it every time if you are too slow in getting airborne after arming.
I did pick up a variance in the outputs on the CW to CCW motors though.

40%20am
Screen Shot 2018-10-04 at 7.07.40 am.png2428×959 523 KB

Have you checked the motor alignments?
Props?
You know, all the usual stuff.
While typing this I thought to have a look at your RCin values.
Have you made sure there are no trims set in your TX?
Have you recently done an RC calibration?
Your mid points are set at 1514 and 1517?
04%20am
Screen Shot 2018-10-04 at 7.12.04 am.png2291×952 256 KB

And there seems to be a bit of noise there when flying.

The default PID’s won’t be correct, just enough to get it airborne.
An autotune is what is needed next.

6

We really appreciate these fast answers! Big thanks! =) Yes, we want to fly, but when this forum is alive, the development continues even faster and we fly more. :wink:

Okay… we have to fly when we start, no timid acting from here on.
After arming (props are turning in idle) and we put a little more throttle and leave it there, the rpm increases and after a while it falls on its side. Why is it increasing its rpm? (We know why it tips over, now).

The variance between CCW and CW. Is it possible it is because one motor is above the others? Or is it because of bad RC Calibration? I mean, is it a problem? Could it be a problem?

Is there a similiar program for the logs for Windows 10 as you use Mike? It looks a lot more detailed.

What do you mean with motor alignments? They are bolted onto the carbon-fiber motor-mounts that looks like these (no rubber between I realised now, should we use that?): http://www.rc-drones.com/images/vulcan-quad-2.jpg

The motors are Tiger Motor MN4014-9 400KV 4-8S 30A 150g 900W.
The propellers are Tiger Motor Carbon Fibre 14x4.8".
ESCs are HobbyWing XRotor Pro 40A 3-6S OPTO with DEO deactivated.

The props are balanced from factory, but I did it again with a balancer to get them better balanced. But I have a magneto-balancer on the shoplist.

Trims, no, I hope not… but what do we know… we flew in the pitch black dark. :wink: Have to check that one.

RC-calibration was done during calibration of the controller after firmware was loaded. Not since that. Are those midpoins a problem? Its only 3… Ah, you mean thats why there is a variance between CW and CCW?

Noise when flying… do you mean from RC-receiver through SBus to the Pixhawk 2.1 or is it interference from something? Vibration noise? Electrical noise?

Autotune… have never used that. Have tuned the PIDs myself if needed. But Autotune are going to be tested with another multicopter any day. Can it flip the Vulcan? It is big. It seems so unstable in hovering. Not vibrating or twitchy, it looks like it drifts slowly and dont know that it is leaning, like it is in manual flight mode. And if you correct it, it drift another way. Guess its because ground effect also. Have to fly more. :slight_smile:

7

The speeding up of the motors when still sitting on the ground is exactly what @oldgazer1 was describing.
There is always variance and drift in the sensors, and if it is trying to control itself based on sensor input but still sitting stationary on the ground it is going to try harder and harder to get to where it thinks it should be.
This is all going to build up until it does what you see.

Have you calibrated the ESC’s?
If you have the means to do a thrust test it might be worth checking the thrust difference between CW and CCW props. Just a thought. There is a variance there thats not going to help so you need to track it down.
Maybe even try it with a standard arm in place of the elevated arm and see if it changes.

I use APM Planner for graphs. I believe there is a windoze version.

The noise in the RCin could be causing some issues.

For an autotune on larger copters read the Wiki and change the aggressiveness and the filter settings as recommended.
I forgot to do that on a larger build quite some time ago and it flipped it into the ground.
You only forget once :slight_smile:

On the RCout graph I threw in the ATT values for yaw and you can see that there some yaw deviation.

8

Okay… todo-list is growing. =) Learning a lot.

