WKM in the wind - Your experiences - What could be improved

[BorisS]

Hi Guys,

one questions keeps on coming up and that is how does the WKM reacts in wind. Some are convinced that its at par with other FCs, some are experiencing troubles and getting hard regulations from their system which makes it less of charm when using for a AV platform.

There are several videos out there which make a statement but at the same time it is a hard one to distinguish what they are actually telling us.
We have this test video from DJI were the Octo is leaning against the fan output:

http://v.youku.com/v_show/id_XMzA2MTcyNzY0.html

or my test with the quad:

At the same time though this is only have the game since a constant side wind is a total different story than actual gust stronger or lighter impacting the frame. So a impacting variable that changes and is not constant. Although my quad reacts nicely in the fight against the wind, it is not that brilliant out in the field dealing with changing wind speeds.

We can see from several videos that the GPS hold and altitude hold is very good in harsh conditions:

Gunter:

Jules_B

Although impressive and probably not the conditions one would expect to get good AV footage from anyways. The good gps hold and the alt hold are probably not the tools one it looking for to get workable footage from in harsh conditions, considering that our gimbal and stabilization systems are mechanically or due to there gyros capabilities not at the state to compensate. It could be though that this is the approach DJI is heading towards with the gimbal from which he have seen ( still to be proven ) amazing results.

Thoughts on what their approach is seen in nature :

I would be happy if you guys join in and give you thoughts what could be improved or how the behavior of the WKM could be changed to get less harsh but more forgiving compensation. If you add videos to show your finding it would be great if you add raw videos not post anything. It is always very hard than to distinguish what really went on and actually brings us right back to the beginning if the footage is not raw.

Also join in with you findings on gain setting prop motor combos that have given you good results.

One thing i haven't seen up to now is the direct comparison of different FCs on the same setup on the same day thus exact same conditions, or even better right next to each other. My point here is not to bash one FC over the other but to learn and understand how conditions are delt with.

Thanks

Boris

 

[BorisS]

I think ken also has a couple interesting flight in wind i just cant find them at the moment. Would be cool if you add them ! Thanks Boris

 

[RTRyder]

I think ken also has a couple interesting flight in wind i just cant find them at the moment. Would be cool if you add them ! Thanks Boris

Boris, I'll should have some very interesting videos to show in a week or so. I'm currently in the process of setting up my CineStar6 with the MK electronics stack that was on my Droidworx AD6, once I finish I'm going to do a few video passes in different conditons, then change it back to DJI and do the same flights in (hopefully) the same conditons, tuned as best I can. Then we'll see how the two stack up against each other, which one really is better in the wind, and how to tune the DJI gains for best results.

It will be a fair amount of work to get all this done so don't expect to see it in a few days.

Stay tuned...

Ken

 

[BorisS]

Cool thanks, Once I have my hercules things solved out I will also but some MK stuff on also reporting what i found. Really want to get to the basis of things on this one !

Boris

 

[GGoodrum]

I have two points to add. First, we know now that the WK-M does not like light disc loading, so making sure a setup is not over-propped for a given AUW is important. With light disc loading, it is extremely hard to get the gains high enough to stop wobbling on descent, without going into oscillation. You might be able to make this better by playing around with the "advanced" settings, but it is far easier, in most cases, to simply drop down a prop size.

My second point is that with a properly "tuned" gimbal stabilization setup, it shouldn't matter how aggressive the platform movements are, to hold position. I also don't believe that it takes a $3k DJI wonder gimbal in order to achieve this level of accurate gimbal stabilization. All it takes is a robust gimbal design with fast servos and no more than about a 2-3:1 belt-drive reduction. Of course, you also need to make sure the gimbal gains are dialed in perfectly, and it would be nice if DJI would fix the damn wander/drift problem when the GPS is locked in. The point is, it can be done.

-- Gary

 

[RTRyder]

I have two points to add. First, we know now that the WK-M does not like light disc loading, so making sure a setup is not over-propped for a given AUW is important. With light disc loading, it is extremely hard to get the gains high enough to stop wobbling on descent, without going into oscillation. You might be able to make this better by playing around with the "advanced" settings, but it is far easier, in most cases, to simply drop down a prop size.

My second point is that with a properly "tuned" gimbal stabilization setup, it shouldn't matter how aggressive the platform movements are, to hold position. I also don't believe that it takes a $3k DJI wonder gimbal in order to achieve this level of accurate gimbal stabilization. All it takes is a robust gimbal design with fast servos and no more than about a 2-3:1 belt-drive reduction. Of course, you also need to make sure the gimbal gains are dialed in perfectly, and it would be nice if DJI would fix the damn wander/drift problem when the GPS is locked in. The point is, it can be done.

