Auto Pilot ordered

I finally order a Auto Pilot for the glider.
I found that there was a $40 discount until the end of next week. That more then makes up for the low $AU

Now to start on an AuduPilot Glider version

More Laser issues

Today I went to use my laser cutter only to find that the output power was nearly non existent.
It turns out that during my last run some smoke particles entered the laser housing and whilst the laser was turned off they landed on the diodes surface.
When it was turned back on it over heated and chipped the fount surface of the diode :(

Now to order a new one....

Back to software

So what do you do with a week and no work?

I got back to software for the visual recognition systems.

1. Fork off a few projects on github
2. Develop and test some changes to make files and sundry scripts
3. Push back to github
4. Merge request back upstream

This is simple :)

why do some people want to do things the hard way and think that merge requests are so hard.
Oh well they can stay in the dark ages....

Happy days are here again

Just picked up from the post office.
New lenses and mounts for the laser cutter.
Be cutting new parts tomorrow :)

It flew !!!!!

Nice still day and so i decided that it was time to stand up and throw the plane.
Standing in the yard out the front of my house, crouched down and full of dread as to what will happen in a few short moments from now.
I throw...
Time stands still as I watch the glider sail off wondering what it will be like when it lands or just crashes.
Slowly it banks right and the wing tip touches the ground.
All is good.
Its in one piece.

6.5:1.5 glide slope. If it didn't bank it might have gone more :)

Happy day.

Next to the sand dunes to try a longer tet flight and balance the glider

At the airfield

So today started by taking the glider to a local airfield for some feedback on the design and I must say I learned a lot from Mitch Gibbs (I hope I got that right...)

The basic feed back was good and I am now encouraged to move onto the next stage after I make some minor adjustments.

1. Fix the nut keeper in the wing that fell out
2. relocate the usb hub as it stops the canopy from going into place
3. Fix the PCB for the wing controls as it failed to function correctly

Update:
All fixed now to go and throw it. Maybe in the front yard first

Charger connection

For a while now I have been pondering how to to charge the twin 12 volt batteries from a single 12 volt supply.
It sounds simple enough, but my issue was that I need to supply 24 volts to the flight avionics for everything to work correctly.

The solution was simple.

A 4 pole 12v relay.

In its de-energised state it supplies 24 volts with both batteries in series. When the charger plugs in it supply 12 to switch on the relay that in turns places the batteries in parallel and supplies the 12 volt charger to them.

Now with the charge circuit working i am off to a local model air field that has a lot of old timer glider enthusiests to see what they say before my first flight tests in a few weeks :)

The start of the avionics and vison computer systems

So it started out as balancing the plane. Testing showed that it required around 1.2Kg of weight in the front of the wings to get it nearly correct. Initially I had a single 12v Gel cell that was about right but latter found that due to its age it was failing under load. The solution was to get a pair of 12v 2A/h seal lead acid batteries. these cam in just under the required wight. 

That is ok as I need to setup a pair of switch mode PSU's. one for the servo control system and one for the on board computers driving the SSD and cameras. Initial static test are ok.

Soon the body will be done and the basic avionics will be in place for test flights.

Buildbot monitoring with check_mk

The following script is used on my buildmaster and report back the status and queue lengths for each individual builder.
Currently uses a script to process out the json response from the server so that it can be processed.

#!/bin/bash
url="http://localhost:8010/json"
json=`curl $url 2>/dev/null`
builder=`echo $json |json.sh |grep state |grep "^\[\"builder" |\grep "state\"\]" |awk -F"\"" '{print $4"="$8}'`
set -- `echo $builder`
for i in $*
do
        export `echo $i |cut -d= -f1`=`echo $i |cut -d= -f2`
done

builder_list=`echo $json |json.sh|grep state|grep "^\[\"builder" |\grep "state\"\]" |awk -F"\"" '{print $4}'`
for q in `echo $builder_list`
do
        builder_curl=`curl ${url}/builders/${q} 2>/dev/null`
        state=`echo $builder_curl|json.sh 2>&1 |grep "\"failed\",\"" |grep -v "\[\]" |grep -v "\[\]" |awk '{print $2}' |sed 's/\"//g' |sed 's/\[//g'|sed 's/\]//g'|sed 's/,/:/g'`
        pending=`curl ${url}/builders/${q} 2>/dev/null |json.sh 2>&1 |grep pendingBuilds |sed '/\[\]/ d'|awk '{print $2}'`
        current_step=`curl ${url}/builders/${q}/builds/-1 2>/dev/null|json.sh -l 2>&1|grep  "\"currentStep\",\"text\",0" |awk '{print $2}' |tr -d [\'\"]`
        export $q="${!q},$pending,$state,$current_step"

done

for q in `echo $builder_list`
do
        state=`echo ${!q} |grep failed |wc -l`
        status=`echo ${!q} |cut -d, -f1`
        step=""
        queue="0"
        current="0"
        if [ $state -ne 0 ]
        then
                STATE="WARNING"
                STATUS="1"
                MSG=`echo ${!q} Failed|awk -F"failed:" '{print $2}'`
        else
                STATE="Idle"
                STATUS="0"
                MSG=""
        fi
        if [ $status == "offline" ]
        then
                STATE="Slave Offline"
                STATUS="3"
                MSG=""
        elif [ $status == "building" ]
        then
                build_step=`echo ${!q} |cut -d, -f4`
                STATE="Building"
                current="1"
                queue=`echo ${!q} |cut -d, -f2`
                MSG="$MSG builder queue has $queue jobs waiting. Current build step $build_step"
        fi
        echo -e "$STATUS Pulse_$q ${q}=$current|${q}_queue=$queue $STATE $MSG"
done

Power for the Glider

Just found the best little switch mode power supply from www.jaycar.com.au the KC5508 (oddly doesnt show on the web site)

It is a 1.2v to 20v 1.5A supply. Works a dream.