Some questions…

  1. When sitting on the ground and that we dont lift off directly, the rpm increases becuase of drifting in the sensors, you say. I understand that now. And then you say that there is a variance between CCW and CW. Is that the answer? =) If I look at the first of the APM Planner-graphs, we are building up rpm during 65000-85000 (variance here becuase its on the ground and sensors are drifting, it tries to yaw, therefore the variance in CCW and CW?). Then we had a jumpy liftoff and are flying between 90000 and 107000 (no variance between CCW and CW, which is normal in flying with 3+3 motors lifting equally?). The ATT Yaw in the first graph shows that there is a variance between CCW and CW, on the ground. Then in the air, I yawed when flying, and I guess it corrected it selt too. Are you sure there is a variance when flying?

  2. The RC In noise. Where is this measured? Is it the internal signals in the controller to adjust stabilize? I mean, IF copter leans right (sensor senses this) -> THEN lean copter to the left to compensate/level up (RCIN channel 1 reacts)? Is it the senors that feels noise or is it the RCin that acts “stupid”? Throttle is crazy in the middle, thats where we had a nervous liftoff. Yaw acts normal. Roll and Pitch is acting “noisy” in the second graph. We didnt fly high, maybe 1,5 max. Ground effect surely made it vobbly? I have put on the list to move the sticks around for a while to get the potentiometers oxided surfaces cleaned a bit. It is a Futaba T14SG V5.0 that mostly havent been used because of mainly auto missions.

  3. How much throttle do you aim for when hovering a big rig? Is it still 40-60%? I mostly try to get it to hover as close to 50% as I can. Changing props until it do.

9

  1. You are way overthinking this. You can’t trust ANYTHING on the ground. This is an AIRCRAFT and it belongs IN THE AIR… So, when you arm the motors get into the air ASAP!!! Unless the aircraft is perfectly balanced in roll and pitch, and unless there is absolutely NO wind, and unless the aircraft is out of ground effect (about 2 times the aircraft diameter), and unless the motors have 0 internal friction and unless the motor windings, speed controller, and their interconnecting wires have 0 I squared R (heat) losses, and unless the props are exactly identical with identical performance and balance, and unless the flight controller has 0 internal I squared R losses, the motors will NEVER be in a perfect PWM balance.

  2. RC noise will come from one of two places: the radio itself, or some outside source of RF (radio frequency) energy (AKA RFI > Radio Frequency Interference). You said the radio has not been used much… DUH… I have also read of situations where firmware version 5.0 causes issues with that radio: http://www.modelbouwforum.nl/threads/futaba-t14sg-software-v5-0-probleem.221711/ (You may need to translate this page); https://www.rcgroups.com/forums/showthread.php?1754925-Futaba-14SG-(See-the-updates-and-warnings-in-the-first-page)/page358 Read post #5358… A Google search will find more… In your example RCIN (From the radio) would stay the same. RCOUT from the FMU changes…

  3. Ideally you want the aircraft to hover at or slightly below 50% throttle, and you NEVER want the aircraft to hover above 60% throttle.

10

Well, the leaning problem is history some posts ago. It leaves the ground as it should (flew today), much better. It was only the scepticism becasue it wanted to lean over and that it flew so unstable in windstill weather, for a big heavy multicopter. Thats why we didnt throw it up in the air. It is an expensive multicopter that have taken some time to build and we want to use it as soon as possible for what it is ment to work with. Dont have time for repairs. I have flown 20 multicopters that didnt have this big problem, we thought. I have also flown over 30 helicopters, they are a lot more stable than this rig in Stabilize-mode. THATS why we were so cautious. Totally unnecessarily… :wink:

The other problem I mentioned in the first post was is unstable flying characteristics. We have recalibrated ESCs, RC, Compass and Accelerometers much more exactly with better results. We flew it today and I took it up immediately, no problem with leaning. =) But, it is not stable when flying. It yaws a lot back and forth sometimes, have to compensate a lot. Roll and Pitch is pretty OK, but very slow on inputs somtimes. It flies as it should between that, but it is all over the place. Wind was only 0-6 m/s. Hovered 10 minutes straight. Autotune next. Autotune-setup is already done (7 0.1 21 26.4 10% …). Learned and tested that yesterday with another multicopter.

  1. Overthinking… thats how I learn. =) But, I didnt agree with that variance-thing. I dont see a problem there. There is no variance in air. Only on ground, totally normal (as you are clarifying).

