Wide Yaw Problem

[rexster314]

Guess I'm another one that has a problem with yawing my bird and it makes about a 4 foot circle instead of twisting on a dancer's pole. I recalibrated the IMU last week after leveling the work area and today finally got a chance to try it out. Apparently the change did some good cause the drifting cut down to about a 2 foot radius. Ran my 3 batteries down just flying around and getting some video/pictures of the great sunset that was happening. Came home and came up with this after a little thought. I had about a 14" wood strip left over from another project. I measured the width of motor prop shaft, drilled out halfway through the wooden strip 14 1/4 wide. I can put the wood strip across the motors with out it falling over. Put my two axis bubble level on the center, shimmed up the legs to a perfect level then recalibrated the IMU. I'll give it a go tomorrow. Here's some pics:

 

[MapMaker53]

Great idea, and simple.

I basically do something similar by placing a kitchen cutting board onto a table, placing a leveler on it and then adjusting the board with a penny or two under the corners where needed to make it level. Then I just place the quad on the board and calibrate it.

 

[Mopar Bob]

I suggest you follow the following posting. It covers the same topic in depth.

viewtopic.php?f=27&t=32454#p296972

 

[JJM]

It's tomorrow already!

 

[Ivan]

rexster314 I love your approach! Measuring level on the motors and not where the phantom is standing. That is much more accurate. However i believe you are introducing some elements of inaccuracies, wood is a questionably level element by itself, are your holes drilled to exactly the same depth and so on. A sheet of glass that goes over all motors I believe would be best. Or some adapater that goes on only one motor and then measuring motor by motor with it.

Also I had(still have) a bubble level exactly like the one you have, and I relized it is not really precise. Test if you have the same reading if you rotate it by 180 degrees on the same spot. I hadn't. That's why I bought a different one, which is considerably more precise:

However maybe it's just a coincidence of what you get in the shop.

 

[Mopar Bob]

Ivan, why do you think leveling of the motors is much more accurate. Isn't the IMU the critical element that needs to be level. It could be just as off kilter relative to the motors as it is relative to the landing gear.

 

[Ivan]

Ivan, why do you think leveling of the motors is much more accurate. Isn't the IMU the critical element that needs to be level. It could be just as off kilter relative to the motors as it is relative to the landing gear.

Because what is important is that the air flow produced by the propellers is pointing straight down, perpendicular to the ground. During a calibration procedure you are basically telling the NAZA-M v2 flight controller to take current values coming from the IMU(Inertial Measurement Unit) as zero. Or in simpler words with a calibration you are telling the phantom "this how I have placed you right now will produce a perfectly straight stream of air from the propellers that will keep you in perfect hover without drifting to the sides". The phantom then knows when it's internal bubble level(IMU) is showing that it leaned to one side, how much does it have to increase the power to the motor on that side and/or reduce power to the opposite one to return the "bubble" in center.
There is one more analogy how this can be described(if you ignore the other sensors it has) IMU can really be imagined as a bubble level(accelerometer sensor) with the option to reset at what inclination the bubble will be at the dead center. And that option is the calibration procedure. The procedure actually lasts for a few minutes because it is doing several measurements just to be sure(to filter out irregularities by averaging). That all is why it doesn't really matter how the NAZA/IMU is positioned inside. However it certainly helps if the IMU is parallel to the legs bottom and parallel to the motors underside. That however implies that that the phantom shell plastic is perfectly manufactured and the assembly process is producing always perfectly aligned components. If that would be true, then you could just calibrate by placing the phantom on a level surface. And mostly it is true and such calibration does yield practically perfect results. In those cases that it doesn't that is because the motors are not perfectly parallel to the rest of the components.

 

[erikgraham]

Yaw circling for me is always a compass issue that I correct in the field. In severe cases I have had to degauss the compass in the shop even when it's not failing basic calibration in the field. I'm sure you've already recalibrated the compass many times, but if you haven't looked at the compass number set in RC Assistant then its worth a look. Plenty of YouTube videos to show you how to fix and what your ranges should be.

 

[rexster314]

It's tomorrow already!

