Tarot650 Quad - Serious jello/vibration problems

ArduCopter Copter 3.4
1

Let`s first start with some hardware:

  • Tarot 650 Sport – modifications: Tarot 680 PRO arms (TL68P05 - 262 mm and anti-vibration motor-mounts (Tarot TL68B33/TL68B34)
  • Tiger MN3110 – 470kv /w Bulltec (equals HK Blueseries/ZTW Spider) 30A Opto /w SimonK
  • Gemfan / Tiger CF Props 15 x 5.5
  • 5S 5000 mAh
  • Tarot T4-3D Gimbal + GoPro 3
  • total flying weight ~2,4 kg
  • Pixhawk 2.4.6 /w Arducopter 3.4.3 on anti-vibration mount, ublox Neo-M8N GPS mounted on a pole

DSC_2878.jpg1500×1000 87.4 KB

DSC_3349.jpg1500×1000 156 KB

DSC_3360.jpg2000×1500 653 KB

Since the maiden flight we are fighting against strong jello in videos and hence a lot of iterations were already done:

  • Static and dynamic prop balancing (> positive)
  • Change to T-Motor props (> positive)
  • Change to cheap 14“ props to change rpm (> no visible change, changed back)
  • Propeller tip adjustment (both tips running the same path) (> no visible change)
  • Autotune (> no visible change, settings too aggressive, changed back – shown log is with default 3.4.3 PIDs)
    The vibrations (according to pixhawk log) as well as jello (as seen in video) were noticeably reduced over time, but are still in an absolutely bad zone. Therefore we did some log analysis which show very good vibeX/Y and a very poor behaviour on the vibeZ. In steady loiter the vibeZ are „okay“ (15 … 45 – mean 30). As soon as the copter starts moving the vibrations are increasing to alarming rates. Even a lot of clipping (but only on 0 and 1 – 2 equals 0?) is happening here.

vibrations.jpg1920×451 87.8 KB

accs.jpg1920×451 75.2 KB

Auto analysis is GOOD (except of the IMU mismatch and of the motor balance – gimbal and gear are also driven by RCOUT and therefore affecting the average here)

We are still at the beginning here and we have absolutely no ideas how we can improve any further – buying a RTF DJI Mavic maybe…? Well: Any help to fight those z-vibrations are highly appreciated!

PS:
Just for documentation – this setup has NO sync issues using low KV motors and SimonK and works very well in autotune.

The landing in the log was hard and expensive – one rod of the gear failed :wink:

2

In my experience, until those factory “carbon fiber”,tubes are changed to something else it won’t take much of a hard landing to keep breaking them. It applies to the skid tubes as well.
I think it took us 3 or 4 times until I just designed a new assembly.

I have a 3D model for the bottom T that you can use some 16mm aluminum tubing for the upper, and I believe I cut up a cattle prod electrode or a fiberglass driveway marker for the skid. If you have access to a 3d printer or whatever.

On the gimbal, did you try using the Tarot software to do some motor adjustments? you may be able to fix some of it that way.
There’s also a bundle of wires going to the little controller/connection box on that gimbal. Could it be catching too much air? Or is it touching on something which might be transferring vibration into the gimbal housing?

Hope you get it sorted.

3

Hi Lance,

thank you very much for your thoughts!
Concerning the Tarot Software, we already have tried different settings for the motor torque values and the rotation speed values. We have tested higher and lower values for the motor torques compared to default, but there were only minor differences visible regarding the visible jello in the video. At the moment we are running with motor torque values of 30/40/40 (roll/tilt/pan) and rotation speed value of 100 for all 3 axis which delivered the best results.
Regarding the cables near the gimbal controller, there is only the main connector attached to the back of the controller. But it is a good idea to check if there might be some cause for additional vibration. We have also already moved the complete gimbal mount back and forth and have also changed the vibration dampers with no major effect.
We are really running out of ideas regarding this issue.

4

I was thinking it was like mine, where there was a small, separate power and control unit that hooks into the gimbal.

You also need to make sure your signal wires are terminated well, and voltage levels are good. I know they really go haywire with just the slightest signal anomaly.

You might set it all back to default, and I believe there’s a calibration routine in that software, but you need to make sure your unit sitting pretty level.

One more thing I did on mine was to sandwich the back gimbal hanger between 2 grommets (on each rail). Then with the the back hanger unable to move, install 1 in the front of the front hanger. If that makes sense.
I felt like there was possible movement since is just smooth plastic hanging on those rails.

I don’t feel like you can ever get rid of all of the vibration / jello. Especially when yawing. But I feel like mine is pretty smooth 95% of the time.

5

Lance, thank you for your tips! The gimbal was already recalibrated when updating from firmware 1.0 > 1.1. Although I think that not the gimbal itself is the problem since the Pixhawk also experiences dangerous vibration levels. Therefor the motor vibration itself should be optimized and the big question is how to get rid of those levels in Z.

In other words: Why are VibeX/Y very good and vibeZ extremely poor?

