High Altitude Arduino at linux.conf.au

This January, for the 5th year in a row and the 6th time overall, I attended linux.conf.au.  This year I did not submit a main conference talk as I was busy finishing my final year project as the call for papers went out.  I did submit and have accepted two miniconf talks; one about my final year project for the Research and Student Innovation Miniconf, and another with Mark Jessop delivered at the Ardunio Miniconf.

Arduino Miniconf

For the past two conferences Jon Oxer, a microcontroller enthusiast from Melbroune, has run the Arduino Miniconf.  It grew out of past tutorials where Jon has provided Arduino boards for the audience to purchase and walked through how to program them.  It also replaced the Embedded Miniconf which was last run in Melbourne 2008.

The morning is set aside for hardware hacking, which for this group involves soldering together various through-hole PCBs containing various sensors attached to an Atmel ATmega microcontroller.  Add the Arduino bootloader and you have something that is commonly referred to as an Arduino – a term that indicates a certain ease of use for those who don’t want to get down and dirty with the specifics of a microcontroller’s programming interface and data types.  As Jon says in his call for papers, “the Arduino Miniconf gives hard-core software devs who typically attend linux.conf.au an opportunity to experience the joy of hardware hacking”.

Our talk was titled Project Horus: High Altitude Arduino, and it introduced attendees to Project Horus and high altitude ballooning in general.  We focused on the hardware and software that is used, and explained how a balloon is tracked following a launch.  That lead nicely into some story telling where Mark and I told the tale of Horus 8, where Mark’s final year project flew as a payload and landed in the ocean, to be rescued by myself in a canoe and our project founder, Terry.


The two boards being assembled was the KitTen starter kit, and the MobSenDat, a flight computer designed by Luke Weston for recording telemetry in the model rockets that many attendees were building the following day.

MobSenDat top layer PCB layout

It is released as an open source design that you can download and re-use for free. The board contains the following instruments:

The software we flew was derived from the balloon telemetry code written by Terry and the sensor test code by Luke.  We named it MobSenDatHab, and it can be found here.

MobSenDat for Hight Altitude Ballooning

The 8MHz ATmega328P can store this information to SD card, or transmit it over either a XBee (yuck!) or a Radiometrix module; we used the 434MHz NTX2 on one board and the 151MHz TX1h. As we explained in the talk, the modules are used as a voltage controlled oscillator to transmit 300baud frequency shift keyed (FSK) RTTY data to a ground station.  The ground station consists of a upper side band (USB) receiver which mixes it down to audio frequencies, and a laptop which takes this audio signal and feeds it into fldigi, an opensource program for decoding various text-based modulation schemes used by amateur radio operators.

Joel with the MobSenDatHab Horus 14 payload

Horus 14

The week before the conference we had launched Horus 14; a multi-payload flight where we tested the first of the MobSenDat fight computers.  The board performed very well, and was able to be received at times that the normal Nut payload was not.  This could be explained by the higher transmit power – 10dBm vs 20dBm – however a small contribution could also be from the lower frequency used, as path loss is proportional to frequency.  The other payload we flew was a special one.  It contained the HD camera playload that had been flown before, but instead of simply filming the scenery, a small plush Tux – the Linux kernel’s mascot – was placed in the field of view.  Tux flew to 30 276m, covering 180km and flying for 2h 19m.

Horus 14 spacenear.us online tracker screenshot

We recorded 1080p footage all the way up and all the way down, but kept it secret until premiereing it at the miniconf.  It was enjoyed by all, and kept under wraps for a further two days until we showed it to the entire delegation of conference atendees.  Unfortunatly a technical error meant only half the video was seen, but everyone enjoyed it none the less.  Since then it has been made public, and has recieved 2 607 views to date.

In addition to showing people a cool video, the purpose of the flight was to create a unique item to auction off at the conference dinner.  In the past money has been raised for Tasmaian Devil face cancer research, and New Zealand emergency services.  This year the money raised was donated to the Queensland Premier’s Disaster Relief Appeal, to help out those who have suffered due to the extreme flooding in Queensland this summer.  The conference managed to raise $23,239, by auctioning off both the plush Tux and a large print of Tux at apogee signed by conference keynote speakers Mark Pesce, Eric Allman, Geof Huston and Vint Cerf, as well as Linus Torvalds.

