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

[RTRyder]

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

From my recent testing I've found some interesting things, the most significant being that a quad of equal size to a hex will perform far better in windy conditions than the hex will with a WKM controller. I built a quad with a motor to motor spacing of approximately 25 1/2 inches which is the same as the Droidworx AD6 I have. Motors are equivalent, Pulso 2814 on the AD6 vs. Avroto 2814 on the quad, both flying on Graupner 11 x 5 props. The quad with 4S 8000 battery weighs in at 6 pounds 1/2 ounce and it performs pretty much exactly as the quad in the video does in windy conditons, maybe even a bit smoother on the reflashed ESCs, and it holds GPS position amazingly well in strong winds. The AD6 on the other hand wants to wobble and bounce and while it holds postion within a roughly 20 to 30 foot circle it is visibly less stable than the same size quad which says to me the majority of the development was done with a quad and the firmware has been optimized to that configuration.

Meanwhile I've been doing some test flights with the CS6 using the MK electronics stack that came from the AD6 and the difference is significant. The CS is quite obviously a lot "looser" with the MK stack than it was with the WKM, it's most noticeable in the autoleveling and altitude hold, neither of which are as solid as they are with the WKM in control. From the camera point of view though, the difference is clearly visible, there is a degree of stabilty with the MK that it just isn't possible to achieve with the WKM in any conditions. Even in calm air there is a very visible difference in the raw video MK vs. WKM but when the wind comes up things don't change much if at all with the Mk stack, it still flys the same and the video is still just as steady as it is in calm air, that is NOT true of the WKM.

There's a lot more testing to be done and I'm currently waiting for some parts to arrive as well as the new 5N. Add that to some instability on the job front in the past day or two and I have limited time available to devote to multirotors at the moment, I'll have more on this once things settle down and I can get back to some degree of normality, if that's possible...

Ken

 

[GGoodrum]

This is where I get confused. If the WK-M's gimbal gains were dialed in, with a responsive mount, why would you see the larger corrections that the WK-M is making in order to do better position keeping?

-- Gary

 

[BorisS]

This is where I get confused. If the WK-M's gimbal gains were dialed in, with a responsive mount, why would you see the larger corrections that the WK-M is making in order to do better position keeping?

-- Gary

I am not quite sure what you are getting at. But i can tell you being through several gimbals now AV 130 AV 200 CS gimbal changing servos doing mods, using picloc, we have to stop believing that there is any way that these hobby servos based gimbals will ever be able to level out the harsh correction like the WKM tends to make. The only things those gimbal can do for us in the state they are in is correct slow smooth movements and already there the performance is night and day between some of them in my eyes. Its like demanding an old ink jet printer to provide the same results as a laser printer, its just not doable. The Resolution is not given and most FCs cant even command the servos at which speed they should reach this approx. point they should reach to keep things level !

I hope i understood you correctly

Boris

 

[Siteline]

@ Boris,
Are there still plans to head to the toy fair for you, this week? Even if the WKM gimbal is not available for purchase, it would be great to hear how it looks from a trusted multi enthusiast.:tennis:

 

[BorisS]

I am going there for sure on saturday also meeting matjaz there from copterworx ! Wanted to go already today but motivation left me to hop in the car and go 800 km back and forward
Lets see how talkative they will be

By the way yesterday i noticed that it sad on the ticket no fotos on the fair etc allowed, lets see what i can deliver

Boris

 

[DucktileMedia]

I don't understand these companies. My Xaircraft would control my av130 VERY well. super smooth and responsive without over correction. When I saw my friend's WKM controlled Av130 I was highly disappointed! I just assumed if the Xaircraft could do it then anything would work. The DJI software is so stupid it boggles my mind how they can leave it this way. 1st they limit you to 1000 units of output. Then they make the gain the only control you have over making the mount level. So if you want it to go faster you also seem to get over compensation of the mount!


Dear DJI....

Please at your earliest convenience change just a few minor things for us:

1) let us adjust the RTH features. how high, the delay, etc. Hover failsafe should work for everyone. For a few of us it never lands, just sits there.

2) On the gimbal controls there should be a higher range of output, the center should adjust just that, the center. The gains should adjust the rate only, not multiply the extents!

3) work on redundancy. Why is the WKM the only one that doesnt do anything different with a motor out?

4) once logged into the assistant software, make it stop asking to log back in. Out in the field where there is no Wi-fi makes this annoying.


