Marc's Public Blog - Rc

2016/08/29 RC Flights over Burning Man, 2016 edition

π 2016-08-29 01:01 in Bm, Rc

It was my dream to do something like this since my first burn in 2002 (you can read my BM reports here: http://marc.merlins.org/perso/bm/ ), and I had my first shot at it the previous year, but despite a lot of preparation, not everything was perfect with my RC aircraft last year. Namely, last year:

I had a small bug which prevented using fly by wire modes while flying in a nice straight line for filming (they otherwise worked fine for safety and return to land, just not for filming)

I had a big motor glider with a front propeller (larger prop that is more efficient), but sadly it of course showed up in the front camera view

Video was 1080p, when I really wanted 4K, but I just wasn't able to put a 4K GoPro on my BFG2600, while I was able to fit one in the front nose of my Sky Eye since it uses a rear prop.

Please read the Q&A section at the bottom if you are thinking about flying a drone at burning man. Drones are restricted and require a permit.

So this year, I took my newer Hobbyking Sky Eye after much testing, and it had these pluses, with extra safety features I added:

Newer version of the pixhawk ardupilot firmware with working FBWA and CRUISE modes including firmware that could recover from upside down flight

Smaller aircraft than last year with less weight and less kinetic energy just in case somehow it were to crash (with still enough room to carry all the electronics and cameras)

Autopilot had redundant sensors, along with a complex EKF filter that can detect bad sensors and fly without them

Triple redundant radio link to control the aircraft, with again the autopilot bring the airplane back on its own if all the radio links are lost. Radio control range was more or less 20km, way more than necessary for extra safety

The autopilot takes over if radio signal is lost anyway

The autopilot also takes over if the aircraft flies too low (altitude deck). This did unfortunately lead to pictures that didn't have some art big enough in the picture, but it was tradeoff for the safety afforded by extra altitude (as a reminder, you usually would want to keep the altitude low-ish to avoid conflicting with full size traffic typically at 1000ft AGL and higher, but in this case, flying within the trash fence is actually safe since it's restricted airspace for full size aircraft).

The autopilot and aircraft obviously continue to fly for quite a while without engine power if the motor dies or the battery were allowed to run low (and there are multiple warnings and radio downlinks to prevent running the batteries low).

While this aircraft has a pusher prop to allow for the front 4K unobstructed camera view, I upgraded the powertrain to provide 800W peak and ability to fly up to 120km/h in case of bad winds. One flight after a storm had winds up to 60km/h, so this extra power capacity was not superfluous. The average drone could have been blown off and unable to come back in that much wind while my plane was able to fly against it without problems, just extra battery draw.

AP throttle controlled modes like RTL were modified to maintain a minimal ground speed even in very heavy winds (some other auto pilots will not notice and kill the batteries flying with insufficient throttle against high winds and fail to return to home).

In prior high speed stress testing at home, I was able to get the wings to fly off the plane under high load at full speed, causing an unfortunate crash, so thanks to that prior testing, I was able to build an new aircraft with re-enforced wings to avoid further crashes. Obviously this is the kind of testing and fixes you want to do at home in a safe place, not at burning man.

But as a nutshell, on top of having 4K video in the front (but only 2K in the rear), plus a 3rd camera on the top in case the 2 recording camera failed, this plane was packed with electronics that are sadly more capable than most airliners out there, and it had a lot of extra power to deal with the low density altitude (up to 33% less performance due to the thinner air) and be able to fly out of a heavy storm and come back to a safe landing (the last day, it indeed flew against a 60kph+ wind and even landed with a negative ground speed, i.e. while going backwards). While it was also able to glide for extended periods of time without power, it had a lot of battery capacity to allow flying for 45mn or more in ideal conditions.

Yes, I realize that this plane does not yield nice steady shots like what you'd get from a drone, but I very much favour versatility and safety over better video, sorry if you were expecting drone-like super steady pans from the air :)

Best of day flights:

Best of sunset/night flights:

Here is a page describing my HK Sky Eye Build:

testing/showing off the video to a friend during preflight checks and tests

Pictures from the sky

you can see from the dust that the wind is often strong, even in the morning

red lightning

lighthouse

pyramids

poor robot :)

there was a big parade of bikes going by

Then, I got super lucky with a sunset to night flight:

sunset turned to night:

art car ring around the man after a heavy storm

Questions and Answers

RC planes vs Drones:
While BM is interested in heavily controlling drones, it does not seem to care about RC planes since not a single one was licensed through the process that seemed meant for drones and both BLM ranger I asked, as well as a BMorg person both told me that their concern and worry were drones. I do support BM restricting the amount of drones to a some pre screened ones operated by trained and responsible pilots, given that drones are way more dangerous than foam RC planes if they fail and fall, and it's way too easy for people without a clue to fly them wrecklessly
Also, putting aside the total absence of glide ratio on a drone, no off the shelf drone I've seen has as many redundancies and safeties as my plane had, so it makes sense to restrict them to areas with a perimeter so that no one gets hurt should one suddenly fail and drop like a stone (RC planes are very unlikely to just fall straight down from the sky unless they break up in flight, so a perimeter doesn't make much sense for them. Also, when they are made out of foam, they are much less likely to cause injury to others, or worse. On the flipside, drones are much easier to fly and yield better video, but personally I'm more interested in safety than image quality.