Wing joint connections

It turns out that having stability in the wing joints is hard and whats harder is to thread the servo control cables through the wings.

So to this end I have put in perminate wiring .

The socket is a 8pin din plug through which I run the 2 servos and a usb connection for the on-board camera in the wing.

 

Then when I go to connect the wing up all there is a plug to plug in, the carbon fiber rod (at rear) to lock in and then screw up the top and bottom plates in the middle. Ready to fly. 

Now the balancing act

So here we are about a year down the track and I have learned a lot.
First thing is that all dxf files are not the same. What I found was that when I selected sections from the cad drawing to be cut and exported from draftsight to dxf files that some of the shapes came out wrong. Curves were the main issue. They would be moved and rotated to odd locations. This means that I then needed to go in and edit them as dxf format for the cam software to pick them up.
This then showed that not all splines joined up and the cam software would not sort correctly. There were also cases of the same spline being repeated so that the cutter would do multiple traces of the same path.
Lasers are brilliant for cutting balsa wood but multiple passes just char to much. You can get away with a dvd laser if all you want is a mark but a 3w is better and a 10w would be excellent.
It is surprising how much weight you need to balance the glider. It's just over a kilo in my case.

Its been too long

After what has been far too long I have finally found time to return to building the plane.
I regained my motivation after visiting one of my planed the sites at Seal Rocks on the NSW coast (about 1 hour away). There I found a very easy to access hang glider launch site that over looks a just easy to access beach. So easy launch, easy recovery.

So today's job is to sort out the nose cone and canopy.

Air brake woes

Whist finishing of the plane i have run into some problems with the air brakes.. They just do not close fully

So I went out to one for the local model air field to see some of the glider pilots and talk over some these issues.
Whilst there I found that the NFR (Nearly Flight Ready) model kits for gliders also were using brakes and that one of the few jobs that you do is to re-enforce the brake.
So my example is the shaping of some 6mm sq spruce spares to glue to the inside as in the following picture.

 

So now with this re-enforcing they close fully.

I also found out some more interesting information on the controls. The air brakes are normally connected to the throttle control and are therefore proportional in control.

Also my total aileron control length it about 50% of the wing width and on the NFR models its more like 25%. I wonder what that will do the flight control, might be very twitchy.

Changes to Arduino 1.0

I tried to make buildbot build my old X10 automation code and found some issues.
The builds kept breaking with errors around WProgram.h
Turns out that the new version needs this to be changed over to Arduino.h

So how to do this? Easy when you run Linux.
1. Change to the base code direcrtory.
2. run the following little bit of bash code
           for i in `find`
           do
                sed 's/WProgram.h/Arduino.h/g' -i $i
           done

Return to buildbot

As my glider takes shape and starting to get towards being "flight ready" I find it is time to start looking at the software that will run on it.

So what will the plane be running?
1. An autopilot and that will be ardupilot as it is very stable and will not need much modification and any that are needed can be contributed back to open source.

2. Some form of image software for building a point cloud from an un-ordered set of images.

3. Sync software for downloading data once it back in range automatically. Then of course is what platform will all this be running on.

The autopilot will be running in an arduino mega and the other software will be on single board Linux PC. It is looking like the Linux platform will be on a RaspberryPI if I can get openCV running smoothly. Time will tell.
So after dusting off my old configurations for buildbot and setting up the new master, gitolite server and gitweb access on my ec2 instance it was time to start testing.
First problem was setting up all those ssh keys for each of the build servers so that I can build code for each of the platforms.
The RaspberryPI and the main x86_64 buildbot slaves were easy and just needed to install via "easy_install buildbot-slave" and then configure. Simple.
The master out on ec2 was harder but only in that I had to install via source from git.

Nearly there

Only got the canopy left for the build

Buildbot and git

I had a problem with the buildmaster when I cut over to git in that svn allowed a cache to be used to specify an old version. This is the cachepath setting in the pollers configuration.

With git this was now called workdir....

why could this not be left as the same setting for all VC pollers?

Wing is covered

The left wing is now finished and I am just finishing off the center wing so that I can cover it.
Should hopefully be done on the other end by the end of next week.

Wing tips

After laminating several peices of balsa and spruce together I created the following wing tips.
The former that I used was a roll of solder.

Wing tip former

Aileron controls

So building the wings and the last thing I wanted was to have contol horns comming out through the wings.
What I have now done is used a throttle linkage kit to connect a 3mm carvon fiber rod to the servo.
The fiber goes through some brass tubes to form hinges for the ailerons.

Air brakes

So after much investigation of different systems I have come up with this one