I have a general question on APM 2.5 Throttle Failsafe, and then will follow up with tlog, log files and OSD video later...
I have been flying a reliable 3DR Robotics Y6 on release 3.1 for several weeks and yesterday had a bad crash from 400ft up. (For equipment Ardcopter 3.1 (APM 2.5), flying a 3D Robotics Y6, Frame 10, DragonLink, 5.8 ghz FPV etc.)
My question is,
- If I had the throttle failsafe setup and tested (fires at loss from DragonLink set to 1026 low)
- and show that Channel 3 raw did go to1026 for two seconds (some reason loss TX from DragonLink... yet to be determined.. may be FPV CAM interference) ....
- and if I see that the MP Tlogs show that the System stat went to "5 - Failsafe" two seconds after the TX loss ....
- and it wasn't a BROWNOUT on the APM board...
- and that the Battery FAILSE in not enabled ....
- and was flying in Stabalize mode at 400 ft up ....
- and I see in the TLOG (and MP telemetry log replay) that the APM board DISARMED immediately after the FAILSAFE event ... INSTEAD of doing an RTL (the main reason I have Failsafe turned on!)
THEN
- what kind of circumstances could have caused the APM to DISARM at the FAILSAFE event, when the copter wasn't sitting on the ground and/or with the throttle set to zero?
I have full OSD video of the flight, along with TLOG and downloaded LOG file.I will follow up with submitting these for help.
In the meantime, I would appreciate it if someone knows of any general conditions that could cause such an un-welcomed response from the APM Failsafe function....
Background:
I had this same failure on my Hexacopter on November 2013. Randy helped me back then on 2.9, but we never found a root cause. I subsequently thought it was a "flaky" GPS, but now on 3.1, a different APM board, GPS, another new Y6 model, new DragonLink TX (instead of 2.4), etc., I thought I had long left that issue behind.
I was fortunate that the fully loaded Y6 feel to its death from 400 feet upside down. Had it hit feet first, my GoPro 3, Tarrot Gimbal, and 8000 mah bat would probably have been destroyed, if not in flames.
In any case, other than everything being sheered off the upper deck, including the APM 2.5 board, all parts seem to be functioning and just need to rebuild.
I just really need to find the root cause of this type of crash before risking again. I know that there are others that have run into this issue and hope that someone might have put their finger on this issue.
Replies
Just to update you - tridge reproduced this in the simulator about 20 seconds ago. There is a nasty bug with the throttle failsafe and there will probably be a release fixing it shortly (we need to find it first).
Ok, update on this:
This is intended behavior. Basically, on occasion people will land and turn their transmitters off before disarming the copter. This would cause the copter to go to RTL, take off and cut that persons face up, hit the ceiling, etc. So to prevent this, when we get a radio failsafe, we check if the last throttle value was zero previous to the failsafe, and then we disarm if it was.
Your problem is that you had set the FS_THR_VALUE to exactly the value that the DragonLink failsafes to. You need to set it at least 10 points higher than that value, or noise and variations in temperature (etc) may cause it to teeter on the edge of failsafe - which could randomly cause a disarm crash. This is documented:
Now, before I get skewered for saying that it is intended behavior, I'd like to note that we will be looking at adding some debouncing to it. So, instead of disarming if it sees that the previous throttle value was a zero throttle, it will instead check that it has been zero for the last half second or similar.
Thanks Jonathan,
for nailing an exact explanation to how APM's Failsafe and the PPM Encoder logic worked together to produce a DISARM while I for 400 plus feet high.
I will be starting the bench test everything on the rebuild with the dual power system on the Y6. Besides a separate FPV power system, the new build also has a dedicated UBEC for the DragonLink receiver providing 6 volts off the main 4s LIPO (wired in right before the APM 2.5 PM.
I will follow the links you sent to implement and test the APM Failsafe function.
I have to admit, I don't remember seeing and/or following the "pull power to the receiver" test procedure.
In any case, in order to truly test that someone doesn't have the same issue I had, I would think that you would have to move the copter past 3 meters away and or safely simulate a running throttle before pulling the power to the receiver,... correct? If not, it will just DISARM when you pull the receiver power, and this would not indicate the issue I had with the wrong PPM FS value while up in the air.. Correct? In any case, I'll test everything this time.
