Two issues with position hold.

Joined
Jan 17, 2014
Messages
344
Reaction score
1
I have had the opportunity to fly my FC40 on three days outdoors now. Each time three flights (3 batteries). It's mainly been testing things work as they should and are reliable. Of course I've also shot video from increasing heights. After these flights I notice two issues about which I'd like to know whether these are 'normal' and others experience these too and may signify shortcomings in the firmware or whether they signify a problem with my particular FC40.

All flights were made from an enclosed playing field which offers protection from the wind near the ground but where the Phantom is exposed to the full force of the wind higher up. All firmwares are latest versions. All flights after CL and HP rapid green flashes and with single green flashes indicating >6 satellites.

The first day the wind was about 9m/s. Near the ground it was much more relaxed but when after a while I went a bit higher the effects became very noticeable. I didn't go very high and everything was too new to make any reliable observations about the two issues.

The second day, yesterday, wind was much less at around 5m/s. I noticed when I let the Phantom hover 1,5m above ground (GPS mode) it tended to yaw left spontaneously. I had to correct it regularly. Two compass calibrations didn't change this. I decided to do an advanced calibration later at home. I did and calibrated the Tx too and flew indoors where I still noticed this effect (ATTI mode).

I reset all gains to their default and flew it again this morning (wind 7m/s) before the weather takes a turn for the worse as it is expected to do (more wind/rain *sigh*). I tested the hover on all three flights (GPS mode). Mostly it went well but during the second flight in one of those tests with hands off the sticks after 40 seconds of holding orientation nicely it rotated about 135 degrees left and remained there until I took over again because time was towards the end of the battery life and I wanted to check the signal light. Indoors it was much more prevalent. I'm getting the impression that it only happens in less turbulent air when the Phantom doesn't need to make very strong corrections. When it is very actively holding position the orientation is held too. When the air quiets down it starts to rotate left. If that's the case this would obviously be very undesirable in still air when you should be able to shoot the best video. Can't have constant spontaneous panning.

So my question is whether others experience similar behaviour and of course any suggestions to fix this are welcomed too. I did every calibration available and don't really know what's left for me to do to fix this.


The second issue is that although near the ground the Phantom holds its horizontal position when I go to greater altitudes with stronger winds (yesterday 30m high and today 50m) the Phantom tends to drift downwind in GPS mode. Much more than should be within its GPS position's accuracy. I regularly have to actively direct it back to where I wanted it to be. This is not what I expect nor want. The whole point of position hold is that you don't have to actively fly the Phantom but can focus on shooting the video. In the stronger wind I cannot do that at the moment. Do others experience this too?

I have since realised both yesterday and today I was at the geofence ceiling level. This could be a factor. Maybe another bug. Maybe when it momentarily goes beyond the vertical geofence limit which it probably does on occasion when at that limit it changes its 'horizontal hold coordinates' values (too). Next time I will try to move below that level and see whether it has any influence. I'd be interested in others experimenting with this too. I don't think there's more flying for me today with wind being 11m/s now (25mph) and rising. Next few days not looking good either. :roll:

With the wind forces I already experienced the benefits of a gimbal become very noticeable in my videos. Today at 50m the bank angle was above 20 degrees in order to overcome the wind and hold position. I'm really waiting for a day without any wind at all. Should be nice.
 
Three more flights this morning before weather turning worse... AGAIN!

The yawing left I experienced again very pronounced during a low hands-off hover (after two more compass calibrations) where it ultimately turned a full 360 degrees in a period of about 1,5 minute before I took over again. Two big 90 degree sweeps and a few smaller ones and in between it is doing as expected. This is not right.

The downwind drifting I have now discovered happens when I rotate the Phantom. Facing in the wind it held position. During rotation it drifted almost as if in ATTI mode. Not quite probably but the wind was so strong that it felt that way. The wind higher up was quite strong today and it drifted downwind pretty quickly when I was yawing. I had set the geofence ceiling to 75m today and probably reached it but went slightly down again so that ceiling wouldn't be a factor. During the last flight the wind was gusting and at that altitude was so strong that after drifting downwind I couldn't really see the Phantom move when I tried to reposition it back to where I wanted it to be while giving full stick forward so I quickly decided to descend. The camera was looking down pretty steeply during that full forward flying against the wind. I'm guessing it was (well) over 10m/s for a short while. Closer to the ground it started to move forward again. The drifting might be stronger during right turning than left turning. I would imagine the two effects I'm experiencing to be somehow linked but I don't know how.

