Hey everyone! This is our first post to DIY Drones, though we've been following this community for some time and have found the depth of knowledge here amazing! For the past few months we've been working on a unique GoPro gimbal design for the 3DR Hexa. We are now at a point at which we feel like we've made enough progress to share some of the results with this community and hopefully get some feedback.
About Us:
We are two guys, Jeff and David, with active careers in the broadcast industry. In addition to being enthusiasts of the UAV hobby, we see enormous potential for cost effective video production solutions. With news that the FAA will soon legislate commercial use, we've started investigating potential video production applications with our 3DR hexacopter. Read more about us here...
As many of you have experienced, we quickly ran headlong into the problem of vibration induced video distortion (better known as jello)! We tried several inexpensive gimbals on the market without much success. Obtaining a camera view that was unobstructed by the copter frame was also problematic.
So we decided to go DIY!
Gimbal Requirements:
Once we decided to design our own gimbal, we set down the main requirements that our prototype should meet.
Vibration isolation: Since our primary application will be production quality video, our main objective is to eliminate as much vibration related video distortion as possible. The video should not require post-processing to remove distortion.
An unobstructed view: The camera’s field of view needs to be completely unobstructed by the helicopter frame. No paying client will tolerate a landing gear or props in the shot, distracting from his or her product.
Plug and Play: Our gimbal system should directly attach to the 3DR frame without the need for additional drilling or modifications.
No Tuning: The gimbal should be balanced and centered relative to the 3DR Hexa frame, resulting in smooth flight without PID adjustments.
Affordable: Through the use of select materials and efficient design, we hope to achieve a reasonable production cost relative to the cost of the 3DR Hexa Kit.
Aesthetic Design: The gimbal needs to be well designed and professional looking. Something that looks cobbled together will not inspire client confidence in the quality of the final product.
The Story So Far...
With these requirements in mind, we set out upon the rocky road of development!
As any good project should, we started with a discovery and research phase where we investigated many different options and approaches. As mentioned above, we started by looking at basic gimbals that were already available. In all cases, we found that in order to achieve an unobstructed view, the gimbal needed to be mounted out on the arms of the hexacopter. Doing so would imbalance the copter, requiring it to be tuned at the software level. This placement also introduced more vibration into the gimbal.
We decided to tackle the vibration issue first, and then design a form that would meet our requirements. As we began researching ways to dampen vibration, we discovered that the three main factors involved are:
1) the weight of the supported object
2) the disturbing frequency (RPM)
3) rigidity of the structure isolated
We started out under the assumption that our gimbal would need to be heavy enough to create a static load within the deflection material, but not so heavy that it would overstress the material. So far we've found that it’s been difficult to find a sweet spot - one that is heavy enough to provide enough compression on the isolator, but not too heavy for the copter to lift.
At higher disturbing frequencies the required minimal thickness of the isolator can be reduced. Because the helicopter operates at variable disturbing frequencies, we determined that the thickness of our isolation material should be based upon the lowest disturbance RPM. Since our motors have a KV of 850 and operate at a maximum of 12 volts, the maximum RPM should be less than 10,200. Since hovering is usually achieved under 50% throttle, the low side of motor rotation for flight should be above 4000 revolutions per minute. For this reason we chose isolators with 5/16” thickness. This should provide us with isolation efficiency of 95% or better. More on the science of vibration reduction here: http://www.easyflex.in/pdff/latest/Vibration%20Isolation%20Theory.pdf
Once we had the vibration mounts, our goals, and the dimensions of the GoPro camera, we were ready to design our gimbal frame. We decided to go with an “H” frame that would be attached to the copters center and yet hold the GoPro carriage far enough in front of the copter to obtain an unobstructed view. Given the APMs native support of servos, we decided to steer away from brushless motors for now. We placed the roll servo in the center, and the pitch servo with additional ballast weight in the back to balance everything out.
Now that we had a form in mind, we needed to work out how to mount everything together. It quickly became apparent that the gimbal carriage would need to be firmly mounted, so that it did not add vibration, while being able to spin smoothly. The meant we would need ball bearing mountings. While looking for an affordable solution, Jeff came across ServoCity.com and found exactly what we were looking for.
Once the parts arrived in the mail, we bought some ⅛” X 1” aluminium strips and started cutting and drilling in our garages. We built the entire “H” frame and assembled all of the components. The gimbal had a very smooth movement on both the roll and pitch axis. The frame was rigid and solid, and protected the GoPro very well. It gave a clear and unobstructed view from the hexacopter
It just had one major problem, it was way too heavy! This gimbal rig was approaching 4.5 pounds and there was no way the hexacopter was going to be able to get off the ground, much less fly in the controlled and stable manner needed for video production.
We decided that we needed to reduce the weight as much as possible and this was going to require a high tech solution. David devised a calculator to help us determine optimal hole placement within the frame. Once we were comfortable with our layout we sent the design off for CNC routing.
How did it all work out? Check out the video and see for yourself. Please Note: there has been no post processing of the resulting GoPro video.
