Neills DIY Gimbal Design



P

PairAir

Member
I've never been really pleased with the gimbal design (AG550), as I think it lacks in rigidity. Plus some other design errors, like bearings being jammed in between washers and nuts....
I think they have corrected some of this in later versions, though. So the NEX5 seemed to be on the heavy side for it, bouncing around a bit in flight. Later tests with a GoPro look a bit better. But for gimbal control, that 30 dollar Multiwii Lite is not too bad.

/Pär
 

B

BIGTACO

Member
I see. And this is what I was hoping you would say. Intially I tried capturing my bearings with screws and nuts but this method doesn't produce good results. After a redesign to precision shafts, machined bearing blocks, bearing shims, and collars, its smooth as butter with zero play (that I can feel or see). Hopefully it will be enough. We shall see.

PairAir said:
I've never been really pleased with the gimbal design (AG550), as I think it lacks in rigidity. Plus some other design errors, like bearings being jammed in between washers and nuts....
I think they have corrected some of this in later versions, though. So the NEX5 seemed to be on the heavy side for it, bouncing around a bit in flight. Later tests with a GoPro look a bit better. But for gimbal control, that 30 dollar Multiwii Lite is not too bad.

/Pär
Click to expand...
 

R

R_Lefebvre

Arducopter Developer
That's a fantastic result man! Congrats! I'll probably be going down the same path as I'm just not impressed with anything on the market at any price (Other than the Zenmuse, of course).

I bought this $20 gimbal from HK, for a "Bumblebee" quad. Just for SnG's. You know, it really doesn't seem that much worse than a $600 gimbal? ;) It's just a bunch of CF plate screwed together. But you know, it actually has bearings installed in proper aluminum bearing blocks?
 

B

BIGTACO

Member
Here is my first flight test...Please try and ignore the absolutely horrible video quality my wife's Sony Cybershot produced. I still need to buy my Nex-5n. In this video I tried to make very extreme roll and pitch manuvers over and over trying to get the gimbal to hit its travel limits.


Problems I see:

1) I have a "bounce" on roll. Since I dont see this bounce when I preform "mirror video testing" holding the gimbal firmly in my hands, I believe this is mostly caused by not enough rigidity on my vertical mounted "roll assembly". I am getting torsional "twishing" in this roll structure. This is causing a resonance type "bouncing" as wind and other external forces hit the roll assembly. Then as the structure bounces the servos try and compensate for it and further degrades the footage. To correct this I plan to add a couple carbon bracing tubes running from the lower corners of my "roll assembly" at an angle up to the rear edges of my gimbal assembly underside. Sort of similar to how RC heli's brace their tail boom. This should increase my torsional stiffness enough to stop this twisting. Ideally I would use slightly thicker plate to help with this but I am really not wanting to cut new plate.

2) When I make roll manuvers with my hex the roll axis doesn't level exactly with the horizon. Its good, but slightly off by what looks like 3 to 5 degrees. I am thinking this may be the roll "scaling" setting within my hoverflygimbal. Looks like I need to increase it some so the gimbal corrects at a higher angle.


Things I am happy with:

1) Tilt. Tilt looks to be almost dead on. I have no bounce in tilt. Also the servo/gimbal/controller seems to be doing an excellent job of keeping the camera correctly stabilized on that axis. The movement looks very very smooth in the footage. Few minor adjustments and I think my tilt is golden.
2) Overall roll movement smoothness looks good. If you ignore the bouncing I am getting in roll and just focus on the overall movement that axis is making it looks very smooth.
3) Gimbal travel looks good. In the video I took my hex to the limits of roll/tilt transition I felt safe with. Somewhere around 30 degrees in either direction (60 total per axis). The gimbal didn't seem to ever hit any mechanical stops so my travel looks good. This only holds true for the flight characteristics of my Hex while in "GPS" flight mode. If I clicked my flight mode over to manual then I am sure I could bottom out the gimbal. But, I don't fly in manual for aerial video anyways!

Another thing that I believe will help things out is my vibration dampning. I don't feel my design if sufficient and that I am sending vibrations to my camera. I am already in the works of beefing up vibration dampning.

http://www.youtube.com/watch?v=8g7kVVBmzLY
 
Last edited by a moderator:

J

jes1111

Active Member
Amazing - and all made "by hand". :strawberry:

The twisting in your stem structure makes sense - the spacers are not an efficient way to bind the two plates into a single, rigid structure. For a quick fix, I'd try gluing small CF plates between selected pairs of the standoffs. This should bind the structure together better. You could also "fill in" the corners where the servo extends laterally with glued-on triangular pieces. In fact, I reckon that's the weak area of the structure. I'm not sure how well your idea of tubular braces would work - the assembly looks like it will be twisting in the middle - the braces might not have much effect on that.
 

