Hi guys, my APM has recently started to reboot mid air for no apparent reason. This first happened on the 9th March when I was flying in stabilize mode and all the controls just froze (no response to RC tx, servos all locked, telemetry reporting no data) for a couple of seconds while the plane slowly spiral downwards and eventually it started responding again but it was too late. The plane ended up being stranded on the tree and this was when I caught a glimpse on my laptop saying it's in "Initializing".
Later that evening when I downloaded the logs via USB, the 2 separate log file confirms a mid air reboot and the starting of a new file. I couldn't figure out the problem and for some reason, have a gut feeling it might be the 900Mhz 3DR radio so I swapped it out with a spare. I'd like to attach this tlog or log file but couldn't find it for some reason.
2 days ago after flying both manual and auto without any problems for about 20 mins I decided to take the plane down as it was getting a little dark. I switched to stabilize and slowly brought the plane down and that's when APM rebooted again. This time it was close to the ground and controls freezing only caused a rougher landing. Attached is the tlog for this particular flight. and it's evident the flight ended abruptly (probably because the radio stopped transmitting when APM resets itself).
I've read through brownout theories causing the reset but I have a 5A UBEC and Vcc plot confirms there is no dip in apm supply voltage. Not sure what else could cause this...
Can someone please have a look at the attached tlog and shed some light on this?
*******Post updated with new tlog I can't find from my first flight********
Doug, yeah this was the one I plotted but there's no dip close to the end of the file when the reset happened.
Nope not using APM power module, the UBEC powers 6 servos and all apm electronics. On bench test I can only get the whole system to spike up to 1.8 amps during very aggressive servo throws and applying loads on the ailerons/flaps so I thought the 5amp ubec is more than sufficient. I used to run it with a 3 amp HK bec for a long time with no problems.
Contrast your plot with a typical Vcc plot from one of my quad flights...(much shorter duration, 3S battery)
Vcc should be as clean and flat as possible...all the time. Vcc is the feeder source for the 3.3VDC regulator of the APM. *Anytime* Vcc dips below the threshold of the minimum input for the 3.3V regulator, you risk a brown out/reset.
With a plot like yours, I am surprised it worked at all.
You say you are monitoring the current in testing. How about the voltage before the UBEC and after?
Regardless, you may be able to stiffen the Vcc with an added capacitor bank plugged into the Vcc rail - (+) of the cap to V+, (-) of the cap to Neg. It will add almost nothing to the weight and will smooth out the spikes. Caution: when adding a cap bank, the initial current surge will be much more than normal. There are 'soft start' methods that reduce this 'arc' effect - usually a resistor in series.
The spikes in the plot are probably more square shaped anyway. The sloped transitions are the result of the sampling points as indicated below (zoomed shot of the worst case in your plot).
With not data between the points, the actual value may be even lower that caught by the sampling circuit. We simply do not know.
This is an interesting problem. Hopefully others will follow along to learn more about their systems. More hopefully you can get a clean power supply and enjoy reliable flying.
thanks for the sample. How is your power setup like? I've actually bought a 4700uF 16v capacitor but worked out theoretically that it's hardly enough to sustain a voltage drop. This fact coupled with the 5A ubec I decided not to install it.
Will probably give it a go and look at the vcc plot. In the meantime, I found the earlier log (first reset that I can't find initially). Hopefully you can take a look at that too and share your thoughts?
What brand of ubec do you have?
My quad uses the four ESC outputs paralleled into the 3DR power distribution pcb.
The 3DR ESCs will tolerate this. Some brands will not.
I will take a look at the lastest tlog above.
Now...the sequence of that event was as follow:
- took off in stabilize
- flew in stabilize and manual for a while
- while flying in stabilize and doing a turn, all the servos locked up regardless of my rc stick inputs
- lucky it wasn't doing a dive at that point so it was just going in tight circles and losing altitude
- this is the point where APM actually resets and that explains why there wasn't any incoming telemetry data for vcc hence the drop
- apm successfully reboots and modem reestablished link, this is when we see the vcc up again
I thought it was a faulty modem, hence the drop in transmission for around 45 seconds but when I downloaded the flight log through USB, there were 2 files, confirming that it was indeed an in flight reset.
The USB downloaded log only contains attitude/gps information so probably not useful for diagnostics at this stage.
This is the UBEC I have installed. The stock ESC doesn't supply any power to the setup, receiving only the throttle signal from APM
Get a new UBEC.. And since they are dirt cheap, get two. One for the servos, and one for APM.
This is a cheap HK ubec. It is the same as the HK 3A ubec, but to make it 5A they piggy-backed 2 of them together and wrapped them with heat shrink.
Do you know that connecting 2 switching BEC's in parallel is a very bad idea?
I don't know why they sell such a thing at hobbyking...
Buy a turnigy 3A BEC, they are proven very reliable.
I'm sorry but I think using 2 UBECS is not a great idea, even when they are not connected together.
It's like reverse redundancy.
let me explain. A bec has a failure rate of one in a million (?)
You plan to use two one for servo's and one for electronics. In this case if your servo BEC shuts down, your RX and APM are fine, but your servo's are left powerless resulting in a crash.
If your electronics UBEC fails, your servo's are fine, but the APM and RX are left powerless resulting in a failure of control -> crash.
This way you have increased the odds of a fatal failure from one in a million (?) to one in 500K (?)
(numbers are for reference only, I don't know the actual statistics)
For real redundancy you want to have 2 UBECS in parallel, with the one standing by and taking over when the other one fails. But that would involve creating your own backup protecting circuitry.