Normal or abnormal flight behaviour? Firmware dependant?

[Big Ben]

I had/have some problems with my FC40 not behaving as expected. It had a problem rotating left spontaneously in hover but that seems to have disappeared after loading the defaults file (twice) and full sets of calibrations. What still remains it something that might or might not be classified as 'operating within normal parameters'.

Since I've had it we've had a fair bit of wind here. Nothing the Phantom can't handle but it limits the circumstances in which I have been able to observe its flight characteristics. I also have only one Phantom and cannot compare it with another one.

What happens is that when I am hovering in GPS mode and just give left stick yaw input the Phantom doesn't hold position but begins to drift pretty hard. Just as fast as it does in ATTI mode it seems but not in the same straight downwind direction. I'm not yet sure whether it behaves 'symmetrically' and its deviation is the same when rotating left as it is when rotating right. An asymmetrical behaviour would indicate something most likely not being as it should be but symmetrical behaviour might be within the norm.

Now I have read about people experiencing changed, less stable hovering behaviour after a firmware upgrade from version 4.00 to 4.02. Mine has the latest. I now wonder whethet it might behave differently if I would downgrade to v4.00.

Its behaviour could be described to look like this if it is indeed symmetrical. The two tracks do appear to be at a 90 degree angle from each other.

After releasing the left stick it returns to its original hover position from where it started its track.

My question is whether this happens to other Phantoms too when rotating? And have people experienced a change in behaviour as a result from firmware update(s)?

I'm not certain my Phantom behaves symmetrically when turning left and right. When I fly forward with moderate speed in GPS mode away from me and make a left turn into a 45 degree wind I see it moving left from the straight line it was on immediately. When I then let it fly towards me and make a right turn with the wind coming from the rear right (135 degrees) it rotates right but doesn't move from the straight line towards me. I have yet to do a test with the wind at 90 degrees both ways. Weather isn't very cooperative today but I hope I'll get the chance to test this today. I would also very much like to do some tests in zero wind but that's not going to happen any time soon according to the weather forecast.

 

[ElGuano]

The short answer is this happens. I've tested it extensively and it doesn't seem to depend heavily on COG as much as weight. I've downgraded to 4.0 to address but haven't found anything conclusive with that as well.

What is interesting is that 1) the direction of drift will spontaneously reverse on certain bootups, and 2) if you maintain a constant yaw, it will come back to last GPS position and eventually yaw in place very precisely. I think it's a flight controller prioritization design, and we don't have much user control over it.

 

[Big Ben]

The short answer is this happens. I've tested it extensively and it doesn't seem to depend heavily on COG as much as weight. I've downgraded to 4.0 to address but haven't found anything conclusive with that as well.

Do you mean weight has an effect on this but not COG?

What is interesting is that 1) the direction of drift will spontaneously reverse on certain bootups, and 2) if you maintain a constant yaw, it will come back to last GPS position and eventually yaw in place very precisely. I think it's a flight controller prioritization design, and we don't have much user control over it.

Huh? Reverse? That's extraordinary. I don't think I've experienced that (yet) but I'd have to study some videos to make absolutely certain. Does your Phantom have the same NAZA-M v2 with the 4.02 firmware?

In how much wind did you test it? Does it also deviate without wind? And then return while rotating? After how long? And/or does it deviate AND return in windy conditions (too)?

 

[ElGuano]

The short answer is this happens. I've tested it extensively and it doesn't seem to depend heavily on COG as much as weight. I've downgraded to 4.0 to address but haven't found anything conclusive with that as well.

Do you mean weight has an effect on this but not COG?

What is interesting is that 1) the direction of drift will spontaneously reverse on certain bootups, and 2) if you maintain a constant yaw, it will come back to last GPS position and eventually yaw in place very precisely. I think it's a flight controller prioritization design, and we don't have much user control over it.

Huh? Reverse? That's extraordinary. I don't think I've experienced that (yet) but I'd have to study some videos to make absolutely certain. Does your Phantom have the same NAZA-M v2 with the 4.02 firmware?

In how much wind did you test it? Does it also deviate without wind? And then return while rotating? After how long? And/or does it deviate AND return in windy conditions (too)?