That would be my immediate wish list.

thanks

Yuri

 

[BorisS]

some info on the nuernberg fair !

http://www.multirotorforums.com/showthread.php?2195-DJI-Nuernberg-Toyfair&p=23106#post23106

Boris

 

[GGoodrum]

I am not quite sure what you are getting at. But i can tell you being through several gimbals now AV 130 AV 200 CS gimbal changing servos doing mods, using picloc, we have to stop believing that there is any way that these hobby servos based gimbals will ever be able to level out the harsh correction like the WKM tends to make. The only things those gimbal can do for us in the state they are in is correct slow smooth movements and already there the performance is night and day between some of them in my eyes. Its like demanding an old ink jet printer to provide the same results as a laser printer, its just not doable. The Resolution is not given and most FCs cant even command the servos at which speed they should reach this approx. point they should reach to keep things level !

I hope i understood you correctly

Boris

I'm going to have to respectfully disagree that you can't make a servo-based solution successfully take out platform motion enough that you still get smooth video. One thing I've learned recently is that most digital servos have a built-in deadband that is set around 4ms. This alone is the cause of quite a bit of "stairstepness" in small servo movements. To get around this, most higher-end gimbals have lots of reduction, usually using a belt drive. The higher the reduction, however, the slower the reaction can be, which can induce a bit of lag, which is most definitely visible in the video. This can look like an "over-correction", but I don't think that is the case.

In the new gimbals I'm doing with Andrey Kim, we are using higher-end Hyperion servos, because one of the programmable settings lets you eliminate the deadband completely. What the jury is still out on, in my opinion, is whether or not the WK-M's gimbal outputs have enough resolution, or are output at a fast enough rate, in order to match the smoothness possible with a 12-bit servo with 0 deadband space. To me, that's the $64 question. I did do some comparison tests before, between using the WK-M's gimbal outputs vs using a XA FC/AHRS in a "PicLoc"-like configuration. With the XA FC's gimbal outputs set at 10ms/100Hz, I could not tell the difference between it and the WK-M. When I switched the XA FC's outputs to 2ms/500Hz, however, it did seem that the movements were smoother. I do plan on testing this further, for sure.

I don't have a PicLoc, but if I did, one thing I would definitely try is using servos that had the deadband space reduced. In addition to the Hyperions, I think the Hitec programmable versions have a deadband adjustment setting.

-- Gary

 

[BorisS]

I dont have the technical knowledge to argument against what you are saying and i must say that i hope you succeed for the sake of all of us ! If you look at new developments though not only DJI, pretty sure or sure other new gimbal will reach the market soon you will not find hobby servos in them that were originally designed to control the rudder on a plane or whatever. And one thing is for sure they will all come out with their own stabilization system !

Boris

 

[GGoodrum]

I dont have the technical knowledge to argument against what you are saying and i must say that i hope you succeed for the sake of all of us ! If you look at new developments though not only DJI, pretty sure or sure other new gimbal will reach the market soon you will not find hobby servos in them that were originally designed to control the rudder on a plane or whatever. And one thing is for sure they will all come out with their own stabilization system !

Boris

Oh I agree, there will be high-end gimbal solutions, like what DJI is doing, that gives the ultimate in "steadicam"-like performance. These will most definitely use dedicated IMUs and will make use of custom stepper-motor drives. That said, I still think there is room for improvement in a "prosumer" setup, and that is the holy grail that I am chasing at the moment.

-- Gary

 

[elossam]

Hyperion servos has futaba output shaft? Does it allow a programed 180° range?

 

[GGoodrum]

Hyperion servos has futaba output shaft? Does it allow a programed 180° range?

Yes, to both questions.

 

[elossam]

Thanks mate. Will have a look at them.

 

[RTRyder]

I don't understand these companies. My Xaircraft would control my av130 VERY well. super smooth and responsive without over correction. When I saw my friend's WKM controlled Av130 I was highly disappointed! I just assumed if the Xaircraft could do it then anything would work. The DJI software is so stupid it boggles my mind how they can leave it this way. 1st they limit you to 1000 units of output. Then they make the gain the only control you have over making the mount level. So if you want it to go faster you also seem to get over compensation of the mount!


Dear DJI....

Please at your earliest convenience change just a few minor things for us:

1) let us adjust the RTH features. how high, the delay, etc. Hover failsafe should work for everyone. For a few of us it never lands, just sits there.

2) On the gimbal controls there should be a higher range of output, the center should adjust just that, the center. The gains should adjust the rate only, not multiply the extents!

3) work on redundancy. Why is the WKM the only one that doesnt do anything different with a motor out?

4) once logged into the assistant software, make it stop asking to log back in. Out in the field where there is no Wi-fi makes this annoying.


That would be my immediate wish list.

thanks

Yuri

You can have #4 immediately, just click the little box that says "keep me logged in" on the screen where you have to login, it won't ask you again and it works just fine out in the field with no internet connection.

I'm going to have to respectfully disagree that you can't make a servo-based solution successfully take out platform motion enough that you still get smooth video. One thing I've learned recently is that most digital servos have a built-in deadband that is set around 4ms. This alone is the cause of quite a bit of "stairstepness" in small servo movements. To get around this, most higher-end gimbals have lots of reduction, usually using a belt drive. The higher the reduction, however, the slower the reaction can be, which can induce a bit of lag, which is most definitely visible in the video. This can look like an "over-correction", but I don't think that is the case.

