Two months ago, we build a quadcopter with a custom frame and equipped with:
Matek H743 Wing V3
T-motor Velox V45A V2 (4 in 1 ESC)
4x T-motor F80PRO 1900KV
6S2P Li Ion battery of 9000 mAh
TFmini plus rangefinder
The whole quadcopter weighs about 2300g (it is also equipped with LiDAR and an onboard computer).
We follow the steps of the MethodicConfigurator to adjust and configure the FCU properly. However, during flights we notice a strange behaviour: in some moments of the flight the drone starts to make a very loud noise (some motors are running faster, much faster). It is like if the vehicle tries to compensate some yaw turn. We did not see a pattern in doing this “yaw correction”, it just happens, all of a sudden, a few times during a 10 minute flight. Taking a look at the logs we see that when this happens the field RATE.YOut is quite big (>0.4) and the unbalance between the motors turning CCW (high) with the ones turning CW (low) is very high (see image below). The rest of the flight (when this behaviour is not happening), in my opinion, the motor speed is very balanced.
Moreover, we realize that if we are using the on board control (using mavros in GUIDED_NOGPS mode sending SET_ATTITTUDE_TARGET messages) if we command a yaw turn during this strange behaviour, the quadcopter first turn a bit to the opposite direction. This is noticeable in the upper image, but can be clearer seen in the following plot:
Both images represents the same time instant. As we see in this image, we commanded a certain yaw speed to turn CCW but firstly the quadcopter turned CW.
We do not understand why is that happening. According to the MethodicConfigurator the RATE.xOut should be less than 0.1 (in roll and pitch this is achieved) but for the yaw is higher most of the time. Why is this happening? How can I address this problem?
We leave the log here.
I want to emphasize that this strange behaviour occurs independently if we are flying using ALT_HOLD or GUIDED_NOGPS, i.e. it is not caused by the on-board control, which in ALT_HOLD is completely disabled.
Sorry I did not specify it, but the Matek H743 V3 does not have magnetometer and we did NOT add an external one, so should not be the problem. Currently the yaw estimation is performed using only the internal gyros (I guess), and comparing with the optitrack estimation is quite good (only have some drift eventually).
The sound it is not caused when I try to command yaw (I mentioned it because you changed the title). The sound is caused due to this strange behaviour for which I am asking for. It happens when I hover trying to maintain the XY position in ALT_HOLD mode, not when turning.
But even with this strange noise and motors speed unbalance the yaw is maintained quite well during the whole flight. Only do strange turns when it is controlled by the companion PC. If the flight is only performed in ALT_HOLD the quadcopter do the loud noise and there is not appreciable yaw turn.
I have not the possibility to test that. The drone was designed to fly indoors.
With different props have exactly the same strange behaviour.
How can I address the high RATE.YOut? As I mentioned, my guess is that the problem is caused due to this value.
I forgot to say that in the first flights (when the PIDs were default tuned according to methodicConfigurator) this behaviour were already present. I don’t know if that can help to find out the problem.
By the way, the yaw estimation does not jump, is quite good compared to the one offered by optitrack.
CCW Motors 1&2 (C8&C9) are always spinning significantly differently from CW Motors 3&4 (C7&C10). This will always limit the quality of the tune because each time a pitch or roll input is made there will be yaw coupling. I’m guessing the arms on this frame are round? The motors need to be leveled and equalized. In a level hover all four motor values should be more or less the same. It’s likely that once you fix this, you should consider re-tuning the quad.
As i remember you struggled with motor inbalance and yaw tuning in your other post about ESC overheat. You need to tune down that Yaw axis and im pretty sure that will fix yor problems.
Yes I have been notified in a past discussion but I tried to adjust it untill the average RCOU of all motors in a flight was quite similar. For this reason, and because I don’t see any appreciable tilt, props in rear anf front motors are exactly at the same high, I thought I solved the yaw imbalance problem. My fault…
No it is not, but, as you can see in the image below, the carbon fiber of the frame is very thin and tend to suffers deformation tightening the screws (this is how I tried to fix the yaw imbalance: tightening some screws more than others and test till the average was very similar). Now I’m working on cut a carbon fiber circle thicker to be placed between the frame and the motors, avoiding that deformations. I will also change the motors pointing uppwards (2 and 4) to be pointing downwards as the others are. Hope that fix the yaw imbalance problem, in any case, facing all motors down I can do a better leveling of the motors using the “props technique”.
Yes I was struggling with that and try to do some fixes, I reduced the imbalance, and thought that was enough, error… I will try to adjust more accurately that imbalance.
I didn’t think that such a tiny (visually unnoticeable) tilt in the motors can cause such a big problem.
There is no point making another post about same problem, we gave you suggestion on your last post how to fix motor inbalance,cg, and yaw tuning. Do what we suggest you without miss any step and you will fix it. Posting same problem multiple time will not fix it, only fix is to follow instructions.
As I mentioned, I thought I really address the yaw imbalance problem because taking a look to the RCOU# values, the average in whole flight is more or less the same for the four motors, after the screw adjustment.
You should not look in that way, the 4 lines of the graph in perfect situation, should be overlap during hover, no pilot input. In real perfect situation, they are overlapped and you only see one line though you call out the graph for 4 motors.
mine is not close to prefect, here an example. outdoor, come with wind condition, no pilot input.
I mean that you need to have all four motors pointed in the same direction to solve this yaw imbalance. You need to get this solved before you work on any other tuning, because this will have an effect on all other aspects of flight. Mechanical issues need to be solved before working on the software/firmware because the best flight controller in the world can’t make a bad mechanical setup fly well.
I do all my best to level precisely the motors, first I put all 4 motors facing down (pusher configuration) and then I adjust the orientation of the motors to be on the same plane and facing completely vertical. The strange sound is still present during flights. I leave a video here for you to hearing the noise what I am talking about. An here the log of this flight: log. Trying to find the error I also changed motor 2 for motor 4 (no change). And then motor 1 for motor 3 (neither any appreciable change). In every change the level and facing was re-checked.
The RATE.YOut seems to be a little lower (before the adjustment reach peaks of ±0.5, now the maximum I see in this flight and others is of about ±0.28). I don’t know how to continue. Can I already autotune or do you think is still too much imbalanced? In affirmative case, do you think that maybe that imbalance is caused by any other thing?
