Flitetest Aeroplanes for Heavy Foamboard

Extract

Foam board has become a common material for scratch building RC aircraft. Flitetest.com provide a good range of free plans using this material.  The problem is that the foam board available outside the USA is significantly heavier than that on which the designs are based.  This leads to heavy fast flying planes that are not beginner friendly.  This article looks at two enlarged FT designs that restore the good flying qualities of the original designs.

Flitetest.com (FT) has been around since 2010 publishing high quality video episodes on all aspects of radio control flight.  They are strongly focused on community with forums refreshingly free of negativity.  They are well known for their range of free plans with designs from beginner to advanced each with a dedicated build video.

The main build material is foam board, which consists of a polystyrene core sandwiched between two layers of paper.  You will see it referred to by names like Dollar Tree foam board or DTFB, Bazics foam board and Adams Ready-Board which is the American manufacturer.  Other makes of foam board are available, but tend to be significantly heavier.  Adams Ready-Board is not freely available outside the United States, so the rest of the world has to be content with the heavy stuff.

To give some idea of the problem Adams Ready-Board weights 297 g/m².  The lightest foam board I have found in the United Kingdom weighs 450 g/m², with some sheets reaching almost 600 g/m².

Before I was aware of this, I bought two 10 sheet boxes of A1 foam board: one in black the other white.   I used the black to build an FT Nutball and the white to build an FT Flyer. both are beginner designs.  Having used the recommended electronics, with a 2S 500 mAh battery, I was surprised to find both designs exceedingly tail heavy.  By using a bigger motor and heavier 3S 1300 mAh battery, I was able to get the centre of gravity (CofG) in the right place, but the models were now twice the design weight.  They did fly, but were very fast and not really suitable for a beginner.

There was no way I could reduce the weight using the same foam board, but having only used two sheets out of twenty I didn’t want to buy anything else.  Heavy models are not necessarily bad.  The problem is the weight in relation to the model’s size or more specifically the wing area.  This is expressed mathematically as the wing loading, defined as the models weight divided by its wing area.

The total weight is made up from two parts.  There is the airframe weight which will increase with the wing area. Then there is the battery, motor and control electronics.  As I was already using a bigger battery and motor to get the heavy foam model to balance, this could stay the same.  As the fixed component is a large part of the total, the wing loading will decrease with increasing wing area.

Nutball

The design weight of the FT Nutball is 212g plus 30.5g for the battery.  The wing is 490mm diameter.  Therefore the wing loading is (212 + 30.5 ) / (3.142 x 490² / 4) = 0.00128 g/mm² or 1.28 kg/m².

It turned out for the Nutball that increasing the diameter to 740mm reduced the wing loading back to 1.28 kg/m² using the same electronics.  This was also a good size for cutting the model from two A1 sheets. I kept the power pod the original size, but moved it forward slightly to help maintain the CofG position.

20141219112756-2

Size comparison of Big Nutball to original FT Nutball.

Big Nutball Full Size PDF Download

Big Nutball A3 Tiled Download

The plans have dimensions for those without access to an A1 plotter or don’t want to stick together the tiled A3 prints.  Construction follows the original video.  You just need to tape on the port wingtip before gluing and use the included gauge instead of the power pod for setting the dihedral.

The push rods are long so use three zip ties to keep them straight.

NutBallSheet1NutBallSheet2NutBallSheet3

Flyer

The big Flyer design followed the same procedure.  The original target wing loading was 1.69 kg/m² with its 660mm span.  Increasing the span to 1000mm got the wing loading down to 1.65 kg/m² with the heavy foam board.  Again I extended the nose to ease the CofG placement.

Flyer4

In a departure from the original I added was a KF step. This is an extra thickness of foam added to the front 40% of the wing, with the it’s leading edge chamfered 45°.  An air bubble forms behind the step, causing the wing to more act like a proper airfoil with more lift and smoother pitch response.  The KF step is optional, but I have been very pleased with its performance and deem it worth the weight.  The extra thickness also makes the wing stiffer.