As you can see our first prototype is still too heavy. The good news is that our vibration isolation method seems to be effective based upon the limited amount of video we captured. That said, there is still work to be done! We've revised the design yet again, and our Mark II version parts should be arriving in the mail shortly.
Changes to the design include a much lighter chassis built from a thinner aluminum. After a few rigidity tests we believe we can reduce weight by using a 1/16” plate. We’ll also be removing as much metal as possible. The end result should look something like a skeleton relative to our current prototype. Another change will be the incorporation of the ability to modify the weight and thus the balance of the gimbal chassis. This will allow the gimbal to work with multiple GoPro versions and accessories. We will do this by adding several weighting compartments on the chassis. We are also planning on relocating the main battery to the rear of the gimbal to contribute to the weight needed for balance. This will further reduce the overall gimbal weight.
Here is the part where we need your help!
Since we only have one hexacopter for testing purposes, any recommendations regarding payload limitations would be appreciated.
- We are interested in how much weight you have been able to successfully fly with?
- What was the effect upon flight time?
- Which frame were you using?
- What motors, props, batteries, ESCs, etc. were used?
We are also interested in any additional methods you may have used for vibration isolation.
- What was effective and what wasn't?
We are very determined and optimistic that we can successfully create a working system with these goals in mind (though we might have a few productive failures along the way). If you’re interested in reading more about our journey, we will be posting regularly on DIY drones and you can also follow our progress on our blog here:
http://www.skyrisfx.com/mission-updates/
Cheers!
-Jeff and David
Comments
I have the same hex and my props all start and stop within a fraction of a second from each other. You tried swapping ESC's, have you tried to swap motors? If the problem follows the motor, then it's the motor. If the pattern stays the same, then it's somewhere in the electronics.
Stephen, thank you for your words of encouragement! We view this journey as one of discovery and believe in the value of trying new things and learning from mistakes.
In regard to the three props you mentioned, those motors are not running at that point, but rather they are spinning down. Those motors seem to have less resistance than the others. We notice this when we spin each motor by hand and compare. We've also done a great deal of ESC swapping with the same result. This leads us to believe that it's a variance in the motor construction? Thanks for noticing!
Jeff and David - I think the people wondering about the weight and reinventing the wheel are missing the point of DIY.
Off Topic, but in the landing, I noticed three props stopped and three still running for a second longer. Would this be an indication of needing the ESC's calibrated?
Thanks everybody for your great comments and constructive criticism! There’s a lot of insight here that’s given us plenty to think about. Please do keep it coming!
Since there was some consensus between commenters, we’re going to respond generally but also to some of the comments individually.
@Forrest Frantz: First off, dude you are awesome! Thank you for your very informative post. We really appreciate the level of detail and brutal honesty. This is the kind feedback we were hoping for!
You've pointed out the depressing reality of weight versus flight time, and the role that weight plays in vibration isolation. In our post we mentioned the difficulty of finding the sweet spot between weight and efficiency. We believe that it is possible to find an acceptable trade off. In such a situation, a prepared aerial photographer would likely bring multiple batteries to the shot site as well as a way to charge on the go.
We would be very interested in seeing the results of your Sorbothane testing. An isolation material that requires less weight to be effective would be great! Please do send us any blog links and info you think might be helpful to us! Many thanks!
Several of you pointed out that perhaps the pitch servo should be moved to the center of the H frame. This is a great idea and something worth exploring. However there is a potential snag that would need a work around. Consider a situation in which the copter pitches forward and yet the gimbal remains level. This opens up the possibility of the props colliding with the GoPro carriage and shattering mid air. A “Y” copter frame setup may be a potential solution for that kind of servo arrangement. Our goal however is to develop something for the more traditional quad and hex frame setup (for now).
In regard to the CMOS vs CCD debate - As John Arne Birkland mentioned, the majority of rugged sports cameras on the market are not built around global shutter technology. Also, when we explored CCD options we found that they had some quality and strange video artifacting issues, especially when the lense is exposed to a direct light source. We also wanted to try and solve a problem for the extremely large base of those who own GoPro cameras.
@John Maffetone: the overall weight of our last prototype weighed in just a hair over 3.06 lbs, resulting in an overall copter weight of 7.5 lbs. This is obviously not acceptable and we have redesigned the gimbal to be lighter and we are now using 1/16” aluminum. We are currently waiting for the newly machined parts to arrive in the mail.
@John Maffetone, and John Arne Birkland: Designing the gimbal to be an extension of the landing gear is a great idea and it makes sense that this approach could reduce overall weight. The extended landing gear we are currently using weighs about 1.2 oz not including hardware. The standard landing gear that comes with the 3DR hexa kit weighs 1.6 oz not including hardware. It seems marginal at first glance but every piece adds to the total weight.
There were some great suggestions in regard to materials. Weight is a primary concern however we have some additional important criteria to consider for our project. The material we use must be low cost and it must be aesthetically pleasing. So far aluminum seems to be the most cost effective material. Carbon fiber is much more expensive (often by a factor of four). Plywood would indeed be very light however it does not meet our aesthetic goals. We are very interested in the G10 material that Jordi mentioned.
