Photohigher skyline rsgs

[nicwilke]

Oh I get it, we are all wrong about this, I'll tell DJI, STK, ACS, Ultramedia, Cineflex, Bob Nettman and John Doyle that we don't need to worry about newtons third law, all we have to do a stick a brake on each axis so it cant move and then everything will be stable relative to the base it is sitting on. Yeah thats a really cool idea. PID algorithms are not needed as they only have negligible effect ( well there's a few thousand hrs. down the drain for a start). We don't need to worry about vibration attenuation because it cant move Woopee what a great solution you have there. I would keep that to yourself though or everybody will be doing it. I guess the only thing that you did fail was to read all of the previous analysis posts. I don't suppose it occurred to you that a heavy camera does not actually move, it is the surrounding camera mount etc. that does the moving during stabilization with all camera mounts not just PH.

Oh well that livens up my otherwise boring day:tennis: Thank God there's a pub opposite

So does damping effect the resolution or not?

 

[flytahiti]

After some further attempts to get the resolution without the oscillations I found this....
The gain setting on the tilt axis needs to be at least 250 otherwise it is way to slow to start the initial correction. (the FSR of your gyro is too high)
That gain setting is much too high to prevent oscillation, however, That can be cured with the fitment of a piston dampener from a buggy. This must be placed on the tray arm next to the IMU not at the opposite end. When looking from the front of the lens the imu should be on the right.
I now have the tilt working with this method.
I expect that the roll could also be improve this way.
Do not fit a friction device to the servo as this is at the wrong end of the drive chain. The wind up in the tray and the belt drive is where most of the problems come from when you place a 550D etc. on the tray. In essence the resolution between the IMU and the tray arm inertia is now tighter.
The PID algorithm needs to control the deceleration better to stop the continuous overshoot.

any photo of your mount

 

[DennyR]

So does damping effect the resolution or not? [/QUOTE
As explained before. dampening is a part of the pid algorithm. Watch James Cottons explanation in the video that I posted. (That is his simplified version that did not cover the the full algorithm) the idea that we need to use a mechanical device is simply a crap way to fix a problem created by someones poor design. If you fit a friction device on the servo output drive then it will definitely reduce the resolution. If you fit it to the input shaft of the IMU then that effect is less noticeable. Try and grasp this. if a body is suspended in space without any frictional losses then a force exerted though its center of mass will not change it's alignment. The more external forces that are applied away from it's center of mass then the further you will move from that objective. Sorry if that sounds blunt but that is how it works. That is the main reason why Zenmuse works and this PoS does not.

To use an old Aussie saying, that camera tray flops about like a terd in a piss pot.

 

[nicwilke]

@Denny
your humor is as dry as a dead dingos donger

Thanks for the explanation.

 

[DennyR]

I like that...

 

[chris hornby]

Thank you all for your comments and suggestions. We would like to acknowledge that we knowthere is a problem with the Skyline and in hind sight we probably released thesystem to early. There is not much wecan do about that but I would also like to assure people that we are trying ourbest to get the firmware to a state that it is acceptable. We are not going to give up and we are goingto work on it until we can get a good result. We only have a small engineering team with two electronics engineers whoare both working on finding a solution. We are also getting some assistance from an external electronics engineeringfirm to try and help us get the issue resolved as soon as possible.
With regards to some comments about us misleading people tothink that the Skyline would be able to compete with a direct drive system likethe Zemuse – this was never intended. The Skyline will never be as smooth as the Zemuse mainly due to servosversa direct drive brushless motors. TheSkyline was developed as a cost effective stabilisation solution for the AV130and AV200 gimbals and possibly others. We have neverreleased videos with post stabilisation to try and cover things up. And we havenever asked people to show up the product. The videos that we have released andadvertised are exactly what they are and show exactly what the Skyline is doing.

