Guorium
![](https://forum44.djicdn.com/data/attachment/common/aa/common_14_usergroup_icon.png) Second Officer
Flight distance : 513061 ft
Australia
Offline
|
Hi all,
Just to get this out of the way first. I don't get paid to do these tests in my spare time. There is not much to gain other than helping other people and me to investigate issues with their mini, particularly with the uncommanded descent incidents. I have a theory about pressure deviation like many others have mentioned. I knew I need to make some tests and data available to back other the theory otherwise it would be prone to scrutiny. I believe theories with evidence in the form of flight data analysis is critical to finding out what is going on with the mini and to help other people to become more aware of their mini in certain risky situations (e.g. low above water). Evidence > some guy claimes to be experienced. So here is my test. It is open to all to discuss and learn. Please be polite and critical. Again, I do not get paid to do this and some individuals on this forum are just too blunt to notice.
Aims
- Confirm altitude deviation with abrupt stops in S and P mode, establish a connection to barometer pressure reading and visual positioning system (VPS).
- Confirm manual descent speed protection is related to VPS.
- Find out if the extended flight in S mode (>12min) causes uncommanded descent.
Methods
Have the IMU, RC and gimbal calibrated at home. Have the compass calibrated on the field. Firmware V01.00.0400.
Aim 1
I get up on this nice Sunday morning and headed to the flat grassy park filled with fresh air. The temperature is about 20 Celsius and no wind.
I takeoff, line the mini up at eye level to get a good visual confirmation of any altitude deviation. Get in S mode. And perform the 1-2-1-2-3-4-3-4 path described in the map below. Full stick on start and full release before the stop. The drone faces the top of the map at all times. So 1 is forward, 2 is backward, 3 is right, 4 is left. Each direction is duplicated with the designed path.
This flight path is then carried out in P mode at eye level. Then at 15m (just outside VPS' infrared sensor range) above ground, in S mode and P mode. Then at eye level with VPS taped to simulate flying above the water in S mode only. I verified that the mini is indeed blind with the black tape applied with my hand, it did not react to my hand under its belly like it normally does.
Aim 2
Full stick punch up in P mode from 0.5m to 40m. Release fully to stop. Full stick descent back to 0.5m. Repeat twice.
Get into S mode and repeat the same procedure twice.
Tape the VPS then repeat twice in S mode.
Aim 3
This test was the least systematic of the three. It was tested because someone whose name (I think Christian U) I forgot here claimed to have isolated his descent issue to over 12min of sport mode action. I performed and a number of full-speed forward flights in sport mode before the 12 min mark, then did a couple more after 12 min. VPS was covered the whole time.
Results
Aim 1
Fig.1 Eye-level, S mode.
Fig.2 Eye-level, P mode.
Fig.3 15m, S mode.
Fig.4 15m, P mode.
Fig.5 Eye-level, S mode, VPS taped.
Aim 2
Fig.6 P mode and S mode punch up 0.5m to 40m and back.
Fig.7 S mode punch up 0.5m to 40m and back, VPS taped.
Aim 3
Fig.8 S mode forward thrusts before 12 min. ~4m high. VPS taped.
Fig.9 S mode forward thrusts after 12 min. ~4m high. VPS taped.
Discussions and conclusions
Aim 1
I can visually confirm the forward and left flights in Fig. 1,2,5 all show noticeable but small climb as the drone is flying to the end. Additionally, a more pronounced climb was observed immediately upon the full release of the stick (what JJB saw in his quick stoptest). Eventually, the mini corrects itself to the start height by lowering altitude gradually if left to hover in place.
In backward and right flights, the mini drops a bit as it flies long, the change is very small. Barely visible by eye and very subtle on the graph.
S mode forward and back flights made a "~" like feature (tilde) in the blue barometer altitude trace. The right and left flights made the "∽" shape (reversed tilde) in the barometer altitude trace. The inversed tildes are way less pronounced and sometimes hard to identify in the graph. The same trend exists in the 15m tests in Fig 3 and 4 albeit I could not visually confirm any altitude change. (Not tall enough sorry) The two types of features exist throughout all test altitudes and VPS state, the barometer reading features appear to be determined by the directions of wind created by the mini's movement. S mode does make the features more pronounced but the shaped are not changed.
I conclude, the fact that this is directional and faster speed does it make it more obvious makes the issue pressure related. The barometer's pressure calculated height gives one type of response in forward and aft and another type in lateral flights. This could be due to the positioning of vents on the fuselage of the body. On the front, we have this big inlet behind the gimbal.
This lets air in fast in during fast forward flights. The air goes into the battery bay where the barometer is located.
For lateral flights, the big inlet does not work effectively. The mini has two lateral mesh vents and vent gaps around the heatsink/mainboard. Those might be responsible for airflow in the fuselage during lateral flights. Why are the pressure features opposite of each other? ~ and∽? I actually do not know, I am not an aerodynamics expert. There might be some complicated flow patterns inside the body dictating what I see. It is very difficult to model even for DJI I wager.
What I did notice for the VPS deactivated flight is the altitude fluctuates quite wildly compared to VPS assist mode. That is normal. However, on occasions the altitude correction (lowering) of the mini after an abrupt stop from high speed can coincide with a downward altitude fluctuation. The combined result is the mini appears to be dropping out of control. You should be quick on the stick to save it because deactivated VPS means mini will keep dropping til it hits something, unless the altitude correction + fluctuation halts before the impact. It would be totally another issue if your mini is irresponsive to stick command in such a situation, and I am not covering that here.
Aim 2
Fig.6 clearly shows there is no speed limit when ascending past the VPS range in P mode. But when the mini comes back down, landing speed protection kicks in (after the vertical orange lines) making vertical speed (blue trace in the bottom panel) drastically reduce to zero. In S mode, the vertical speed reduction kicks in faster than P mode, as soon as VPS sensor picks up the signal it starts braking.
Fig.7 shows VPS was indeed at work for the descent braking. With VPS taped, I manually braked at about 3m above the ground or my mini would have slammed into the grass. Fig.6 shows the mini started automatically braking at about 5m above the ground. Again, be real careful when VPS is not working.
Aim 3
I did not observe any noticeable flight behavior changes before and after 12min of operation (overheating) in S mode, both when visually observing the flights and the data in Fig.8 and 9. But Christian U did say he would not expect that from every mini as he did not get that from all of his 3 minis. Welcome to comment if anyone sees anything I did not.
My final words: Thank you for looking through this rather lengthy report. I hope this helps people out there. Please be critical to this report and let me know about anything you find suspicious. I spent my Sunday afternoon doing this out of the goodness of my heart and interest in the hobby and the truth for all users. I do my best to be logically sound. Special thanks to JJB for making such a useful analysis program!
TLDR: Flying low without VPS can be really risky especially when going fast! Keep your eyes peeled and go easy on the brake.
Logs: Dat file for eye-level S & P mode abrupt stop tests
DAT File for 15m S & P mode abrupt stop tests and descent speed test has been lost on my phone due to unknown reasons.
Dat file for VPS taped tests on S mode vertical descent and eye-level abrupt stop tests and 12 min tests
|
|