Big Flyer PDF Download

As the Flyer is all straight cuts, I haven’t bothered with a tiled plan.

BigFlyerSheet1BigFlyerSheet2bigflyersheet3

I wanted to add an 808 Key-chain camera on top of the fin to record some flight video so I made a small platform from foam board and taped it in place.  The extra tail weight was balanced by replacing the 3S 1300 mAh battery with a 2200 mAh.  There was no noticeable change to the flight behaviour, but the resulting video showed continuous wing wobble.  This was actually an illusion of relative motion as it was the fin that was wobbling.  I braced the fin with some diagonal BBQ skewers  to the horizontal stabilizer and the wobble disappeared.

Conclusion

I’ve been very pleased with these two models.  They fly well and have proven very resilient.  The Flyer survived two tree landings and extractions with only a broken prop.  The only problem has been summer heat.  It turned out the black foam board’s paper was attached with hot melt glue.  After sitting in the car for a morning it all detached. The white foam board doesn’t have this problem and is still going strong.

Flyer1Flyer2Flyer3Flyer5Nutball1Nutball2

I have since managed to obtain some Bazic branded Adams Ready-Board from which I have built an FT Bloody Wonder (Mk3), another great flyer.  I’m thinking of scaling it up for the heavy foam board, possibly with twin engines so I can use it as an FPV platform.

Analogue FPV First Flight

I set up a small quad with a contemporary FPV system.  In this case an 25mW ImmersionRC video transmitter using my GoPro 3 Black as the camera.  I had wanted to compare the latency of this system against that of the Raspberry Pi.  The problem is that I can’t use the GoPro to measure the latency of itself.  I need to see if I can borrow another one.  For now it was just nice to get out and fly.

The quad is a Quadrixette 30 from http://www.eyefly.info.  It’s quite an old design now, but I like it because it is small (300mm), but still bale to carry some weight.  The flight controller is a OpenPilot CC3D and I’m using a Spektrum AR8000 receiver with telemetry module to report the flight battery voltage.

IMG_20150223_230817 IMG_20150223_230827

One of my club mates was also at the field with his Tricopter so we had some proximity flying.

Comparison of 2.4 GHz WiFi, 5 GHz WiFi and 100 MBit LAN

Extract

Up to now I have been using an Edimax EW-7812UAn V2 USB WiFi Dongle.  This has been very reliable and reasonably simple to install and configure.  The only problem is that it operates on the 2.4 GHz frequency band, which means it will interfere with the majority of modern radio control systems.  Up to now I have been using a 35 MHz RC system which was fine as long as I was flying on my own.  However, I’d like to fly with others who will be using 2.4 GHz systems.

There are now a number of Dual Band Wifi Dongles that operate in the 5 GHz band as well as at 2.4 GHz. This provides a solution so long as they work with the Raspberry Pi.  A couple of candidates were obtained, configured and tested on my Model B Raspberry Pi.  For an extra comparison I also ran a test over the wired LAN interface.

The results were interesting in that there was no clear winner in the speed stakes, there was however a clear loser.  What is clear is that useable 5 Ghz WiFi is possible but for now it’s not plug and play.

Setup

For these tests I used the following network devices.

  1. Raspberry Pi Model B Ethernet (100 Mbit/s)
  2. Tenga W522U Dual Band (2.4 GHz & 5 GHz)
  3. Edimax EW-7612UAn V2 (2.4 GHz)
  4. Edimax AC600 Dual Band (2.4 GHz & 5 GHz)

I setup my Model B, with the latest release of Raspbian (2015-01-31) and the current scripts.

5 GHz WiFi Prerequisites

To get 5GHz WiFi working you need to install the Central Regulatory Domain Agent (crda).  Basically, the 5 GHz band is more tightly regulated worldwide and you need to specify your regulatory domain so that hostapd knows which channels are available.

