I doubt you will find as simple answer as raising or lovering D or something similar, because it simply depends of so many things.

Gimbal mechanics and vibration dampening are my suspect number one, but if those are ok it could also be the Power/PID/gyro sensitivity/ect.

I was tuning my handheld gimbal couple of days ago and tried something like this with good results. http://www.rcgroups.com/forums/showpost.php?p=28045237&postcount=9447 This could work also with your gimbal. I have not tried this yet with any smaller or airborne gimbal.

'I' is also important, especially on 2.4. On 2.3 it affected RC response, but I think other things too. What do you have on 'I'
'
It has been a while I have been tuning 2.3 but I think your D is too high, probably P too. On 2.4 at least idea is to get P high, D low with 'I' at small value (0.03) and then 'I' as high as possible. 2.3 is different on 'I'.

How is your gimbal acting if you test as per the link on my previous post?

"like in windy conditions, too much speed, or getting the multicopter down too fast." Any gimbal will fail on these conditions regardless of the PID settings, if it is too much. Vibration dampening and mechanical construction are the primary things to improve performance on these conditions.

How much difference are you seeing on the on flight videos depending of the PID settings?, I suppose you have always tested first 'handheld' that the settings work. Have you tried just well working PID combination (when holding the multicopter in hand) i.e. no high or low values, just simply normally tuned gimbal.

I tried finally in flight the gimbal with the P = 50, D = 50, high gyro sensitivity (sorry for writing the word badly before).

Still some vibrations, although some less perhaps, and only when going backwards (gimbal facing forward) and sideways, especially with high speed and/or strong wind.

I noticed now the vibrations are slightly more high-frequency.

I'll try doing what Garug explain in the link, and maybe I'll try to low the D parameter, I hope that will do the trick for me. I usually look at the graps when tuning, but didn't know how exactly to interpret the lines, except maybe for the flat line meaning no vibrations when idle. The lines usually are not so much straight, maybe too much P or too little D... Anyway, I'll give it another shot.

I'll comment something more when I'll be able. Best regards.

I'm trying to reproduce the situations where the roll shakes without the need of going in flight, since I can't fly the octo every day.

At first I thought the vibrations could be the result of a little stiffness from the yaw axis, that could transfer vibrations to the other axes, or some yaw motor vibrations in some moments. But while all that can cause some vibration, I then recalled that when I used the gimbal by hand, in the land, in some movements the vibration was also happening because of some of my steps, the less smooth ones (when I was tired, or passing some steps).

I also tried stabilization with the gimbal turned off (not in flight yet), and the vibrations are much less, but of course the roll loses horizon quickly.

So it must be something about the settings, not a physical/mechanical thing (I hope), nor a multicopter issue (no vibrations with the old gimbal AV200).

I finally updated to 2.40 b7 for good, and using patience and knowledge I think I've achieved a good tuning, looking at the realtime data and testing the movements. It wasn't so hard to get it right, and this version seems a little more stable, maybe. Now I'll like to try these settings in flight.

Another question for you, guys: Can a wrongly calibrated accelerometer cause vibrations? I think that's not what happens, but maybe I should start thinking about redoing the 6-point recalibration... But it's so complicated. :/

So I'm glad I'm closer and closer to the solution. When I'll solve the issue, I'll be very happy!

@Dronelab We have a few videos on YouTube and linked to our webpage, although the parts with vibrations are cut, obviously. Fortunately, the vibrations only occur in some extreme situations and most takes can be used perfectly. :)

So I tried in flight with some other PIDs for roll (P = 22, D = 44), and it lacked some P since it couldn't hold the horizon/angle too well (in the realtime data the roll was doing really well, but...). I'll try with more P and less D and explain how it goes. I'm trying to figure if for my gimbal it's better more P than D, or more D than P, or both basically the same... It seems D shouldn't be as high as P, for the moment.

I was wondering if someone could tell me... If the gimbal battery is at one side of the roll and the motor is at the other side, with the camera well balanced, could the weight of the sides (battery and motor) make more inertia than the usual and be causing the vibrations, or at least making them worse? There's not much space to put the battery on another place, but if you think it's possible that could be possible, I could try it and move the camera more on the contrary side of the motor to see if it balances well enough and the roll gets less vibrations in extreme movement.

