Well today I got the Be-Gez scared out of me. Lifted off my 550 class Alien that is the sum of about two months work and over a $1000 only to lose control a few seconds into my flight. This build has about 10 flights on it. Most worked very well from a pilot's point of view. It did what I said and came back in one piece.
Pretty sure it went into failsafe caused by Vcc fluctuations. I have attached the log if you'all would check it for me to make sure. (I am new to many of the features of Mission Planner).
I had lifted off and was planning to execute an auto tune so I brought it into a stable hover at about 15 feet up and flipped on ALTITUDE HOLD. It jumped a couple of times and seemed to recover as I advanced the throttle to 50% then it jumped up to about 30 feet. I tried to bring it down slowly by retard the throttle gracefully and then the throttle hit bottom. Still climbing, switched to STABILIZE and it started down at a pretty good rate but as I neared the ground I could not slow the rate so I switched back to ALTITUDE HOLD and it shot up again. I have no X or Y control and am pretty panicked at this point and just want it back rather than flying away and getting lost so back to STABILIZE and zero throttle where it descended until it hit the ground hard enough to flatten the landing gear out and knock the battery loose.
My Question: What could be causing the boxcar effect on my Vcc graph? Pretty sure the only load on the 3DR module is the APM, GPS and minimOSD. Should not be anywhere near overloading the power module should it?
My setup:
Alien 550 folding frame
Hobby King 16000 4s lipo
Hobby King 40amp opto esc's
Sunnysky x3508S-16/700kv motors
15x5.5 CF props
mininOSD
Ublox NEO-M8N GPS Module
400mw 5.8g tx
4line CCD camera
headlights from Eachine racer 250
neopixal tail lights via on board Adafruit Trinket
external super bright status lights from Aliexpress
3DR power module
3.1 mini APM
5v UBEC for tail lights and trinket
12v UBEC for minimOSD, headlight, camera, vid TX and external status LED's
Replies
Ok, I did connect the radio 5v and GND directly to the 3DR power module then I added a 25v 470uf cap across the power bus of the DTFUHF.
The new traces look much better! New one is on top, old on the bottom.
More tech stuff and chat here http://discuss.ardupilot.org the new home of the ArduPilot forums
I can't explain why those voltage drops seem to last for several seconds. Only thing I can think of at the moment is that the log just happens to be showing a harmonic (logs data at just the right time to land on one of those 25ms pulses low or high every time for those seconds)
Well there ya have it. It IS the DTFUHF RX. Here is a pic of a three minute log of the Vcc.
First minute: power on drone, TX off
Second: powered on the TX
Third: removed power from the RX
Now what the heck to I do to stop this. Big Capacitor on the radio power input? Have to check the schematic and make sure that won't cook something on the power connection current surge.
Ya'll feel free to chime in here. Dang quiet out there......am thinking if this problem has not gotcha yet it will sooner or later for those running a LRS.
Picture0003.jpg
Just saw the note where it said post .bin file not .txt
22 31-Dec-69 4-00-00 PM.bin
scope shows a pulse every 25ms, amplitude was .15 v dropped across about one ohm. Duty cycle was close to 50%. Well I think that only amounts to about 0.0225 additional watts. I don't trust my setup that well. No precision dropping resistor and hokie jumpers.
Well how about the ole brute force method? Clear the logs, fire up the drone system, log for a few, disconnect the RX, log and repeat. See if boxcar goes away.
Had to turn on the TX to get the RX to transmit. Shows 0.069 amps now but that is still only 0.345 watts. Am betting the pulse is pretty short and still may be my power spikes. Scope next.
I may have found a culprit. The power for my TX is supplied by the APM. What? lil' ole TX? Ya, a DTFUHF one watt TX.....am betting when I get an amp meter on that....well, data to follow.
Well that didn't check out. DTFUHF pulling 0.026 amps......little bit O nuttin'.....Hmmm. Me thinks the cheapo Innova walmart meter might be full of it. Mr. Ohm's says that only works out to 0.13 watts. Don't think I am going to get one watt out with a tenth in....more checking!