Phantom 3 out of control at 81 mph and no wind!

A difference in motor speed causes a constant rotation around any of the 3 axis, it does not result in a constant (pitch) attitude. Equal motor speed theoretically results in a constant attitude (roll/pitch/yaw) at a constant speed and heading. Flying forward at 50 degree pitch attitude backwards or sideways at full speed, all 4 motors will run at the ~ same speed. Only a change in attitude requires a temporary difference in motor speed.

If I understand what you're saying, you conclude an attitude based on relative motor speed. You cant do that. There is no correlation between them. Thats like concluding the direction a car drives by looking at the steering wheel. All you can say is that the heading will remain constant when the wheel is centered, and will change when its not. But its not heading north east just because the wheel is slightly to the right.

Similarly, if you fly a quadcopter in acro mode (which the phantom doesnt have, but bare with me), stick neutral results in equal speed for all motors - Im simplifying and ignoring external factors. The quad will happily continue flying forward, backwards, sideways or upside down with all 4 motors running at the same speed. A "thrust angle" as you call it would correspond to a stick input in acro mode in any direction, which will cause a difference in speed between front/back or left/right which results in a constant roll or pitch rotation for as long as there is a difference in speed; until its upside down and back again and upside down again. Well, or hits the ground ;) But you can not tell attitude/orientation from motor speed, or vice versa, without accelerometer data, the quad can not maintain and has no idea of its own attitude relative to the world. It can only detect and cause changes relative to where it was before.

Or Im misunderstanding what you're saying.
The following is from part of a flight I did a couple of months ago. I think it covers all the points we're talking about.
upload_2016-7-7_17-30-20.png

It begins in a hover with pitch about 0. The elevator is pushed forward and the front motors slow down. Pitch decreases to about -7. This combination of slower front motor speeds and -7 pitch continues for about 2 minutes. The P3 does not continue pitching down and tumble, it stays constant. At about time 252 the elevator is eased back, the motor speeds become the same, and speed reduces to 0.
upload_2016-7-7_17-37-20.png

Here is the thrustAngle
upload_2016-7-7_17-38-28.png

thrustAngle is a prediction of which way the P3 is moving relative to it's frame. Of course it isn't used by the flight controller. It's only useful for guys like me who want to know which way the P3 is flying relative to it's frame of reference. BTW, it's more accurate than I would've expected.

Your analogy about the steering wheel needs to be modified. If the wheel is turned right you don't know the heading but you do know that the car is turning right.

I see you removed the claim that neutral sticks will result in a continuation of any pre-existing velocity and attitude. Any P3 pilot will tell you the result is roll, pitch, and velocity all equal to 0. Have you flown a P3?

I don't know much about acro mode and can't really address that point. But, the P3 doesn't have acro mode so what's the point?

Could you do me a favor? If you have further claims please provide data like I've done.
 
Yes, Ive flown a P3 as well as probably 3 dozen other types of multirotors ranging from single motor ducted fans to octa copters. You missed my point on the acro mode. Its not about stick movement. In acro mode the flight controller has no knowledge of its attitude vs ground, in many cases, it doesnt even have the sensors to measure it. So it doesnt know if its flying level or at a 45 pitch angle. And yet, if you ignore external influences, it doesnt require any pilot corrections to maintain any given attitude. The quad will not return to level either fast or slow, nor will it substantially pitch or roll away from its current attitude unless you apply differential thrust. And differential thrust will result in a constant rotation at a rotation speed that is proportional to the speed differential. Ive suspended enough quads on strings for flight controller PID tuning to know this for fact.

Of course, there are secondary and tertiary effects. I'll admit its even possible that due to aerodynamic effects from the frame or gimbal, there is on average a tiny constant difference in motor speed necessary to maintain a certain attitude. You might even be able to measure this. But the effect could vary vastly, for instance simply by installing a gimbal protector. And more importantly, your interpretation is going to be completely wrong if you dont understand the primary effects. Your original assertion where you said that a thrust angle of around zero means its moving forward along its X axis, is simply incorrect.

Tell me, what value would you expect if it where moving backwards, sideways, or for that matter, yawing at a constant speed? What value would you expect if the quad is level, but drifting in the wind?

