Here is the basic setup for my Quad.
1200 mm motor to motor frame.
Pixhawk Flight controller.
Castle Creations 50A ESC’s
6s 15C Lipo Battery
AttoPilot battery monitor
U8 Tiger Motors
28" Tiger Props
Taranis 9XD Radio
Minim OSD with a 5.8 ghz video transmitter from 3D Robotics
Telemetry module from 3D Robotics
12v/5v Dual BEC
I was flying pretty well in Stabilize and Alt Hold prior to a rebuild. I rebuilt so it would be easier to transport, I was suffering from high levels of vibration affecting the flight controller, and to improve flight stability in Loiter mode. My goal to reduce the vibration readings on the .bin logs to improve the ability to fly in loiter mode I felt could be accomplished at the motors and at the flight controller. Prior the the rebuild Loiter in inconsistent conditions the Quad would become unstable and rock out of control. I have been working with others to find solutions to provide stable flight characteristics in calm to windy conditions for these large platforms.
The items I changed were hard motor mounts to Tarot Anti Vibration motor mounts. I added a 3D printed Pixhawk anti vibration mount for the flight controller. The frame I changed from the V2 PRO QUAD/X8 frame with 25mm CARBON arms to the FOLDABLE PRO QUAD frame with 25mm tubes from quadframe.us/ . I added another main section plate and aluminum standoffs for a battery bay to mount dual 16000 mah 15C batteries.
Now that the rebuild is complete I went though all of the same procedures of calibrations to be flight ready. Once complete I connected it to my Galaxy Note 4 using Droid Planner 2. I noticed that I had some problems with Pre alarm Velicity alarms High GPS hdop issues at first. I re-calibrated and removed the Velocity alarm. After 5 minutes the High GPS hdop came down to 2.5 and I was able to Arm the motors. I launched the Quad and noticed some twitching while in Stabilize. I was able to land and Disarm. Now I picked up the quad and held it over my head and Armed the Quad. I powered up to hover wile still maintaining hold of the Quad. The twitching was less pronounced. I did this again with the same results. I Disarmed and set the Quad on the ground. I armed the Quad in Stabilize and launched. for a short time I flew but the twitch came back and this time rocked out of control and flipped.
No all this said Please review my logs and help me figure out what I’ve done wrong. I will be attaching pictures of the current build to help visualize the project.
Thanks in advance
Will
Hi Will,
So this is not such an easy problem. What you are doing with these huge frames is more advanced than most people use, and it comes with quite a few challenges. Fact is, very few people are successfully flying these very large multirotors.
First, problem, is vibrations, which you are having a problem with.
The log you posted is after the rebuild? What was it like before? Because in this log, the vibes are definitely too high, and the system will struggle. You have vibes about double what we recommend as being the maximum. Can you show pictures of the vibration damping mount?
Looking at your frame on the vendor’s website, I’m not really impressed with the design, and this may be part of the problem. It does not look very strong for such a large frame. And the grade of G10 used might not be great. I bought a heli frame made from G10 that looks just like that, and it’s very soft.
Using 28" props, you need everything to be strong, but even then, this is going to be difficult. Those are very big props turning very slow. I can try to help you, but this really is an engineering project unto itself.
Now, I notice you are not using the EKF (AHRS_EKF_USE) but you may want to consider it. I use it on everything, including my helicopters, and it helps. It’s a new feature, but most of the devs are using it and we haven’t had any problems lately.
Now, looking at the Attitude control, it looks pretty bad. It’s really not in control at all. It could actually be a problem with vibrations, hard to say for sure. But it could also be badly tuned. Did you do an autotune? It sort of looks like you did. And if you did, it doesn’t look like it went very well. We’ll probably have to work with Leonard and get him your autotune logs to look at.
Your Rate Roll numbers are quite a bit higher than your Rate Pitch numbers, and it probably shouldn’t be that way. Neither axis is really under control. I’ve helped another person tune a very large octocopter, and it turned out it needed very high P terms. But your Rate Roll P terms are already fairly high. In fact, in the event where it flipped, your motors were actually hitting the limits. I’m not sure if this caused the crash, because you don’t have enough power to stabilize it, or if it was simply caused by too much P-term.
There’s sort of too much going on here, I don’t know where to start. High vibrations, Tuning, Motor Power.
Your GPS does look good, once if finally gets a lock. It gets a solid lock, and stays there, it just takes a while. This may be due to the FrSky X8R receiver you are probably using. There are some reports that it emits a lot of noise which disrupts the GPS signal. I’ve done some testing myself, and it does seem to take longer than normal to get a lock, but once it does, it’s good.
You should try moving the Rx as far away from the GPS as you can. If you could put it on a landing leg, that would be great.