It's weird, because tilt seems pretty stiff on most settings, while roll isn't as hard even with power at the maximum, P = 50 and high gyro sensitivity... Could be needing a more powerful motor? It's too bad, because a bigger one wouldn't fit inside the part that holds the motor... I'm using RCTimer 4114 brushless motors with 22 poles for a camera weight of around 700 g, with a 3 S lipo battery...

Best regards.

Hi @Octovisuals
I have the same problembut it feels like that the shaking on the roll is less when the voltage of the lipo is not full powered. So maybe try to lower the Power value if you haven't already.

Someone solved the problem?
It also occurs in handheld gimbals at extreme movements..

@GregorPQ There are so many possible setting combinations I'm still testing and trying to achieve the level of stabilization I wish to achieve someday without spending like 10 000 € in a brand brushless gimbal. I think that's what I'll end up doing if this still doesn't get better soon already, though. I think the roll oscillations, and some other little horizontal vibrations (when not using a gyrostabilized lens), are caused by a combination of factors, some of them mechanical, so...

@austin32 All three have 22 poles.

@Garug Hi, Garug, thank you so much again for the reply. It's really encouraging to see that someone is always ready to help...

My board is 8-bit with just one IMU attached with screws to the horizontal pitch boom. Firmware is 2.4 b7.

Vibration is pretty damped through several carbon fiber booms and plastic and rubber pieces. They are not very stiff, though, I still am wondering if they flex too much sometimes.

For what I tested so far, I think the motors are just strong enough for my GH4 with a pancake lens. They are RCTimer 4114, 3 S lipo battery. They seem to hold well with a power of around 200.

I think I'm beginning to see that all my roll problems are PID related. Starting with power at 200 and I at 0.01, which seem to be the correct values for both parameters, I realized I used to set P too high, and also D. I think the problem is P, though, since low values seem to not cause the infamous roll oscillations. The problem is, I need to find the exact balance between a too low P with too much movement in roll, and a too high P with roll oscillations. That “perfect value” is killing me (someday I'll like to try Autotuning... :P). In order to discover this exact value, I must go outdoors to fly (since the gimbal behaviour is different on the table and in flight), and that requires time. So I'm still testing, and I hope I find the best parameters for the stabilization I want (without using OIS lenses, or stabilization software).

I was just wondering, how do you guys tune your PIDs and find the correct values? Tap with the finger the axis, move the gimbal handheld and aggressively, just looking at the realtime tab and blue numbers...

Because in the official guide it says: “Slowly increase gain factor "P". At occurrence of oscillations increase factor "D" (it is responsible for the damping of vibrations) in small increments until the oscillations have gone”, but I don't get any oscillations or weird behaviour with my gimbal, if it's still, even using P 50, D 50 and Gyro high sensitivity (which I normally don't use). So how can I know P and D are right before I go to fly? I tried everything and the only way I found is just go to fly... No tap with the finger, no graph can emulate what my gimbal really behaves when in flight.

Nice testing video. One must have time, knowledge and patience to tune manually these brushless gimbals... :D

Best regards.

Hi Octovisuals,

did you manage to work this out? I'm having very similar issue. Inside while tuning my gimbal is working just great but every time I go to the field I can't stop thinking that something completely different is attached to my hexacopter. It's losing horizon, shaking on all axes and time to time going completely crazy. This is happening in windy weather .. although 5m/s is not a wind I would expect to destroy my footage so badly.
I went through lot's of auto tuning (with some manual tuning after all) but I ended up with same result in the end. I guess I will now try some manual work - hope it will work better.
Btw. I'm using v3 board, 2.43 b9 SimpleBGC and this gimbal:
http://quadframe.com/collections/frontpage/products/medium-size-bldc-gimbal-v2
and these motors for roll and pitch:
http://quadframe.com/products/bldc-motor-5010
yaw motor is bit bigger
http://quadframe.com/collections/frontpage/products/bldc-motor-gbm5208-024-200t

everything is balanced perfectly ... but maybe these motors are not giving me enough power to keep nex5 camera stable in wind ?

Thanks a lot in advance

My gimbals work on huge range of PID and power settings, but certainly not with any settings. Typically I just auto tune them one axis at the time. There is always big enough PID settings that will cause vibrations.

P 8 is quite low, but sounds you currently can not tune it higher. Next step is to see the mechanical reasons for this and then tune again.

I guess it is more like a belt, i.e. not flexing like rubber. flexing belt would be bad.

Is it 1:1 gear ration? If not that affects the number of poles.

Calculate how many teeth there is on the gears. set poles on GUI to ('bigger gear' / 'smaller gear') x 'motor poles'.