Why would you even use this "thrust angle" to try to deduce some marginal secondary effects, when you have accelerometer data that will flawlessly tell you the quads attitude?
 
Last edited:
At last, a proper intellectual discussion, keep it coming, I'm learning all the time from your (Vertigo and BudWalker) 'back and forth'. Just please keep it civil and we'll get a resolution..
 
The DatCon battery current bug discovered by @Vertigo has been fixed in the newest version of DatCon which can be obtained by going here.
 
A difference in motor speed causes a constant rotation around any of the 3 axis, it does not result in a constant (pitch) attitude. Equal motor speed theoretically results in a constant attitude (roll/pitch/yaw) at a constant speed and heading. Flying forward at 50 degree pitch attitude backwards or sideways at full speed, all 4 motors will run at the ~ same speed. Only a change in attitude requires a temporary difference in motor speed.

If I understand what you're saying, you conclude an attitude based on relative motor speed. You cant do that. There is no correlation between them. Thats like concluding the direction a car drives by looking at the steering wheel. All you can say is that the heading will remain constant when the wheel is centered, and will change when its not. But its not heading north east just because the wheel is slightly to the right.

Similarly, if you fly a quadcopter in acro mode (which the phantom doesnt have, but bare with me), stick neutral results in equal speed for all motors - Im simplifying and ignoring external factors. The quad will happily continue flying forward, backwards, sideways or upside down with all 4 motors running at the same speed. A "thrust angle" as you call it would correspond to a stick input in acro mode in any direction, which will cause a difference in speed between front/back or left/right which results in a constant roll or pitch rotation for as long as there is a difference in speed; until its upside down and back again and upside down again. Well, or hits the ground ;) But you can not tell attitude/orientation from motor speed, or vice versa, without accelerometer data, the quad can not maintain and has no idea of its own attitude relative to the world. It can only detect and cause changes relative to where it was before.

Or Im misunderstanding what you're saying.
Trying to understand what you are saying and what the datcon values show.
Here is a flight of mine without compass errors using the thrust angle values from DatCon. Its this gauge which I renamed Thrust direction.
upload_2016-7-8_22-34-23.png

As you will note in the video the arrow points in the direction in which the P3 is flying relative to the two red stickers.
Compare it to the GPS track and moving P3 that is using the Yaw value as heading . Pretty much matches. This also matches the actual video footage which I can add if you want.
You will also note the Mag Yaw value gauge matches the Yaw Value gauge in this flight.
upload_2016-7-8_22-50-42.png


To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Now compare it to the compass error flight that has been posted here. (the gauges with blue background are from the TXT log compared to the brown gauges from the Dat log)
To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

And finally here some basics on Quadcopters.
How Do Quadcopters and Multicopters Fly | Quadcopter Multicopter Basics

So where do you see the issue on thrust angle not being able to determine actual heading?
 
Last edited:
Lets clarify a few things things first, because I dont see a 'thrust angle' column in my CSV and I dont know how you calculate it. From whats been said before, I understood its essentially the difference in motor speed between L/R, F/A or CW/CCW motor pairs. Is that correct?

As for the yaw angle, again I do not know exactly what it refers to. I do see that column, but I dont know if it is the result of the flight controller's own calculation based on compass, amp corrected compass, compass + gyro integration, or if its something else.

I will say this; if you want to determine the attitude of the quad, so its pitch and roll angles vs ground, just look at the accelerometer data. Unlike compass data, or data derived from motor speeds, I can think of no plausible way that data would be incorrect, other than a faulty sensor which for sure would cause a quad to crash in attitude mode.

As for how a quad flies; I like to think I know that much. Ive built my own designed VTOLS and wrote custom flight controller mixers to achieve thrust vectoring in forward flight and for transitioning between forward flight and hovering. I'll just note the article correctly states that rotation speed, not pitch or roll angle, is achieved by differential thrust. .
 
Last edited:
I didnt want to hijack someone thread but I just wanted to share this, I was taking video of fireworks near the lake, The first battery went fine, but the second battery just went crazy after launch with compass errors and such. After a big fight with it, I managed to get it back in one piece without going into the lake. If someone could just take a look at the log and let me know what happened. Thanks.

