I am on mobile connection and the video does not play well :-( , sometimes the written word is better ;-). (but I enjoy watching your videos.)

Balancing looks good. About power, you need first to get the gimbal working well, then worry about the battery life. The thing to look after with Power is not to overheat motors, but other than that, it is usually better to have high power. It is difficult to tel from the video, but power looks ok.

But why do you disable the other two axis when tuning?? They should all be active when tuning. They interact with each other, and the two axis rotating uncontrolled would be confusing for the controller.

I would suggest using the auto tuning with good stability setting ( I usually use the slider 1/4 up). first for all axis, and then second time for each axis at the time, and with this I mean keeping all the axis active, but tuning just one.

If tuning manually, just set all the PID to low values so that it is stabile. By stabile I mean does not shake, has some basic stabilization working. Then tune one axis at the time like you are doing on the video.

The vibrations, do not bring down the power. Retune the vibrating axis on that position so that vibration goes away (Auto tuning works also for this.), or/and use filtering. http://www.levitezer.com/styled-2/blog-2/files/archive-oct-2014.html

I did not get longer than 20 minutes, need to contact the service provider, mobile data has usually worked good here...

Yes it looks like problems on editing SW, but gimbal is not perfectly tuned either. Did you try the autotune? That looks like quite ok gimbal, the autotune should work on that.

Ps. Here is video from last weekend, taken with well tuned, but a bit different gimbal.

Some up and down movement but that is because I am still learning to walk steadily :-).

@Garug After some testing I've discovered that the power for my roll axis (for example) must be 180, which is pretty high, I think (I'm using a 3 S battery). Lower than that the axis would lose steps in slightly harsh conditions (especially in flight, like when speeding up), even with good balance. Given that power value, I apparently must use a somewhat low P, like 15, because my D is almost 50 (I have an 8-bit controller) to be able to get rid of the oscillations/vibrations caused by P (looking at the graphs)... Is this right?

Thank you very much for the reply and the advice, Garug.

It's true D looks weirdly high, but after all these hundreds of tests I did, using all types of tuning possible, and looking at the graphs, they seem to be the best values...

I parameter seems to be working well at 0.01 or 0.02 (0.07 for yaw), more produces bad results in the graph.

While stabilization and general behaviour right now is pretty good already, it still shows some really little vibration I can't get rid of. I think it's from the gimbal frame itself. I'll try the result using a lens with stabilization...

I tried “Compensate voltage drop” once, it gave me weird behaviour, disabled it since. Do you recommend it? I may go up to 240 power for yaw in follow mode...

Dave, which lens with stabilization do you use with your GH4? Is it effective against that kind of vibration?

Best regards.

I want to make soon a little video and post it in one of my threads here, showing my gimbal and some stabilization results right now. (Sorry, Dave, for “stealing” your thread!) :)

Currently the stabilization is fairly good, but there's some horizontal, little movement when running or in flight, but for the looks of it I think it's because some parts may flex slightly and are prone to that kind of vibration, especially when carrying some relevant weight like the GH4 plus lens and a camera plate (it's a not-so-big gimbal).

I already changed some parts to improve the gimbal and achieve better balance, and therefore general performance. It's much better now than a year and a half ago, when I purchased it. Still some more work can be done, and the level of perfection I'm looking for is when the camera seems to be levitating, no oscillations or vibrations whatsoever, and good behaviour in all (normal) situations and axes.

That high D I got when I tried to completely remove low-frequency oscillations caused by P, looking at the graph and trying to get sine waves, without much disturbance and peaks, which is I think the correct way to check proper tuning before testing. I can however lower it and get somewhat good results, but the stabilization seems to look better with that kind of a high D. If I lower D below P, even using a relatively low P, the oscillations seem to increase.