View Single Post
Old Feb 26, 2012, 01:43 PM
GGoodrum is offline
Find More Posts by GGoodrum
Registered User
GGoodrum's Avatar
South Orange County, CA
Joined Mar 2003
8,270 Posts
Quote:
Originally Posted by TimIndy View Post
Not a bad idea. It seems to me Gary has the tilt jitter waaay improved with the new servo, but the controller seems slow on roll compensation. Roll compensation correction seems pretty reliable in post, so I think the tilt only option might be a good way to go without a Picloc or similar, although maybe a higher DJI refresh rate can help. The tilt jitter/twitch was impossible to correct in post. Gary's latest video (the one right above your post quoted above) looks like he has really nulled the twitch very good (good job Gary- you have done a bunch of work that will help us all... thanks).

I am slowly working my way through this thread. I am late to this party since winter and my workload had me sidelined for a while. I am anxious to reload for the spring season when in a couple weeks I catch up with my over loaded work schedule.

Good to see you back Askman. I still have a couple of heli mounts and one pole mount of yours.

I just had a great warranty experience with DJI on a GPS unit ... I just wanted to mention it here just to get that said since it is natural to only talk about problems and not the good experiences. Thanks Sidney, Fiona and Robert! Sorry for the off thread mention.

Tim
The mount, with the second layer of isolation (between the LG and the F550 lower plate...), does a great job of eliminating the "jelly roll" that typically comes from props and motors not being perfectly balanced. What no isolated mount can do, however, is eliminate the same sort of vibrations from getting to the FC. In the video above, you can hear the servos "buzzing". This is what is causing a little vibration, occasionally, mostly while it is hovering. This buzzing is caused by the FC's gimbal outputs trying to make "corrections" for the platform vibrations. I have the props perfectly balanced, but I haven't yet done anything about the motors. The only way around this is that I can see is to use a separate gimbal controller, like a PicLoc, or maybe one of the new HoverFly Gimbal boards, mounted on the isolated side of the gimbal/mount. I'm not sure about this, but maybe another option is to move the WK-M's IMU to the isolated side. This might work, or it could screw up the basic flight characteristics.

I'm convinced that the gimbal outputs do not have an inherent lag. What is causing what appears like a roll lag is actually not having the gains perfectly set. I used a bubble level on this last adjustment, and got it pretty close, but with the GoPro's fisheye effect, roll "errors" are more readily visible. I will adjust this one more time today, before my next test flight.

There is another issue with the WK-M and Naza gimbal outputs and that is the update rate. From my tests, it appears they are using an update rate of 100Hz. This is too slow by a factor of about 4 or 5. What I have found is that the faster the update rate, the smoother small movements will be. At slower rates, small movements will be slightly jerkier than they will at fast rates. I have an original XA FC/AHRS V1 combo that I have used as a separate gimbal controller, and one of the controls you have is to set the update rate, from 2-20ms. This equates to a rate adjustment from 500hz down to 50Hz. When I set the rate to 10ms/100Hz, the FC/AHRS' gimbal outputs behave exactly like the WK-M' or Naza's outputs. If I max out the rate to 2ms/500Hz, and I zero out the deadband timing in the servos, small movements are buttery smooth. I'm putting together a video that shows the differences, to send to DJI. I'm hoping I can get them to make a change to speed up the update rate, because it does make a noticeable difference.

I've had a hard time finding a NEX-5N locally here, but yesterday I did pick up a Sony DSC-HX9V that I'm going to give a "shot". () What intrigues me is that like the NEX-5N/7, it does 1080p/60fps. What I'm not sure about is that it has a two-level image stabilization function, that has two settings, "Normal" and "Active", but I don't see a way to turn the function off. I have no idea whether one or the other would be helpful, or whether it will just make things worse, but I will try both.

-- Gary
GGoodrum is offline Find More Posts by GGoodrum