For the rest, please read my Q&A from last year, but like last year, the pilots and ground spotters will remain anonymous for a variety of reasons explained there, although they were properly licensed to fly RC as per the FAA's requirements and as already explained employed multiple methods to maximize safety.

This shows the autopilot applying 72% power (i.e. power reserve left to get 88kph airspeed, and only 45kph ground speed during a very heavy 61kph headwind, the highest I had flown against so far). This used only 250 Watts out of a maximum of 800W available in case of real need:

Tips and recommendations for RC flyers

Like last year, I beg non expert pilots not to fly at burning man without proper training and permission if you are flying a drone. BM is not a place for you to learn to fly or get to try this new toy you just bought and didn't spend weeks or months building and testing yourself (including its failure modes like low battery, RF being shutoff and so forth).

My aircraft description and its safeties, including months of building and testing, should make it clear that even flying a foam airplane without a propeller in the front, is still serious business if there is any chance you could crash into people, so you really need to know what you're doing and have a very reliable, very tested aircraft with as many redundancies as possible, and obviously you should still refrain from flying at low altitude over people outside of what's required for takeoff and landing.

If I'm being that cautious about flying a foam RC plane, you'll understand that I just wouldn't want to fly a drone at burning man in any other fashion than a ladder over a perimeter that's been cleared of people in case the drone fails and falls out of the sky. This is why drone usage is heavily restricted at burning man, for everyone's safety.

Please read the rest of my tips from last year

last, but not least, if you don't have your full contact info on your aircraft because you're scared of what might happen if you crash and someone else finds it, please don't fly. Thank you :)

See more images for RC Flights over Burning Man, 2016 edition

2016/08/05 120kph Capable HK Sky Eye FPV build with Pixhawk/Ardupilot 3.5

π 2016-08-05 01:01 in Rc

I needed a new plane for my plans of overflying and filming burning man 2016 with 4K resolution, and I ended up choosing the HK Sky Eye for this job (follow the link for the result)

My BFG2600 was my big range, big load FPV platform so far. It's a great platform, but as big as the plane is (2.6m), the cargo bay is not actually that huge, and it's totally packed with electronics, to a point that it's hard not to have them conflict with one another.
The next problem with the BFG is the front propeller, making it hard to get forward facing video without getting the prop in the way (even if it's actually capable of gliding for quite a while without motor power and with the prop folded). Related to this, I wanted to switch to 4K video, and fitting a 4K camera in the plane, preferably forward facing, and without sticking it out the plane where it would foil airflow, didn't seem to be very possible.

So, the previous year I had tried the X-UAV Talon but while it looked good on paper, it was a bit of a handful to fly, and it meant more for long range FPV at high speed than low speed loitering and filming, and due to its wing design and wing loading, it needs a pretty long runway for landing, so it's not ideal for difficult terrain.
Enter the Sky Eye, which is a good compromize between the two by having a pusher prop which is sadly not in the rear like the X-UAV, but still allows decent power (more than 600W if needed), and it comes with an open view front dome:

Power Upgrade

Here are some details on my build. While it comes ARF, I trashed the undersized ESC which I wasn't going to trust and wasn't compatible with a 4S setup anyway (3S was not going to yield enough power for my needs), and while the stock motor did work on the bench, I was not going to trust it to output 700W on 4S when it's rated for 430W on a good day.
So I used a Castle Edge 50 which I knew I could trust and had data logging, along with a bigger motor: NTM 3542 1250KV, 700W/50A peak. The stock setup allows 50-60kph, while my upgraded setup allowed 120kh, which is way safer when flying in high winds.