-- Gary

Overall weight is a lot more important than many people realize with the WKM. I just moved the WKM from the Droidworx frame to the beater frame I built for the Naza to do some waypoint testing on a platform that can absorb some abuse should there be a problem. Initially I could NOT get the thing to fly right regardless of the gain settings, in anything other than dead clam air it would oscillate as though it needed the gains turned down, and that was at the default settings! I won't bore everyone with the details of what I tried to fix it but ultimately it took the addition of roughly 1 pound of additional weight and suddenly it was cured, flys beautiful now on the same exact setup that flew like crap when it weighed less...

On teh subject of gimbal stabilizagtion, it's not just the fast hard movements that matter, it's also at the other end of the spectrum as well. The problem is that most flight controllers treat gimbal stabilization as an afterthought and also have to give priority to flying the platform, there is always going to be some degree of error in the gimbal compensation that will become visibile in the final result in less than optimum flight conditions. Certainly a dedicated IMU for the gimbal can eliminate or minimuize the problem but for those people that rely on the flight controller only it's never going to be perfect. Doesn't matter how fast the servos are or what the gear ratio is, if the F/C doesn't give the command fast enough or any at all for very slight motions, then there will be perceptable "imperfections" in any video recorded, that's kind of where I am with the CS6/WKm at the moment. The WKM either doesn't sense very tiny motions or maybe it doesn't send a compensation signal unless the movement is over a certain threshold value, don't know what the answer is but there is an extremely small but visible rocking motion that cannot be eliminated during FF in dead calm air that just bugs me and defies all attempts to make it go away. This may be one area where the DJI "magic gimbal" would do the job but I certainly don't want to spend $$$$ to find out it doesn't! For the less stable conditions I can make do with a combination of flying style and TX adjustable ATTI gains to get useable video, but again, the controller has to give processing priority to keeping the platform flying, stabilizing the camera is secondary and those parameters simply are not tunable.

Ken

 

[UAVproducts]

Smaller props with WKM make a big difference I have found on any platform.

 

[GGoodrum]

Overall weight is a lot more important than many people realize with the WKM. I just moved the WKM from the Droidworx frame to the beater frame I built for the Naza to do some waypoint testing on a platform that can absorb some abuse should there be a problem. Initially I could NOT get the thing to fly right regardless of the gain settings, in anything other than dead clam air it would oscillate as though it needed the gains turned down, and that was at the default settings! I won't bore everyone with the details of what I tried to fix it but ultimately it took the addition of roughly 1 pound of additional weight and suddenly it was cured, flys beautiful now on the same exact setup that flew like crap when it weighed less...

I've seen the same thing. Once I installed the vibration isolation plate on my Naza-equipped F450, I had all sorts of crazy oscillations with the basic gains set at around 125% and the attitude gains set at 85%. I had to reduce the basic gains to about 90% and I have the attitude gains up to about 95%. I'm sure this has a lot to do with the fact that I'm using the Graupner 11x5 props, and I'm running on 4s, but this is a pretty heavy F450 setup, at around 4-1/2 lbs/2.1kg. I'm not getting any wobbling to speak of, and it flies fine.

On teh subject of gimbal stabilizagtion, it's not just the fast hard movements that matter, it's also at the other end of the spectrum as well. The problem is that most flight controllers treat gimbal stabilization as an afterthought and also have to give priority to flying the platform, there is always going to be some degree of error in the gimbal compensation that will become visibile in the final result in less than optimum flight conditions. Certainly a dedicated IMU for the gimbal can eliminate or minimuize the problem but for those people that rely on the flight controller only it's never going to be perfect. Doesn't matter how fast the servos are or what the gear ratio is, if the F/C doesn't give the command fast enough or any at all for very slight motions, then there will be perceptable "imperfections" in any video recorded, that's kind of where I am with the CS6/WKm at the moment. The WKM either doesn't sense very tiny motions or maybe it doesn't send a compensation signal unless the movement is over a certain threshold value, don't know what the answer is but there is an extremely small but visible rocking motion that cannot be eliminated during FF in dead calm air that just bugs me and defies all attempts to make it go away. This may be one area where the DJI "magic gimbal" would do the job but I certainly don't want to spend $$$$ to find out it doesn't! For the less stable conditions I can make do with a combination of flying style and TX adjustable ATTI gains to get useable video, but again, the controller has to give processing priority to keeping the platform flying, stabilizing the camera is secondary and those parameters simply are not tunable.

