Anyone tried F550 NEX-5N/7 2-Axis Gimbal....yet?

B

BIGTACO

Member
Hi everyone my name is Neill and I am new to these forums. This is my first post actually! I am a long time RC'er, long time heli guy, and mechanical engineer by trade. I recently purchased a DJI F550 for the specific purpose of aerial video for use with a small company I own with a couple buddys. Its a videography and production company. Who would have guessed?!

www.gearheadflicks.com

I understand there are many more "pro" systems out there than the DJI F550 but it fits the budget of our flegling company right now...

Currently I modded my F550 with some landing gear and a self fabricated antivibration "hard" camera mount. As seen in the pics below. Soon I plan to upgrade the bird with the Naza add-on GPS and a nice camera gimbal. I plan on using the Sony NEX-5N as our camera. I have done a fair amount of research in gimbals and recently came across this one: http://tppacks.com/catalog/product_info.php?products_id=33&osCsid=30bb2fa6ebd540f3804ed5573bc05e19

This one is designed specifically for the F550 and the guy who designed and built it looks to have really done his research in regards to servo selection, ratios, and overall design. He also suggests to use the XA V2 FC/AHRS for gimbal control instead of the onboard Naza gimbal control due to limitations he has uncovered. If you follow his thread over at RC groups and actually look into the specifics of why his system gets the results his demo video seems to show it looks as though this is a good choice for a gimbal. But, as with anything you never know till you try. I would like an unbiased opinion of someone who has actually tried this system out themselves...

Has anyone?

DSC02551.jpg

DSC02548.jpg

DSC02549.jpg

DSC02550.jpg
 


J

jamienz

Member
yes I have the F550 NEX-5N/7 mount from Askman/Tppacks.com. With the MKS Servos and updated gear ratios it is really nice and smooth when stabilized from a DECENT FC (not a Naza or WKM).
 




steph280

steph280

Member
I had one on my MK Okto and currently fly the updated one on a DJI hexa with Rusty's frame. I believe this gimbal system is catered to the "prosumer" market, but the performance is very close to those pro level units costing twice as much. They've done a lot of R&D to get the perfect combination, so the product does work really well.

And yes that Sony CX730 is awesome.
 





DucktileMedia

DucktileMedia

Drone Enthusiast
I can tell you that the DJI anything makes the WORST gimbal controller. My Xaircraft was superior in comparison and that should not be the case when speding $1200! Also, the way DJI has the assistant laid out, you have no real control of the necessary functions. You can assign end points but the gain control only has one point it works otherwise the gimbal over shoots. You should be able to control rate, end points and center separately. Buy a Skyline, Hoverfly board, etc. I also thing that having a dedicated control is the way to go regardless.
 

GGoodrum

GGoodrum

Member
Sorry, I really need to spend more time over here. :)

I think the only way you can get "zen-like" gimbal performance is by using a separate gimbal controller that is mounted on the isolated side of the gimbal mount. I have tried pretty much everything out there as a standalone gimbal controller, including an FY-30A, XA v1 and v2 FC/AHRS units, one of timecop's Naze32 boards and a HoverFly Gimbal board. The problem with the Naza and WK-M gimbal outputs are two-fold. First of all, they have a 90 degree limit on the travel, which translates into only +/-18 degrees after our 2.5:1 belt drive reduction. Not near enough. The second issue is that the outputs were originally designed for use with inexpensive small gimbals that use analog servos, so the output update rate is very slow, at around 100Hz. This is slow by a factor of 5-8 times what it should be in order to get smooth movement for small corrections. On top of this, the WK-M also has had a problem where once the GPS is initialized, the gimbal outputs start this weird twitching/wandering problem. DJI has totally ignored this problem, or has tried to mask over it. Nonetheless, you still will get better results with a separate gimbal controller that is on the isolated side of the gimbal. With effective vibration isolation, the main FC is going to be sending small corrections that the gimbal is just not seeing. :)

The FY-30A has tiny little 1/4-turn pots for adjusting the gains, which makes it pretty much impossible to get dialed in properly, but it does work better than the Naza/WK-M. I've had very good luck using an XA FC/AHRS as a standalone gimbal controller, both the v1 and v2 versions. With these, you can open up the travel to the full amount the servos can handle, and you can set the output PWM rate to 500Hz, which makes small movements buttery smooth. The only slight nic against this option is that they use a slider control in their PC app, to set the gains, and you are limited to integer values. The gains for our gimbal want to be somewhee between 73 and 74, but you have to pick one or the other. The difference is quite small, but detectable nonetheless.