81.83 mph is that some kinda record?!
I haven't forgotten that the original purpose of this thread was to look at your flight. This incident is like several incidents that have happened recently. A "compass error" occurs and the AC becomes hard to control if you're lucky or does a fly away if you're not so lucky. Some of these incidents have a compelling explanation as to the cause of the compass error. Some don't. I haven't been able to determine the cause in your flight.

There was speculation that a large current spike (due possibly to a motor issue) caused a problem with AC's electronics that manifested as a compass error. I don't think this was the case. Rather, the compass error caused the flight controller to attempt a correction that required maximum motor speeds which then caused the current spike. Not knowing much about motors I had asked if the current spike observed could have been normal for this scenario. I didn't hear any response so I suppose that means it isn't a totally ridiculous supposition.

The compass error isn't necessarily a problem with the compass. Usually, it's a disagreement between what the flight controller expects to see and the values measured by the sensors. In your incident there was a difference in what the FC computed for yaw and the yaw derived from the magnetometers. There was also a difference in the velocity derived from successive GPS locations and the velocity that, I suspect, comes from the IMU.

I had noticed that the gyroZ drift rate (0.438 degrees/sec) on your AC seemed high. But, I don't actually know what's acceptable. I just know that I had that mine was about that value until I did an IMU calibration. I don't think that was the cause of your incident but I'm going to look at some other flights to see if a pattern emerges.

My advice? Send the AC to DJI for repair. This is the second incident and it still isn't resolved. Maybe they will accept the .DAT and tell you if there is an issue. If you manage to extract any info from DJI please share it.
 
....
I will say this; if you want to determine the attitude of the quad, so its pitch and roll angles vs ground, just look at the accelerometer data. Unlike compass data, or data derived from motor speeds, I can think of no plausible way that data would be incorrect, other than a faulty sensor which for sure would cause a quad to crash in attitude mode.
.
Nobody has ever said that motor speed data is being used to determine attitude. I use it in situations where the Yaw value coming from the flight controller is suspect. It's useful precisely because motor speed data is completely independent of gyros, accelerometers, magnetometers, etc. It's a really simple diagnostic tool. That's it.

I gotta ask you. You seem like a really smart guy. Why is this so difficult?
 
Its so difficult because you're not explaining what you are measuring/calculating and deducing from those calculations. This discussion started with you saying this " The thrust angle is close to 0; i.e. the P3 is flying forward along it's X axis. "

Flying along an axis to me is a statement about its attitude. I asked you repeatedly if I understood correctly that you are deriving a pitch/roll attitude from motor speed, and said that if you are, its incorrect.

Now rereading, Im guessing instead you may be trying to deduce yaw rotation rate from motor speeds? Now that should be doable. I always said that from motor speed, you can calculate rotation speed along its 3 axis, just not attitude.

Whether its useful is another matter, since any fault in the gyro will translate immediately to a change in commanded motor speed. Gyro stabilization is the flight controllers most inner PID control loop. Attitude (accelerometer) stabilization is next and compass/GPS/navigation is the most outer PID loop. So a quad just cant fly if the gyro's arent working. And your motor speed measurements, if Z axis rotation is what you are trying to measure, is not at all independent from the gyro, since the flight controller ties them together at >400Hz.
 
BTW, again assuming Im reading you correctly this time; dont forget that a quad can change its GPS ground course in any direction with absolutely zero change in heading, so no Z axis rotation. Just fly backwards and your ground course changes 180° with no yawing. Fly sideways and it changes 90 degree. Thats why I keep hammering on attitude being so important, and no amount of motor speed info will tell you that. Accelerometers will.
 
The phantom doesnt have an airspeed sensor, it derives its speed from GPS location, and is therefore prone to error. Far more worrying about the logs is the battery voltage. You took off with a full battery and 30 seconds in to the flight, you reach critically low voltages. It recovers later, but something must have been drawing excessive current. Ive seen this before, I dont know what causes it, but the high currents could explain the compass errors.
I thought it had an IMU or an accelerometer...
 