I'm currently using an aeronaut cam 8x6 prop (8x7 works too but needs a bit more of a throw), and it's making plenty of power
Here's a quick table I wrote then choosing motors and props:

Stock motor, 3536 1200kv, 430W/30A max RPM Watt meterV meterA 12500 295W 11.8V 25A 3S, 8x6, 50-60pkh? 15000 540W 14.8V 38A 4S, 8x6 (>max allowable power 30A/430W) 690W 15.0V 46A 4S, 8.5x6 (way over max allowable amps), 110kph+

NTM 3542 1250KV, 700W/50A peak RPM Watt meterV meterA 13000 250W 11.6V 21A 3S, 8x6 cam folding prop, 60kph? 12300 700W 14.4V 50A 4S, 8.5x6 prop, 60A ESC cutoff 10600 500W 15.0V 33A 4S, 8x6 cam folding prop, 110kph

The space for the motor isn't that big, but I'm thankful the NTM 3542 fits in there. For prop, you can go to 8.5" and if you go to 9" or bigger you need an adapter, or you need to notch the boom, which may or may not be a great idea:

new motor with foldable prop for gliding

Making the Airframe Safe with 100kph+ speeds

The reason why it's important for me to have a plane that can go at 100kph or more, is that I may fly in wind that is up to 50kph, and want enough reserve power to fly against it back to the landing point. At Burning Man, I actually encountered a wind that peaked at 60kph right after a storm, while I almost lost my bixler2 as it was flying full throttle against a wind that was more than 50kph and was actually flying backwards (i.e. it was getting further and further away from me as its battery was draining while it was trying to fly back, and failing).
Now, the next problem with the Sky Eye is that the plane wasn't designed for those speeds, and sure enough, after a light crash during flight testing, the wings got slightly bent, and the very little that was keeping them snapped into place, failed during a test run at 110kph+, the wings got ripped out in flight, and my plane had a very bad crash
Thanks to my multiple radios and GPS coordinates, I did find the wreck, but I never found the wings, they were taken away by the wind at high altitude, and despite a 2H search around and downwind from the crash site, I never found them:

i'm impressed the lipo did not blow up, it was a pretty high speed crash straight down (130kph)

To avoid this from ever happening again, I changed the wing design to allow a strap to go around the entire plane, including the wing roots. Sure, it makes the plane a bit more dirty to the air flow, but I'll take that for wings that won't fly off:

made a hole in the wings, and added reinforcing wood

with a velcro strap

initially I secured the wings like so, until I realized that in my crash the top part with the ESC broke off in flight

so instead I'm wrapping around the body top...

to bottom, for extra strength

With that fix made, the next sizeable modification I made was to allow the flaps to travel down a lot more for crow flaps. The only issue I found was that it wasn't suitable for landing, so I now release some amount of crow to avoid smack landings:

I shaved off the flap hinge

allowing for severe flap down and better descent rates for a short field behind a fence

Avionics and VTX

For the avionics, while the cargo bay has a lot of room, but I opted for putting the pixhawk in a cramped and inconvenient space below the wings. The reason for this madness is to avoid electronics around the flight controller, and not have to worry about putting a battery on top or something else that would avoid free movement on top of the anti vibration foam pads. Cabling does suck as a result, but it really feels like the best place for it:

As a result, I have plenty of room for the electronics and a big battery (4S 5000mAh, but bigger is possible). In there:

3DR power module 90A

RCCCv2 3 channel video switcher plus switch for night lights and strobe

Battleswitch RC controlled relay to switch 12V for VTX and 5V for the 3 onboard cameras

Front Camera: mobius or GoPro Black 4K (both 5V)

Rear Camera: mobius with detachable lens

I2C Airspeed Sensor

MinimOSD with 5V mod, heatsink, and nightghost firmware

Brotronics Diversity RX with backup lipo and lost model finder buzzers

RC controlled buzzer to help finding aircraft in flight

2W 5Ghz VTX in the tail

This is what it looks like:

fit inside nicely, more easily than in the bigger BFG2600

2W 5Ghz VTX gives 5 to 10km range in non ideal environments

Sending the power and video to the tail in a prebuilt plane wasn't too hard, I just punched through 2-3 foam walls with this tool and pushed the cables through:

nicely got to the other side

all good

Video

Now, let's look at the video issue. Like other such blames, it comes with a plastic dome. Let's be honest, all those domes give very unsightly reflections from the sun while in flight. I didn't waste much time trying to fit a camera behind the dome, even if I did a quick test just to verify anyway:

it fits, but as expected I got dome reflections on the video and didn't want that

so I made a ghetto solution with foam padding that I carved. Looks like crap but better than an open hole acting as an airbrake

Now, the other way to do this if you don't need 4K, it's much easier with a mobius. Someone made a nice 3D model for a mobius dome that I was able to 3D print:

it weighs the same as the transparent plastic dome

I later printed this without the rear piece so that I could slide the mobius back, or change its angle more easily

not too bad

I then also looked at other options for getting better than 1080p, and the Joovuu X almost gave me a 3K solution, but in the end the lens extension cable that took way too long to arrive, was too short, and also managed to short my camera and kill it:

so I use a trusty 2K gopro with detachable lens in the rear

I also found an interesting mod for this plane, which was a 3D printed lock pin tail hook that allowed taking the tail off for transport. It was appealing in theory, but after trying it out on my broken plane (it required carving the tail surfaces that weren't big to start with), I concluded that it worked in theory, but there was no way I wanted to take the risk of flying loads with that setup and potentially lose the tail in flight due to weight and G load:

this was the iffy part, lots of foam had to be removed, I didn't feel what was left would be strong enough