Timecop has a nifty little 32-bit Naze32 board that uses ported over multwii code. It has a mode where you can use it as a standalone gimbal controller, and TC has added a bunch of commands that let you control more than just the basic gimbal gains. You can, for instance, set the PWM output rate to 500Hz, for instance, just like the XA FC. You can also adjust how smooth the outputs are, and can open up the travel limits. In short, it is pretty close to the perfect gimbal controller, at least until you get it in the air. :) It has a weird twitch problem, primarily around the neutral point. This is being caused, we learned, by noise from the acc chip. To make a long story short, the code is having to be re-written, first to use full floating point calculations, but also to add kalman filtering to the inputs, which is what most FC's do in the first place. No idea how long this is going to take, because it is still basically an open source project, but a lot of smart kids are working on it. :) It might take awhile, but it will probably be worth the wait because these boards are relatively inexpensive at around $65 each.

Right now I'm using one of the new HoverFly Gimbal boards, which is uber-expensive, but it flat out is the best gimbal controller I've tried, period. I'm sure the also expensive PH Skyline board will have similar performance, but these re just now starting to get into the first users hands. Also, the HFG has already had one firmware update, and will soon get another. I would guess it might take a couple of iterations with the PH board as well, but we'll see. In any case, the HFG board runs at 800Hz, and I can tell you that there is a noticeable difference in smoothness between 500Hz and 800Hz. It also has precisely adjustable gains, plus the ability to open up the travel. What it also has is a setting for controlling the servo accelerations, which really helps in getting the performance tuned to the particular characteristics of each servo. Where this really comes into play is when there is a big correction, where the gimbal has to make a big movement. With the acceleration control, you can get it to react just fast enough without over-correcting.

Here is an example video I did the other day, using the HFG on my F550 with our AG550-NX gimbal and a Sony NEX-5N with the 16mm pancake lens:


This was in some strong gusty winds that were moving the F550 around quite a bit. We've been making some small improvements in the gimbal design, including the mounting scheme, in order to reduce the mechanical "bounces" that occur when a wind gust hits the side of the camera/gimbal. We still have a little bit left in the tilt axis, which we will address next. In any case, the HFG board worked well. I have the gains and the accelerations set pretty close.

Just for grins, I ran the NEX video through FCP's stabilization function with the settings pretty close to the minimum (1.0 with the max setting possible being 5.0...):


As you can see, it doesn't take much to make this pretty close to "perfect", even in 15-20 mph wind gusts.

I do agree, however, that the PJ710V/CX730V camcorders, with the Balanced Optical Stabilization System (BOSS) is a great option. It doesn't do anything in the roll axis, so you still need a good stabilization setup, but it will totally absorb movements in the yaw and tilt axis. I will post another video I just did in a little while. Same exact setup as the one above, just with the PJ710V yoke instead of the NEX-5N version. There is no need for any post stabilization whatsoever. :)

More later...

-- Gary
 
Last edited by a moderator:

N

nisouck

Member
Thanks for that, a very interesting state of the art.

Could you give me the full name of HFG gimbal controller?

Ok I find it's Hover Flight Gimbal controller but by the way: do you sell your V2 Gimbal?
 
Last edited by a moderator:

GGoodrum

GGoodrum

Member
nisouck said:
Thanks for that, a very interesting state of the art.

Could you give me the full name of HFG gimbal controller?

Ok I find it's Hover Flight Gimbal controller but by the way: do you sell your V2 Gimbal?
Click to expand...