To install the necessary packages use:

sudo apt-get install hostapd crda iw

Then edit /etc/default/crda and set the REGDOMAIN variable to match your region.  Examples are US, JP, EU and in my case UK.

crda

Devices

The amount of work needed to setup networking varied by device.  I’m not going to give a detailed description of configuring each device.  I’ll do that in separate posts.

1. LAN

This simply required plugging the Pi and the Laptop into a switch and running ipconfig on the receiving device to get its IP address.  This was then used in remote.conf to direct the UDP stream.

2. Tenga W522U

The W522U is nl80211 compatible.  This means it works with a standard install of hostapd.  No special configuration is needed.

3. Edimax EW-7612UAn V2 (2.4 GHz)

The EW-7612UAn V2 uses the RTL8191SU chipset.  The kernel supports this chipset for client networking, but not for AP mode.  However, Realtek provide a custom version of hostapd that needs to be downloaded, compiled and installed which does work.

4. Edimax AC600 Dual Band (2.4 GHz & 5 GHz)

The AC600 is not supported by the latest kernel.  All is not lost, it just takes a little more work.  You need to download the kernel source code and the code for a Realtek 8812U driver.  Then you need to install the latest version of GCC from the Jessie repository.  This lets you build the kernel module.  Finally you need the same hostapd build as for the Edimax EW-7612UAn V2 device.

Hostapd Configuration

There are four combinations of hostapd.conf depending on the driver and band.  I’ve set them out in the table below.

Hostapd Version 2.4 GHz 5 GHz
nl80211
interface=wlan0
driver=nl80211
ssid=Pi_AP

hw_mode=g
channel=13
macaddr_acl=0 
auth_algs=1 
ignore_broadcast_ssid=0 
wpa=2 
wpa_passphrase=Raspberry 
wpa_key_mgmt=WPA-PSK 
wpa_pairwise=TKIP 
rsn_pairwise=CCMP
interface=wlan0
driver=nl80211
ssid=Pi_AP

hw_mode=a
channel=44
ht_capab=[HT40+]
ieee80211n=1
macaddr_acl=0 
auth_algs=1 
ignore_broadcast_ssid=0 
wpa=2 
wpa_passphrase=Raspberry 
wpa_key_mgmt=WPA-PSK 
wpa_pairwise=TKIP 
rsn_pairwise=CCMP
Realtek
interface=wlan0 
driver=rtl871xdrv
ctrl_interface=/var/run/hostapd

ssid=Pi_AP
channel=13

beacon_int=100
hw_mode=g
ieee80211n=1
wme_enabled=1
wpa=2
wpa_passphrase=Raspberry
wpa_key_mgmt=WPA-PSK 
wpa_pairwise=CCMP 
max_num_sta=8 
wpa_group_rekey=86400
interface=wlan0
driver=rtl871xdrv 
ctrl_interface=/var/run/hostapd
ssid=Pi_AP
channel=44
beacon_int=100 
hw_mode=a 
ieee80211n=1 
wme_enabled=1 
ht_capab=[SHORT-GI-20][SHORT-GI-40][HT40+]
wpa=2
wpa_passphrase=Raspberry

wpa_key_mgmt=WPA-PSK 
wpa_pairwise=CCMP 
max_num_sta=8 
wpa_group_rekey=86400

Procedure

These tests were a straight comparison between the four devices using the same settings and the same testing methodology as previously.  The resolution was 1280 x 720 pixels with a 6Mbps bit rate.

Results

The table below shows the resulting latency by device.

Device Latency
Edimax EW-7612UAn V2 – 2.4 Ghz 132 ms
Tenga W522U – 2.4 GHz & 5 GHz > 10 s
Edimax AC600 – 2.4 GHz 122 ms
Edimax AC600 – 5 GHz 126 ms
LAN 124 ms

Here they are as a graph.

