High Altitude Copter for Mountain Flying, Design, Construction, and Setup

[Michail_Belov]

Yesterday I have finished real world test flights of my copter, and so far I feel I am on the right track, so it may be now time to give some details about the built if someone would want to replicate this copter.

What follows is data which was tested in real world flights in the mountains (more than 25 flights, over 7 days, total flight time 14 hours, at altitudes of 2000 … 6500 m).

I will provide complete info about “as built” and then will make some suggestions for improving (I am working on the next version already).

Legal Disclaimer: Where I fly, the highest mountain is close to 6900 meters, I fly no higher than 120 meters above ground, so legally I can fly to 7020 meters, and if it feels that the copter is higher than 120 meters above the ground, that is an optical illusion…

Objective: Design and build High Altitude Medium Range Copter for flying in mountains in adverse meteorological conditions with limited payload capacity for filming.

Flight to a Glacier at 5000 m

[Michail_Belov]

Tested Specifications:

Control Range RC: at least 20 km under ideal conditions (clear line of sight, antennas perpendicular to each other)

Video Range: ++20 km.

Flight Range under ideal conditions: 20 km. (i.e. out and back for a total of 40 km., no wind, at altitude of 500 meters, no reserve, Liion battery cell voltage down to 3.0 V without load)

Flight Time under Ideal Conditions: 55 minutes (ideal conditions in this context, sea level, slow flight at 15…25 km/h, Liion battery cell voltage down to 3.0 V without load)

Flight Time under Ideal Conditions with extended 30 cell battery: 73 minutes (ideal conditions in this context, sea level, slow flight at 15…25 km/h, Liion battery cell voltage down to 3.0 V without load)

Hover Time under Ideal Conditions: 50 minutes.

Useful Range: 10 km. (Useful range being defined as range where the copter can fly out and back, starting at altitude of 0 to 3000 meters, climb +2500 meters, stay/hover for 10 minutes, return back with a reserve of 10 %, with winds around or slightly below 20 knots.)

Tested Wind Resistance for Landing: 20 knots

Tested Wind Resistance for Flying: 30 knots (at 1000 meters)

Tested Wind Resistance for Flying: 35 knots (at 4000 meters)

Tested Flight Behaviour: to altitude of 5500 meters.

Max Speed at sea level: 80 km/h (limited by Z vibrations) level flight

Max Speed at 5000 meters: 105 km/h (limited by common sense).

Max Tested Rate of Climb in Acro Mode: 17 m/s

Max Tested Descent Rate in Acro Mode: 19 m/s

Max Tested Descent Rate in Alt Hold Mode: 3 m/s

[Michail_Belov]

Parts:

Motors: Sunnysky 3508 600 kV

Props: EOLO 13 x 5

Standard Battery: 5S 4P Samsung 35E or similar, weight 1050 grams

Extended battery: 5S 4P + 5S 2P, Samsung 35E or similar, weight 1600 grams (No comprehensive high altitude tests were run with this setup).

ESC: 4 x 1 60 A Speedybee BLS

FC: Matek H743-slim

RC: Speedy Bee 2.4 Nano LRS (with Radiomaster Ranger Micro 1 W TX, standard TX antenna)

VTX: Analog AKK 3 W transmitter, run at 2 W, with stick antenna, receiver RC305 with 17 dbi maple wireless antenna, mounted on a simple tracker.

Frame: DIY balsa/fiberglass/carbon fiber sandwich construction with carbon fiber 250 mm arms with 16 mm o.d., with root reinforcement of 18 mm o.d.

Payload: HD camera, 100 grams

[rmackay9]

Hi @Michail_Belov,

Nice video!! thanks for that

[Michail_Belov]

Well, here are two more flights of the same type…

Somewhere at 5000 meters...

Another flight...

Next urgent thing is a more civilized gymbal. The videos here were filmed on a cheap HD runcam camera with mechanical stabilizer.

[Michail_Belov]

Some construction details of the copter:

The arms: permanently glued to the main board, it looks rather shabby, but very effective and very light weight, much better than bolted connections, about 6 grams per arm. The motor mounts are also glued in using carbon fiber tows and epoxy and weigh an extra 12 grams per mount. The frame/arm joint is very rigid, virtually impossible to twist.

Now I have very big issues with Z vibes which appear when flying above 80 km/h, where Z vibe may be around 10 at speed of 50 km/h and 300 W and above 70 at speed of 95 km/h and 600 W.

I implemented a LUA script which quite succesfully reduces power when Z vibe vlues go above a certain point.

I have two theories for that: it is either due to a too rigid connection between frame and arms or the arms are too slender. I am building another copter with bolted arms just to test out this problem. I am even thinking about placing a very thin soft material between the frame and the arms so as to prevent propagation of vibrations from arm to the frame and dampen them.

The main frame is made from balsa wood, covered in 0.5 mm fiberglass cloth and strengthened with carbon fiber tows.