  2. Noise… 2.4GHz is used in almost everything today, so that is a possibility. But it seems it is only on roll and tilt according there is noise in that graph. But today the yaw was insane. Well, have to look at this some more. The transmitter have worked splendid. We dont use those cheap receivers like FrSky (like them in the link) or Orange/Corona whatever their name is (I have had big problems with all of them) and dont even old Futaba-receivers. It also seems that there is different firmwares (version 5.0 from Hobbico for example) from different places that you have to watch out for. Futaba all in for me now (since 2009). No problem ever for me or the others on our field when using matching same-year pure Futaba-products. I do like that a lot.
    After we used the sticks around on the transmitter for a while to get the potentiometers fresh from oxide, the midpoints stopped flickering. Have had that “feature” on other radios during all the years I have been flying. Old technique with pots…

  3. Thanks! =)

ANYHOW!!! It flies! :wink: Not as good as I thougt a heavy big rig would do from default values, but it flies.

11

And the bin-file from todays flying (10 min straight).
https://1drv.ms/u/s!ArRUW2agVpZnjE59QIY99-It_Y21

I am not the best to find the right graps in Mission Planner… but it seems there is a lot happening when flying (RC OUT).

12

The log files shows it is not flying well.
Here is desired against actual

45%20am
Screen Shot 2018-10-08 at 8.46.45 am.png2461×956 313 KB

Definitely needs an autotune.
But I can’t quite pick where this is coming from, either the RCin noise

34%20am
Screen Shot 2018-10-08 at 8.49.34 am.png2472×958 329 KB

or some frame flexing.
Your RCin does not reflect your RCout as you can see above your throttle is only at 1300 but the the motors are at 1600 to 1700. Some calibration required here as well.

36%20am
Screen Shot 2018-10-08 at 8.52.36 am.png2482×957 397 KB

Also notice the separation between the motor outputs.
Some issue there as well I think. Your CCW motors are working quite a bit harder than your CW motors.
But this graph of the Actual/desired yaw against RC4in indicates either a very bad tune or some major flex.
The controller is fighting real yaw oscillations.

47%20am
Screen Shot 2018-10-08 at 8.55.47 am.png2482×956 361 KB

13

I’ve seen similar plots with a 1040mm quad. I was using 5010-350kv motors swinging 1755 props on 6S. This aircraft was REAL sloppy in roll, pitch, and yaw. Motor PWM values and temps were uneven. Ultimately there were three things resolved these issues.

  1. Shorten the arms. I used 16mm diameter arms that are approximately 17 1/2 inches long.
  2. Use shorter props.
  3. AtuoTune.

The arm length was causing some frame flexing. The long props were very efficient, but the low blade speed was actually having a negative impact on control authority.

So I shortened the aircraft to 662mm, dropped down to 1655 props, and ran Auto Tune. Now the aircraft is stable, agile, rock steady in Loiter and motor PWM values temperatures are much closer together.

14

We have gone through the logs and what you have written and showed above. Havent had the time or weather to do an autotune yet. Probably any day now. I think we have answers and “why’s” to all. Except for example why RCINs vs RCOUTs:

  • RCIN channel 7 (Autotune) = 1100 and RCOUT is pwm 1100 (to test, no noise, its rock solid)
  • RCIN 6 is not used, and that one is also steady at 1540 but RCOUT 6 for this one have noise at 1600-1700?!
  • RCIN 5 (Flightmode) was at 1810 at the test, steady. But RCOUT 5 have noise at 1600-1700 that one too?!
  • RCIN 3 (Throttle) moves up and down relative to my flying (around 1340-1420), but RCOUT 3 also has noise between 1600-1700?!
  • RCIN 1, 2 and 4… are everywhere (Im flying) but same same here, noise at 1600-1700.

I dont know where this is coming from?! Should RCIN = RCOUT?

Where is RCIN measured? On RCIN-“sbus/ppm”-port on the Pixhawk 2.1? It seems like that according to the steady PWM-values. But where is RCOUT measured? And why is it at 1600-1700? Same logs as above, I just looked at the bin in Mission Planner.