Conclusion

I like the Sky Eye, it's very reasonably priced ($150-ish), mostly pre-built even if I had to replace some bits and carve the flaps a bit, and the cargo space is fantastic, loads of room.
I was dubious of the 8" rear prop at a weird angle, but it works well enough and I was able to get 800W out with an 8.5" prop and without a motor or prop adapter, so that worked fine for me. The only downsides are that you still have a whiny noise from the small prop and you're going to use more amps to fly against the wind, but I'll take that for a nice unobstructed few in the front.
I think I only really wish it came with more custom made domes for popular cameras where the lens is in front, unobstructed by a dome.

I'm very happy with the results I was able to get while overflying and filming burning man 2016 with 4K resolution.

See more images for 120kph Capable HK Sky Eye FPV build with Pixhawk/Ardupilot 3.5

2016/08/01 Brotronics PowerTowerRX Documentation for Pin Mapping

π 2016-08-01 01:01 in Rc

After trying and documenting the Brotronics Broversity RX, I tried out a Brotronics PowerTower Rx. While the github page has schematics, sadly it does not even include a clear labelling of each pin, or the fact that output 5 seems not to be wired to a pin at all.

You would buy the PowerTower Rx because you want a lipo backed up receiver which will sound an alarm on a loud buzzer (that you provide) once the signal is lost. It will also switch to transmitting a distress find me beacon after signal has been lost for 45 seconds, so you can use the radio signal to go find it, even if your main battery has been ripped out after a crash (happened in all my crashes).

For more details on how that works, see my post on Brotronics Broversity RX, the missing manual.

This receiver is more compact, cheaper, but does not have diversity RX, which is likely ok for most uses. But as I mentioned, it comes with 0 documentation, and not even a proper pin mapping. I ended up documenting this in this rcgroups post, but I thought I should make a proper page here with that info:

There are 8 outputs in the RX configurator, but I believe only 7 are wired (#5 seems to be going nowhere?)
If you use the buzzer for loss signal and loss model finder (you should) and don't solder to use RX/TX, you're really only left with 3 servo ports since #1 is used for PPM to your flight controller
Port 2 can be connected to directly and 3 and 4 (RSSI and SCL) need to have a custom cable made because the layout did not include enough pins to just replicate Gnd/Vcc in the right places.

Here are the channel outputs you can configure in the RX tab of the openlrsng configurator:

PPM
labelled SDA, can be analog
labelled RSSI on board, can be sending different signal to lbeep pin if you solder a jumper
labelled SCL, can be analog
may not be wired anywhere
link loss indicator (buzzer port, labelled BZ+ BZ-)
RX
TX

The labelling on the board is pretty poor though. Here are the pins:

Pin1: PPM (port #1)

Pin2: Vcc (+ on the board is not aligned with it)

Pin3: Gnd (not labelled at all)

Pin4: Gnd (labelled -, off center a bit)

Pin5: RSSI (port #2)

Pin6: lbeep (I believe it's also port #2 and activated by a soldered jumper)

On the other side of the board, you have:

Pin1: Buzzer -

Pin2: Buzzer + => connect a buzzer to those 2 pins and set output port #6 to 'link loss'

Pin3: SCL is port #4, but you'll have to write your own servo cable with GND and VCC from other pins

Pin4: Gnd

Pin5: Vcc

Pin6: SDA => this allows to plug a servo cable between pins 4,5,6 to get Output Port #2

The wiring does allow to plug a servo cable directly into PPM on one side and Gnd/5V/SDA on the other side.
This means you can connect to port #1 and port #2 directly with a servo cable.

Port #3 require that you route your own wires to make a servo cable out of pin 5 inthe back, Vcc and Gnd

Port #4 is the same with SCL, you have to make your own cable and route Vcc and Gnd

Port #5 does not seem to exist on the pins I could find

Port #6 is nicely wired with Bz- and Bz+, you just connect a 2 pin powered buzzer there

Ports 7-8 for RX/TX are on the other side of the board, you have to solder on them for some reason.

The battery port is also missing a header, but at least it has holes to add one. To be honest, I have no idea why anyone would buy this receiver if you're not going to use the battery backup. There are otherwise some cheaper or better wired ones you can buy (I did buy it for the battery backup which I think is a fantastic idea)

Marc's Public Blog - Remote Control Aircraft