Mark and Joel with the print signed by Linus and the keynote speakers


Project Horus: High Altitude Ballooning

One of the hobbies I’ve taken up over the last few months has been High Altitude Ballooning with Project Horus. A group of interested people design, simulate, launch and recover a helium filled latex balloon. The flight and recovery generally last a full afternoon, and the balloon regularly reaches above 30km, with 35km being the highest achieved so far.


The ballons fly a telemetry payload which provides tracking through the reception of GPS signals and transmission of position, altitude and speed to receivers on the ground. The payload modulates the carrier of a narrow band FM transmitter in order to produce two tones, a modulation technique known as binary frequency shift keying (FSK).  Data is encapsulated in RTTY strings; 7-bit ASCII characters with coded start and stop bits, as well as parity information. The strings themselves carry a CRC16 checksum; 4 symbols of error detection for each string of approximately 50 symbols.  The transmission rate is 300 baud, at 10mW on the 70cm band using a quarter wave monopole above a ground plane.

The two FSK tones carrying RTTY as seen by fldigi

The receiver is a scanner capable of decoding Single Side Band transmissions. This decoding produces a two tone audio frequency signal that is fed into a computer, and proceed by a modified version of the fldigi program, called dl-fldigi. The modifications provide convenient presets for high altitude ballooning, as well as the facility to upload to spacenear.us, a web based tracking application.  The spacenear.us site allows the team to open up balloon tracking to anyone able to receive the telemetry signal, which can cover a large proportion of the state as the balloon reaches altitude, and provides attribution by listing the call sign of anyone who uploads a sentence.

Web based tracking software spacenear.us

Chase cars are able to also submit their GPS coordinates to spacenear.us, which helps with the chase teams knowing each other’s location, as well as for spectators watching online.

Other payloads have been flown such as APRS beacons, voice repeaters, a few experimental flight computers, and some multimedia payloads for recording the flight.

Horus 8

The first launch I attended was back in October where Horus 8 was flown in order to test my classmate’s final year project; a telemetry payload designed to operate in the Antarctic and transmit data back to a receiver in Australia.  Mark’s project was flown along with the normal Nut flight computer, as well as a HD camera intended to record 1080p video footage for the duration of the balloon flight.

Horus 8 Launch at Grahams place outside of Mt Barker

Some misshaps occured right at the start; the full balloon escaped our clutches and released an amount of it’s helium.  With limited reserve supplies, we flew with the bare minium required to obtain lift, and this meant a very slow accent rate.  Possibly due to the slow ascent rate, the balloon began to float at apogee – not gaining altitude, and therefore the normal drop in external pressure that causes the balloon to expand and busrt did not occur.

Chasing Horus 8: The front seat car computer running dl-fldigi

As it headed out over the coast towards the open ocean, it did burst, but the landing site was 50m out to sea at Carrickalinga.  As the sun went down, a strong offshore wind was pushing the payload further out of reach.  A dramatic recovery involving Terry, the project founder, employing his swimming tallents, and myself jumping on a “borrowed” canoe that we found on the beach resulted in a reocvery of all the payloads, but the electronics did not fare so well having come into contact with the sea water.

Terry on the left, myself canoeing on the right with balloon in tow
Landing path in the sea at Carrickalinga

Horus 12

Since Horus 8, I attended a second launch that was much more successful.  This launch was funded by Tony Wheeler of Lonely Planet, and the idea came from artist Chris Lansell. It flew a HD camera and a still camera, and it obtained some stunning shots at burst of the balloon fragments passing between the lens and the sun.

Horus 8 capturing balloon fragments in the sun as it tumbles

The balloon also captured some images of the recently refilled Murray River system, including Lake Alexandrina. It landed in a field of cows about 100m to the west of the river, just outside of Murray Bridge.

Coming in for landing over the river at Murray Bridge
Yorke Penisula; Gulf St Vincent; Adelaide; River Murray

I look forward to more adventures with the Horus guys as time goes by. Look out for more blog posts about experiments we fly, and the adventures we have whilst recovering them.


Back to blogging

Whilst upgrading the servers that host Tim’s blog, my wiki and this blog, I managed to break wordpress.  If you were listening to the RSS feed you would have seen all kinds of duplicate posts, drafts and empty posts.  Sorry about that!  The purpose of this message is to tell you that it’s safe to come back now, and service should resume as usual.

A note to those using Google reader; it appears that the old RSS has been cached, so you may still see the old spammy feed.  Hopefully Google refreshes it’s cache soon.