Thanks for the great support!
Mike Governale
"root cause", after responses from Mike with DragonLink about keeping the voltage supply to their receiver above 5 volts along with another discussion contributor who was pulling the receiver supply from the APM output rail (no jumper) and documented the voltage dropping to as low as 3.2 volts causing an apparent receiver brownout.
Well, to ensure that you have it set up correctly you simply have to set it 10 points higher than the RC3 value when you transmitter is off, and then you will not have this problem again. Your copter can and should be disarmed for that test.
We're working on it: https://groups.google.com/forum/#!topic/drones-discuss/o-GbCC5SAqM
Is it too complex to consider descent rate as a part of the "debounce". If it sees zero throttle check descent rate for half a second, or a bit more, If throttle is still zero after that time and descent rate has increased. THEN disarm.
Jonathan - Is it possible for code to check what you just said and correct the setting in case someone doesn't read the fine print?
Know you are busy, but was hoping you might still be planning on taking a look at the logs from my crash that you asked me to send. I'm in the process of rebuilding my Y6 but hoped to identify the root cause. I also would like you comment on the possibility that I might have gotten a short power disconnect, or brownout to my Dragonlink receiver, because Hugues said and I think Randy verified there is still this bug in 3.1.
In any case, I really appreciate any help in avoiding another crash. I would really like to use APM's Failsafe, but after two horrible crashes, the medicine seems worse than the cure...
I'm sure you'll understand .
Thanks,
Mike
Just want to be clear that it's not a bug in AC3.1. It's an issue with the software on the ppm encoder chip (the 32U2) that is responsible for decoding the input from the receiver. It can temporarily go to 1100 as it loses power.
Randy,
If you could take a look at my logs from the crash, I would sure appreciate.
It seems that my DragonLink receiver either had a temporary "short" or "brownout" causing the FailSafe/DisArm with the know ppm encoder bug.
Before the crash, I was running all of the following on a single 8000 mah 4s battery (with a 12 volt adapter).
On the 12 volt supply was a Sony FPV CAM, GoPro 3 CAM, video switch, Tarot Gimbal with pitch controlled from AUX 7 on DragonLink receiver, JD-board / boom led's. On the APM PM supply side was the DragonLink receiver, 3 way Y-cable to MinimOSD, JD-Board, MAVLink TX/RX.
I am rebuilding using a separate 3S battery supply on the 12 volt items (removed the jd-board/led's).
Putting a separate 5amp UBEC from the 4s main battery to the DragonLink, letting APM's PM supply the 4.5 to the board, minimosd and Mavlink.
I would just like some assurance from someone that my old arrangement did invoke the issue with the DragonLink, and thus raised the only bug I've ever heard about that matched the two horrible crashes I've had todate.
If from looking at the board and tlogs, you can verify somehow that I am rebuilding to result that allows me to use the APM Throttle Failsafe, I would be happy. Ironically over the last couple of years, the only two bad crashes I've had with my APM 2.5 boards seemed to have been cause by trying to use the throttle FAILSAFE feature.
I just want to get finally past this issue and gain confidence in flying my expensive models.
Thanks again Randy for you gracious support in the past and for anything you may provide now and in the future.
Mike Governale
Was wondering if it's so important to disarm in stabilize so quickly (ms) when it sees zero throttle.
I would think that there are very few of us that would shut off their transmitters the instant it touches down.
Most of us take at least a second to come down to ground after the drone has landed, assess the situation and start walking back towards the drone before they absentmindedly shut their transmitters off.
I ask because maybe it would be worth checking rate of descent before actually disarming. Even if it's a second, we could prevent a situation like this from happening.
I use a Dragonlink, and at this point, I'll have to tweak my procedures... I've already had a jump or two in the air from a faulty RC Tx cable to the Dragonlink Tx. At least I use Pixhawk, so shouldn't have to worry about the Dragonlink losing power. It should go directly into RTL right?
-
1
-
2
-
3
-
4
of 4 Next