@Silvio: Ha Ha! Yes. We jokingly refer to our last prototype as the “tank.” Our next version will be much lighter. The size and layout of this design is largely dictated by our goals of achieving an unobstructed camera view along with proper weight and balance for vibration isolation.
@Daniel Allen: You were astute to point out that the battery could be used for balance. This is exactly the approach we are taking with our next prototype. Our next version will also include weighting chambers. This will allow us to be much more flexible in our vibration testing and will also allow us to maintain compatibility with other GoPro versions and accessories.
As we mentioned our latest design revision parts are in the mail and should be arriving shortly. We will definitely take some of your suggestions into account when we design our third prototype iteration. We look forward to keeping you all posted on our progress. Thanks again for all of your comments so far!
Jeff and David
www.SkyrisFX.com
Good start. Have you considered using laser cut 3/16" plywood? It will be at least 1/2 the weight.
Its a good start to have the camera away from the props. But I think the way it tilts the camera is no good. IMO the tilt should happen at the base of the gimbal, just like the roll axis. This way the camera doesn't move as much. Plus I would use the battery as a counter weight instead of the extra bulk at the tail end. I think this design should be used more Glyder 600
2 Tons of steel and all for Gopro? :)
Great to see a nice educated discussion on vibration dampening/isolation. Rigidity and mass are important factors that many forget when trying to isolate vibration.
Also +1 for global shutter sensors (like CCD sensors), but sadly consumer/pro-sumer cameras with global shutter sensors and a size/weight range for copters are more or less non existent.
John Maffetone also makes a good point. If you build the gimbal into the landing gear and isolate them both, you get better utilization and "save" some weight that has to be included regardless.
Beautiful work. Nice machining.
Vibration is an interesting ghost. So let me be helpful first by passing on how weight penalizes flight time as you asked. Then the bad news become evident I'll tell you what you may not want to hear but will also offer a direction if you believe me.
Amp increase approximately for this class of multi-copters by about 0.002ish amps per gram per motor. So lets say the gimbal weighs 1 kg. Then the amp increase is 2 amps per motor, effectively more than doubling what it takes to hover an efficient multi-copter built for photography. So you have cut flight time in half. Being photographers, you know what that means.
Now even worse news. Call up any reputable isolator manufacturing company and tell them that you want to isolate a something less than a kilogram. They will tell you that it's not possible using existing science and materials but they will also tell you that if you can get the weight over 3ish kg then they might be able to help. They will tell you that the best passive way to reduce vibration to light objects is to simply add weight. So yes, your platform did reduce vibration a lot. If you don't believe this, take a video in hove using your gimbal and then take another video with same-weight tripod screwed into the tripod mount. Lift that and compare videos.
So, how can I be helpful after either putting you in a zone of disbelief, being upset with me (sorry ... but I wouldn't have taken my time with you if I wasn't impressed by your work), or just saying that Frantz guy is a nay sayer, or who knows, maybe your reaction was ... I was afraid of that?
So what can you do that would be really cool. Build you frame from really light materials like carbon braided tubes (all other types of carbon are too fragile). If you are interested I can lead you to the blog where super light weight and fast construction techniques using round tubing is discussed. Or replace the metal with glass fiber plate (that will cut the weight by a factor of four). Don't worry about vibration. The best way to solve vibration is to deal with the battering of prop wash in the design of the copter frame. Get rid of harmful prop wash and the problem is solved. I'm building a photo ship now that cuts down on vibration by more than a factor of a hundred by accepting the reality that isolation isn't where there is low hanging fruit on the vibration tree. It can point you to the blog on the construction of that ship if you like.
But, there is really good news. Ready?
The best way to get rid of jello is to not use a GoPro. GoPro uses a CMOS sensor like most cameras do. Most all the professional cameras use CMOS. My still/video camera is CMOS. But because CMOS scans line by line, with fast motion, by the time the scan goes from the top to the bottom of a single frame (very fast by the way), the lower scans are progressively offset from the upper and you get tilt and jello. A CCD sensor solves this as every pixel is shot at the same time on every frame. There are many videos showing this, so here is one. I'm looking for the right CCD camera for the photo-ship.
https://www.youtube.com/watch?v=BN6xw4b4nkw
My next focus after getting the Octa-V photoship tested is trying to find out if the lowest density Sorbothane (the best isolator material for lower weights) can somehow be used to isolate light objects. The way I test is is by putting an accelerometer on the source vibration (a motor and prop) and another accelerometer on the camera platform and just trying stuff (we call this try-storming versus brain-storming). The signals are fed to my oscilloscope where I can objectively see and quantify the impact on vibration.
Collaboration? That would depend on if our goals are similar. I'd say that if you think you two are smart enough to come up with a 200 - 300 gram gimbal, then I'd say yes, our goals are aligned.
I've given you a lot to process. Hope you will see this in a positive light. Best wishes and really cool design and concept.
@John, Rumors are that 3DRobotics might be adding a Y6 to their store soon... :)
http://diydrones.com/profiles/blogs/thirty-days-of-awesomeness-5-so...
-
1
-
2
-
3
of 3 Next