Thank you,
Kimberley Attwell
Photo Higher

I am with you all the way Kimberley, - cant speak for theothers! I am getting decent results now, and look forward to onwards and upwards

 

[psionic001]

Two days ago you were telling us dampers would not work. Now they do?
You've put damping where you said it would not work.

After some further attempts to get the resolution without the oscillations I found this....
The gain setting on the tilt axis needs to be at least 250 otherwise it is way to slow too start the initial correction. (the FSR of your gyro is too high)
That gain setting (250) is much too high to prevent oscillation, however, That can be cured with the fitment of a piston dampener from a buggy. This must be placed on the tray arm next to the IMU not at the opposite end. When looking from the front of the lens the imu should be on the right.
I now have the tilt working with this method.
I expect that the roll could also be improve this way.
Do not fit a friction device to the servo as this is at the wrong end of the drive chain. The wind up in the tray and the belt drive is where most of the problems come from when you place a 550D etc. on the tray. In essence the resolution between the IMU and the tray arm inertia is now tighter.
The PID algorithm needs to control the deceleration better to stop the continuous overshoot.

Kimberley Your advert clearly states Rock Solid Gimbal Stabilization. That statement is not withstanding what ever else is on the market. It clearly never has been rock solid and is not fit for purpose (YET). With any luck we may be able to help you achieve that goal as we are all stuck with it.

Perhaps you should reimburse us all or offer a significant bounty for the one who gets you out of this mess.

You have to provide a vibration dampening between the pitch shaft and the IMU. NOT the shaft and the frame otherwise the imu will not control the movement correctly. The shaft must remain free to move and be mass balanced or it will not work. killing the jitter by holding the shaft is not the way to do it. You have kill the jitter at the source of the problem which is incorrect gyro mounting.

If you strap down the pitch shaft, how can it possibly move in small increments. It will stay put until it overcomes the friction and then jump.

It is a known fact that a high sensitivity gyro setting is required to stabilize a camera, the trade off, is that it will be sensitive to all other movements including its own feedback if it is not mounted correctly. A pid algorithm cannot deal with bad mounting if it is to maintain it's designed sensitivity level.

Also once the imu has been mounted correctly it will have some small amount of free play in which to move against any vibration. This makes it sensitive to any loose wires etc. that are attached to it. These must be strapped down to the same base or removed as even a very small movement will produce an unwanted reaction at the camera.

 

[ChrisViperM]

Two days ago you were telling us dampers would not work. Now they do?
You've put damping where you said it would not work.

I guess Danny will reply in his own way, but more interesting: What solutions did YOU come up with....?

Other than that, it looks like the cooking pot is under serious pressure right now and the lid is about to hit the ceiling....it's time that Bart comes up with some sort of "Steam-Off Corner"....:tennis:

 

[DennyR]

I stand by what I said, the imu cannot work correctly if the any of the shafts are strapped down. Especially the servo output. You guys want a solution to fix the problem don't you. It is not correct but it works to a level which the average hobbyist would now be happy with. As I cannot access the firmware code another way has to be found to stop it overshooting. A problem compounded by flexibility in the drive chain. i.e. the imu should have been mounted on the same end as the servo wheel. And the servo should have been direct drive. Owners of the 200 model probably do not get the same problem.

This is better than chucking the whole thing in the bin where it belongs. If you have a better solution please don't keep us waiting.

In any case what I have done is nothing like the concept of squashing material between two moving surfaces to try and create some form of controlled friction.

Chris I expect Bart like many others is quietly laughing his head off.

 

[pixvertex]

Could somebody please share the 1.7 beta link with me (PM)
Will go out for testing today. I must say it getting better and better. I did the H loom. I have no jitter on the roll only some on the tilt (130AV)
The Sony 730 compensates a lot of the jitter. Way better than the mk stabi.

This is a link i found on the mikrokopter side
http://www.youtube.com/watch?v=9b8LZU7fIZg&feature=youtu.be

 

[psionic001]

I would have shared my solutions if they worked really well, but they didn't.

We are on 116 as 117 fails if you are using 2 skylines.

