For the past several months we have been working closely with an infrastructure inspection firm to develop a multirotor system for them. There were many goals, but COTS parts and cost were major drivers. We decided to use a DJI S1000 frame but selected the 3DR Pixhawk controller. We went with the Pixhawk for a variety of reasons, but flexibility was a primary consideration. The ability to plan complex missions, geotag photos, and a variety of flight modes ultimately sealed the deal. On the gimbal side we ended up using the brand new DJI Ronin-M. Equipped with remotely controlled zoom, the Sony AX100 provides crisp 4K video as well as 20MP still photos (even while rolling video). Mounting the Ronin was no small feat, requiring our dedicated fabricator to spend a few late nights in the shop. On the downlink side, we used a Amimon Connex, sending 1080p footage at 60 FPS. A Futaba 14SG provides flight control as well as camera control.
I must say, the S1000 is an impressive frame. The performance with the Pixhawk running Arducopter 3.2 is staggering, especially when compared to the DJI A2 that would normally occupy this frame. Youtube videos coming shortly!
Comments
I swapped the 2 red arms, be sure #1 = CW, the rest fall into place.
S1000 frame motor # ---> Pixhawk M out#
1 1
2 5
3 7
4 6
5 2
6 4
7 8
8 3
Matt,
How did you connect the motors to the pixhawk? I'm currently at this point in my retrofit and just want to make sure I get everything right the first time. I'm thinking that I just need to swap motor connectors on the S1000 frame map to the Pixhawk as shown in the two images below and swap CW props with CCW and vise-versa. Any pictures of your setup would help greatly!
Regards,
Jordan
Thanks a lot Matt. Yep, i have to read a lot yet but ... well ..... i need time .. hehe
We get about 1 minute flight time per amp hour of battery capacity. So a 16 amp hour battery will fly for around 15-16 minutes all said and done. We keep flight times to 9-10 minutes so as to comply with the terms set forth by the FAA in the COA/333 waiver.
I can post a screen shot of the PIDs tomorrow, but they will not work well with other setups. In fact, bird 1 and bird 2 needed slightly different tuning in order to get smooth video at high zoom levels. When adding the gimbal, I needed to reduce the rate roll and rate pitch I and D terms significantly below the auto tuned gains. This gives up a small amount of ultimate control authority but is always a compromise between video stability and airframe performance. Long story short, anyone building something like this needs to understand how tuning works in order to get good results in the wind.
How are you? i would like to know how many figlth time you get with that setup? and also if you guys can share your PID settings would be great too !! i know auto tune is a great feature but is my first time with Pixhawk so i so not feel too confident with it ....
I've found my s900 flies ok without a gimbal, but with one it needed very different PIDs to get it to fly well. Even then it is quite picky about which gimbals it will shake or not.
Problems with the performance?
Imho is only question of good tuning, that hardly is the same as it can depend on many factors.
With "APM: Copter V3.3 RC4" you can have significantly higher performance, must surely you must insulate the board from abnormal vibrations, this is imperative, then you can play with some filters.
I have a pixhawk controlled X8 that i nearly bought a Zenmuse z15 gimbal for, second hand mind you. However i understand that the Zenmuse is designed to work with that companys flight controllers only.
Can any of you knowledgeable guys provide any guidance on whether this is accurate and/or point we towards any info that would make this practical.
Thanks in advance
keith
I used auto tune with a ~3kg ballast firmly mounted in place of the gimbal. After tuning it was slightly over-tuned and I had to back off the rate I and D terms as well as stabilize P on roll and pitch. With the auto tuned gains it flies like it is on rails (like Marco's videos) but has a high frequency ring (looks terrible in the logs, didn't even try video). I reduced the max pitch angle a bit. I had to reduce the gains on the yaw axis probably due to the DJI frame's angled motors giving good yaw authority. I think it could be even better if I spend more time yaw tuning, but with 3.3 around the corner....
I have an S900 sitting in the box and plan to build it up with a Pixhawk as well over the long weekend. I will post the results for sure.
-
1
-
2
-
3
-
4
of 4 Next