In the new gimbals I'm doing with Andrey Kim, we are using higher-end Hyperion servos, because one of the programmable settings lets you eliminate the deadband completely. What the jury is still out on, in my opinion, is whether or not the WK-M's gimbal outputs have enough resolution, or are output at a fast enough rate, in order to match the smoothness possible with a 12-bit servo with 0 deadband space. To me, that's the $64 question. I did do some comparison tests before, between using the WK-M's gimbal outputs vs using a XA FC/AHRS in a "PicLoc"-like configuration. With the XA FC's gimbal outputs set at 10ms/100Hz, I could not tell the difference between it and the WK-M. When I switched the XA FC's outputs to 2ms/500Hz, however, it did seem that the movements were smoother. I do plan on testing this further, for sure.

I don't have a PicLoc, but if I did, one thing I would definitely try is using servos that had the deadband space reduced. In addition to the Hyperions, I think the Hitec programmable versions have a deadband adjustment setting.

-- Gary

I have a set of programmable Hyperions in one of my AV130 mounts, they're not the wonder servo you think they are. I've tried all sorts of combinations of travel, speeds, and deadband settings, they're no better or worse than a good set of hi-res (4096) servos properly sized to the load that they'll be moving. Actually, my AV130 with Savox hi-res out performs the one with the Hyperions regardless of how I program them, and that's with both Mk and WKM flight controllers in charge of the servo movements.

The biggest problem I have the WKM and a camera gimbal is that it seems like the stabilization code has a built in deadband. By that I mean it takes a certain amount of movement of the frame before the WKM recognizes that it needs to command the servos to compensate and the end result is a tiny but visible wobble in recorded video as the multi is flying in smooth level FF. I've noticed this quite a bit with the CS6 and NEX 5N on the faster AV130, not quite so noticeable with a GoPro but still there. A simple swap of WKM for MK electronics with all else being the same and the problem is gone, so the issue is with the WKM and not the servo deadband or the mechanics of the mount.

Ken

 

[GGoodrum]

I have a set of programmable Hyperions in one of my AV130 mounts, they're not the wonder servo you think they are. I've tried all sorts of combinations of travel, speeds, and deadband settings, they're no better or worse than a good set of hi-res (4096) servos properly sized to the load that they'll be moving. Actually, my AV130 with Savox hi-res out performs the one with the Hyperions regardless of how I program them, and that's with both Mk and WKM flight controllers in charge of the servo movements.

First of all, I'm not saying these are "wonder servos" by any means. You are right, there's lots of 12-bit choices out there. There were a couple reasons why I picked them, however. First of all, although the newer Hitecs I've been using let you program the deadband space, when you put them in the "hires" mode, it limits the throw to 120 degrees, not 180. Another reason is that the "mini" sized DH16s are 11-bit, where as the Hitec and Savox minis are 10-bit. Not sure that makes a difference, or not. Just out of curiosity, what model Hyperion servos did you have in the AV-130?

The biggest problem I have the WKM and a camera gimbal is that it seems like the stabilization code has a built in deadband. By that I mean it takes a certain amount of movement of the frame before the WKM recognizes that it needs to command the servos to compensate and the end result is a tiny but visible wobble in recorded video as the multi is flying in smooth level FF. I've noticed this quite a bit with the CS6 and NEX 5N on the faster AV130, not quite so noticeable with a GoPro but still there. A simple swap of WKM for MK electronics with all else being the same and the problem is gone, so the issue is with the WKM and not the servo deadband or the mechanics of the mount.

Like I said above, I too saw smoother performance when I used the XA FC/AHRS, but only when I set the faster 2ms/500Hz update rate on the gimbal outputs. At the standard 10ms/100Hz rate, I could not tell the difference. What that tells me is that it might be simply that the WK-M/Naza gimbal outputs need to be at a faster rate. Your built-in deadband theory would also cause a similar "step" in the outputs. Do you happen to know the ouput rates of the mk's gimbal outputs? Just curious.

In the video below, which I did an hour ago, it was too windy to get any data that would shed more light on this issue. I had to yank it around quite a bit, just to keep it close, so this is anything but smooth flying. :shame: I added a CC 10A BEC, which I programmed to run the servos at 7.4V. This speeds up these to .10s, and it has eliminated the slight lag I was seeing before. Anyway, I've got the gains dialed in about as good as I can get them. I also think the Basic gains are a bit high, because I saw some oscillations, during a couple gusts.

I'm hoping tomorrow that I can try this again, without the wind, and see how smooth I can get it. I might also give the XA FC/AHRS combo another try.

-- Gary