131 replies to this topic

[peabody124]

So I'm creating this thread to try and start an open design (or set) for a camera gimbal that can be made on a reasonably accurate CNC machine and performs well (fine movement resolution, low vibration, variety of weights or scales well). This everything with OP let's shoot for top quality at a hobbiest cost

Ideally we would have a set of CAD files in whatever is a common format (not my domain) as well as the BOM (bill of materials) for the miscellaneous hardware that is required with sources for them. Ideally it could be cut out of something like G10.

This thread is a continuation from here

Just throwing out some designs for starters I'm partial to something like these
http://www.rcgroups.com/forums/showthread.php?t=1488264
http://www.droidworx.com.au/shop/products/camera-mounts/64-pro-mount-miniav130
http://photoshipone.com/products/3xpro/

But it would be great to get input here from people who have used a variety of form factors.

[StarRider]

Hi,

Are we aiming for 2 or 3 axis ? Is Yaw needed ?

Regards,
PA

Edited by StarRider, 08 September 2011 - 01:47 AM.

[peabody124]

Are we aiming for 2 or 3 axis ? Is Yaw needed ?

It would be good IMO if yaw could be added on. I almost imagine something nice and cheap like mounting a heli block for holding the main shaft onto it with a gear to a servo. That's partly my selfish interest in getting a 3-axis gimbal to work on camera stabilization code though

[jes1111]

IMO the AVerticalView type would be easier to make, and (according to someone on multirotorforums.com who's tried them all) it performs better than the expensive PS1/PH types. Going direct drive takes a lot of the complexity out of it. I'd be interested to hear some specifics from Rusty on the disadvantages.

[Ramboky!]

"Camera Lock" brings to mind fixing the camera at a particular attitude. A better name (in my opinion) would be "Target" or "Point of Interest".

Names aside, it's an excellent feature opening up the possibility of a range of awesome shots. Looking forward to hiding the bill from my wife when the Pro is available

I'm currently looking into a number of gimbals (the DroidWorx one particularly) so will follow this with interest.

[joecnc2006]

I think Rusty stated that direct drive always had issues no matter what servos he tried but I do agree, its simple and elegant if we can get it to work.

For me, 3 Axis is really required, with the pro we are planning an advanced feature called "Camera Lock", this enables an operator to click on the map and position the camera, it will then lock on that position no matter what. The position the camera points can be set using the map or hopefully by even clicking on the video window, so you can lock it on to a target you are looking at.

Very good point.

People can post sketches and we can discuss the pros and cons of each, the final goal would produce a downloadable DXF/PDF files in the wiki with a BOM (Build of materials).

[StarRider]

IMO the AVerticalView type would be easier to make, and (according to someone on multirotorforums.com who's tried them all) it performs better than the expensive PS1/PH types. Going direct drive takes a lot of the complexity out of it. I'd be interested to hear some specifics from Rusty on the disadvantages.

Hi,

I don't know, but looking at the AverticalView I think that the roll pivot will have some problems dealing with heavy cameras, all the weight is suported in the horizontal by a single swivel.

PA

[peabody124]

I don't know, but looking at the AverticalView I think that the roll pivot will have some problems dealing with heavy cameras, all the weight is suported in the horizontal by a single swivel.

And certainly with my little gopro mount where it's directly coming off the servo that's seems to pick up a lot of vibration. If it were on a decent bearing axis it might work... but then it's not direct drive.

[joecnc2006]

For belts and gears here is source i have used before for cnc machines.

http://www.econobelt.com/

[Rusty]

I'm officially declaring this cruel and unusual punishment

Rusty (nah, nah, nah, I'm not listening....nah, nah, nah...)

[peabody124]

I'm officially declaring this cruel and unusual punishment

Rusty (nah, nah, nah, I'm not listening....nah, nah, nah...)

LOL. I'm gonna send you pics of one of these mounted to your quad. In fact I'm gonna fly it in front of a mirror to taunt you!

Updated logo for you:
 Screen Shot 2011-09-07 at 11.22.24 PM.jpg   34.36K   50 downloads

[Mat Wellington]

Avertical View, tried - belt drive - not happy with it , sloppy and jerky.

The best so far seems to be the "bell" style.

will dig up some photos of one that i have at the moment

[G-unit]

This should be good..

[SeismicCWave]

>>IMO the AVerticalView type would be easier to make,<<

Absolutely! You are talking about a shaft pivoting on bearings. A much easier proposition than CNC some G10 plates and expect the edges to be perfectly smooth to ride on some bearings. Too much linear movement on the PS1 design. I have been slaving on a way to get around that issue. The simplest is to do like A Vertical View.

[SeismicCWave]

>>I don't know, but looking at the AverticalView I think that the roll pivot will have some problems dealing with heavy cameras, all the weight is suported in the horizontal by a single swivel.<<

You can increase the depth of the support and length of the shaft. I have also thought about counter balancing the camera weight with the battery pack on the other end of the shaft. Two birds with one stone. You increase the mass so it is easier to isolate vibration and counter balance the weight of the camera so the shaft support is not subjected to a "tilting" load as much.

