S800 alternative props

[DennyR]

The flight controllers should never open up all the motors to stop the descent, or control the yaw. #1 priority has to be maintaining attitude. Let it fall straight down, but come down straight. There's a pretty good chance that nobody is standing directly under it, but if you allow the attitude to get that bad, then open it up to stop the fall, it shoots off in some unknown direction.

There's also a really bad yaw right at the end too. I wonder what it was doing there...

The motors are not aware of what the other motors are doing or not doing, so they respond in the first instance to the primary movement which is a drop in altitude, the slowest response will be in yaw, which happens at the end because the motor 4 is not responding. Motor 1 will not slow down because motor 4 is not responding, the loss of height will be seen as a global response albeit out of balance. Only motors 3 and 5 could arrest the tilt but they may be affected by a short.

The key to knowing what really happened is what state the motor/esc of number 4 or 5 is in. if it is all working fine then then the problem may be with the code. If there is a short then that may impact on the other motors because that red wire should not be used ...... .It does not reach the ESC anyway

The weakest part of the design is the routing of the cables through the arms and the gauge of the power supply lines, that leads me to suspect that or the capacitor mounting which is a common fault with vibration issues on other frames. They should be supported with a silicon bridge.

FWIW when my stuff came back from DJI the capacitors were almost flush mounted and the whole ESC was covered with a deep silicon conformal spray. (Nice job) the first ones were not like that. If you just take out the top motor screws and carefully lift out the motor you can look inside and check to see if you have this also. My arms are stiffened up with filler in the webbing which makes a significant difference to the torsional strength.

I'll post some footage from a shoot last week that shows how close I flew to people. I was under pressure from the director and did not want to do this. Having seen that a problem still lurks I should not have done that. I would have used a SR. Makes me cringe at what could have happened..
View attachment 7134
This image shows that the Capacitors are nicely anchored and the high temp solder joints are not showing any signs of overheating. Note that the PMW and the Ground only get to the ESC. so your red wire is not required on the cables.

 

[anyvision]

hi boris, sorry for the breakdown! Can you please tell me what is your left hand is doing at around 17 sec? and denny, are you still flying? i am thinking about throwing a coin ;-) just nervous as anybody here and thats is no good
condition to loose faith even after preflight check. unfortunatly i am not an electronic/propeller/aeronautic/physic superhero so again thanx to all of you specially denny cause without this forum and your input all we non electronics are totally lost.
hope somebody will find the bug soon and something tells me it will not be dji ;-) Thanx to all master beta testers out there ;-)

 

[3dheliguy]

Man Down, Man Down... Dang that suuuuccckkksss! I was really hoping this all stoped with version 2... I remember the first V1 that took off at my buddies shop and five seconds later it looked like that.

Thanks Boris for Posting this, this needed to come out, and I hope DJI sees it to. They should be sending damn near everything brand to you, and a camera.

That def makes me think about how DJI is building and releasing firmware updates lately.

Sorry again to who's Multi that was.

 

[cs8]

FOD from a friend of Boris is the first one with different props?

 

[BorisS]

Boris

I hope you can help me with this theory. I think what happened then was that the pwm signal wire shorted into the carbon reinforced plastic arm. With any luck that arm is not broken so we can check it. I cant think of any other reason why this is occurring after a period of time. It must be the way in which some arms are rubbing through the shellac coating. Never did like that idea. It is also very strange why the main power lines are such small gauge. How can they not get hot. It has to be said that the FOD has not been happening to other DJI frames on such a regular basis. so that must tell us that it is in the frame somewhere. If not the arms then the Power board. When my S800 came back from DJI repair they rebuilt the arms so I guess they took some time over them. They also replaced the top plate.

I would be much happier seeing all of these arms being rewired with silicon coated wires. And then reinforced to make them more rigid. If I were redesigning them I would also reduce that dihedral angle. :upset:

Think Marwon will show up at my place at some point this weekend to test the zenmuse if its still alive etc. Than i will take a closer look if I can find anything !

 

[ChrisViperM]

The most annoying thing about the crash was not the damaged hardware but the expression in Marwon's face....I really feel sorry for him and wish I could do something for him.

Sorry to say so direct....but slowly I think: f*ck DJI and f*ck the horse on which they came....

Chris

 

[JDM66]

U are not the only one and this guy bought everything from us. So i really want to see what will DJI answer tomorrow.

Hi,

any update from DJI, please?

jdm66

 

[Stacky]

Hate seeing this but one thing struck me right away and that is all those videos out there of people showing how stable their machines are by wandering over to them and pushing them while they are in position hold etc. Wonder how many people consider the possibility of this happening whilst pushing their quads by hand....

hit another person ;( I was actually building/setting up the bird together with him and he was testing it the last couple days. Really hurts to see this !

Boris

 

[BorisS]

I love it how i manage not to think about focus

Okay but couldn't reproduce the Speed up whats interesting though is if PWM and PWM GND is shortened there is a speed change in the motor. Don't know how the MC would react if this happens if a speed change happens that it didn't command. Makes me think i should shorten the PWM GND again and see what happens if during shortening i change the pulse.

Enough s800 v3 esc around to go crazy

Boris

 

[JDM66]

Hi,

the video does not work

jdm66

 

[BorisS]

hmmm strange for me it works what error message do you get ?