FWIW the guys at PH are working really hard on this and have been giving us some great help to get us out of a bind with various issues we had during the 117 install on separated skylines.

I'll be trying the rotary damper soon. Anyone have any idea about the damping torque level required? I'd guess it would be as low as 100-200mNm.

I guess Danny will reply in his own way, but more interesting: What solutions did YOU come up with....?

Other than that, it looks like the cooking pot is under serious pressure right now and the lid is about to hit the ceiling....it's time that Bart comes up with some sort of "Steam-Off Corner"....:tennis:

 

[chris hornby]

OK finally got it cracked. Throw away the gel, felt washers and springs. Replace the H harness with 3 Y leads as per attached diagram. Basically youare powering the servos and the skyline with completely separate supplies. 3 small eneloop batteries are ideal.
With this setup I can crank the gains right up with no evidence of jitter in tilt or roll.View attachment 5609 As for pan. Control that separately with a cheap heli headlock gyro (if you want)

 

[psionic001]

How much roll and tilt drift do you have.

OK finally got it cracked. Throw away the gel, felt washers and springs. Replace the H harness with 3 Y leads as per attached diagram. Basically youare powering the servos and the skyline with completely separate supplies. 3 small eneloop batteries are ideal.
With this setup I can crank the gains right up with no evidence of jitter in tilt or roll.View attachment 7459 As for pan. Control that separately with a cheap heli headlock gyro (if you want)

 

[chris hornby]

How much roll and tilt drift do you have.

No drift on roll, a bit on tilt but not enough to be evident in a flight video so I am not bothered about it too much, but could probably get rid of it in the servo fine tune

 

[DennyR]

I don't understand this, about a week ago we said that the power should come from three different sources. So I did that back then and saw a slight improvement but I had already cured the jitter. I could get no more than about 80 on the tilt gain which was nowhere near the required pointing accuracy. So today I tested the latest idea and it worked far better than I ever expected. The tilt gain was working at 255 and showing me an image in the LCD that looked like a still. It was still delivering the goods with the motors running which is a major step. I am going to do the same idea on the roll axis and then test fly it. I am expecting top quality results but I will also expect some slight movement at around the neutral position in comparison with a Zen. but not anything that could not be cured in post. Certainly miles ahead of the DJI FC outputs. This is the result of all of the mods. that I have discussed and I don't think it would work if any of them were changed.View attachment 5610 Some loss of the 90 travel volume due to eliminating ineffective crank angle. When testing this stuff it is vital to use a heavy camera like the 550D or the results will be misleading. Certainly the biggest step that I have made thus far..:tennis: 99% would be thrilled with this resolution. It goes without saying that there must be no free movement in that linkage. Even if the firmware is improved I think this will still be the way to go as it can only get better from now on.

 

[RancherPete]

I don't understand this, about a week ago we said that the power should come from three different sources. So I did that back then and saw a slight improvement but I had already cured the jitter. I could get no more than about 80 on the tilt gain which was nowhere near the required pointing accuracy. So today I tested the latest idea and it worked far better than I ever expected. The tilt gain was working at 255 and showing me an image in the LCD that looked like a still. It was still delivering the goods with the motors running which is a major step. I am going to do the same idea on the roll axis and then test fly it. I am expecting top quality results but I will also expect some slight movement at around the neutral position in comparison with a Zen. but not anything that could not be cured in post. Certainly miles ahead of the DJI FC outputs. This is the result of all of the mods. that I have discussed and I don't think it would work if any of them were changed.View attachment 7461 Some loss of the 90 travel volume due to eliminating ineffective crank angle. When testing this stuff it is vital to use a heavy camera like the 550D or the results will be misleading. Certainly the biggest step that I have made thus far..:tennis: 99% would be thrilled with this resolution. It goes without saying that there must be no free movement in that linkage. Even if the firmware is improved I think this will still be the way to go as it can only get better from now on.

