Bummed. New P3A Crashed. Pls Advise

[syotr]

Just for fun, I did some more testing to see what other situations could cause prop loss in flight. Some have speculated that air pressure above or below a prop (such as a Phantom falling without power) would make it unscrew. It seemed obvious to me looking at the shape of the blade, that upward pressure would have the same effect as the props normal rotation, that is to tighten it. Downward air pressure should spin the prop backwards so that it would unscrew if completely free to turn on the motor shaft.
I strapped the Phantom to the deck railing and tested by blowing air from a shop vac, both from above and then below the prop. The motors were free to turn. The prop was placed on the shaft completely loose, not down against the shoulder. As expected, air from above would loosen it and air blown from below would tighten it. Of course, it behaves the same way for CW and CCW props.
This could only happen if the motors were never started. As soon as the motors were started, the props were tight enough that air flow would not budge them. For the record, these are 9443 DJI props on a Phantom 1.1.1. They do not have the composite hubs as with the P3.
Next I tried to see if a battery disconnecting would stop the motors quickly and throw off a prop. (In previous tests, I was able to consistently make a prop fly off by disconnecting one of the three wires from the motor to the ESC.) In this test, I just unplugged the battery while the motor was running at different speeds. In all cases, the motor simply coasted slowly to a stop with no chance of losing a prop. Since all phases lose power at the same time, there is nothing to stop the motor.
With an ESC, two motor phases are energized at a time, one pushing the back and one pulling the front of the adjacent magnet. The third phase acts like a position sensor, detecting when the magnet has passed the center of the coil. This signal tells the ESC when to switch current to the next set of coils. With one phase disconnected, the signal never comes and the motor stops instantly, held by the coils and magnets in its current position.

 

[III% Streve]

Just for fun, I did some more testing to see what other situations could cause prop loss in flight. Some have speculated that air pressure above or below a prop (such as a Phantom falling without power) would make it unscrew. It seemed obvious to me looking at the shape of the blade, that upward pressure would have the same effect as the props normal rotation, that is to tighten it. Downward air pressure should spin the prop backwards so that it would unscrew if completely free to turn on the motor shaft.
I strapped the Phantom to the deck railing and tested by blowing air from a shop vac, both from above and then below the prop. The motors were free to turn. The prop was placed on the shaft completely loose, not down against the shoulder. As expected, air from above would loosen it and air blown from below would tighten it. Of course, it behaves the same way for CW and CCW props.
This could only happen if the motors were never started. As soon as the motors were started, the props were tight enough that air flow would not budge them. For the record, these are 9443 DJI props on a Phantom 1.1.1. They do not have the composite hubs as with the P3.
Next I tried to see if a battery disconnecting would stop the motors quickly and throw off a prop. (In previous tests, I was able to consistently make a prop fly off by disconnecting one of the three wires from the motor to the ESC.) In this test, I just unplugged the battery while the motor was running at different speeds. In all cases, the motor simply coasted slowly to a stop with no chance of losing a prop. Since all phases lose power at the same time, there is nothing to stop the motor.
With an ESC, two motor phases are energized at a time, one pushing the back and one pulling the front of the adjacent magnet. The third phase acts like a position sensor, detecting when the magnet has passed the center of the coil. This signal tells the ESC when to switch current to the next set of coils. With one phase disconnected, the signal never comes and the motor stops instantly, held by the coils and magnets in its current position.

??

 

[BudWalker]

@Luap @TheRealNick Luap pointed out that that roll and pitch data coming from flylog.info were switched. That's been fixed now. If you're using DatConverter you'll need to get a new copy.

It makes a difference in this case because some have been looking closely at the beginning of the event trying to determine which corner pitched down first. Then correlating this with knowing which prop came off. Roll and pitch being reversed would make that determination impossible.

 

[TheRealNick]

@BudWalker, if you look at the post I made of a frame in Luap video it matches the prop, front right. The front dropped and it banked right. Only missing info is why that corner dropped. Any chance there is motor RPM in the .dat file?

 

[III% Streve]

The direction of rotation as the event started tells us what color prop came off. I just thought about it, but if a motor locked, there would be a jolt to one direction. Then a spin to the other. What we see is nothing but a spin. Motor couldnt have ceased. The torque from that rotating mass would have had to be transferred to the body.

 

[BudWalker]

Could someone please tell me exactly how to access the flight logs other than the one included in the DJI GO app with the overplayed map. Where is the raw data? How do I download the data? Do I need another app? Please be specific.

Thanks!

Visit flylog.info. You'll find info on how to retrieve these files, how to upload them to be converted, or how to download an app that will convert them locally.