6

I’ve been fortunate in that I’ve not had many vibration issues. But I’m obsessive to a fault almost about thinking out my builds. So I don’t have a simple answer for you. I can tell you how I approach it though - other than the obvious things like balancing props, checking motors and cabling to them are secured etc.

Z is vertical relative to the FC I believe.

Its been my experience on our Tarot that even though there are a ton of screws, the folding design lends to them loosening. I don’t ever fold it up. Put some BLUE (low strength) thread lock on those arms, and on the folding gear.

Now that I’ve looked close at yours, I’d look at the cabling around your mount, and especially look to see if that lid is contacting ANY of those cables. I can’t imagine it wouldn’t transfer vibes from air movement right into the FC, and it sure looks like there wouldn’t be much clearance left between lid and FC. Have you tried flying without the lid to see?

I personally mount as much as I can below and in between the deck around and above the battery tray and fish them up through the top plate. It’s a bit frustrating sometimes but I think it negates a lot of issues like that. And it looks cleaner to me.

I also don’t buy into all the special vibration damping materials. I think simple is better. All you’re doing is de-coupling it from the frame, and so long as it isn’t super spongy, it should be fine (my opinion). I use simple 2 sided foam mounting tape by 3M, I think it’s 1/8" thick, and secure the mount to the deck with that.

I’ve used the mounts, and I’ve also just secured the FC directly to the deck with foam adhesive, and I’ve noticed no real difference. The mount gives more area to secure to the deck though.

Secure the power module and battery leads so they don’t move.

Where the cabling connects to the FC, make sure and move those leads slightly…see if they induce any movement on the FC. Especially that black one that wraps around the FC…those seem to be way less flexible which would easily transfer into the FC. Some of them even look like they could touch the FC if moved slightly.

Here’s some pics of my Tarot - it’s my backup so its kind of a mess atm but it gives you an idea what I’m talking about.
I take as much slack as I can out of them at the connection to the FC, keeping them low, but leave enough so that they aren’t pulling on the unit, and can move a bit without moving the FC.

TarotPic1
TarotPic2

These practices work for -me-. None are scientifically proven :slight_smile:

7

@Lance_B has given you some good advice.

Check EVERYTHING for movement.
Make sure everything is tied down and nothing dragging on your FC.
Secure the camera, making sure there is no looseness in the mounts.

Z vibration can be frustrating to reduce, as it is sometimes just the aeroelasticity of your frame.
I had a cheap Octo frame presented for build up that literally flapped its arms while powering up, but once up was smooth.
Be wary of thin wall tube arms and thin centre plates.
Flex your arms and see what sort of movement they have.

The Z axis is where you will get the most variance in vibration as it is a function of prop lift and aeroelasticity in the props, arms and frame.
In some frames its just better to isolate everything as you may not be able to reduce it without a rebuild.

8

Thanks Mike…and thanks for the better Z explanation

There’s one other thing I forget on the Tarot. When my brother first got this frame we did see some odd flight behavior.
(he’s since upgraded, and given it to me so I use it to test).

The more I looked at it, I noticed just some motors that weren’t quite level in relation to the frame - due to the design which is a separate mount that slides onto the tube.
Then began the long process of finding an easy way to level them.

I wound up cutting up an old beam level and used it as a straight edge - laying one end on the motor mount and the other on the deck, rocking them back and forth until I “felt” both ends laying flat. Then tighten the mount. Rinse and repeat.
This assumes your frame, like Mike has stated, is stable. So make sure all screws are tight, there’s no severe flex or movement in the parts and that it’s all squared up. I level a TV tray, and then make sure my frame is level relative to it, as it’s sitting on it. Then I calibrate level sitting on that.

The only reason I say this is in one of your pictures, it almost seems like the back right one is off kilter, but it may be perspective.

Anyway hope you get it solved.

9

Hi Christian, I have the same Quad frame as you, 4114 motors.
I had same problem, bad Z-Ax vib, where as XY acceptable. Tried several approaches to resolve Z. This worked for me : The FC hard-mounted on mounting-plate, then the plate on damping balls, similar to yours. I removed the FC & mounting plate from the balls, stuck a small lead weight (arbitrary 20g) [You can of course use diff weight for your rig] down the center axis of the FC mounting plate and remount it back on the damping balls. This fixed the Z-ax problem and provide good XY-ax results as well. I hope it works for you. Pic of Lead mounted underneath the FC with mounting plate.

IMG_6110_FC_LeadbarFitted2_sz.JPG480×553 106 KB

10

First of all happy new year to all of you!

There are 2 very important tips in here:

@Lance: the mounts are adjusted by jusing a spirit level and additional secured against rotation by a screw through the carbonfibre tube. So the photo might be a bit misleading here. The hole build is also done with loctite and the bird had its c-check about 5 flighthours ago. So everything is fine - even on the flexarms. But the thoughts with the lid giving some vibrations to the wiring and fc are very new to me - we haven’t checked that yet > to be done!

