Photohigher skyline rsgs

D

DennyR

Active Member
Jeremy
I think that stuff is a bit heavy going but looking back to the first PID algorithms I think they were created as a solution to the steer large ships in high current flows to arrive at a track made good destination.

With regard to the friction devices that I have made They are quite adjustable with the type of silicon oil used plus I play with the hole dia. to adjust the friction value. Pitch and roll are now quite different.
 
Last edited by a moderator:

R

R_Lefebvre

Arducopter Developer
jes1111 said:
As Denny pointed out a few pages back - one tends to forget that behaviour on the bench is very different to what happens in the air. On the bench, the airframe stays still and the gimbal moves. In the air the gimbal is effectively stationary and the airframe moves relative to it. Thus the contribution of gravity (not to mention, momentum, inertia and so forth) is very different for a flying craft compared to one sitting on the bench. Logically, therefore, "bench tuning" has to be tested/verified by the "in the mirror" technique at the very least.

btw - Interesting academic paper I stumbled on last night: http://alexandria.tue.nl/extra2/200412510.pdf - all about how friction can worsen limit cycling (a.k.a. jitters) :)
Click to expand...

Somewhere around here I've got a really good paper about the effect of friction on a PID controller for a small gun turret on a ship. Almost the exact same application. It was actually fairly easy to follow, at least in the beginning. I wish I could find it again but I've misplaced the link. It was to do with Fuzzy Logic superimposed on a PID. Anyway, it talked about how the friction made it almost impossible for the the turret to follow slow rolling motions of the ship's deck, while still having enough gain to respond quickly when needed. It's pretty obvious what happens, is that the actuator loads up against the friction, and then it breaks loose suddenly causing an overshoot. They used Fuzzy Logic to implement gain scheduling based on speed.
 

J

jes1111

Active Member
Yep - very stodgy stuff, that paper. Certainly not comprehensible to me, all those equations!

When does "damping" become "friction"? Artificially introduced friction/damping may well help so long as the amount is carefully controlled in relation to the rotational torque available/needed. So a felt washer whose influence is calibrated by loosening or tightening a screw(s) - probably not "good engineering" ;). A hydraulic damper with carefully chosen oil viscosity and orifice size - much better idea. But best of all, of course, is an adequately stiff mechanism with zero backlash and no need for any introduced friction/damping :)
 

R

R_Lefebvre

Arducopter Developer
Damping ends and friction begins when the resistance is non-linear. So basically anytime you have dry friction. If you replaced the felt washer with a plastic or rubber washer that was greased both sides, you'd have a cheap rotational damper. I'd suggest a heavy, heavy grease like, about the consistency of honey.
 

D

DennyR

Active Member
jes1111 said:
Yep - very stodgy stuff, that paper. Certainly not comprehensible to me, all those equations!

When does "damping" become "friction"? Artificially introduced friction/damping may well help so long as the amount is carefully controlled in relation to the rotational torque available/needed. So a felt washer whose influence is calibrated by loosening or tightening a screw(s) - probably not "good engineering" ;). A hydraulic damper with carefully chosen oil viscosity and orifice size - much better idea. But best of all, of course, is an adequately stiff mechanism with zero backlash and no need for any introduced friction/damping :)
Click to expand...
Dampening and a tight mechanical movement helps a lot but I can see by the temperature build up in the tilt rotor motor that it is working overtime when it is not actually moving. The noise feedback issue has to be rectified in the IMU. because that tilt motor behaves normally when the roll power line or it's PWM signal is taken out. That suggests to me that the IMU is causing the problem by sending a corrupt V.ref return signal to the tilt servo. If you simulate that with an external supply of 1-2 volts the servo reacts normally.
 
Last edited by a moderator:





B

barnytj

Member
Interesting stuff on the friction thanks Jes1111 and everyone else.. and the chicken is pretty good also!

I haven't yet bothered with the H or Y harnesses as I am not struggling from jitters whilst bench testing - but perhaps I should..

I'm currently on location in East Africa for a series on one minute TV commercials - which are fortunately filmed in the hand held style - but with get on with these mods as soon as I get back to town.. for now I have just added a squib of wd40 bedhind the servo cog/belt and that seems to have helped a little..
 


B

barnytj

Member
hdj912 said:
Any updates on how to prevent the rolling when panning? 1.1.7 still suffers from that.
Click to expand...

Although I am running 1.1.6, I seem to have overcome this problem.. the steps I took were as follows:

1. Make sure the skyline is absolutely perpendicular to the camera tray.

2. Use a spirit level (I used a mobile phone app) to ensure that the camera tray is perfectly level when setting the level in the software.

3. Make sure your multirotor is stationary when powering up the skyline and during the start up sequence.

If anyone can send me a copy of 1.1.7 I would very much appreciate it.. barny@highlyvisual.com
 

R

RancherPete

Member
With 1.1.7 I can level my gimbal only by using the auto feature. I can no longer adjust my level a little at a time in roll and pitch with the drop downs. In order to use these drop downs in the past it required that the servos be active (duh) but now when the servos are active the RSGS' green led is on and therefore the gimbal does not respond to any adjustments from within the software. Are any of you able to use the drop downs for fine level adjustment? I am stuck with only AUTO now in 1.1.7
 



H

hdj912

Member
Just a quick question, on the AV200 which way does the white servo gear needs to be mounted? In either way it seems to be touching the servo housing when i fasten the screw.
 


D

DennyR

Active Member
DennyR said:
Dampening and a tight mechanical movement helps a lot but I can see by the temperature build up in the tilt rotor motor that it is working overtime when it is not actually moving. The noise feedback issue has to be rectified in the IMU. because that tilt motor behaves normally when the roll power line or it's PWM signal is taken out. That suggests to me that the IMU is causing the problem by sending a corrupt V.ref return signal to the tilt servo. If you simulate that with an external supply of 1-2 volts the servo reacts normally.
Click to expand...

I think we need a low pass filter on the V.ref return. But right now it's 40 degrees and I'm staying on the boat.
 



You must log in or register to reply here.
Top