 

[BudWalker]

Based on the last 6 or 7 lines in the log from the tablet.. This is only a small sampling of the data that shows what Im saying..
View attachment 35809
Based on this data... at 37:26:6 we have impact
rate of fall, and altitude indicated at C16 supports this as well
we also have rapid changes in non-trending directions for the data in D17 E17 F17 G17 H17 I17
Also, withing a tent of a second there is a GPS anomaly at C18
I included the gimble data because it too shows an HUGE spike in movement at my proposed impact point
Recording continued for .5 seconds or so after all of those huge changes.. leading to my saying that the battery remain intact for at least that half a second because it was still transmitting data to the tablet to be recorded.

I guess I'm puzzled how you can look at that and conclude that the P3 was powered after impact. If anything it shows the opposite. Line 17 is described as being when impact happened. But, an additional 4.6 feet of altitude were lost. How is the P3 going to lose altitude after hitting the ground? The fact that the altitude readings are negative is because the point of impact was below the HomePoint.

The following plot shows the barometer readings of the last 1.59 seconds before the recording stopped. It clearly shows recording stopped on (or before) impact.

Am I missing something?

 

[BudWalker]

Can someone check if negative roll is left bank? Do you know Bud?

Negative roll is left bank. I checked that while I was confirming (and fixing) that roll and pitch were switched.

 

[TheRealNick]

The direction of rotation as the event started tells us what color prop came off. I just thought about it, but if a motor locked, there would be a jolt to one direction. Then a spin to the other. What we see is nothing but a spin. Motor couldnt have ceased. The torque from that rotating mass would have had to be transferred to the body.

Agree....but it also seems if the prop spun off it would also rotate with the loss of torque on that corner....that combined with the constant current could mean maybe the control system simply failed and the prop spun off on impact...still thinking about it though...

 

[BudWalker]

@BudWalker, if you look at the post I made of a frame in Luap video it matches the prop, front right. The front dropped and it banked right. Only missing info is why that corner dropped. Any chance there is motor RPM in the .dat file?

I'm sure the prop RPM, etc. is in the .DAT. However, I don't know where. It's on my to-do list.

If you or anyone else has an idea of the update frequency for these values that'd help narrow the search. The present data excluding the battery data is 200 HZ. The battery data is 50 HZ.

 

[erkme73]

What happens if a motor gets shorted out while at speed? Would that not cause it to brake? Could a glitch with the ESC or a foreign metal body rolling around inside the arm not have caused a 'temporary' short explaining both the loss of power, braking, and prop fly off? As to the current draw, if the data is heavily filtered, it may not see an instant spike. Once the ESC is shorted, the power is killed immediately - and in my case, on all four props.

 

[BudWalker]

What happens if a motor gets shorted out while at speed? Would that not cause it to brake? Could a glitch with the ESC or a foreign metal body rolling around inside the arm not have caused a 'temporary' short explaining both the loss of power, braking, and prop fly off? As to the current draw, if the data is heavily filtered, it may not see an instant spike. Once the ESC is shorted, the power is killed immediately - and in my case, on all four props.

The current is updated at 50 HZ. It could be filtered but why?

 

[erkme73]

The current is updated at 50 HZ. It could be filtered but why?

I don't know, noise? Just grasping at straws.

 

[BudWalker]

I don't know, noise? Just grasping at straws.

That's a good point.

 

[Ptr]

Maybe not perfect but close, YOU bought it, considering, she is ahead of the game, your substitute please

 

[TheRealNick]

The current is updated at 50 HZ. It could be filtered but why?

It is probably slower because it is only needed to determine the battery state of charge. I worked on some hybrid vehicles and it was similar, also the monitor was always most accurate if you started with a battery that was fully charged....in fact that is how we calibrated them.

 

[flyNfrank]

Is there video of this flight?

 

[III% Streve]

What happens if a motor gets shorted out while at speed? Would that not cause it to brake? Could a glitch with the ESC or a foreign metal body rolling around inside the arm not have caused a 'temporary' short explaining both the loss of power, braking, and prop fly off? As to the current draw, if the data is heavily filtered, it may not see an instant spike. Once the ESC is shorted, the power is killed immediately - and in my case, on all four props.

As I stated earlier, we know that the bird started spinning in one direction at the moment there was a problem. A motor shorting would cause a twitch in the opposite direction as the P3 absorbed the torque cased by loss of rotation in that motor. The logs to not show any rotation in that direction.

 

[III% Streve]

Agree....but it also seems if the prop spun off it would also rotate with the loss of torque on that corner....that combined with the constant current could mean maybe the control system simply failed and the prop spun off on impact...still thinking about it though...

I guess I can see a flight controller glitch potentially causing the result. But I would really expect that kind of flaw to be evident in the logs

 

[JamesBandes]

Well, that's probably it. One was missing completely from the bird and no where in sight. Dang. Must have been my fault then. I spun them on, and did a slight twist on all. I wouldn't characterize any as too tight. Maybe I didn't slight twist one.

Don't they self tighten as they spin?