WiFiAdapters

Analysis

The big surprise is how bad the Tenga W522U adapter was.  The latency was not just two or three times higher, it was two order of magnitudes higher.  There are a number of threads on various forums where other people have found similar issues, although not to this extent, but then they are probably not swamping the feed with a continuous video stream.  Whether this is just a compatibility issue with the Pi remains to be seen.

The Edimax AC600 was 10 ms faster than the previously used Edimax 7612UAn V2 at 2.4 GHz.  At 5GHz it was still faster, but only by 6ms.  Interestingly, the LAN connection was not any faster than the AC600.  On the Pi the LAN connection is 100 MBit/s as is the USB bus, so although the AC600 is theoretically capable of 433 Mbit/s it is never going to manage more than 100.

Conclusion

What we can take away from this test is that there is a limit to how much the latency can be reduced by changing the WiFi adapter.  It is a shame that to get a 5 GHz WiFi link requires a lot of compiling and configuration.  Hopefully this will improve with time.  For now we have another small reduction in the video latency to add to the current optimisations.

Adendum

Just as these tests were being done, the new Raspberry Pi 2 was released promising a x1.5 to x6 speed increase.  As I had used the latest release of Raspbian, I swapped the microSD card from from the Model B to the Model 2B.  What I found was that the Edimax AC600 was no longer detected and I couldn’t find a way to compile the module so that it would.  So that’s my next task.  I may run a quick comparison of the B versus the 2B with the Edimax 7612UAn V2 to see how much improvement the new processor has made.

Raspberry FPiV – Flight 4 – 1, 2, Tree

Extract

The aim of this flight was to test the airframe and the electronics after the unplanned lawn dart into the mud.  It was also the first test of the CRIUS CN-06 V2 U-blox GPS module.  I didn’t bother to stream the video to the laptop.  Instead I used the JuiceSSH app on my phone to access the Raspberry Pi and start a local 1080p HD and GPS recording.

Setup

The Crius GPS cable terminates in a standard 0.1″ pitch, four pin socket.  The pin outs on the Raspberry Pi are spread out over five pins, so I borrowed a single socket housing from a PCB development jumper and re-arranged the connections as shown in the photo below.

2014-05-23 13.11.08

The GPS was held on to the case with BluTak.

The only other change was to mount the camera on an extension so I could point it further down and to the right.  The aim being to get the propeller out of shot.

I didn’t plan for any big tests on this flight so didn’t bother setting up the laptop.  Instead I used the JuiceSSH app on my phone to start recording the GPS and video output on the Pi using a short script.  As I wasn’t streaming the video, I bumped the resolution up to full 1080p.

#!/bin/sh
cat $0

# Zero duration is continuous
DURATION=0
WIDTH=1920
HEIGHT=1080
FRAMERATE=25
BITRATE=10000000
NOW=`date +%Y%m%d%H%M%S`
FILENAME=$NOW-Pi.h264
gpxlogger -i 2 -f $NOW-Pi.gpx &

/opt/vc/bin/raspivid -t $DURATION -w $WIDTH -h $HEIGHT -fps $FRAMERATE -b $BITRATE -n -o $FILENAME

The Flight

Almost immediately I felt something wasn’t working properly.  Whilst the model was flying at normal speed, the climb rate was very low.  The wind wasn’t as strong as previously, but there still seemed to be some significant turbulence tossing the plane around.  After only one and a half circuits it started descending whilst over the far boundary and landed in a tree.

PlaneStuckInATree

The only visible damage was a broken tailplane and a broken prop.  I still had a WiFi connection from my phone to the Pi and was able to stop the recording and shut it down.  I had a 4m long strap in the car with a heavy buckle on one end, but it wasn’t long enough to reach the plane.  As this was lunchtime, I eventually had to leave it up the tree and return to work.