B

BIGTACO

Member
I see what you are saying now about rigidity and it makes sense. Not sure why I didn't see this in the first place as its a pretty basic engineering principle. The contact area between the ends of the aluminum standoffs and the plate is too small and localized. I need a contact area between prependicular members that runs the entire width (ideally) of the structure to keep it rigid. *Sigh*

How can I glue carbon plates together? Seems like CA glue would crack off since carbon isn't porous at all... Maybe epoxy?

jes1111 said:
Amazing - and all made "by hand". :strawberry:

The twisting in your stem structure makes sense - the spacers are not an efficient way to bind the two plates into a single, rigid structure. For a quick fix, I'd try gluing small CF plates between selected pairs of the standoffs. This should bind the structure together better. You could also "fill in" the corners where the servo extends laterally with glued-on triangular pieces. In fact, I reckon that's the weak area of the structure. I'm not sure how well your idea of tubular braces would work - the assembly looks like it will be twisting in the middle - the braces might not have much effect on that.
Click to expand...
 



J

jes1111

Active Member
Yes, epoxy - preferably slow-cure. Rough up the surface first with sandpaper. Clean surfaces thoroughly with isopropyl alcohol. Use firm pressure until it goes off.
 

anaka

anaka

Member
yes for sure, with so small jitters it'll come out perfect
you should try AfterEffects 5.5
the new warp stabilizer rocks!!
 

B

BIGTACO

Member
I am looking into getting afterEffects 5.5. All I have at the moment is 5. In the meantime I ran this same raw footage through Sony Vegas stabilizer @ 25% setting for s&g... I think I could have ran it lower. Maybe 10-15% with good result. Took the jitters right out. Regardless, I have to resolve my jitter on the gimbal for real or I won't stop thinking about it.

http://www.youtube.com/watch?v=YdL9fZcqDg4
 

Tomstoy2

Tomstoy2

Member
I see what you mean when comparing the the 2. The question is, how do you go about it? The obvious solution is to add some type of dampning, but that presents it's own problems, not to mention defeating making everything run so smooth.

Tilt is almost perfect! Roll seems too aggressive at extremes, tilting the horizon slightly. I imagine this is just a matter of fine tuning the HFG.

Other than that, I would say you got almost everything sorted. Damn good job!
 

R

R_Lefebvre

Arducopter Developer
I think the result is excellent. It's hard to say if it's top-shelf or not, because you are intentionally flying it aggressively to try and find the problems. I'd like to see you try and do a real smooth flight, and see how good it is. You might find it's perfect just the way it is. As it is, I can barely see the bouncy roll you're talking about, but it's hard to watch too long without getting seasick. ;)
 


R

R_Lefebvre

Arducopter Developer
BIGTACO said:
Hey thanks! I'll try that today at lunch and see how it turns out.
Click to expand...

Just in case it's not clear, what I'm saying is I think the gimbal could be equal or better to the top gimbals out there, save the Zenmuse. It's just hard to say for sure until you try smooth flying with a decent camera.

If your concern on the roll is that the vertical member of the gimbal is twisting about the Z axis, I think it could be stiffened up simply and easily by replacing a few of the aluminum standoffs with machined aluminum bulkheads. ie: take two standoffs that are in a line, and replace them with a single aluminum bulkhead with screwholes for both sets of screws. Must stiffer than two discrete standoffs.
 

B

BIGTACO

Member
Wow that statement makes me feel great! I cannot wait to get a real camera to try out.

In regards to the stiffness issue. Your idea and jes111's are both good. I am currently thinking through all the options. Another needed change that will help is clocking the servo down instead of sideways. This way I am not cutting a huge spot out of my structure in the center.

R_Lefebvre said:
Just in case it's not clear, what I'm saying is I think the gimbal could be equal or better to the top gimbals out there, save the Zenmuse. It's just hard to say for sure until you try smooth flying with a decent camera.

If your concern on the roll is that the vertical member of the gimbal is twisting about the Z axis, I think it could be stiffened up simply and easily by replacing a few of the aluminum standoffs with machined aluminum bulkheads. ie: take two standoffs that are in a line, and replace them with a single aluminum bulkhead with screwholes for both sets of screws. Must stiffer than two discrete standoffs.
Click to expand...
 



You must log in or register to reply here.
Top