[muralha]

Let start with the requirements:

• 3 axis ?
• is it aim mainly for multi rotors and helicopters or also for planes ?
• maximum rotation of each axis ?
• maximum camera size to be used ? (it will scale, but let's aim at some camera size...)
• is it mainly for video or photo or both ?
• should it rotate for "portrait mode" or should it have vertical holders ?
• should it have space for 2 cameras, for 3D or stereo video/photography ?
• does it need a trigger ? (servo or cable/electronics)
• does it need zoom ? (servo or cable/electronics)
• does it need a wireless comms ?
• does it need a video feed ?
• does it need a vibration absorber "plate" ?
• is it going to use a dedicated CC board for stabilization or a OP Pro future gimbal cable ?
• include the landing gear outside the gimbal ?
• should it use direct drive or simples gears or pulleys or gear systems (planetary or worm drive or harmonic drive) ?
• does it need a cover for extra protection ?
• ...

[dolby]

This is the new hotness in camera gimbals (simple and strong)...

http://www.quadrocop...-camera-gimbal/

Notice how large the wheels are. I'm no expert but maybe that is important for nice fluid movement?

Edited by dolby, 08 September 2011 - 10:04 AM.

[SeismicCWave]

>>is it aim mainly for multi rotors and helicopters or also for planes ?<<

I vote for multi rotor only. Single main rotor helicopter can haul up more payload so the camera gimbal can be larger. There are quite a few larger gimbals on the market but very few smaller ones. Fixed wings have yet another set of requirements.

>>maximum rotation of each axis ?<<

I will take how ever many degrees available. 15 degrees is better than none if it is easy to make. Of course the more the better but we have to understand servo rotation contraint. I do not like using a continuous servo so the servo rotation will be limited. Once geared down with gear or belt we loose servo rotation even more. Even if the servo rotates 180 degrees after a 4 to 1 gear reduction we will only get 45 degrees and that is for both ways on roll. Tilt may have more down than up.

>>maximum camera size to be used ?<<

I vote for a camera about the size of the Panasonic GH2 which is around 700 to 750 grams and 5 inch wide. Maybe the Canon T2i at the max. I know most poster on RCG always ask for something to lift up the Canon 5D Mk2 but not that many of them are serious. If I have to lift up a 5D Mk2 I will use the Maxi Joker instead of a quad. Yes I say quad because we cannot stabilize the camera with a hex yet.

>>is it mainly for video or photo or both ?<<

If there is a camera gimbal it will do both. Video is a lot more difficult to do well than stills. As long as there is enough light any one can crank the shutter speed up to 1/1000th of a second and get nice clear images. Video takes a lot more everything.

>>should it rotate for "portrait mode" or should it have vertical holders ?<<

I vote for KISS. If it starts to get complicated we are adding weight and complexity. I much rather see something simple that is popular than something complex that is a pipe dream.

>>should it have space for 2 cameras, for 3D or stereo video/photography ?[*]does it need a trigger ? (servo or cable/electronics)[*]does it need zoom ? (servo or cable/electronics)[*]does it need a wireless comms ?[*]does it need a video feed ?[*]does it need a vibration absorber "plate" ?[*]<<

All that will add complexity. Next thing you know the gimbal will be just like those big boys offered by Droidworx and such. Then we are no better.

>>is it going to use a dedicated CC board for stabilization or a OP Pro future gimbal cable ?<<

That's an interesting thought. I have been thinking about that. The only time that is necessary will be because the main CC board is flying more than four motors. Maybe the Pro will make that a moot point.

>>include the landing gear outside the gimbal<<

If the design is similar to the PS1 roll mechanism then yes the landing gear is part of the gimbal because it holds the bearings for the roll plates to ride on. Most likely the overall weight will be lighter if the landing is part of the design.

>>should it use direct drive or simples gears or pulleys or gear systems<<

It will have some sort of reduction to smooth out the movement. Gear or belt will be fine as long as they are adjustable.

>>(planetary or worm drive or harmonic drive) ?[*]does it need a cover for extra protection ?<<

That will add complexity again.

The most important point to me is weight. A heavy multi rotor simply doesn't fly as nice as a lighter machine. So the camera gimbal will make the machine fly worse than without the gimbal. If we can make the gimbal as light as possible that will be better. Hopefully most flyer involve in AP that use the gimbal will have enough flying skill to not suffer too many hard landings.

[SeismicCWave]

I have a dream. This dream will work if the Copter Control camera stabilization function can perform mixing.

Here is my dream of a simple, light weight, no frills camera gimbal.

Imagine a camera mounting plate with two short edges to make a U shape when view from the front. There is another U shape when looking down. So the first U shape is to hold the camera. The second U shape is to create two tilt pivot points for the first U shape. The second U shape will have a pivot point in the back to give the gimbal the roll pivot. So far very much like the one from A Vertical View. Here is the difference. The servos are not mounted in the usual place. What if the servos are mounted on top with a pushrod coming down from one of each servo. The attachment point will be the two front corners of the first U shape. When the servos both move their perspective pushrods up and down you get a tilt on the camera mount. When the servos move one up and one down you get roll. So the trick here is to have the signal coming out of the CC board mixed like an elevon.

I have tried it with an electronic V tail mixer and two gyros. It works. The disadvantage is the very limited gimbal travel. However if there is just enough travel to take out the bumps and wobbles from a multi rotor I am satisfy.