 

[BorisS]

by the way long contact between GND and signal results in a shutdown, contact take away again motors speed up again

 

[DennyR]

hi boris, sorry for the breakdown! Can you please tell me what is your left hand is doing at around 17 sec? and denny, are you still flying? i am thinking about throwing a coin ;-) just nervous as anybody here and thats is no good
condition to loose faith even after preflight check. unfortunatly i am not an electronic/propeller/aeronautic/physic superhero so again thanx to all of you specially denny cause without this forum and your input all we non electronics are totally lost.
hope somebody will find the bug soon and something tells me it will not be dji ;-) Thanx to all master beta testers out there ;-)

Yes I am still flying it. I have done a lot of testing so I have to think it is safe but what I don't understand is why some are not safe at the moment. Based on the time in the air that was trouble free, I have to balance the possibilities against the probabilities and I made the decision that I would have to be very unlucky for it to fail at the precise time I was overhead somebody. These rehearsals were an example of that risk management. Those kids were briefed and could in all probability jump out of the way. I still bucked the request to go lower and faster. I do have quite a tough preflight routine before we go. Most of the actual video was shot with a Red epic so the rather flat raw footage will get the same color treatment across the board and look a lot brighter. It was a shame that we had such dull and windy weather which is very unusual. Having done the establishing shot alone the Greek director played around with the Zen himself. Shooting mostly U/C. He was amazed at the stability after he soon got the got the hang of it. Oh and BTW the horizon only moved when someone touched it. I am looking fwd to see what a GoPro three can do with a Zen.

 

[RancherPete]

Denny: Can you please tell me which shipping service you have used to ship our prop adapters to the US? Many thanks.

Pete

 

[HeinzABF]

@ Boris

What Software version did you use at the crash?
5.12 or 5.14?

Regards
Heinz

 

[BorisS]

Hi Heinz,

Fw 5.14 was used ! But they seem to have found the problem !

http://dl.dropbox.com/u/53205026/DJI FINDS FIRMWARE BUG CAUSING SPONTANEOUS S800 SYSTEM FAILURE.pdf

Boris

 

[R_Lefebvre]

The motors are not aware of what the other motors are doing or not doing, so they respond in the first instance to the primary movement which is a drop in altitude, the slowest response will be in yaw, which happens at the end because the motor 4 is not responding. Motor 1 will not slow down because motor 4 is not responding, the loss of height will be seen as a global response albeit out of balance. Only motors 3 and 5 could arrest the tilt but they may be affected by a short.

This may all be besides the point in light of the recent announcement from DJI. I hope they have found the problem, though I can't make any sense of it based on what they wrote. This appears to be a fairly extreme system failure, and I'm not sure how the compass unit could cause it. Anyway, we simply don't know enough about the inner workings to say...

However, what you wrote above appears to be based on the assumption that each motor operates independently from the others, and simply responds to pretty basic input from the gyros. At least in our code, it does not. There is an over-arching observer which attempts to intelligently control the motors as a system, and has a hierarchy of control authority. Attitude always wins, then throttle, and yaw last.

I assume DJI is now also doing this, as evidenced by the video demonstrating the failsafe performance. When they have 1 motor fail, you see the S800 doing a pirouette, because the yaw controller has given up.

Thinking about it more, my best guess at this point about the FOD failure, and the compass firmware, is that they had something like a "Not a Number" in the code coming from the compass module. NAN then spreads through the code like a cancer and just breaks everything, which is sort of how that looked. We had this problem too, but it was easier to find because of datalogging. It only took us one crash to find it.

Anyway, I hope this is it, failures like these make all multirotors look bad. And it's just plain unsafe for users.

 

[DennyR]

Rob, I think we can now move fwd. to the next level of mechanical refinements to make it even better than it is. I doubt that it will be the universal cure for all maladies but having a solid base to work from is sure going to help. I recently had a bad V.sens unit supplied for my Naza, it was giving out noise and spikes all over the place. the model was dangerous to fly. I changed the unit and it was then as smooth as silk again. Skookum are about to release a GPS unit for their SR 720 model. Do I want to trust it? do I really need it? Probably not is the answer.

 

[R_Lefebvre]

Interesting, I just read on RCG that you have the old GPS, and that might explain why you have not had FOD.

 

[snurre]

.... NAN then spreads through the code like a cancer and just breaks everything, which is sort of how that looked. We had this problem too, but it was easier to find because of datalogging. It only took us one crash to find it.

Anyway, I hope this is it, failures like these make all multirotors look bad. And it's just plain unsafe for users.

I wonder why DJI chose not to include data logging into the systems design? Sure, data logging capacity would cost some processor capacity and take some programming effort, but the engineers at DJI seems to be skilled people so they could do it.

To me, choosing not to provide data logging may seem a little arrogant; like implying "no need for it - we only release error free software" and "flight tuning our system is SO easy, thus flight data analysis is never required".

I am not saying this is the DJI stand (I was just thinking loud), but then again they never commented on the absent logging feature... Maybe in next generation MC? :encouragement:

On a side note, I look forward to the day our MR systems come with fault tolerant software (at the extent possible in this application).
/ Tomas