I am helping a friend of mine diagnose a problem with his quad.
Flight controller: APM 2.6
Battery: 3S 25C 4000mAh
Battery failsafe: disabled
It was working pretty well - he had been flying it in Stabilise, Loiter and RTL. However, sometimes it would inexplicably drop from the sky like a rock.
We inspected it for visible hardware faults (broken wires, poor connections, etc) but could not see anything obviously wrong.
I have been having a look over the dataflash logs to see if I can identify what the problem is, but I am not getting very far. I was wondering if there was anyone out there would could help.
Out of the sixteen logs I was able to download, three are flagged as having brownout events. I assume these are the ones that were generated during the crashes. I will attach these four for your reference.
(Note that the crashes occurred several years ago, hence the strange datestamps on the files).
Of the brownout files, I have been paying particular attention to ‘2014-11-09 15-13-19.log’ as it seems to be a good example of the problem. This is the log I will discuss below.
When I graph CURR.vcc, I’m seeing fluctuations of only about 0.044v, which seems ok to me.
When I graph CURR.volt, it looks like the battery voltage never drops below 10.47v, which seems reasonable, too.
Does anyone have any idea what the problem might be, or what areas I should be looking into?
I can’t get any of your logs to load in the log broswer. All I get is a list of parameters, and it looks like you are using AC3.1. The latest firmware for an APM is AC3.2.1.
I can get the logs to load in to Auto Analysis. and all three show a brownout.
I need a complete hardware list with specifications, frame, motors, ESCs, GPS, telemetry, the works, ESPECIALLY BATTERIES, how old, what voltage, what mAh, and how they are being stored and charged. If any are puffed, trash 'em…
Enable the low battery fail safe and log EVERYTHING.
I upgraded the firmware as soon as I took delivery of the quad - it is now 3.2.1.
The battery is a Turnigy nano-tech (3S 25C 4000mAh). The battery was quite new at the time the problem occurred, and still has no visible problems.
I am happy to find out the other details you requested if required (I don’t have them to hand - I didn’t build the quad myself, and the parts were purchased several years ago), but first I would like to find out why you can’t view the logs.
I can view them successfully in a recent version of Mission Planner (1.3.56). I am using the ‘Review a Log’ button under the ‘DataFlash Logs’ tab. Works like a charm.
Are you using Mission Planner, or something else? Is your version up-to-date?
Ok, to get an idea for what is happening I tried an experiment in to see if I could reproduce the brown-outs in a controlled environment.
I flipped and rotated the props on the quad (so they push down) and ran it at half throttle for a few minutes to simulate a vigorous flight. At around 3.5 minutes I lowered the throttle a bit and continued for another couple of minutes.
I have attached the .bin of the experiment, and a screenshot of the battery voltage vs throttle.
Note that the test took place in a shed, so the GPS was not able to get a connection.
I find it pretty suspicious that the battery voltage drops down to 9.5v almost straight away. That seems really low to me - is this an indication that the battery is stuffed?
(EDIT: I did not read your plot correctly. This response of mine is outdated, I’ll re-post in the thread below)
To make sure I understand your plot correctly: Around Line Number 12*10^3, the battery voltage drops from 11.4v to approx 9.5v, but holds pretty consistently there.
I think drops like this are common. I fly a 3S battery beginning at 4.2*3 = 12.6v with a full charge. On takeoff, applying almost full throttle makes the (healthy) battery read ~11v while under load. That 1.5v drop seems consistent with your drop from 11v to 9.5v, so I wouldn’t immediately read that as a sign of bad battery health.
I think I misread your plots, sorry for that. Over the course of your test, the battery is depleted from 11.5v to 9.9v, but (of course) goes lower than that while under load, as the props are spinning.
The correct interpretation of your plots:
your THROTTLE (blue line, left axis) is raised up from 0 to 800 around line 110^3, and it remains there until line 1210^3, where it is lowered in steps to a value of 300. Around line 16*10^3 the throttle is reduced to 0. (I’m ignoring the final momentary spike.)
The battery voltage (red line, right axis) is observed to drop from ~11v to 9.3v under load, and we watch it noisily decrease from 9.3 to 8.6 while running the props. When you decrease the throttle to 300, we see the voltage rise to 9.8v but then continue to decrease to 8.9v. When you cut the throttle at line 16*10^3, we see the voltage rise to 9.9v. (I’m ignoring the final momentary spike)
A 3S battery at full charge should read ~12.6v. Yours started at 11.5v. So I agree that it probably wasn’t charged, and you should have done this test with a fully-charged battery, not a nearly-depleted one. I hope you didn’t damage your battery.
Yes, I was surprised to see the battery wasn’t shown as fully charged on the plot - I thought it was. I was several minutes into the test when I connected telemetry and saw that it was below 10.5v
I’ll pay more attention to the starting voltage next time.
For the next test I’ll take it to the local park next time the weather is good and cut a few laps at low altitude. Hopefully I can repro the problem that way.
Really should make sure the voltage and current sensors are correctly calibrated and set the failsafes (and test) - that way you know there’s either something wrong with the battery or some other component.
If you voltage sensor readings are not perfectly linear, when compared to a good quality digital multimeter, then calibrate to suit the low battery voltage such as 10.8 volts.
Also I have ARMING_VOLT_MIN set to 11.0 (3 cell lipo) so you cant arm and fly if you’re not going to get any actual flight time before triggering battery failsafe.
The voltage sensor was calibrated correctly (to within 0.2v as per the documentation).
The battery failsafe was left off deliberately as the test was an attempt to reproduce the conditions of the brownout in a controlled environment. I thought that perhaps CURR.volt went low enough it might cause CURR.vcc to drop enough to cause a brownout. Unfortunately this was not the case. Even when CURR.volt was dropped down to 9.5 volts, I could not repro the problem. So while the battery may or may not be faulty (I may even have damaged it), I think I can be a bit more confident that low voltage was not causing the brownouts.
Perhaps the problem only occurs during flight for some reason? I’ll take it out for a (carefully controlled) flight first chance I get.