Actually, it is HoverFly Gimbal. Here is a link. :)

We don't have an "official" v2 version of the AG550 just yet, but we've been making incremental changes. The main one is a new set of stiffeners that were added to the adapter plate "wings" that hold the special silicon isolators we use. These plates now sit out wider now, so that all four isolators are in a 5" square configuration. Anyway, all gimbals ordered since last week include this latest change. We will also make these changes available to existing users for a nominal fee (around $25 + shipping...). I'm also testing a slightly beefier roll asembly, which is patterned after our soon to be released full DSLR gimbal's roll design. This will have less of an effect for existing users, but my own setup was the very first prototype, with thinner roll plates, so it flexed a lot more than the current ones do. Anyway, the whole point of these upgrades is to remove every last bit of unwanted mechanical movement during strong and gusty winds. In light winds, even my original prototype did fine, but we want this to work as good as possible in the worst conditions. :) Anyway, the last change we will do before "officially" switching to a "v2" configuration is a little consolidation of yoke types. Mainly, we are going to combine the basic NEX-5N/7-GF3/5 yoke with the version for the PJ710V. We've had several requests for this, including from me ( :) ), in order to be able to quickly switch between a PJ710V and a NEX-5N. We will still have a dedicated GoPro yoke, but we are probably going to move those wanting to use a 550D or GH2 up to the full-size DSLR yoke, which will be able to carry pretty much anything. I'm getting the first DSLR gimbal to test later this week.

Here is the video I took this morning, using the PJ710V:


Here is a good example where the the acceleration settings can improve performance. This is a much heavier camera than the NEX-5N, but I did not change the gimbal gains at all in the HF Gimbal app. For slower corrections, this is fine, but you can notice right at the beginning that thee is a quick larger roll to the left, and it shows up in the video as a slight lag. By increasing the acceleration value slightly, you can eliminate this slight lag for large/quick movements. Anyway, this is raw footage, right from the camera. Very "zen-like" I think. :)

-- Gary
 
Last edited by a moderator:


B

BIGTACO

Member
Ggoodrum. I must tell you. Your informative responses are very valuable. Thanks for the input. And great work on the systems you build. As a practicing engineer I understand the untold hours you spend thinking, building, testing, over and over.

I see your gimbal as a less expensive option that rivals the performance of a AV130 for instance. This is what I am hoping atleast. For me, it will come down to a purchase of yours or that one...

Do you have any video of the gimbal itself in action? Like a video from directly behind the gimbal as it is functioning? Or a video of the gimbal working as someone holds the bird in hand?
 

D

DennyR

Active Member
IrisAerial said:
I can tell you that the DJI anything makes the WORST gimbal controller. My Xaircraft was superior in comparison and that should not be the case when speding $1200! Also, the way DJI has the assistant laid out, you have no real control of the necessary functions. You can assign end points but the gain control only has one point it works otherwise the gimbal over shoots. You should be able to control rate, end points and center separately. Buy a Skyline, Hoverfly board, etc. I also thing that having a dedicated control is the way to go regardless.
Click to expand...

Strange comment. In this video all of the shots were stabilized with DJI Naza and WKM standard gimbal outputs. The trick is to have gains close to 100% and then have your servo drive ratio to about 5:1 I have used the outputs from the AMP-2 that also work fine, you just have to know and understand how the system works. The GoPro is as always the weak link. This camera mount moves only 10 deg. across the two stabilised axis. the tilt is controlled by a separate small slew servo that does not need to be stabilised twice. It is if you like an inner axis system which is more accurate than a full travel system. The best (most accurate) stand alone IMU is the Diydrones ArduIMU V3 but again you have to know what your doing to get it working as it should.
 
Last edited by a moderator:

D

DennyR

Active Member
GGoodrum said:
Sorry, I really need to spend more time over here. :)

I think the only way you can get "zen-like" gimbal performance is by using a separate gimbal controller that is mounted on the isolated side of the gimbal mount. I have tried pretty much everything out there as a standalone gimbal controller, including an FY-30A, XA v1 and v2 FC/AHRS units, one of timecop's Naze32 boards and a HoverFly Gimbal board. The problem with the Naza and WK-M gimbal outputs are two-fold. First of all, they have a 90 degree limit on the travel, which translates into only +/-18 degrees after our 2.5:1 belt drive reduction. Not near enough. The second issue is that the outputs were originally designed for use with inexpensive small gimbals that use analog servos, so the output update rate is very slow, at around 100Hz. This is slow by a factor of 5-8 times what it should be in order to get smooth movement for small corrections. On top of this, the WK-M also has had a problem where once the GPS is initialized, the gimbal outputs start this weird twitching/wandering problem. DJI has totally ignored this problem, or has tried to mask over it. Nonetheless, you still will get better results with a separate gimbal controller that is on the isolated side of the gimbal. With effective vibration isolation, the main FC is going to be sending small corrections that the gimbal is just not seeing. :)

