A2 FC Firmware 2.3 - Here We ALL go Again

[BirdsEyeProd]

The A2 system has taken out my copter. DJI is saying I had an ESC fail and go full throttle (M1) which caused the copter to start spinning... only problem was it started spinning way before the esc went full throttle according to the iOSD Logs. Here are screenshots of the logs from the motors along with the video showing it going down. I am also including my iOSD file so you can take a look for yourself and see if you can find anything. I honestly don't believe that an ESC caused this. The copter started spinning clockwise then stopped and went counter clockwise. I will be inspecting all my motors and escs today and then sending them back to KDE for them to inspect and tell me if they find anything wrong.

Video: http://youtu.be/bgHHrnVx9HM
iOSD File: http://www.filedropper.com/fly011


Hey birdseyeprod,
your video link wouldnt work right. can you please re-post the link again?! would be nice.

Here it is with the graph in the video as well.

https://www.youtube.com/watch?v=5sPrZy6ZqrY

 

[Jimi Handmix]

this looks like the same thing happend to my skyjib6 / wookong-m the oter day. WHAT IS THAT BEHAVIOUR OUT OF THE BLUE?!?!
first i thought it might be a vortex ring problem. but the second time it appeared while hovering. i cant really figure out what it is but im tipping on bad/wrong esc´s. or maybe vibration at some certain propeller speed, that might manipulate the imu?! im going to use different tape to fix the imu.

 

[kloner]

how many succesful flights were on these rigs?

 

[R_Lefebvre]

this looks like the same thing happend to my skyjib6 / wookong-m the oter day. WHAT IS THAT BEHAVIOUR OUT OF THE BLUE?!?!
first i thought it might be a vortex ring problem. but the second time it appeared while hovering. i cant really figure out what it is but im tipping on bad/wrong esc´s. or maybe vibration at some certain propeller speed, that might manipulate the imu?! im going to use different tape to fix the imu.

Yes, definitely vibration can screw up an IMU, and cause a loss of control. That was a common problem with APM because there is no vibration damping on-board. But, IIRC DJI logs the IMU data and I looked at it and didn't see anything that would indicate that. But I'm no expert on DJI logs of course. You'd think if that was the case, DJI technical support would point that out.

 

[JLO]

Hi has any one able to figure out the posible specific problem with the A2?
is it the a few bad units
is it something related to the IMU like vibration
is it the software
is it the builtin receiver
is it a incompatibility with some components
are most of crashes yaw related
I know many many that have had expensive (suden and mysterious) crashes are super angry and frustrated specially that DJI sims to be mostly MIA, very detach from customers, not addressing this head on, and blame other factors
I my self have one that is working ok.... so far.... am I just another crash waiting to happen

 

[kloner]

i don't read every thread on it to detail but the ones i have seem to have a common theme...

there not using 8fg using the built in receiver.... I'm convinced all these telemetry radios are making weird things happen like 14fg
it seems to kill models more when people use it's advanced features..... IOC, ground station, etc.

 

[JLO]

interesting... I am using the built in A2 receiver with a Futaba T14SG radio, I wonder, if using the original futaba receiver instead of the built in may be a safe thing to do... I know part of he reason for the A2 are the many features but perhaps in this case its better to limit them to the basic stuff.... have not use the IOC...

 

[ACP]

Hi All. I have just joined this forum to reply MombasaFlash's post "A2 Problems Catalogue" as I bereave it is important to log these issues.

I mentioned IMU lock in my replay listing some of my recent A2 experiences (and ACE Wavepoint)

What I mean by IMU lock is I was lucky enough to be watching my OSD at the time my A2 went nuts. I noticed that... I put in BOLD as this is important...

The pitch and Roll data at the top center of the OSD went blank, seconds before the A2 went NUTS (OK 0.8 seconds apparently but felt like seconds) and shut my motors off.

When I say 0.8 seconds apparently!, is because I look at the data log I see the IMU data stopped 0.8 seconds before the PWM to the ESCs was dropped to 1.2ms (shutting the motors down).

This is interesting and what I call IMU lock and wonder if this is the vibration issue mentioned in this thread.
But looking back at my frame camera DVR which is hard mounted to the frame i.e. no anti-vibration.. I don't see any excessive / unusual vibrations in the video.

I hope this helps.

 

[PeterLester]

I have over 18 successful test flights to date thru builtin a2 rx using my Futaba T8FG Tx.
But remain scared as to what might happen next.


Sent from my iPad using Tapatalk

 

[ACP]

Ah I have just had an idea. I happen to have a vibration plate I use to test the systems I develop. The plate's frequency is controllable through my frequency generator.
Since I have another 3 A2s which I WON'T BE USING now....

I could mount an IMU onto the plate and simulate a flight.
I will monitor the ODS (since there is no way to read the logs live) whilst slowly increasing the frequency and see if I can reproduce the "IMU lockup"

I will try this tomorrow and report back.

Regards
Anthony

 

[ACP]

Peter, I too have had countless flight with several Multies with the A2 with out ANY issues but it only takes one (or 3 ) hick-up(s) to destroy a dream. (and a business)

Yours
Anthony

 

[JLO]

Peter, I too have had countless flight with several Multies with the A2 with out ANY issues but it only takes one (or 3 ) hick-up(s) to destroy a dream. (and a business)

Yours
Anthony

Anthony you couldn't have said it better I for one have not had an incident yet.. related to a suden, with out warning, surprise A2 crash .........but Im so terrify to use it

 

[R_Lefebvre]

Hi All. I have just joined this forum to reply MombasaFlash's post "A2 Problems Catalogue" as I bereave it is important to log these issues.