Rob,
k you for taking time to help. I know that I have a mess n my hands right now. I have been chasing my tail on this project. I am using this frame for R&D for a future frame I am designing. The vibration totals before the rebuild we double that I have now. The crazy thing was that in Stabilize and Alt Hold, it flew pretty well. Loiter was my problem. With the slightest wind it would start rocking and I almost lost it a few times. I thought that I could address the vibration problem in two ways at the controller and at the motor. I tried to change out the motor mounts fro the Tarot Anti-vibration mounts but this caused further instability issues. There was too much movement at the motors which caused the twitching to turn into a problem quick. As far as the mount for the Pixhawk board. I used a Omnimac Anti Vibration Mount for the 3DR Pixhawk Flight Controller found on ebay. In between the mount and the Pixhawk I used the foam provided by 3D robotics. I cut the foam into 1/2" squares an applied them to the four corners of the Pixhawk. I attached pictures of the current build in this post. You will notice that the GPS is knocked off as a result of the crash. I will try the EKF but truly I don’t understand it yet. I guess I am still a bit over my head in this project. The attitude problem you mentioned was the twitching that turned into rocking. Again this was with the Tarot anti-vibration mounts. I have since changes them to solid mounts. The anti-vibration mounts i found were http://www.ebay.com/itm/like/151479621062?lpid=82 . I am using the same mount but i have removed the rubber gromments.
To be honest I am nervous about PID settings because I do not know enough about them and with this large of a copter it can get dangerous quick if it goes wrong. That is why I am thankful for the help. I am hoping that this will be a stepping stone to my final build made of all carbon fiber that is all weather resistant. Please look at the pictures closely the rx/OSD is mounted one platform below the Pixhawk. If you feel that something could be arranged better I will try anything.
Lastly i wanted to show you a video of a flight after I changed the motor mounts out. Its better but if you look closely you can still see the twitching. Also I about lost it at takeoff. [youtube]https://www.youtube.com/watch?v=jvFi4OgeQsI&list=UUlC9K8OSdznj-PVhMR6RYCw[/youtube]
Thanks again
Boy, where to start, this is almost overwhelming. It’s such a large, complicated machine…
Where in that video is the twitching/rocking? It looks pretty decent from what I can see before and during the autotune. But then on your second flight starting at about 10 minutes, I see it twitching pretty badly, not sure why. Was that on the autotune settings? You should be working with your GCS to see what the PIDs are before and after the autotune. It’s possible it didn’t save the autotune settings for some reason. I’ve seen that before.
It’s hard to comment on the vibration system, I can’t quite make it out. I get the basic idea, with the 3D printed platform, and then using the foam blocks. It’s hard to say if this is a good idea or not. I’d have to play with it a bit. I’ve never done a dual-layer system before. A few tips though. You want to minimize the amount of force that the cables can exert on the Pixhawk. So, minimize the number of wires, make sure they are flexible. Make sure none of them are tight. Do not “cable” them together with zip ties near the Pixhawk. They should all be as loose as possible. You want about 1" of loose, floppy, flexy wire between the frame and the Pixhawk. And mass is your friend. Some of my most effective systems involve adding mass to the bottom of the Pixhawk. A large aluminum plate, or even lead. This drives down the natural frequency of the damped system, and also reduces the amount of vibration the wires can transmit.
Are the props balanced?
You also need to make the frame as stiff as possible. What you have there looks pretty flimsy. For example, the way you have the batteries mounted to a plate which is suspended underneath the frame by a couple of really long aluminum standoffs… that’s not great. I imagine those batteries are squirming around under there. I either build some stronger bulkheads to replace those standoffs, or just strap the batteries very tightly to the bottom of the frame, with velcro used to help secure them.
I have a tendency to overbuild things, and they can be heavy, but they also fly well, because everything is strong and stiff.
One thing you should do, is try to get a log from an autotune session, and then we’ll try to have Leonard look at it to see what he thinks.
Two big issues I see. First you need a lot stiffer frame. Second, big quads are inherently unstable! I’m running a big quad frame (1300mm) with U8pro motors in an X8 configuration. I gave up on the quad. Even when it was tuned and flying well, there was always to possibility of a flight with a good gust of wind. I went through a whole year of large quad trials, just cut you losses and buy 4 more motors : ) BTW, the D130 frame is incredibly stiff!
Maybe three issues… Big motors and props means big vibration. I use a Vortex dynamic balancer, really smooths everything out. Of course if you have flex in the arms, you still get vibration. 30mm arms are substantially stiffer than 25’s. Here’s a pic of my D130.
So an Octo in that size is inherently more stable, even with the same size props and motors? I have seen a few successful very large Octos. I am curious if they can ever really be stable however. I have seen a lot built, but they are rarely seen in operation.
Ah, that’s the Devourer from Foxtech? I’ve wondered how stiff those folding arm mounts are.
I’m actually waiting delivery of a new Tarot X8 frame (unfortunate name they chose, it’s actually a conventional Octo) with folding arm mounts like that. Not nearly as big as what you guys are using, just 15" props.