Yes, on some bootups it will drift in the direction of the yaw, in other cases it will shift and drift against direction of yaw.

It drifts less at 900g than at 1400g. At first I was certain it was cog so I reversed my phantom (so the led faces forward ). That centered my quad's cog with 4400mah batteries and gimbal installed, but did nothing for the drift.

My phantom has the later naza-m v1 that is hardware/firmware identical to the v2 but comes in the older red/white housing. The led v2 works with it.

Tested on 3.16, 3.18, 4.0, and 4.02. On some days I think 4.0 is better, but on others I think it's in my head. One of the DJI guys said that in a previous firmware change, precise pirouettes were changed and the drift started (ostensibly to enable some other feature) but I haven't been able to pinpoint if that's true.

I think I have more hours of flight time testing this issue than some people have flight hours total on their phantoms....I've tested in no wind and light wind, and indoors, but I don't fly in high wind so I've never bothered there. It does drift more with wind, but if it doesn't TBE it always returns to the last GPS lock location and eventually spins on a dime. While rotating. Every time (unless it tbe's). You can try it, yaw moderately until RAP gets triggered, it'll start, drift away, come back, and eventually spin around one motor tightly.

 

[PoolGuy]

I have and FC40 also and have experienced the same thing. We've had horrible weather since receiving the unit, so I have only put it up 2 times before crashing it into the snow.

I'll be keeping my eye on this post.

I haven't found a way to subscribe. Am I missing something?

 

[davemcm]

Also noticed this with a stock phantom 1.1.1 (not FC40) and always just put it down to the wind and the fact that the phantom's body will go through an aerodynamic change when yawing, so when it rotates so far, the wind hits a higher/flatter side of the phantom and it drifts...

Not really had much chance to test it out on low / no wind, as we've not really had any days without wind (24mph with 37mph gusts today )

Reading this has made me think it might be something else though....

 

[mikeydaddio]

I guess I'm not of the norm here because I have not noticed anything like this. Phantom 1.1.1 with FW 4.02. This even prompted me to go back and watch a few videos where I was panning in a 360 to see if there was any drift and I can't detect any. Wonder why mine doesn't do it? Now, so far I've only flown in calm conditions so I wonder if it has anything to do with not being able to (proper) compensate for wind speed while yawing?

Mike

 

[The Editor]

Obviously assuming one is in GPS mode.......

The Naza isn't cleaver enough to calculate yaw off set dynamically whilst it's being actioned.

It is simply a combination of the counter rotating props giving rotational torque and the Phantom drifting with the wind (which can be different up high than at ground level) !!

Once the yaw maneuver is complete (you release the rudder stick) the Phantom will return to it's previous position immediately before the yaw since it has received no aileron or elevator input but 'knows' it shouldn't be where it's drifted to.

You will notice the drift is far more pronounced in windy conditions with it being virtually non existent on calm days. Also a slow yaw will show this drift up more since the Naza is not correcting until the yaw is over. I quick yaw will see virtually no drift.

 

[ElGuano]

I guess I'm not of the norm here because I have not noticed anything like this. Phantom 1.1.1 with FW 4.02. This even prompted me to go back and watch a few videos where I was panning in a 360 to see if there was any drift and I can't detect any. Wonder why mine doesn't do it? Now, so far I've only flown in calm conditions so I wonder if it has anything to do with not being able to (proper) compensate for wind speed while yawing?

Mike

Yours may not be doing it, or you may just not be seeing it. If your Phantom is higher than 30ft up it's difficult to detect, because the drift is usually within 6-15ft. Think of it this way - can you yaw your Phantom in a 10ft wide hallway without having it hit the sides?

 

[ElGuano]

Obviously assuming one is in GPS mode.......

The Naza isn't cleaver enough to calculate yaw off set dynamically whilst it's being actioned.

It is simply a combination of the counter rotating props giving rotational torque and the Phantom drifting with the wind (which can be different up high than at ground level) !!

Once the yaw maneuver is complete (you release the rudder stick) the Phantom will return to it's previous position immediately before the yaw since it has received no aileron or elevator input but 'knows' it shouldn't be where it's drifted to.