If people would try rotating in GPS mode in a good breeze to see whether they could replicate this phenomenon I'd be very interested to hear the results. Or have people observed this too during their flights?

Question is what to do? Since DJI apparently cannot be reached for adequate support it seems I have only one option. To return it for an exchange. Or does someone have a suggestion how to fix this?
 
The yaw problem sounds compass related to me. Mine doesn't yaw at all with hands off even with a little wind although I haven't tested in wind like you had: seems like a lot of wind. Is there anything on your person or on the craft that could be causing EMI that could make the compass wacky? I know I've had some strange effects (knocking out GPS) just from the RF leak on my GoPro!

Mike
 
The only electrical/electromagnetic thing is the FC40 camera which is transmitting its WiFi signal to my smartphone. I doubt it to have any effect if I take the camera off.

And the yawing seems to be worse with little or no wind. In stronger wind when more actively making corrections it seems to keep its orientation better.

The last compass calibration I did holding the Phantom away from me and the transmitter to minimise any effect that I might have.

Regarding the drifting the Phantom seemed to remain in a more horizontal position and wasn't leaning against the wind so much while changing its orientation. Maybe I'll up the basic gain again. Regardless of all these things the Phantom should simply hold the coordinates in its memory as long as I don't give any right stick commands. Even when it drifts while rotating it should return to that position in memory after it stopped rotating. It doesn't do that. I looks like the coordinates where it is keeping its position have changed during drifting while rotating. This might suggest a firmware issue but then others should experience this too.
 
Have you connected it to the assistance software and watched the stick positions coming into the controller? Just trying to rule out the transmitter (even occasionally) sending out a yaw/tilt signal. Also, you should probably go to the IMU tab in the software and make sure the X/Y/Z values aren't jumping around. Set it on a flat surface and X and Y should be close to zero and Z should be close to 1.0 (force of gravity). Then move it around a bit and set it back down and make sure the X/Y/Z values go back to 0/0/1.0.

Mike
 
Do the basic first. Then advanced. If that fails redo the whole firmware update. It's not hard and takes about ten to fifteen mins. The download and import dji default parameters and calibrate again as you first did the day you got your phantom.
Good luck.
 
I have noticed that while the Gyroscope values for the X-axis and Z-axis fluctuate around the zero value the Y-axis value tends to favour a slightly positive value in the beginning but this seems to disappear after a minute or so. Since the Y-axis is the pitch axis and the Z-axis the yaw-axis I don't really think this is significant. The sticks stay rock solid at the zero value and don't fluctuate ever, unless I move them.

A random moment shows all good values in my opinion.

21msoyr.jpg



Unless I have a good reason to expect a firmware reflash would have value I'don't intend to do it.

Three more flights today. A bit less wind allowing me to permit a longer period at height before I compensate again. I now noticed that although the Phantom drifts downwind while yawing once I stop the yaw command the Phantom slowly returns to its initial position. In a later flight when the wind had picked up a bit this didn't seem to happen.

Setting aside whether the Phantom should lose position or not while yawing I now have the theory that position hold might actually work but that the speed with which the Phantom autonomously corrects the deviations from its position is relatively low. Possibly the same speed with which a RTH procedure is executed which is by no means the maximum speed it's capable of. Although that low RTH speed makes sense to prevent high speed collisions with obstacles in its path it might be acceptable or even desirable to use more aggressive speeds during position hold especially when the Phantom has risen considerably since take-off where the wind might be (much) stronger. During my tests in stronger winds the Phantom might have returned to its original position if I had given it more time provided the wind speed had fallen below its autonomous 'manoeuvring speed'. I didn't want it to drift too far though because I wanted it to stay above the small field and not drift over the surrounding houses so I didn't wait too long before I took over again.

An interesting test would be to let the Phantom hover at shoulder level and then grab it and pull it away some considerable distance and observe at which speed it returns and whether it indeed does fully return when released again. People with a GPS logger would then be able to determine this manoeuvring speed.

The yawing in hands-off hover is still there after both a basic and advanced calibration. I also notice it tends to yaw left during take-off. I don't think that's done by me involuntarily. Maybe a motor/ESC hardware issue? If the drift is indeed caused by the strong winds and the low manoeuvring speed what still remains is this annoying yawing. If I can't fix it I'm afraid I'll have to return it and get another one.
 