It is definitively too big, because I had that idea of using very large batteries initially, and I was wrong about it. So ideally, the length of the frame can be cut by at least 8 cm, and that would result in weight reduction of avbout 50 grams.

[Michail_Belov]

The underside of the board - the antennas go here. The VTX antenna is close to center, but due to the board length - 30 cm `- it can become obscured at high angles of inclination, especially when flying back, because the HD camera is mounted low.

I definitively would decrease the length of the board and increase the height of the antenna (i.e. placing it lower).

[Michail_Belov]

Here you can see the mechanical stabilizer for HD camera, it works extremely well, and cuts off all the oscillations, but obviously not the turbulance. It weighs 50 grams, of which about 35 gram are the counterweights. The idea was to put them as far out as possible, and this idea worked really well. I see zero oscillations in video, although the copter has more or less constant oscillations with amplitude of 0.2 … 0.4 degrees.

[Michail_Belov]

Finally, the GPS mount, away from Liion battery

[Michail_Belov]

The copter is running on 4.6 Beta 3 version arducopter, and here is the parameter file for 13 x 5 EOLO props and total flying weight of 2.2 kg.

ESC is loaded with BlueJay 21.0 48 kHz.

LR SummySky APC 13 x 5 21032025.param (35.2 KB)

I do not want to go into too much details as to how it was setup, but I wanted to point to a few unorthodox features.

First, RTL has a climb of 80 meters, so as to be able to clear any potential loss of signal if the copter flies behind a mountain wall. In fact, the 80 meters could be too little, possibly I would have to change it to 150 m to be aboslutely safe. Ideally, cyclic smart RTL would be the best solution instead of having to shoot up, but vertical climb always saves the day.

Second, under no circumstances there is a RTL for low battery or critical battery. It already happened to me that the copter tried to do weird things when I was landing, once, when I was at 50 cm. from ground the copter started to climb, due to battery failsafe which causes RTL which causes 80 m vertical climb… Not a very good idea to do when you are low on battery.

My conviction is that if you need a battery RTL, then you should not be flying.

Below is typical OSD setup necessary for mountain flying. It is cluttered, but really everything is needed. Not sure if MinimOSD has better fonts, but the stock fonts in ardupilot are really not the best, and the Pitlab AP OSD which I was using for planes has a much better visual interface:

Some points:

• Vertical speed is extremely important! One can not really judge well from the screen where one is going. This indication is the most important for me.
• Power: I have both the Amperes (8.07) and Watts (136).
• Throttle percentage is useful only to know when one approaches the limit, otherwise I am not using it.
• pitch/roll useful to have an idea of the wind
• The EFF: line is short LUA which gives both vertical and horizontal efficiency (mah per km).
• Compass and heading virtually useless, I think I will eliminate them.
• Home arrow and ground speed arrow, also extremely useful in high wind situations, because one can not judge well where one is going with respect to the home. If both are aligned, we are going home. I may put the home arrow much closer to Ground speed arrow.
• Voltage, Mah used, permanently monitored

[Michail_Belov]

One thing I plan to do is a LUA script which will give me vectors for flying just on the verge of loosing signal.

The flights start often from narrow valleys, and from time to time I have to look back and check where the Home is and whether I am going to loose the Line of Sight behind some mountain. Often the most interesting things are behind the ridges or to the side of the valley…

The LUA script I have in mind is the following: I fly out, look back, and inch myself to the point where I am about to loose LOS. There I press a button, and from now on LUA will indicate by means of < > and UP/DOWN if I am on the same line, i.e. I fly 5 km out, press the button, and continue roughly away. If I deviate from that straight line LUA will tell me…

Unlike planes, I have lost LOS many times already, but with copter I always recovered manually, i.e. without RTL. With climb rates of 10..15 m/s and the ability to brake on the spot it is not very complicated provided that you are ready for this. The LUA script will make thing much more relaxed in this sense.

[amilcarlucas]

Nice pictures, good descriptions and amazing scenery. Thanks for sharing

[Michail_Belov]

I have modified the Lua scripts for the Glideslope feature.

Here is a short video.

Basically, I fly out to some reference point, usually that would some gap in the mountain ridges. There I press shortly the special push button type switch, which sets the reference point, and the LUA now gives me indications as to how to stay on the line. The idea is when you want to fly as far out and as low as possible without loosing the Line of Sight beyond some ridge, and if you follow this line, you should be safe in terms of signal.

There is a vibe activation indication, which kick in above 12 (mostly for the sake of video quality), i.e. when you try to get best footage, this indication should not appear.

There is also a short high speed flight, where another Lua script, reduces the power when vibes becomes excessive.

When the copter face the home position, the lateral arrows invert their indications…

Glideslope LUA script test.

[prsh8ya (Kevin)]

Interesting, thanks for sharing.