The FY-30A has tiny little 1/4-turn pots for adjusting the gains, which makes it pretty much impossible to get dialed in properly, but it does work better than the Naza/WK-M. I've had very good luck using an XA FC/AHRS as a standalone gimbal controller, both the v1 and v2 versions. With these, you can open up the travel to the full amount the servos can handle, and you can set the output PWM rate to 500Hz, which makes small movements buttery smooth. The only slight nic against this option is that they use a slider control in their PC app, to set the gains, and you are limited to integer values. The gains for our gimbal want to be somewhee between 73 and 74, but you have to pick one or the other. The difference is quite small, but detectable nonetheless.

Timecop has a nifty little 32-bit Naze32 board that uses ported over multwii code. It has a mode where you can use it as a standalone gimbal controller, and TC has added a bunch of commands that let you control more than just the basic gimbal gains. You can, for instance, set the PWM output rate to 500Hz, for instance, just like the XA FC. You can also adjust how smooth the outputs are, and can open up the travel limits. In short, it is pretty close to the perfect gimbal controller, at least until you get it in the air. :) It has a weird twitch problem, primarily around the neutral point. This is being caused, we learned, by noise from the acc chip. To make a long story short, the code is having to be re-written, first to use full floating point calculations, but also to add kalman filtering to the inputs, which is what most FC's do in the first place. No idea how long this is going to take, because it is still basically an open source project, but a lot of smart kids are working on it. :) It might take awhile, but it will probably be worth the wait because these boards are relatively inexpensive at around $65 each.

Right now I'm using one of the new HoverFly Gimbal boards, which is uber-expensive, but it flat out is the best gimbal controller I've tried, period. I'm sure the also expensive PH Skyline board will have similar performance, but these re just now starting to get into the first users hands. Also, the HFG has already had one firmware update, and will soon get another. I would guess it might take a couple of iterations with the PH board as well, but we'll see. In any case, the HFG board runs at 800Hz, and I can tell you that there is a noticeable difference in smoothness between 500Hz and 800Hz. It also has precisely adjustable gains, plus the ability to open up the travel. What it also has is a setting for controlling the servo accelerations, which really helps in getting the performance tuned to the particular characteristics of each servo. Where this really comes into play is when there is a big correction, where the gimbal has to make a big movement. With the acceleration control, you can get it to react just fast enough without over-correcting.

Here is an example video I did the other day, using the HFG on my F550 with our AG550-NX gimbal and a Sony NEX-5N with the 16mm pancake lens:


This was in some strong gusty winds that were moving the F550 around quite a bit. We've been making some small improvements in the gimbal design, including the mounting scheme, in order to reduce the mechanical "bounces" that occur when a wind gust hits the side of the camera/gimbal. We still have a little bit left in the tilt axis, which we will address next. In any case, the HFG board worked well. I have the gains and the accelerations set pretty close.

Just for grins, I ran the NEX video through FCP's stabilization function with the settings pretty close to the minimum (1.0 with the max setting possible being 5.0...):


As you can see, it doesn't take much to make this pretty close to "perfect", even in 15-20 mph wind gusts.

I do agree, however, that the PJ710V/CX730V camcorders, with the Balanced Optical Stabilization System (BOSS) is a great option. It doesn't do anything in the roll axis, so you still need a good stabilization setup, but it will totally absorb movements in the yaw and tilt axis. I will post another video I just did in a little while. Same exact setup as the one above, just with the PJ710V yoke instead of the NEX-5N version. There is no need for any post stabilization whatsoever. :)

More later...

-- Gary
Click to expand...

Timecops little 32 bit controller is a great little board that does what it was designed to do. However the camera gimbal outputs will not do the job properly. You will see a cross axis output interference across roll function. I think the guys at the Afro site are working on that.
 
Last edited by a moderator:

jonnydonut

jonnydonut

Member
Gary, can you share your hoverfly gimbal settings?

On adidas advice, to eliminate jitter I've gone down to -10 on tilt and roll acceleration, but it's still not where I want it to be. I need to upgrade to graupner props I think, but the jitter is only there when the servos are plugged in.

 
Last edited by a moderator:

You must log in or register to reply here.
Top