Ken

I do agree that small smooth movements are just as important. To get that, you pretty much have to have some sort of reduction in each axis, preferably belt drives, in my opinion, as they have less backlash/"slop" than geared or linkage-based reductions. Too much reduction, however, and there's too much of a speed penalty for even the fastest servos. I've found 2:1 to be a good tradeoff between speed and smoothness.

Where I don't necessarily agree is whether or not the WK-M's gimbal outputs are fast enough, and whether they have enough resolution. With high resolution (12-bit), blinding fast (.06s), high-end digital servos, in a direct drive configuration, I can get a WK-M or Naza driven gimbal to keep a camera perfectly still, for sharp/quick platform movements. This proves to me, at least, that the gimbal outputs don't have a lag in reacting to platform movements. Even with these high-res, 12-bit servos, I can't, however, get smooth small movements. There's always a bit of "stair-stepness" to these small movements, in this direct drive configuration. With a belt drive reduction of 2-1/2 to 3:1, the movements are a lot smoother, but I started to see some lag, so I wasn't happy with the speed penalty. At 2:1, there's zero lag, but like you said, there is a perceptible lack of smoothness that gets into the video. I think I've found a pretty good solution for this, however. All digitals come with a preset amount of deadband space, around the neutral point. This typically is around 4ms. With the higher end Hitec servos, you can change this deadband to be between a relative number of 1-5, which I believe translates to around 2-10ms. If you set it to the lowest setting, the smoothness factor is greatly increased. On the new gimbals I'm doing with Andrey (Askman...), we've switched to the Hyperion high-voltage/hi-res Hyperion DH16 and DH 20x series, mainly because you can adjust the deadband timing all the way down to 0ms, via the PC programming app. I've found a setting between 1-2ms works quite well, and I can get very smooth movements, even with a 2:1 belt drive ratio.

I certainly agree that a separate IMU and/or a mechanical gyro/stepper motor solution is going to be the "ultimate", I'm just not yet convinced that you can't get close with a proper FC-based setup.

-- Gary

 

[GGoodrum]

Smaller props with WKM make a big difference I have found on any platform.

Smaller props equal higher disc loading. Anyway, I've found this to be true as well.

 

[vislaw]

Boris, I'll should have some very interesting videos to show in a week or so. I'm currently in the process of setting up my CineStar6 with the MK electronics stack that was on my Droidworx AD6, once I finish I'm going to do a few video passes in different conditons, then change it back to DJI and do the same flights in (hopefully) the same conditons, tuned as best I can. Then we'll see how the two stack up against each other, which one really is better in the wind, and how to tune the DJI gains for best results.

It will be a fair amount of work to get all this done so don't expect to see it in a few days.

Stay tuned...

Ken

I really REALLY appreciate this kind of effort and want to thank you for undertaking the task. This is the kind of stuff that makes this forum so valuable to all of us. THANKS!!!!!

 

[Vortex Aerial]

Boom length/motor spacing may be another factor as the longer the boom moment the further the motor/prop assembly has to travel in the vertical plane to both induce and fulfill both nick and roll corrections.
Sense the CS8 now has smaller props, the motors can be slid inboard. This theory perhaps goes against the generality that bigger is more stable. Frankly i don't see how this is possible when your talking about an aircraft that has virtually no inherent stability of its own and relies solely on a computer to fly.

 

[BorisS]

I have found the same that smaller props bring better results, in calm or challenging conditions. At some point I will slide the motors in on the CS8 and see if i can notice a difference. But i have to stay with Tabb on this one. The loss of flight time choosing 11x5 over 14x4.7 is a lot. Give and take a little (also not the right props, cant choose any graupners ) from the ecalc calculations but somewhere there we will end up in a real life test ! So the question is how much is 1 min, 30 secs flight time loss worth ? Given that its not possible yet to really pinpoint it and say for plan AV flying with wind the WKM is superior over other FCs.

View attachment 2399View attachment 2400

 

[La Turbine]

Things have been said about global weight, but not about the position in Z axis for the CoG. I would like to test this parameter, as there is perhaps some benefits to have a low CoG when there is a lot of wind/gusts. A bad thing with a lower CoG is that the gimbal is underslunged, and the wind have a lot of effect on it because of the global shape of the bird that is thick...
I would add that it should be a good update on a next firmware, to add the "longer of the booms" in the sofware. Perhaps it could provide us a better personalisation of our setup.

 

[redridinghood]

After a few tests with a very heavy platform of AUW 7.4 kilos and on a very light one of 1.8 I have come to the conclusion that it's not the weight that creates the wobble and instability. I have the wobble on both platforms. I have tried overproping and underproping. DJI control loops perform acceptably for only when you are in the 65-75% throttle range. So if you are not in that throttle range no ammount of gain settings will stabilize it.
When you are in that sweet throttle range spot DJI flies better than MK. For everything else MK flies better.