If you have any theories about the above, we would be grateful. I hope we fly soon… too hard winds or too dark to fly Autotune.

We have also changed to 13" Tiger CFs to raise the “throttle-stick-hovering-position” from 40% to … hopefully just under 50%.

15

Hi, RCIN 1-4 are the stick input from the tx.
RCOUT 1 is the pwm output from FC to motor 1 …RCOUT 4 ,output to motor 4

16

Hari, that one I did not think of, you are absolutely right. Big mistake by me. =)

Anyway… we have been flying a bunch of autotunes and it flies really great now.

Here is the .bin for a 1 minute flight earlier today (one of the flight):
https://1drv.ms/u/s!ArRUW2agVpZnjFG2MhcEzPGiPf-A

Can you see anything alarming?

I am learning and have som questions regarding the flight (the log)::

  1. Armed, some seconds later starting to throttle up and the motors jumps up to very high revs. The copter dont stand perfectly level on the grass. It seems that the highest motor revs a little and the lower the motor is, the higher rpm it gets. Hard PIDs makes it do this? It takes off good shortly after that.

  2. When looking at the “Power issues” 5,35V, its between 5,29-5,32 but have some spikes between 5,21 and 5,41. Is this a problem or pretty normal? Should a bunch of capacitors even these out?

  3. Compasses… how can they have X identical, but Y (and Z) differ? Is this even out after calibration and these are the raw-values?

  4. Vibrations seems good (below -3…+3) and VibeX, Y and Z are below 15 m/s. Pretty good. But are those vibrations only 1-2 Hz? Seems more like shaking to me? Or are these filtered to even them out? Are these “vibrations” from the controller using hard PIDs to keep the mulicopter stable? Or flexing arms? Legs whipping in the wind? No loose parts as far as we know.

  5. Two internal barometers…almost 1m difference. Raw-values. Is there an combined smoothed altitude used instead? Like CTUN - Alt?

  6. Have I read right when it says CURR (not CURR2) are around 23 amps in hover? 23000mA per hour? We use max 70% out of the 2x10000 parallelled LiPos which is 14000mAh. (14000/23000)/60=36 minutes and 0,9x36 to level out the current/voltage-change = 32 minutes. Nope! It flies 10 minutes (3,85V a while after per cell). What have I missed?

  7. CurrTot… 300 in 1 minute? What is that? Using Mauch-sensors. “Total current drawn from battery” it says. 300mA? 3000? None is right?

  8. The BIG question we have is… If we hovers stable with the stick at 49% (RCIN3 = pwm 1512, around 1485-1540), the RCOUT1-6 are around 75% (pwm 1725 around 1650-1800). 1094 is 0 and 1934 are 100%. Why does it hover att 75% motor-output? We skipped the 15" and went for 14" but with 14" the stick is at 40% at hover. We changes to 13" and now the stick is at 49% and it hovers very stable. But with motors at 75%. I dont get that. When looking at the specs for the Tiger Motors there are 50%, 65, 75 85 and 100% throttle. I thought 50% was motor-speed/power… or not? I dont get those values, anymore. When using 15x5" Tiger CF props on our Tiger-motors the specs says it would be perfect for a 7,5 kg hexa. This one is 7,2. But nope. It seems we have missed something.

Other flights:

  1. Got EKF_Error-02 a few seconds after ARM and a few seconds before starting to throttle up. Some seconds later before liftoff, we get EKF_Error-00 (EKF errors cleared?) and it flies as it should. Why did we get those errors?

  2. Flat hexacopter with very long legs on three arms (as in the photo in the first post). When trying to lift off it seems one legs get stucked to the ground. I guess this is normal. If that arm is higher than the others, that motor dont rev up as much as the others and it tries to push outwards when lifting, then it revs up the motor on the stucked leg and it cant push the leg outwards, and the copter starts to tilt. It works better on o more level surface. How do you get around this? I dont find these legs convinient for takeoffs at uneven surfaces. I have seen other manufacturers use legs pointing inwards.