You will notice the drift is far more pronounced in windy conditions with it being virtually non existent on calm days. Also a slow yaw will show this drift up more since the Naza is not correcting until the yaw is over. I quick yaw will see virtually no drift.

I initially thought this as well, but there's more to it than just that. Regardless of what causes the drift, as I mentioned before, you don't have to stop yawing for the drift to be corrected. The Phantom can, and does correct during drift, but it takes a bit more than 5 seconds for it to kick in. It 'knows' it shouldn't be where it should be and reacts even with rudders activated. It's just delayed.

Allegedly, it also used to NOT drift with earlier firmware, suggesting either the cause was absent, or more likely, the correction was not as delayed as it is now, for one reason or another.

 

[The Editor]

I have to confess I haven't seen that !!

I will have to investigate now you said that

 

[mikeydaddio]

I yaw all the time just 15-20 feet above the ground and just haven't seen it. When the wind stops blowing 30 MPH I can take it out and take a closer look. What should I do to maximize the effect: yaw slowly in one direction and continue doing that?

Mike

 

[ElGuano]

I yaw all the time just 15-20 feet above the ground and just haven't seen it. When the wind stops blowing 30 MPH I can take it out and take a closer look. What should I do to maximize the effect: yaw slowly in one direction and continue doing that?

Mike

That's the way to do it. Moderate yaw speed, constant rotation. It's more noticeable at lower altitudes of course. If it drifts less than 6ft that's very good. If it drifts 10-15ft that's very noticeable.

Or, try this: be thankful it doesn't happen and think no more of it I've been trying ways to get rid of it for months!

 

[Big Ben]

Yesterday weather was too bad to fly. Today three more flights although wind was pretty strong. I flew with the wind coming from the rear so I could observe the direction of the drifting path better. I still have to do a full analysis of the videos but basically it behaved very erratically. Rotating left it veered off left... but rotating right it did so too... and rotating left it also veered off right... and it also drifted straight downwind sometime... it veered left and then it changed course and crossed from left to right until RAP kicked in.

I also did simple 90 degree rotations. Mostly it then moved backwards away from the centre line. I started to analyse...

In GPS mode the Phantom constantly has a position it wants to return to if it gets moved away from that position which the wind will cause. Apart from that it has to react to the control inputs. It has to combine the two. When it is hovering in (strong) wind it will basically always be in the process of correcting its position. Depending on the direction and the distance it was removed from its 'hold' position at any time it will be flying in a different direction and with a different speed making 'micro-movements'. When I then tell it to rotate that command will come at different moments while making different micro-movements. One time I might tell it to rotate while it is rolling to the right and pitching up and moving at a 160 degree course and another time I do so while it is pitching down, not making any rolling movement and moving at a 250 degree course. It will also be making different rotational corrections. It is imaginable the path it will follow will depend on the combination of these factors at the moment it starts to do the rotation.

Rotation is also a fairly subtle motion the Phantom makes. It must be achieved by accelerating the two props that turn in one direction and slowing down the other two. This will result in a rotational force while keeping the total lift constant. While doing so it also has other demands on the relative speeds of the props for keeping it level or in the position required by the control inputs to make it move in other directions. All these requirements have to result in coordinated and constantly varying speeds of all four propellers. Not a minor task and pretty complex. Also the thrust forces are MUCH greater than the relatively small rotational forces of the propellers. No doubt the algorithm for doing this can be tweaked in many ways. Indeed a higher priority for one of these demands is likely to affect how well the other demands can be met.

Ideally you would test two Phantoms at the same time in the same conditions flying 20 meters apart from each other in a sort of 'synchronised flying' exercise to compare the flight behaviour to find out how much is attributable to flight algorithms and how much to individual differences in hardware or calibration.

I'm beginning to feel as if I'm a DJI test pilot since so far 90% of my time flying my Phantom has been spent testing and analysing instead of flying just for fun and shooting nice video.

They should pay me for this. :roll:

And ehh... I've uploaded a 4 second clip of my second flight today. Less than 2MB.
It's here. What do you people make of this?