 

[BorisS]

Red i dont quite understand what you are getting at. What do you mean with throttle range certain RPM that has to be reached, 65 - 75% to hover or you mean in general if the bird reacts better if its climbing and not hover at a spot ?

Thanks

Boris

 

[redridinghood]

Boris I believe that if the hover rpm is around 65-75% DJI gain adjustments perform better.

 

[DennyR]

Things have been said about global weight, but not about the position in Z axis for the CoG. I would like to test this parameter, as there is perhaps some benefits to have a low CoG when there is a lot of wind/gusts. A bad thing with a lower CoG is that the gimbal is underslunged, and the wind have a lot of effect on it because of the global shape of the bird that is thick...
I would add that it should be a good update on a next firmware, to add the "longer of the booms" in the sofware. Perhaps it could provide us a better personalisation of our setup.

I have done extensive research on Z axis center of mass. The lower you have it then the more inertial response from lateral movements. The more inertia you have the the less can be the gyro sensitivity gain because it will cause oscillations that have to be compensated for in the PID Algorithm. The higher the gain the earlier the movement is seen and the better the stabilisation works. As was said, arm length does not contribute to stabilisation because stabilisation is a function of the computer. The response from the Gyro is mostly effected by the prop. design and it's weight. the timing of the ESC, KV. factor of the motor and the pid algorithm. The more global inertia you have in the system the worse will be the stabilisation. Generally the bigger the model the worse the problem gets. Bad oscillations in the descent are caused by too lower rpm which causes partial blade stall. This also can happen with large yaw movements when some of the motors slow down.

 

[elossam]

I tested de octo with a second 4S 5000 on board and 500 gr load without any noticeable difference. That overloads the octo and generates more rpm due to the less C required but.. no changes. Have written DJI asking the if they are happy with the results in octo V to know if the have plans to improve anything or to see if they are completely ignorant of the problem every day more and more people are showing. The only thing I hope can make a difference is to try the Octo I vs. the octo V configuration I'm currently using. Anyone has tested the Octo I configuration?

 

[BorisS]

I have done extensive research on Z axis center of mass. The lower you have it then the more inertial response from lateral movements. The more inertia you have the the less can be the gyro sensitivity gain because it will cause oscillations that have to be compensated for in the PID Algorithm. The higher the gain the earlier the movement is seen and the better the stabilisation works. As was said, arm length does not contribute to stabilisation because stabilisation is a function of the computer. The response from the Gyro is mostly effected by the prop. design and it's weight. the timing of the ESC, KV. factor of the motor and the pid algorithm. The more global inertia you have in the system the worse will be the stabilisation. Generally the bigger the model the worse the problem gets. Bad oscillations in the descent are caused by too lower rpm which causes partial blade stall. This also can happen with large yaw movements when some of the motors slow down.

So what would DJIs strategy look like when it comes to heavier loads on the copter ? Eventually prop size would have to increase to stay in a reasonable range of efficiency resulting in us ending with 800 or 900mm frames again, lower Kv etc. Of coarse one can argument that a 5D with a big lens is not the optimal AV cam, but on the other side people want to use them and considering the weight we heard the magical DJI gimbal will have they are on the road to hit these payloads even with a lighter cam than a 5d.

In general i dont know if these fast regulations that everyone is striving for really make sense. The catch up game to the impact from the wind/gusts onto the frame doesnt seem reasonable to be won or ? It seems like there is a missing link in the whole strategy and that is to give the user the ability to smoothen out the regulations to being slow but camera and camera gimbal friendly movements that can be compensated and not result in a rodeo, to the costs of position hold which i could life with. Giving the user the possibility to choose. Some might say this is possible by bringing down the gains, but this again makes the copter hard to control in the wind compared to an MK.

Considering that many DJI owners have chosen their components according to what works with other FCs I am somehow confused why DJI hasn't put any effort into informing us which setups are necessary to get the best out of their FCs. Shouldn't it be in their interest to have the multirotor AV celebds also shooting their vids everyone drools over with their setups ? Up to now i hear most of them complaining that the risk is to high to having to deal with a windy day and then WKM will not give them expectable results ! If its all about going for a complete system solution frame gimbal FC etc. than i am not against it. But at the same time pushing out an FC first that will obviously not work to the fullest with the given setups out there is a strange one !

Boris

 

[BorisS]

I would be fine with it doing these kind of regulations in GPS hold and wind, that ones choice to use it at a given situation

But give us a flight modus with auto-leveling and alt hold that makes the best out of windy conditions demanding more skills from the pilot maybe but at the same time reasonable stable results.

Boris