After work I stopped in at a local hardware store and bought 15m of chord and a pack of 10mm shackles.  The plane was still up the tree when I got back to the field, so I tied two of the shackles chord and started launching it skywards.  I soon found out I needed to tie some shackles to the loose end to stop the whole lot ending up in the tree.  After about 10 minutes, I managed to get the chord over the fuselage and pull it free of the branches.

After disconnecting the battery, a more detailed examination revealed the only other damage as a dent in the nose.  As the ESC had been connected for five hours, the battery had continued to drain and when checked was well past the minimum voltage.  One of the cells was reading 2.15V, so that battery will be going for recycling.

Post-Mortem

I was able to extract the video and gps files from the SD card.  It turns out the propeller was still partly visible in the video and because of this I could see the low voltage cutout had activated just as the plane crossed the far boundary. This is why it started descending and ended up in the tree.  Why the battery went flat so quickly was another matter, as was the lack of power and the inability to climb.

Once I got everything on the bench I found the motor felt a bit rough.  I can only assume that grit had got into the bearings after the mud bath.  This could account for the increased current draw with the reduced power output.

Results

I think pointing the camera off to the side was a bad idea, even for a non FPV flight.

The one positive result was the CRIUS GPS.  There was an excellent correlation between the video and the GPS track, as the two pictures below show.

Flight4-PowerLine

Flight4-Trees

Additional benefits for the CRIUS are:

  • Doesn’t require a separate power supply as it doesn’t draw too much power from the Raspberry Pi.
  • It’s a compact all-in-one unit.
  • Freely available from eBay.

Conclusion

From now I shall be using the CRIUS GPS.

As the aircraft needs some rebuilding, I’m going to rework the internals so that I can mount the Raspberry Pi internally, with the camera above the nose.  I also need to build a new power pod the will mount above the wing.

As an alternative platform for some actual First Person View flying, I’m also building a tricopter.

Raspberry FPiV – Flight 3 – Recording on the Ground

Extract

After much trawling through the forums, I found a way to record the video stream at the receiving end using gst-launch-1.0 in such a way that it can be converted to a playable format later on without re-encoding and the resulting loss of quality.  Eventually, replacing gst-launch-1.0 with a dedicated gstreamer application would allow a valid video file to be recorded straight from the stream without the current post processing. This flight also included a test of the external amplified antenna for the Ultimate GPS Breakout board. The results were disappointing, not attaining sufficient accuracy for the desired application. The flight was cut short due to a combination of high, turbulent wind and dumb thumbs, but was long enough to get meaningful results. You need to watch the video at 720p for a meaningful comparison. Please bear in mind that the video has been re-encoded at least once, possibly three times between the Pi and your browser. https://www.youtube.com/watch?v=eCazn-JbR_0

Hardware Setup

The amplified GPS antenna comes with 5 metres of cable and a heavy magnet.  I shortened the cable and removed the magnet, but the parts were still unwieldy when strapped to the top of the plane.  Hardware wise this was the only addition to the previous test. 2014-04-25 13.15.03

 Software Setup

I had already generated and swapped ssh keys between the Pi and the laptop to remove the need to enter a password when connecting. While I’ve been testing different GPS setups , I’ve started gpsd using a small shell script

#!/bin/bash
echo Starting gpsd
sudo gpsd /dev/ttyAMA0 -n -F /var/run/gpsd.sock

I want to use the current time and date for the video and gpx file names.  As the Pi does not have a real time clock, I set the time from GPS using a  script based on one I found at  http://blog.petrilopia.net

#!/bin/bash
sudo date -s '01/01/2014 00:01'
sleep 1
GPSDATE=`gpspipe -w | head -10 | grep TPV | sed -r 's/.*"time":"([^"]*)".*/\1/' | head -1`
echo $GPSDATE
sudo date -s "$GPSDATE"

Next I run cgps to check the GPS accuracy.  Once it stops improving I exit cgps and run the third script to start gps logging and record the video stream.  The video is recorded as the raw h264 stream with a Tx (Transmitter) suffix.

