Hello,
In a search to film with a HD camera I made different tentatives with all kinds of materials and devices to avoid the famous "Jello" effect. I tried moongel, I tried elastics with foam, I tried rubber grommets...everything lead to Jello!
I though I would be unable to get rid of it until I got the following idea, inspired from military strategy: not to try to block an ennemy wave at first defence. Rather, build successive defences in cascade : a bit of the attacking wave goes through the first line of defence to the second line; a bit of the bit of the wave goes through the third line, a bit of the bit of the bit of the wave dies on the third line, etc. These cascding lines of defence will utimately get rid of most of the vibrations to the camera. That was the idea, so how to apply this on a quadcopter ?
First decision was to attach the camera gimbal on the bottom plate because I think intuitively that is the best place to start with less vibrations (alternative was to use the sonar mount between two arms, but then I am way to close to the vibrations sources).
First line of defence, the bottom plate
Thus the bottom plate had to be the first anti-vibe line of defence. As shown in the picture below I decoupled this bottom plate from the rest of the frame with 8 M3x20 rubber dampeners (instead of screws):
Second line of defence, transversal booms to hold the camera gimbal:
I purchased a one meter U shape aluminium boom (10mm x 10mm) that I cut in two pieces of the length of the bottom plate plus a few extra inches to hold a camera plate (will be shown in next picture). These two booms are attached in parallel to each other and screwed on the bottom plate with two M3x10 rubber dampeners.
Third line of defence, the camera plate:
The camera gimbal is suspended on a aluminium plate using the two parallel booms. This aluminium plate is attached to the booms with four M3x15 rubber dampeners as shown below,
The extra trick is to keep an extra rigid link between the camera gimbal and the frame. For that I reused an idea that was posted previously on this forum and bought at servocity the famous servo blocks. This is not ideal to avoid jitter of servos in mouvement but this is the best I found until now. Because now the best would be brushless gimbal and that shall be my next project (waiting to do the investment, it is not cheap).
So with all that I did a quick test flight that you can see in this youtube video. This video was not corrected nor transformed. You will see that except mouvements of the gimbal (i flew in stabilize with lots of pitch/roll corrections), the video is absolutely without any Jello.
Mission accomplished.
By the way, the music of the video is a credit for Jake Wells who inspired me.
CU,
Hugues
Comments
Neill, they have independent screw holes on both ends and nothing but rubber in the middle. There's no bolt going all the way through, which is probably what you're thinking of.
I don't think those dampeners are doing anything at all... As far as I am aware they only function in compression, not tension....
The video looks very nice. I use a similar setup on my hexacopter from Rusty (AGL Hobbies) and it works very well.
http://www.shop.aglhobbiesllc.com/Camera-Mounts/UAP-8-Damper-Kit.html
It's also worth noting that unbalanced props can cause the "jello" on CMOS cameras.
Excellent approach, it does seem like no jello at all.
Multi layer vibration damping also works really well for the flight controller boards.
Problem now is to get stabilization up to snuff, in fact, the current stabilization algorithm updates at 50hz.
Plenty fast except servos, especially digital servos are a lot faster which means they inevitably update with a lot of starts and stops hence jerkiness (which is why often slow mushy analog servos actually work better.
The low cog brushless gimbals, unfortunately are also really a band aid over this problem substituting slow analog approximation for a truly adequate full speed update. (They won't be jerky, but they could be mushy).
The proper solution is for the digital servos to be updated at full servo speed and smoothing to be performed in firmware.
Unfortunately the APM may not have the bandwidth or memory capacity for really fast servo update or smoothing - - but the PX4 does.
Those really nice rigid and tight ball bearing servo city gimbals lend themselves to this use really nicely, although the offsets from gimbal center that you have will show up in slightly imprecise stabilization maintenance.
My friend Oliver is the first one to have built a gimbal out of them and that is probably the blog you reference.
I believe the addition of a few brackets to place the optical plane at gimbal center could be useful and that is the bracket I will be making.
Thanks, Hugues!
Nice job! I've been contemplating how I was going to mount my camera, and you've inspired me. I have a question, though: what brand of pan/tilt mechanism is that? It's *nice* and I want one!