It does have those, but they cant measure airspeed (or ground speed for that matter). An accelerometer essentially measures gravity and will be able to point towards the earth. Thats it.
 
No i didnt record the flight, after about a few seconds I knew something was wrong and was only worried about getting it back. I do remember seeing that thing lean like crazy. Maybe that is where my speed came from.

Tonight I took it back out in a field and there were no problems at all.

1) Warning Propulsions output is limited to ensure the safety of the battery.

2) Flight mode is set to Atti Please set flight mode to GPS to ensure a safe flight

3) Warning:Abnormal compass function or GPS signal has been detected. The aircraft has switched to ATTI mode

4) Flight mode is set to Atti Please set flight mode to GPS to ensure a safe flight

The 4 messages above were displayed across the screen in the first 43 seconds of the flight. I'm curious why you did not abort the flight knowing hell was on the way based the 2 warnings up to that point?
 
...
Now rereading, Im guessing instead you may be trying to deduce yaw rotation rate from motor speeds? Now that should be doable. I always said that from motor speed, you can calculate rotation speed along its 3 axis, just not attitude.
.
No. I'm not trying to deduce a yaw rotation rate. This really simple. I think the Yaw reported by the FC is incorrect (in this flight it was incorrect). I can confirm this by using the thrustAngle and the directionOfTravel to calculate a yaw that is independent of the FC. Go back and read post #35 in this thread.

You've said several times that attitude can't be determined by motor speeds. Of course it can't. I've never said that it could. I'll say it again. It's really very simple. thrustAngle is being used to determine the ACs heading relative to the AC. If you don't know why I've underlined "relative to the AC" please ask.

Just to be proactive. directionOfTravel means direction-of-travel. It doesn't mean AC heading, roll, pitch or yaw. If the AC moves in a easterly direction then the directionOfTravel is 90
 
If its so really simple, then either im an idiot or you are not doing a good job explaining. Does anyone else reading this understand in detail what is being measured here? What is "thrustAngle " exactly? A quadcopter has 4 controllable degrees of freedom, you just use a single number? Which motor speeds are you comparing and in what pane does the "pushing" occur?

The other parameters need defining too. Like "yaw". I know what heading, bearing, track and course mean. But yaw is what? if by yaw you mean the direction the front of the quad is pointing in the Z pane, then its a heading, which to me, is an attitude indicator. If you mean a change in heading, then its the yaw rotation which you said its not. And heading relative to the AC? Heading is by definition relative to an external frame of reference.

So yeah, Im more than confused. But surely someone will be able to explain this to a simpleton like me.
 
  • Like
Reactions: Numone
I just converted one of my own DAT files. "Yaw" is clearly heading, relative to north (be it magnetic or true, doesnt really matter), as calculated by the flight controller. Its therefore an attitude indicator. There is no way you can calculate that based on motor speeds.

You can calculate a change in heading by comparing CW to CCW motors, which will give you an approximate yaw rotation speed and direction. Could you use that to determine if there is a problem in the calculated heading ('yaw') ? Perhaps, but its no different from using gyro data.

Can you compare that somehow to the quad's GPS ground track like you did ? No, not at all. Like I said many times, just fly sideways and your GPS track will be 90° different from your yaw, yet indicating no problem at all and requiring absolutely no yaw rotation that you could measure from motor speed.
 
Last edited:
I didnt want to hijack someone thread but I just wanted to share this, I was taking video of fireworks near the lake, The first battery went fine, but the second battery just went crazy after launch with compass errors and such. After a big fight with it, I managed to get it back in one piece without going into the lake. If someone could just take a look at the log and let me know what happened. Thanks.

81.83 mph is that some kinda record?!
I had the same problem. My battery was still at 58% and all of a sudden it tell me ( critical low voltage the craft is landing). It started to go down by it self over the sea and with great despair i just managed to land it on land but not with out a great fall. Lucky on soft land . It's the second time that this happens. I'm now to scared to even fly the dam thing. Any suggestions? Regards to all.
 

Members online

Forum statistics

Threads
143,066
Messages
1,467,356
Members
104,934
Latest member
jody.paugh@fullerandsons.