#!/bin/sh
cat $0

# Duration in milliseconds. Use 0(Zero) for continuous
DURATION=1800000
WIDTH=1280
HEIGHT=720
FRAMERATE=25
BITRATE=2500000
NOW=`date +%Y%m%d%H%M%S`
FILENAME=$NOW-Tx.h264
gpxlogger -i 2 -f $NOW.gpx &
IP=$(ip -o addr show wlan0 | sed -n 's/.*inet \([0-9.]*\)\/.. .*/\1/p')
PORT=5000

/opt/vc/bin/raspivid -t $DURATION -w $WIDTH -h $HEIGHT -fps $FRAMERATE -b $BITRATE -n -o - | tee $FILENAME | gst-launch-1.0 -v fdsrc ! h264parse ! rtph264pay config-interval=1 pt=96 ! gdppay ! tcpserversink host=$IP port=$PORT

Finally, back on the laptop, in a new terminal window, I run a script to capture and display the video stream.  The h264 file is multiplexed (muxed) into an mpeg transport stream (ts) container with a Rx (Receiver) suffix.

#!/bin/bash
# Read the Settings file
source ./settings.conf

# Set file name from current time
NOW=`date +%Y%m%d%H%M%S`
FILENAME=$NOW-Rx.ts

# Read the stream.
gst-launch-1.0 -v tcpclientsrc host=$IP port=$PORT ! \
 gdpdepay ! rtph264depay ! \
 h264parse config-interval=96 ! \
 tee name=t ! queue ! \
 video/x-h264, framerate=25/1 ! avdec_h264 ! videoconvert ! autovideosink sync=false t. ! \
 queue ! mpegtsmux ! filesink location=$FILENAME

# Try to copy the file from SD card to Hard Disk
if [ ! -d $TARGETDIR ]; then
 sudo mount /dev/sdb4 /mnt
fi
cp $FILE $TARGETDIR/$FILENAME

Here is a sample settings.conf

#!/bin/bash
export IP=192.168.42.1
export PORT=5000
export FRAMERATE=25
export TARGETDIR=/mnt/home/user

The Flight

When I arrived at the field the wind felt to be blowing at a constant 10 mph.  In the short time it took to get everything ready it had started gusting much harder.  Having driven to the field and got everything setup I was determined to fly so opened the throttle and gave it a good chuck. My first realisation was that the wind was stronger than anything I had flown in before.  In order to get the aircraft to track up the field and not be blown out the back, I had to point it 20 to 30 degrees past the desired direction of flight.  Also, as the aircraft got further up the field and closer to the trees, the turbulence made it harder to control.  Eventually, it got bucked around to the point I lost orientation, dumb thumbed the rudder flipping it upside down and I wasn’t able to recover it before it hit the ground. When I checked later, the met office were quoting 26 mph gusts.

The Aftermath

The previous months rain had softened the ground enough that the aircraft nose was stuck in three or four inches with no apparent damage.  The rubber bands had separated, releasing the wings.  Any electronics that weren’t screwed down were scattered around the crash site and given a liberal coating of mud. Everything was dismantled and cleaned in fresh water, then left to dry.   When I did reconnect everything, it all worked.  The only problem was that camera now has a green tint.  I suspect that that the mud has abraded the lens coating, either on impact or during cleaning.  As they are cheap I bought a new one.

Results

I was worried that the crash would have lost or corrupted the on-board video recording.  As it turned out, only the last few seconds of the flight were lost, as you can see in the video.  After collecting the scattered components, I found the X terminal on the laptop reporting that the SD card had become read only.  It was still receiving and recording a video stream, although this was mostly a totally brown frame which turned to a greeny orange when I removed most of the mud.

