My quad is once again functioning after my last crash! f450 frame replacement arrived in about a week, but the replacement motor took over 3 weeks. Hard to wait for. So as soon as I got it in the mail, I flew the next morning at a beach in laporte - got some nice footage! html5 video link 1, and html5 video link 2.
Despite the platform being flyable, I am still concerned due to RX/TX reliability, low voltage performance, and vibration.
RX/TX reliability:
In order to test range of my FS-TH9X, I went out to the field across my house, powered up my copter, and left it disarmed. I then left my transmitter on the ground and walked out into the field holding my quad. Viewing telemetry data with my phone using andropilot, I found that my throttle dropped to failsafe value consistently at 400ft.
When I got home, I opened up both the TX and RX to check for antenna continuity, which looked OK. Therefore, I'll be picking up a 9X compatible FrSky TX and RX module to increase my range.
Low voltage performance:
Due to the divebomb performance (in RTL mode due to loss of transmitter) during my last crash, I thought that it might be due to low battery. In order to test, I flew until a low alarm buzzer at 3.6v/cell. Then I changed the buzzer to 3.5v/cell, and hovered for 60 seconds after the buzzer engaged. While it required an increase in throttle to maintain hover, it did not cause any unpredictable behavior, so I am not too worried at the moment. In the future, I plan on picking up an attopilot so that I can view live battery voltage and current.
Vibration:
The swerving behavior in an automatic mode (RTL), also could have been caused by vibration in the internal measurement unit, as arducopter software actually uses acceleration to calculate altitude, and then uses barometer and GPS data to compensate for any errors. So today I turned on IMU logging, and flew a quick hover session indoors. The arducopter wiki article on measuring vibration states that an acceptable vibration level for Z is between -15 and -5 m/s/s. The following graph shows raw data (X and Y at the top, Z at the bottom), and I do fall in that criteria.
However, the vibration damping page states that "You should strive to get in the region of + and – 1/10 G in all axes". In order to check, divide the raw IMU accelerometer data by 9.8 m/s^2 (1 G) and add 1 to calculate variation from standard gravity. My results are shown below - I've reached the 'required' vibration levels of +/- 0.5 G, but about 20% of the time I have vibration greater than 0.1G.
Obviously, I can do a lot better than a bootleg F450 frame and one piece of stick 3M foam as a mounting platform. Until I do a complete platform revamp, I will be staying away from any automatic modes.
I've copied over this content which I posted on my site, http://wrlab.com
Comments
When you get ready to move up frame wise, consider Hoverthings Flip FPV.
I have a Flip Sport and My Freind Oliver has the Flip FPV Pro.
And they are absolutely great and even more indestructible than a real DJI Flamewheel.
okay, if anyone reads this and cares to weigh in...
looking at my last log, i had repeat loss of transmitter signals (ERR 5, 1), paired with an RTL failsafe (MODE, RTL), followed by a return to normal (ERR 5, 0), (MODE, STABILIZE).
HOWEVER! in the last 10 seconds before my crash, here is the timeline i see in the log:
Event Time
MODE, LOITER 0
ERR, 5, 1 3.8
MODE, RTL, 559 3.8
ERR, 5, 0 4.2
MODE, STABILIZE, 559 4.4
ERR, 5, 1 6.6
MODE, RTL, 559 6.6
ERR, 5, 0 9 <=== throttle reception OK, but no mode change??
ERR, 12, 1 13.8
Between the throttle OK (ERR, 5, 0) and crash (ERR, 12, 1), I jerked the throttle around a bit, but the altitude continued to fall. Is it possible that the APM got stuck in some limbo between RTL and the STAB?
@austin, thanks for the feedback.
I haven't tuned anything, i am using all default parameters. using 3.1.2.
The circumstances of my crash were a little strange - all within a few seconds, i switched into loiter mode, my low voltage buzzer started (set at 10.8v) and my transmitter connection dropped out (as I discussed above in the tx/rx reliability section), and per my config, the copter went into RTL mode. it then attacked the ground pretty vigorously.
you can read a writeup and see the KML illustration here: http://wrlab.com/1st-catastrophic-crash/
.....man..... now that you mention it, if it was caused by vibration, RelAlt would have shown an increase in altitude, but its actually an accurate representation of how it crashed. now i'm not really sure what happened.
What was the behavior of the copter like in RTL? I have an SK-450 frame from hobbyking with over 300 hours on it, 160+ of them autonomous. It's my favorite thing to fly, just the right size to strap onto a backpack and take with me, and robust as they come. It has an APM 2.6 on it and just straight double sided foam holding the APM onto it. Needless to say, the vibes aren't quite as good as they could be. The only result of this that I've seen over the last 160+ hours is a slight (1m-1.5m at most) drop in altitude between waypoints, and I have yet to even rule that out as a barometer issue. Even so, it corrects itself every time and locks in at a slightly lower altitude. However, even with my vibes at the higher end of the spectrum, I have not had a single problem with swerving. How long have you been tuning? I spent 2 days tuning on a suspended rope test stand, then another 2 months getting it locked in. Is it possible something just isn't quite right yet? I still don't have everything as perfect as I would like it after 8 months!
I did at one point have it on Kyosho Zeal Gel, and it was brilliant. Absolutely no vibes, just had to do some maintenance one day and ran out of it. I'm a broke college student in a moneyless world, what can I say ;) Helped some with the altitude problems as well, did not get much testing on it. However, if your quad dropped further than that, something is inherently not quite right. What was your default altitude set to? The quad will fly to that altitude, and I've had problems before with RTL in that respect.
Another thing, make sure you're using 3.1.2. I upgraded from 3.0.2, which I had 100 hours on, to 3.1.2, and the difference was immediate. It instantly locked in much more solidly and fluidly than before. It now floats around perfectly locked in and does exactly what I want it to do with absolutely no fussing or stuttering. I cannot not even wrap my head around how much of an improvement it was. Hope all this helps, keep us updated!
Moving too soon is expensive hobby. One thing is sure that persistence pays.
I bought a cheap spider frame from RCtimer. haven't completed the build as the ESC and motors didn't come with needed 3.5mm bullet connectors, so waiting for it.
cheers and good luck
Thanks guys.
@gary and mdisher; are the designs universal? Is it definite that the screw holes will line up, or should I just buy a whole new frame?
Frankly, I'd kind of like to move to an H quad, any suggestions?
Hi Joe, the official line is that + and - 3 (~3/10 G) is OK but I think that is simply a concession to the fact that it can be hard to achieve 1/10G using easily available damping materials.
I use a 3/4" square of Kyosho Zeal Gel under the corners of my APM and PX4's in my 3 Quadcopters (including a real DJI F450 and F330 and get sub 1/10 g.
I highly recommend the Kyosho Zeal Gel.
The other thing is as mdisher says above, get real DJI arms, most of the others are way too flexible, the real DJI ones aren't.
Best Regards,
Gary
Do dump the HK arms and get 4 Legit DJI arms. You'll magically see some issues just disappear. The HK arms are way to soft/flexible. People complain about DJI arms too but swapping them is like moving from a 4 to an 8 on a scale of 1-10. UAVProducts.com has them for $4 each and ships quickly. (No affiliation other than a happy customer).
Good work. More blogs like these is what many want to see.
Cheers