Haha, I had too many irons in the fire but got a chance to test fly the new settings and had pretty good success. Now I can do a slow yaw within the space of 8-10 inches or so which to me is good enough for government work. But I'll probably do one more recalibration. I also drilled the holes for the prop shafts through my leveling wood strip where it would lie on the base of the shaft below the threads. Then I noticed that the exact center of the a/c has a hump for the GPS that interfered with the wood strip, causing it to see-saw. So I got two matched washers with an ID exactly the size of the threaded shaft, cemented the washers over the drilled holes when then raised the leveling strip just enough to clear the raised center of the a/c. I also checked (not just by eyeballing) but using a carpenter's level, insured the wood strip is not warped or misshapen and is totally square. Haven't done a recal with this set up but think I will get closer even more. Going to try and set it up tomorrow and have another go at it

 

[Mopar Bob]

Based on the comments made about leveling against the motors versus the landing struts, I re-leveled my Phantom accordingly. I first leveled my platform, then placed the Phantom on it. I then took a sheet of glass and placed in over the motors. I then used a 2 foot digital level and measured it in all directions. In my case the level values off the motors were no different than the bare leveling platform I made. But if in fact there was a difference, I would adjust the correct bolts accordingly until it was in level. Lock them in and it would preserve the setup for future calibrations. So based on my situation, I cannot tell if leveling off the motors is more accurate than off the landing struts. But it would be interesting to hear what others are finding. Also, if you think some part of my setup provides inaccuracies that are not apparent to me, please let me know.

 

[fastsmiles]

Looks like an ideal test. I will check mine and post results.

 

[JJM]

Based on the comments made about leveling against the motors versus the landing struts, I re-leveled my Phantom accordingly. I first leveled my platform, then placed the Phantom on it. I then took a sheet of glass and placed in over the motors. I then used a 2 foot digital level and measured it in all directions. In my case the level values off the motors were no different than the bare leveling platform I made. But if in fact there was a difference, I would adjust the correct bolts accordingly until it was in level. Lock them in and it would preserve the setup for future calibrations. So based on my situation, I cannot tell if leveling off the motors is more accurate than off the landing struts. But it would be interesting to hear what others are finding. Also, if you think some part of my setup provides inaccuracies that are not apparent to me, please let me know.

Members in other forums are leveling the IMU (or whatever it's called) by opening the phantom and placing a small level on it.

 

[BlackTracer]

Members in other forums are leveling the IMU (or whatever it's called) by opening the phantom and placing a small level on it.

This is an unneeded step in the process. And if that is how they are leveling the phantom and then calibrating the IMU, then that is just plain wrong.

By leveling the phantom first, and then calibrating the IMU, you are telling the IMU what baseline level is. That is the goal. That is why some pilots are going to great lengths to level their phantoms, not the IMU unit inside. Some are even designing leveling sticks/boards that span the motors because that is the part of the phantom you want to be your baseline level.

 

[JJM]

Members in other forums are leveling the IMU (or whatever it's called) by opening the phantom and placing a small level on it.

This is an unneeded step in the process. And if that is how they are leveling the phantom and then calibrating the IMU, then that is just plain wrong.

By leveling the phantom first, and then calibrating the IMU, you are telling the IMU what baseline level is. That is the goal. That is why some pilots are going to great lengths to level their phantoms, not the IMU unit inside. Some are even designing leveling sticks/boards that span the motors because that is the part of the phantom you want to be your baseline level.

Well that's what I was thinking too...

 

[rexster314]

In my way of thinking, leveling the IMU by removing the top shell might be counter productive. By doing it that way, the IMU might be level but the bird itself won't be, creating unnecessary tilting of the aircraft while flying. By leveling the a/c itself, you are telling the IMU that this is a level flight and the IMU adjusts itself to keep everything level. When I first flew the bird a couple of times it could yaw on a dime, but being me, I fooled enough with the settings (gain, IMU) to put it out of kilter. After a few skittery flights (always under control) and more reading on this forum I finally realized what I was doing wrong and what I actually needed to do. I've cut the yaw drifting down tremendously and every time I recalibrate I've refined the leveling process beforehand. I've recalibrated last night but haven't had a chance to test it out, hopefully later this afternoon I can check it out.

 

[rexster314]

Based on the comments made about leveling against the motors versus the landing struts, I re-leveled my Phantom accordingly. I first leveled my platform, then placed the Phantom on it. I then took a sheet of glass and placed in over the motors. I then used a 2 foot digital level and measured it in all directions. In my case the level values off the motors were no different than the bare leveling platform I made. But if in fact there was a difference, I would adjust the correct bolts accordingly until it was in level. Lock them in and it would preserve the setup for future calibrations. So based on my situation, I cannot tell if leveling off the motors is more accurate than off the landing struts. But it would be interesting to hear what others are finding. Also, if you think some part of my setup provides inaccuracies that are not apparent to me, please let me know.

I like the leveling platform. Homegrown or purchase somewhere. Looks fairly simple to make, hardest part would be to find perfectly true flat material so as not to introduce another variable

 

[Mopar Bob]

hardest part would be to find perfectly true flat material so as not to introduce another variable

I used cabinet grade plywood. Which is what I had laying around. Its a very flat material and will not warp easily. Any plywood with lots of plies (layers) will work to ensure its flat. However the key element that needs to be flat is the top that has the level on it. I used $2 sheet of glass. Won't get much flatter than that.