GPS

I’ll cover the GPS first as it’s going to be short. The results using the larger amplified antenna were better, but still not good enough.  There seems to be either a lag or an offset in the computation.  The two images below show the problem.  For the next flight I’m going to use the CRIUS CN-06 V2 U-blox GPS module. Flight3-CrashSite Flight3-CricketSquare

Video 

The video recorded on the Pi (Tx) was copied to a Windows PC using WinSCP.  Then I used MP4BoxGUI to mux it into a MP4 file.  I have not had any problems using the generated file. With the video recorded on the laptop (Rx) as a transport stream, the first task was to convert it into an mp4. The following script takes the file name as an argument and creates two new videos.  This need to be run on a machine using Gstreamer version 1.2 (Check using gst-launch-1.0 –version).

#!/bin/bash
echo .
echo Remux
echo .
gst-launch-1.0 -v filesrc location=$1 ! tsdemux ! h264parse ! video/x-h264,framerate=25/1 ! mp4mux ! filesink location=$1.remux.mp4
echo .
echo Re-encode
echo .
gst-launch-1.0 -v filesrc location=$1 ! tsdemux ! h264parse ! video/x-h264,framerate=25/1 ! avdec_h264 ! videoconvert ! x264enc bitrate=2500 ! mp4mux ! filesink location=$1.reenc.mp4 

The first gstreamer pipeline (Remux), simply extracts the h264 video from the transport stream container and them multiplexes it into an mp4 container.  The h264 stream is unchanged, so there is no loss of video quality.  To make this file viewable on Windows required a couple of extra steps.  With the file copied on to the Windows PC I used MP4BoxGUI to demux and extract the h264 stream (again) and then to Mux it back into an MP4 file.  This new file was playable in Windows media Player. The second gstreamer pipeline (Re-encode), re-encodes the video as a new h264 file and multiplexes it into an mp4 container.  While re-encoding has to reduce the picture quality, it is not obvious without close inspection.  I use the re-encoded video in Sony Movie Studio 11.0, which had trouble reading the first remuxed file.  I also had to manually set the frame rate in Sony Movie Studio 11 to 25 fps as by default it was using 29.97 fps making it impossible to sync the Tx and Rx videos together.

Conclusion

The video posted on Youtube shows the Tx and Rx videos side by side.  They are fully synced and without a gap between I don’t think you can spot a join or see any difference in video quality.  Although a bit of a faff, for now I have a working pipeline to record usable video on the ground. As for the GPS, the search goes on.

Raspberry FPiV – Flight 1 (Line of Sight)

Extract

Having established from my tests with gstreamer that a low latency video stream over WiFi could be achieved from the Raspberry Pi and that a usable range could be achieved,  I decided to mount the system on an aircraft for some line of sight testing.  In order to extend the range test, I added an Adafruit Ultimate GPS Breakout Board.

The results were mixed: the video stream remained consistent throughout the flight although with higher latency than expected and the GPS accuracy was poor.   The reasons for the problems have been identified and fixed, so the overall results are still promising.

Computing Setup

To save booting from a thumb drive every time, I resized the Windows partition on my laptop and installed Lubuntu to the end of the hard disk.  This was fine for a while, but then an update to gstreamer made it impossible to install the bad and ugly plugin packages together.  In the end I gave up on Lubuntu and installed Arch Linux on both the Laptop and the RPi.  The driving force for this was that version 1.2 of gstreamer was available on both platforms.  Having never used Arch Linux before it took me most of the Christmas holidays to get everything running properly.

To save a bit of weight I replaced the Raspberry Pi model B with a model A.  The SD card can be moved between the two boards without changes.  An  Adafruit Ultimate GPS Breakout board was added to extend the range testing.  This was connected to the Pi’s UART pins.  I didn’t solder on the battery holder for the real time clock.

The board was mounted in a bright raspberry pink case, so it would be easier to find should it become detached from the plane.  I trimmed the plastic around the SD card slot so the camera cable could be run out above the card.  The camera was mounted in a PiBow holder and bolted to the case.  As the PiBow holder clips together, I glued the joints to secure everything.