These vibration problems can be solved. I have a Pixhawk running on a large gas powered helicopter that shakes like hell. I purposefully chose a large unbalanced engine to test for worst-case scenario.
The Devourer looks kind a spindly in the photo, but the arms and pivots are crazy stiff. Surprised it doesn’t weigh a lot more lol. How’s you gas heli working with the pixhawk? There a new lidar unit that’s pushing 1km range, but at 8kg, not really appropriate for a multi. A big gas heli would be perfect.
I had a crash on the gasser early on, because the frame vibrations broke the GPS unit. I’ve now soft mounted all the electronics, and first tests are good but now I have magnetometer problems. Still sorting that out.
8kg is a fair amount, but I’m not sure how well this one would lift that. It’s actually going to be power-limited more than anything, as this engine only has about 2-2.5 hp. With a more powerful engine, this airframe is probably capable of it though. 5kg should be fine.
8kg is a lot. Been out of heli stuff for a while so my product knowledge isn’t so good. I remember the Bergan Gas Trainer was a good heavy lift unit. Have to check it out. Know of anything new?
No, there’s really not that much new to report on. Nothing significant anyway. I’m not a fan of Bergen. Miniature Aircraft has closed. I’m not a fan of Century. There’s RJX, but I haven’t seen their helis in person. I’m not impressed with Airstar either. My Helix 800 is based on Trex 700 mechanics, with aftermarket frames. The frames are not very good. Eventually I’ll probably make my own frame, but using Trex mechanics.
There have been a number of professional heli UAV companies, but it seems most have gone under. IMO, they are simply too expensive, largely because of the custom developed electronics.
That was actually a Vario Benzin Trainer, not Bergen. It would be under powered anyhow. You’re right, there isn’t a lot out there.
The twitching started in the beginning of the video (not very noticable). Since it was minimal I decided to try to Autotune while it was behaving. Once the Autotune was complete i disarmed it to save the settings. I run the GS on all flights so I will pull the log tomorrow and upload it. The second flight was consistant to the twitching I have seen before. Not wanting to chance jt I landed and parked it. I like you idea of adding weight to the Pix. I have read that the lack of mass challenges vibration dampening. It was for this reason I used the dual system of foam and the Pix mount. My desire to be tidy could have hurt me with the cables. I will check and adjust the setup to make sure there is slack on all cables.
As far as the props I have checked them but in a crude manner. I noticed from the factory there are drill holes on the bottom side of the props, which are sealed over. I take it this was there way of balancing them. At $170 each I am nervous to drill additional holes. Can you offer suggestions?
For the frame i can only go off of what i have seen as far as stiffness. There is no vertical movement in the arms in relation to the frame. Once i lock down the bolts the arms are solid hirizontaly. I do notice with some force I can twist the arms though. I can video this and submit. Again, I will pull the log of the Autotune in the YouTube video and submit it for review that way there is an additional form of reference. I know that I am trying to set up a gocart cart when my finished project will be a Corvette, but this frame was economical for R&D. My final frame will be constructed of carbon fiber and simular to the MD 1000 by Microdrones. I have very interested parties who want such a build for service inspection purposes. Your help is appreciated. I will post the logs tomorrow.
Thanks
Will
What you describe about the blades indicates they are factory balanced. That is how high end helicopter blades are done. They measure the imbalance, and then drill holes, insert led, then seal the holes over with epoxy. At $170/each, I’d hope they did that, because I pay the same for much larger heli blades that have this done.
It would be good to check the balance though, but this does require good tools, especially for blades of that size.
As promised here is the log file for the autotune the other day. Any help would be appreciated. In the video you can see me run the autotune, land disarm, arm and fly. Serious twitching after. Im stumped.
Thanks
Will
Hi,
Seeing i’m currently working with the same frame than yours (but in x4 config), i wonder if you have experienced stiffness problems with the attach landing gear-main structure.
Thank you very much
Pablo
Seeing your log, though you’ve done an althold flight moving around in X and Y, i think your throttle response is quite irregular, in my opinion due to bad X,Y and Z accels. I had the same problem and still testing, but curiously the ±3 m/s2 limit doesn’t seem to work properly in these kind of big propellers quads
@pabloneira Yes there is a stiffness problem with this frame Using these motors and props. I have replaced the booms with 2mm thick booms that are 600mm long. I mainly choose this frame for the transportation ease. It is only a test frame to get all of my gear working. I am designing a custom frame that will have removable booms. I have since changed some of the parameters and have solved my vibration issues. I had a 58 minute flight last week where I maintained vibrations between 1-4 g’s. I will try to load up the log for the flight, but due to the length of the flight it has been giving me trouble processing it.
Will
@wuzfuzwill that’s great, and i’d like to see your last log.
I remarked placing batteries on top it helped too to minimize vibrations and in general adding mass in the center plate, if that can help you. Could you please share your PIDs in stabilize?
many thanks
Pablo