  3. How do I activate Vibe Clip? They are 0.

I really like to write a lot, apparantly… Sry… =)

17

To answer a few of your questions:

  1. You can see here that there is an offset in the Desired and actual roll and pitch that causes a wind up when armed.

    39%20pm
    Screen Shot 2018-11-11 at 3.52.39 pm.png2433×950 427 KB

    This would also cause a lean on throttling up.
    The actual is quite tight with the desired, but there is a fair bit of ringing.
    The PIDS may be a bit too tight.

  2. The power fluctuations are quite good except for the few spikes.
    It might be worth checking all your connections.

  3. Many things affect compasses. There is plenty of reading to do about them in the Wiki
    Here is a good place to start

  4. There is quite a bit of calculation behind Vibe. I still tend to use raw IMU to see the real picture.

6-7. Seems like you need to calibrate your power module, as most do.
Search the forum, there are some very good posts about calibrating the power module.

  1. You seem to have the wrong end of the stick here.
    Forget RCin, it is the PWM out that you should be trying to get at 1500or close to it.
    The RCin can be adjusted, in fact it does it automatically on first flights.
    So don’t worry about the RCin on throttle being exactly middle, this will explain how to adjust it.
    At the moment you have the MotorOut at around 1700.
    You must consider 1800 as your max, which does not give you much headroom.
    So re-prop your copter to get the lift back up to hovering at 50% PWM output.
    If you read the link above you will understand the difference now between throttle in and out.
18

Hi,

Sry for the amount of questions but big thanks for the answers!

  1. This offset you mention is before liftoff. And as I understand from what has been said in this thread, we should not care and just fly and the problem is gone? We are trying 14" today and re-autotune. That may soften up the PIDs.

  2. Those spikes arent that high right? We do have four 220uF 63Vs on the 6S-PDB and one 3300uF 16V on the RC-receiever. Is that enough? We usually solder as much as we can, but all the Mauch-things have connectors. But they are all redundant with 2 or 4 cables. The Pixhawk have 4 plus and 4 grounds for the power (Power 1 and Power 2) from different BECs.

  3. Compasses… have to read some of that to understand more. Always something to learn.

  4. If you look at the raw IMU-values, arent they under recommended levels? Shouldnt that be okay? Or have I missed something?

6-7. PowerModule… not what we exactly use. We have one Mauch BEC (5,35V) for powering the Power1-port on the Pixhawk 2.1 and it also powers the Sensor Hub. The two Mauch sensors are connected to the Sensor Hub that keeps track on the two LiPos. There is also a Backup-BEC connected to Power2 (only power, no sensor-signals). They are all made for use with Pixhawk 2.1.
The sensors (that measures voltage and current) are calibrated at factory and the values that came with them are entered in Mission Planner. They measure exactly the same voltage as external LiPo-testers. How do you see that they need calibration? They are spot on as we can see.
We were tired of calibrating all the other power modules which never got good anyway (AttoPilot, 3DR, 3DRobotics, and some others) and went for Mauch. Soo easy. Soo perfect!
But I do have a question. We have two sensors with their own setup of values. But which value do we put in Mission Planner? I didnt get any respons from Mauch on that.

  1. This one made my day! =) I have totally got it all wrong reagarding throttle-stick and PWM-out. It is not throttle-stick that should be 40-60% or just under 50%. It is the PWM-out. Ofcourse! I have been mislead all these years and then got it as a rule. Doh! Thanks for the link. Wow! :slight_smile:

This forum is goldl!

19

One little weird thing… When arming, motors revs up to a low idle-rpm. But almost immediately the throttle-stick is touched, the motors revs up a lot!!! Just under hover. It doesnt take much to liftoff. It did this before too. I have no idea why it does this.

20

Look at the offsets I mentioned.
If the flight controller is not level (is your artificial horizon level?)
Or there are trim values in the Tx
The flight controller will try to get to what it thinks is level the moment you give it any throttle.

ESC’s calibrated?
Do all the motors wind up or just some?
Do you have a log file of this happening?