I mentioned IMU lock in my replay listing some of my recent A2 experiences (and ACE Wavepoint)

What I mean by IMU lock is I was lucky enough to be watching my OSD at the time my A2 went nuts. I noticed that... I put in BOLD as this is important...

The pitch and Roll data at the top center of the OSD went blank, seconds before the A2 went NUTS (OK 0.8 seconds apparently but felt like seconds) and shut my motors off.

When I say 0.8 seconds apparently!, is because I look at the data log I see the IMU data stopped 0.8 seconds before the PWM to the ESCs was dropped to 1.2ms (shutting the motors down).

This is interesting and what I call IMU lock and wonder if this is the vibration issue mentioned in this thread.
But looking back at my frame camera DVR which is hard mounted to the frame i.e. no anti-vibration.. I don't see any excessive / unusual vibrations in the video.

I hope this helps.

Huh, I wonder if it could have been a NaN problem. Not a Number. It basically happens if you do a divide by zero. Pretty common bug with all the complex math going on in these things. We constantly have to fight to prevent this.

I always suspected this was the cause of the original Wookong FOD.

 

[PeterLester]

Only inexperienced programmers do not gain control of the division by zero issue as well as unsigned, signed integers, floating point numbers etc. We call such programming "ppp".... Pizz poor performance!


Sent from my iPhone using Tapatalk

 

[R_Lefebvre]

It's true that experienced programmers probably wouldn't make this mistake. However, from what I've seen, experienced programmers can't get their heads around the inviolable laws of physics. This isn't like programming a database or application. So the project requires a mix of both types, people who understand the physics, aerodynamics, and the electro-mechanical systems they have to interface with, and then experienced programmers to help glue it all together. And sometimes these mistakes slip through.

 

[ACP]

Huh, I wonder if it could have been a NaN problem. Not a Number. It basically happens if you do a divide by zero. Pretty common bug with all the complex math going on in these things. We constantly have to fight to prevent this.

I always suspected this was the cause of the original Wookong FOD.

I should certainly hope Dji would be beyond floating point, stack overflow and divide by zero stage well before starting the company.
Perhaps this phenomenon don't effect underwater photography. (WETPIXEL just in case you did not know where Dji started from).

 

[Robwik]

I like R_Lefebvre's strip. I do recognize it from my daily work as IT proffessional. Programmers meeting users meeting support meeting bosses etc..

I have been a lurker at most forums, Drones Discuss, here at Multirotorforums, ******** etc. I began with a Parrot drone, then modified it for RC. Then I went to Naza V1, Zero UAV YS-X4. That YS-X4 I moved to a well tested Tarot T810 with countless hours of successful benchtesting. At first flight it lifted off to about 2 meters then suddenly veered of to the left and dropped to the ground, the frame destroyed. A Gemini was sent to me, but that one just stopped working correctly at benchtesting. So many jobs losts and hours, and, of course, Money.

Now I have a Naza V2 on a F800 Evo (successful flights), an A2 on a F800 Evo (benchtesting), and a 3DR Pixhawk on a F550 (successful flights).

The RC hobby is old, but advanced FC's are expanding quite fast in development, and many of the large companies want to be on the edge and take a market share. Often as with all other Electronic stuff (BluRay players, cell phones, TV's etc), the hardware spec's are quite good, but the software have faults or really don't take advantage of the hardware.

Personally, I was afraid to lift off with the Pixhawk. All the settings seemed to be made for engineers/software developers, that had countless years of experience with APM. But it was not as dramatic as it looked I was so happy to see the F550 shake (yes, a bit shaky liftoff in manual, prop wash), then take of in the sky. It was steady as it was!

Now I am afraid to liftoff with my A2 due to all negative posts... I have a 3-axis gimbal with a Sony A6000 and soon a BMPCC. I will not fly acrobatics, but just slow steady filming. But I'm lifting some Money...

And, programmers make mistakes. Just look at the Hubble telescope metric error. Everyone can make a mistake, and I'm not judging. I'm not faultless either. Hopefully we'll see a quality improvement overall regarding both hardware and software on FC's. BUT, I think the main issue is *education*. HOW to use a specific FC, and understand how THAT specific FC works. The developers can help out with that with as foolproof installation guides as possible.

Best regards,
Robert

 

[ACP]

Just to let you know I have managed to replicate the A2 failure I had as mentioned yesterday.
I have posted the results under MombasaFlash's A2 Problem Catalogue, as I felt this was a more appropriate location.
http://www.multirotorforums.com/showthread.php?21630-A2-Problems-Catalogue&p=180054#post180054

thanks
Anthony

 

[Tahoe Ed]

I should certainly hope Dji would be beyond floating point, stack overflow and divide by zero stage well before starting the company.
Perhaps this phenomenon don't effect underwater photography. (WETPIXEL just in case you did not know where Dji started from).

I am sure Eric would be flattered that you considere his company as the start of DJI but that is not true. Eric is the Director of Aerial Media for DJI. DJI was started with the Ace One controller for Single Rotor Helicopters by Frank Wang and Swift Xie.

 

[MombasaFlash]

... DJI was started with the Ace One controller for Single Rotor Helicopters ...

... or rather, what eventually evolved into the Ace One !