Does sound like a balance problem or maybe one of the motors being out of spec and it's fighting that "weak" motor to try to balance itself. Only other thing I can think of is to play with the gains. The gains are what determined how fast it tries to correct for course changes.

Mike
 
I wish more information was made available by DJI about HOW exactly the Phantom determines and holds its position. I would expect that the compass input would be used to determine and hold its orientation along the Z-axis. If it would be facing SW for instance I expect it to keep facing SW no matter what I do with the right stick. Only when I tell it to rotate using the left stick to give 'rudder' it should deviate from that SW facing position. How difficult can it be to measure its compass direction and stick to that? Even if a motor wouldn't be functioning the way it should that shouldn't matter. As long as the NAZA 'knows' in which direction it is supposed to face then it could always regain that compass direction... IF that's how it is (supposed to be) implemented in the firmware. But we don't know.

My personal wish is that DJI would open source the firmware so (groups of) people could contribute to improving it and could help to develop additional useful functionality. Then we would be able to actually KNOW the inner workings of the Phantom/NAZA and this yawing could be understood better.
 
Yeah, to say documentation is lacking is an understatement! And I agree, that's the way it should work. That IS the way mine works! There are only two things I can think of that would cause a problem. The first would be "leakage" of other channels such as pitch into the yaw channel. Carefully observing the sticks in the software, have you confirmed that you can move the right stick wherever you want and it doesn't inadvertently show left to right changes on the left stick?

If that's not happening, it's almost got to be a bad compass. I would do as suggested earlier and reload the whole firmware, reload the parameter file for that firmware, recalibrate the compass, IMU, and even calibrate the transmitter. Colin Guinn explained in the "how to avoid fly aways" video that bad and strange things can happen if the compass and the GPS are fighting each other. For example, you may know it is facing SW but when it drifts a little and GPS wants to compensate, if the compass is giving conflicting info (versus GPS), it can do odd things as it tries to use the compass to undo the drift but in doing so, ends up heading the wrong way.

One final thought (really grasping at straws now), when I updated to 4.02, my transmitter didn't connect AT ALL. After hours of fiddling, I finally found out that you have to change the RC type in the software from PPM to D-Bus. Apparently they changed something but forgot to set the default to D-Bus for the latest firmware. Now, I just assumed that it wouldn't work at all in PPM but I suppose it's possible that on a different receiver/transmitter, PPM might "work" but not properly. Is yours on D-Bus?

Edit: and yeah, Ozzy has a good idea too. Try the degauss. After doing all the things mentioned here, if you end up at the same point, I think it's time to return & exchange it as something in the hardware must be defective. But do all you can (that is within your control) to recalibrate everything first.

Mike
 
The current firmwares are the ones it came with from the factory. At least it came out of the box that way, assuming nobody did anything to it since it left the factory. Which made me remember...

I bought it from Conrad. A German firm which also for decades has had a shop and later a web shop in The Netherlands. It was shipped from Germany. The box had most likely been opened by the importer because I found a European mains cord for the charger inside plus a German manual (but no English manual) plus some RC stuff leaflets.

Now I think maybe THEY very 'helpful' also did a firmware update to prevent customers doing it wrongly saving them many returns... provided they do it correctly!!! MAYBE they did but didn't load the defaults and recalibrated. So now I DID load the defaults (stupid they don't supply defaults with the correct transmitter setting - you load the defaults, change the Tx to D-Bus and save them again and hey presto... you now have correct defaults to put on the web site). Also clicked the defaults button in the Assistant. I was thrown an Error[27] which seemed only fair since I hadn't yet done a compass dance nor acquired GPS fix. So I calibrated the IMU, basic and advanced, the Tx, got a GPS fix and did the dance and checked every setting.

I didn't want to stick a magnet near my compass module yet because from what I've seen if there's nothing wrong with it you potentially create more problems than you solve.

I then flew it indoors for about 5 minutes and haven't seen any inclination to rotate left during that flight. This is hopeful. I'm now recharging that battery, for the first time set at 2A, and plan to do some outdoors testing once it's full.

I'll keep you posted.
 
Back from three flights outside. Wind higher up 8m/s. Near the ground very doable. In ATTI the drift is no more than walking speed at low altitude.