Yes, it will give more tongue, but less speed. For yaw the speed is often enough. For rotating landing gear that could be good, it requires a lot of torgue, but can not say, I have not used and would not use normal rotating landing gear. there is just too much that the wind can shake and it is part of the stabilised construction.

DIrect driven motor is better for many reasons, actually I can not think any other reason it would be worse than tongue. Big motor and encoders provides plenty of torgue, but I still would not use normal rotating landing gear.

Note: with encoders I believe the motor poles are set normally, but there is separate setting for gear ration. Non encoder FW does not have gear ration, so the number of poles need to be set to the number of poles that actually pass on 360 degree rotation, calculated as mentioned on previous post.

89 is fine, remember to retune, and maybe worth experimenting also with the power. Usually high power is better, but if the PID is very low, it is worth trying lower power. and retuning higher PID

I have lately been working on follow yaw

So, here's some testing video I made using the octocopter, the 3-axis brushless gimbal and the GH4, from a testing flight some months ago:

https://youtu.be/UpMGaHpZdVQ

This is without lens stabilization and of course without software stabilization. It was follow yaw mode on if I recall correctly.

Since then I improved yaw movement and stability both in follow yaw on and off, in flight and handheld, and slightly also the other axes, and added a lens with stabilizer, which improves the overall stability, most especially the high-frequency oscillations, but not so much the roll movements (but something it does too). I'm still testing some things, though, to achieve the best performance possible.

So, what do you guys think of the stability you see in the video as it is? Any advice?

The camera is OK (I'm a professional ;)), but it's true that particular panning isn't very smooth. I'm guessing it was done when the yaw wasn't that well configured yet, and the movement by RC was this jittery. It's maybe a little too fast for the frames per second used (25), too.

In later videos the panning is smoother again (unless when done too fast, of course):

https://youtu.be/sxkve7fwjpM

This is from the last testing flights, three weeks ago, with lens stabilization, and without software stabilization.

I'm checking the originals right now and I remembered some parts of the first video were made with follow yaw on, but on the others it was off. That pan was done using Profile 1 (follow yaw off) and RC command, not using the octo movement.

Yes, it's the same gimbal with the belt. The vibration dampers are some rubber rings and two small carbon fiber booms, but it's true they aren't very stiff.

That's why I've been trying to get dampers and octo-gimbal assembly stiffer, but it's not so easy as some parts need to be changed.

Do you see PID/power and other settings for stability OK from the videos?

I'll keep working and testing.

Thanks a lot.

What frames per second did you use in that panning, with which shutter speed? Some jitter when doing pan movements is normal when using less than 30 (29.97) FPS, especially depending on the movement speed, the shutter speed, the distance to the subject and the shape of objects in the composition.

The panning in the second video, I think it's pretty OK for 25 frames per second progressive... It's not completely smooth as you comment, though, and I'm thinking maybe the yaw could need some more adjusting, at least in the RC part, or more likely the mechanical parts of the axis itself need a little bit of improving, as I'm trying to do now.

Don't mind the noise, I was doing some tests with film grain to see the results after the YouTube thing...

It's true Internet recompressing and streaming isn't the best thing for video, but... we can't help it.

The gimbal seems now pretty responsive and not shaky (no noticeable vibrations) when holding it.

I think I still need to work out some mechanical things, and perhaps do to a couple values a really minor readjustment, and I'll be OK with this gimbal already... Until I purchase another, better one. :D

Thanks a lot again.

I'm reviewing the videos, even frame by frame, and I definitely don't see any relevant jitter, judder, stuttering... It must be YouTube, but I don't see anything myself there either. It's not a perfect horizontal movement because there will always be turbulence and some wind, and I need to review the mechanical parts again, but the panning is definitely good overall.

For both videos I used 25 FPS, progressive, with a shutter of 180°, which is the “standard”, edited of course at 25 FPS as well. Not so sure if YouTube changes framerates, though. Shutter at 360° would look too blurry as there would be too much motion blur, and less than say 120° is too little and would stutter (strobe effect).

Just curiosity, what do you think about that “film grain” I added to the videos? I didn't think anyone would notice, but I believe a good little bit of grain gives a more organic feel to the digital video. I'm still testing, though.

The yaw feels pretty strong and smooth doing what you suggested, as far as I can tell. Doesn't feel it's stuttering too noticeably, but I'll definitely keep testing, especially the pan movements.

I'll explain some more when I know something else. Thanks.