@Gerhard: we also had the idea of adding some extra weight to the fc. Your input is extremely helpful to get things realized on the copter. I’ll give feedback as soon as its done.

11

Hi, I have a very similar setup - 650sport frame with pixhawk mounted on the same kind of antivib platform. The standard arms are frustrating as you can only just use 14" props with the canopy and not 15" props. I got fed up with the standard plastic mounts which kept splitting where you do the bolts up so changed for the aluminium antivibration mounts which are waaay better:


These don’t rotate at all and don’t break, and do a decent job of removing vibrations from the motors before they reach the frame. I’ve also used the extended 650/680pro arms with bigger props with the same antivib mounts with no problems.
Z-axis vibs can be quite difficult to track down and eliminate scientifically, I find it easier to just try different ways of mounting the pixhawk. What works for one frame doesn’t work for another and vice versa. Try orange foam, harder rubber ball thingees in your existing mount, harder foam pads. I think high z-axis vibs tend to indicate that your mounting is too soft and is allowing too much vertical movement, so you may find by stiffening the mounting it helps, although it may then increase your x/y vibs. It’s also worth checking your motors for play in the bell, that can also cause z-vibs.
I just have my pixhawk on the same mount you have, using 3508-580kv motors with cheap 14" props and don’t have any problems at all with vibrations. It’s a nice frame, if a bit heavy. I like the landing gear and the canopy, makes for a neat package.

12

Adding mass to the FC mount has been the most effective solution I have found.

These pics are of mounts I use generally in all builds, these being in “Cattle Quads” used to muster cattle that get a ridiculously hard life and might be flying with chipped propellers for some time before being fixed. They use the old APM2.6 as they don’t require much more.
The PixHawk setups use the same system with great success.

I have tested the iso balls of varying densities but found the “O” ring suspension to be the best isolator of all. You just have to make sure there is good tension in the “O” rings, and silicone “O” rings are the way to go, the other materials don’t like being under tension.

There is a 1mm sheet of lead under the PixHawk between foam iso pads.

IMG_1106.JPG1704×2272 624 KB
IMG_1102.JPG2272×1704 847 KB

13

In my opinion your problem is the placement of your gimbal. I make sure all my aircraft are carefully balanced. That means having the ginmbal near centered in the frame. Hanging it out the front exposes it to prop wash which causes the vibrations in the z axis. Putting it under frame not only keeps the balance right but shieldsthe gimbal from prop wash. I run two identical 8000mah batteries, one on each side, to maintain perfect balance. cheers

14

Mike, I obviously cannot share your opinion, since we flew the copter without the gimbal for autotune purposes. The vibration levels were the same for the pixhawk… So the gimbal is not responsible for the high vibeZ-values. When the gimbal is attached the battery is also mounted off-center to keep the copter in good balance.

Shielding the gimbal might indeed reduce the video-jello itself > put it on the todo-list.

15

Hello Christian,

I own an Tarot 680 pro - stock - for two years with a similar setup.
First of all I would recommend you to read Igor88’s post in:
https://www.rcgroups.com/forums/showthread.php?2133373-Hexa-and-Quad-and-X8-w-e-AV-flying-time-30-minutes
and also there are other tips on the thread:
https://www.rcgroups.com/forums/showthread.php?2186236-Tarot-680-Pro-Hybrid-Version-2-Heavy

Besides it is a brilliant frame, very strong and robust, it shows vibration problems. Double check your motor mounts, all folding clips, the first screw that hold the folding arms. Try to force them up and down and see if something moves.

I’m fighting also to reduce micro vibrations in my frame. Have tried everything with mixed results, however the best solution was to complement t4-3d dampening solution with Kiosho gel ( will post pictures tonight ). For pixhawk I ended up printing the 3d model from GuyMcCaldin which is in the wiki http://ardupilot.org/copter/docs/common-vibration-damping.html#common-vibration-damping with astounding results.

Regarding Z-vibrations I also recommend that you verify if your motors has any play on the bells as @fnoop suggested. That is well know to cause Z vibrations.

My gimbal is also offsetted to the front and the batteries to the back , just like you did. Prior to printing the dampening solution I had decent results adding mass to the flight controller.

Will post a few pics when I get home.

1 Like
16

I haven’t read the entire post but in my experience, the vibration dampening foam (now being sold by mRobotics) is a very good and simple solution. http://store.mrobotics.io/product-p/mro-pxfoams-mr.htm

17

The original pixhawk foam was the best but I could never source it other than the 3dr store which was too painful to order from. Great it’s available elsewhere now but $12.90 + shipping for 4 tiny bits of foam?!!

18

I’ve been searching for other suppliers, but can anyone who has this exact foam please advise exactly how thick it is in mm?

19

Don’t know about exact this foam, but used Kyosho gel and happy with it: 5mm thick, 14.5x4.5 pad for $13.41 at amazon: https://goo.gl/WnfVdY

20

Oleksii, do you just cut 1cm square pieces one for each corner of the Pixhawk?