Finally, I created some scripts to get everything configured and talking.

Aircraft Setup

I retrieved my old “Sorcerer” slope soaring trainer from the loft.  This design dates from 1969, although mine was built during the eighties.  It’s not pretty, but it is tough.  The radio system was a Futaba Attack 4 channel system on 35MHz and so wouldn’t interfere with the 2.4GHz WiFi signal.  As it hadn’t been used for 25-30 years, I replaced the transmitter battery, but apart from that everything worked fine.

As this was a glider, I built a power pod from birch and lite plywood to sit in place of the hatch.  Having the motor up front put the centre of gravity in the right place without adding additional weight.  The flight battery was replaced with a 1300 mAh LiPo which powered the receiver and servos through the ESC’s Battery Eliminator Circuit (BEC). I used a propeller that had previously been used on a multirotor that was well balanced.

2014-01-10 13.28.19

The Pi was powered by a dedicated 5V 3A BEC connected to the LiPo battery’s balance connector.  Everything was held on with rubber bands.

2014-01-10 13.28.28

The Flight

The first launch turned into a powered glide.  I hadn’t allowed enough throttle for the extra weight and drag.   The second launch had it climbing and zig-zagging up the field.  The Sorcerer has a huge rudder on a fin that is little more than a post.  With the added prop wash over the rudder, turbulence from the everything strapped on top of the wing and the gusting wind, it was difficult to keep it straight and smooth.  Once some height had been achieved and the throttle reduced everything settled down. I flew for about 5 minutes doing big circuits of our local playing field, before bringing it in to land.

Flight1-ScreenSetup

Results

As I was flying without a spotter, I didn’t look at the laptop screen until the flight was finished.

After retrieving the glider, I found the stream was still running which was a good start.  I hadn’t checked the latency with the new system before the flight, so it was disappointing to find it was running noticeably higher than I had achieved in my earlier tests.  I still had the original Raspian SD card and Lubuntu Thumb drive and a quick check showed that that setup still ran with low latency.  Research suggests the higher latency is due to incompatibilities between Arch Linux and the particular Intel chipset in the laptop.  It could also be that the hard drive speed is slower than the thumb drive, especially as the linux partition is at the end of the available drive space.

The next check was the GPS track.   I used WinSCP to get the video and gpx files off the RPi.  I found a really useful website that animated the track in real-time: www.gpsanimator.com/StarTraX.   I did a video capture from the screen using Applian Replay Video Capture which I then synced with the on-board and head camera footage in Sony Movie Studio 11.  As well as producing the video, this also allowed me to get frame grabs to calculate the gps error.

The animation showed the track jumping all over the place.   If it is to be believed, most of the flight was spent on the wrong side of the trees and I was flying behind where I was standing.  When synchronized with the video it is easy to see the inaccuracy.

Here is an example.  This image shows the glider crossing left to right in front of me.  In the bottom right hand corner is a tree.

Flight1-GPS1

 

Now look at the GPS track. I’ve marked on where I was stood, the position of the tree and the observed track.  The recorded track shows aircraft moving right to left, well behind me and over a small wood. The error is about 60 metres.

Flight1-GPS3

 

An even clearer example is the landing.  The error is about 130 metres.  Seeing as the current maximum tested range is only 240 metres, this is not good.  After being stationary for a little while, the GPS eventually caught up.

 

 

 

Flight1-Landing

I’ve got a Crius CN-06 V2 U-Blox based GPS which has a much larger antenna as the picture below demonstrates.  I also have the high gain remote antenna for the Ultimate GPS breakout, which is the same size as the Crius.  Finally, I’m going to add the real time clock battery holder, which should improve the warm start times.  At least one of these should improve things. 

 

 

GPS Size

 

Conclusion

While there were issues with the latency and the GPS, as a first flight the main purpose was achieved.  The streaming from air to ground worked and remained stable throughout the flight.  There is certainly enough promise to keep the project moving forward.