Sorry, I'm a little lost. Which latest idea are you referring to that seems to be working so well? Is it the idea of using 3 separate power sources? Just trying to keep up. Also: I can't make heads or tails out of this diagram above for separate power sources. Are we just talking about using 3 separate BECs to provide separate power to each servo and the RSGS? This diagram looks lke somethign different all together so I don't understand what it represents. ALso: I don't see how 3 Enelope batteries could work. Wouldn't that add up as follows: 1.5 X 3 = 4.5 volts? Unless 4.5 is okay.

 

[Soaringclauz]

OK finally got it cracked....CUT

that was exactly what i was going to try this evening, yesterday i tried with two separate supplies (one for the rsgs and tilt, and one for the roll) with little benefit, now i'll set this up with more confidence. Thanks!

 

[pixvertex]

the 3 Y loom really makes a big difference. Both for tilt and roll.
Wanted to try the setup before flying.
I power the skyline with 5 volt (scherrer uhf receiver - 5 volt recom). Tilt 6.5 volt Traco (should be replaced since only 1A), 6 volt 6a switched bec for roll.
My sony 730 is mounted on gel dampers.

Tilt gain 100, RollGain 650. I have more or less no jitter now. Only on tilt when i use values over 120. Due to the gel damping.
Great stuff now. I hope to get a video over the weekend.
It is getting better :tennis:

edit:
The H loom made thing better, the 3 Y loom is even better concerning the jitter

 

[chris hornby]

Sorry, I'm a little lost. Which latest idea are you referring to that seems to be working so well? Is it the idea of using 3 separate power sources? Just trying to keep up. Also: I can't make heads or tails out of this diagram above for separate power sources. Are we just talking about using 3 separate BECs to provide separate power to each servo and the RSGS? This diagram looks lke something different all together so I don't understand what it represents. ALso: I don't see how 3 Enelope batteries could work. Wouldn't that add up as follows: 1.5 X 3 = 4.5 volts? Unless 4.5 is okay.

Actually I have 4 power distinct power supplies.
1. 6V via a Bec for the Airframe RX, the DJI WKM and the pan servo,
2. A 6v eneloop pack to power a RX for the Skyline and the skyline itself
3. A 6V eneloop pack to power the Tilt servo
4. A 6V eneloop pack to power the Roll Servo.

All these power sources whether BEC or battery MUST have a common ground (0v) If not you will get ground loop issues.

All the Y lead loom arrangement does is ensure that the power supplies are kept separate, whilst having a common signal connection and common ground. You just need to literally plug 3 Y leads together and snip both the red wires in the first Y lead so that no power gets from the Skyline to the servos, also the servos do not get any power fluctuation from each other.

I think that it is about time to replace the RSGS loom and all the other wiring with a PCB to tidy things up. SOrry my diagram is not clear, when we have got clear of all our Olympic visitors I will make a video.

I am pleased with what I have got now but YMMV.

 

[DennyR]

Sorry, I'm a little lost. Which latest idea are you referring to that seems to be working so well? Is it the idea of using 3 separate power sources? Just trying to keep up. Also: I can't make heads or tails out of this diagram above for separate power sources. Are we just talking about using 3 separate BECs to provide separate power to each servo and the RSGS? This diagram looks lke somethign different all together so I don't understand what it represents. ALso: I don't see how 3 Enelope batteries could work. Wouldn't that add up as follows: 1.5 X 3 = 4.5 volts? Unless 4.5 is okay.

The idea that has transformed the performance is shown in the picture. All this crap about wiring is not going to help much. Two things have made this work. One is the mounting of the imu and the other is the damper on the tray arm. I did explain why it has to be this way. When I say work I mean work at the top level. The point that you probably missed is the fact that the AV200 has the imu on the same end as the servo pulley which makes it much more rigid and could be the reason why some are not having so much trouble. The camera tray bends with the weight of a heavy camera on the AV130 and this produces a wind up. If you cant get your tilt gain to hold 255 without an oscillation it aint gona work.