MUCH better if not gone! Did some hands-off hovering and it didn't rotate. Higher up the drifting seemed to go better too but I had changed the basic gain for pitch and roll (back) to a higher setting hoping that might have a positive effect on the drifting too. I lowered basic gain to defaults after I experienced the left yawing so the higher gains shouldn't be the determining factor for that. Since tomorrow winds will be very/too high I wanted to test the higher gains too although ideally you only change one thing at the time. I'll do a test with default gains again later.

Only once in the first flight after having come down from higher up when in ATTI mode drifting with the wind just over the sheds in the adjacent gardens it suddenly turned left again. I deliberately hadn't done a compass calibration in the field because I wanted it to be the same as during the flight indoors. Before the second flight I did recalibrate and didn't see it again. Maybe something in those sheds interfered. Quite possible.

One thing I have tested now because I started to notice it. When you make rapid circular movements with the right stick making your bird 'rock and roll' it does gradually rotate in the same direction as you make the circles with the stick. It is reproducible and is possibly caused by the constant movements with both the X- and Y-axis of the Phantom constantly not being in the horizontal plane. I expect others to experience this too (please try it and report) and it could be considered a bug in the firmware since without left stick rudder commands it shouldn't yaw.

It would appear my bird will stay in her new home and will not be sent back if during subsequent flights she will keep performing as today. This makes me happy because I wasn't looking forward to all the hassle and the wait for a new bird and then the testing again. Now I'm finally past the first testing phase I can look for interesting flights in different locations.

Thanks for your input guys. :)
 
Glad that's worked out! Today is the first day in about 4 days that we've had good flight weather so I'll try the quick circles with the right stick today and will let you know. I've run slow wide circles like that and it didn't turn but haven't tested quick ones.

Mike
 
The 'rock & roll effect' isn't something you'll immediately notice. It isn't very strong. After 8-10 rapid stick circles the deviation is perhaps 10-15 degrees but it is there and it works in both directions. It didn't give the impression as being related to my earlier problem, which now should be gone anyway, so it potentially seems of interest to all.

I hope more people will try it and give feedback.
 
Big Ben said:
The 'rock & roll effect' isn't something you'll immediately notice. It isn't very strong. After 8-10 rapid stick circles the deviation is perhaps 10-15 degrees but it is there and it works in both directions. It didn't give the impression as being related to my earlier problem, which now should be gone anyway, so it potentially seems of interest to all.

I hope more people will try it and give feedback.

Hey Ben,

I haven't tested that exactly yet (but plan to now). However, to me logic would dictate that doing rotations on the right stick like that will naturally cause a build-up of torque on the yaw axis, enough that you eventually see the aircraft turn in that direction (corresponding with the direction of stick rotation).

Think about it this way...if you spin a coin on a tabletop, after it loses the bulk of its momentum but just before it falls flat it will "wobble" much like the Phantom does when you first start rotating the stick like that. Now picture the physics of the coin in reverse...if you start with inducing that wobble (and continue to add to the forces), eventually it will build yaw force enough to spin.

This is purely speculation on my part, but it seems perfectly normal to me that you'd see the yaw effect from that type of stick input.
 
Just in from a 7 minute flight. For the first 60 seconds, I tested this. I rotated the right stick around in a circle and the Phantom did a wobbly top type motion. Sure enough, it does exactly what you describe! It would always slowly yaw in the direction you are circling the right stick. Not much: as you say, 10-15 degrees after 8 or 10 rotations of the right stick.

Not sure I agree with the buildup of torque explanation. I can see how that might cause it to yaw initially but the compass still has the proper bearing so once you stop it should return to the initial heading because there has been no input on the yaw stick. The only thing I can figure is that they've programmed in some yaw effect if you do certain things with the right stick: maybe because some movements of the right stick "should" cause a small amount of yaw?

Mike
 
OI Photography said:
This is purely speculation on my part, but it seems perfectly normal to me that you'd see the yaw effect from that type of stick input.

Yes... if it weren't for the fact that the NAZA is supposed to have a bit more 'brains' than a spinning coin and is supposed to actively keep its orientation with the help of the many sensors it has, one of them being a compass which should give it the possibility to 'stick' to a given magnetic course until the pilot tells it otherwise. Otherwise being left stick rudder input. In the absence of such input I expect it to maintain its compass orientation.
 

Members online

No members online now.

Forum statistics

Threads
143,066
Messages
1,467,358
Members
104,936
Latest member
hirehackers