A number of people have asked for some more technical details on the AVA tracker hardware. The full name for the board is PAVA/ATLAS and it is a prototype based on my existing PAVA boards developed with the help of James Coxon, Phil Heron, Nigel Smart, Ed Moore and Dave Akerman.

PAVA is a 70cms tracker board based on an ATMega328P Microcontroller, Ublox MAX6 GPS and a HopeRF RFM22B radio module. PAVA was born from a desire to make a tracker small enough and power efficient enough to be run from a single battery for enough time for a flight and possibly to reside in the neck of a balloon. To facilitate the use of the single cell a Texas TPS61200DRC boost converter is used.

The board was designed to initially run at 3.3v but have the option for running at 1.8v for increased power savings. 3.3v boards use a TPS61201 with a 0R0 in place. 1.8v boards use a TPS61200 with a 510k / 197k resistors to set the voltage. When running at 1.8v the AVR uses a 4Mhz crystal, 3.3v uses an 8Mhz crystal. Extensive work was done on the software and with a combination of the 1.8v hardware and the power saving in code the battery life was increased from 11 hours to about 47 hours.

A combination of the the hardware and software meant the PAVA boards were a very light, robust and good base for future projects.

PAVA/ATLAS was born out of Project Swift. The Swift board was a powerful but large board designed to do concurrent APRS transmissions outside of UK airspace and transmit RTTY telemetry as well. Although a good board there were a number of issues with Swift, namely the power requirements, weight and cost. James Coxon proposed a light weight version that could do the same but potentially under a foil balloon.

Taking the existing design for the PAVA boards a modified version was made to facilitate mounting a Radiometrix HX1 transmitter piggy back on it. Retaining the existing 1.8v board additional features were added, the GPS was wired to the batteries for hot starts. The GPS has a FET based power switch so it could be turned off entirely. The HX1 was mounted on a separate mezzanine board.

With its own 5V step up that could be switched on and off from the AVR it only needed to be on when transmitting then it was turned off again. To get the voltage up to the right level a transistor was used. The HX1 board mounted on top of the microcontroller board. To ensure the GPS antenna wasn’t obscured by the HX1 the board was extended slightly.

The board is very prototype and there are a number of errors on it :

a) RFM22B must be pulled down not up. This is fairly easy to fix as SDN is right next to GND so you can bridge it with a 0603 10k.
b) HX1TXD pin fouls the ICSP header. This was just a stupid mistake, you end up bending the pin to get the programmer on.
c) Holes for the Samtec headers are too large.
d) A polygon priority mistake left VBATT connected to GND. Pretty much a show stopper fixed with some careful scalpel surgery.
e) HX1 APRS TXD is on pin 11 which is use by the RFM22B. Cut the track and rewired it to the other interrupt pin 3.

Microcontroller board on the left, HX1 board on the right:

Underside (RFM22B & GPS to be attached)

The board weighed in at 16g and could run for way over 24 hours from a pair of AA’s with the APRS transmitting every 2 mins.

Hardware is only half the story. To comply with the UK legislation the APRS transmitter could not be used in the air (this also applies to other countries in the EU). Taking geofencing code from Project Swift & work done by Steve Randall a map of Europe was built up using Google Earth ( See this map ) and converted to code. See this post for more info: Preparing for a European Floater.

Once the payload knew where it was it could disable the APRS transmitter to comply with the local legislation. Additionally it could generate the correct ITU prefix for the callsign. This code was tested on two PAVA flights without the APRS module attached to iron out any bugs. The first test indicated the UK Geofence wasn’t correct.

The code use is here : https://github.com/Upuaut/APRS_Projects/tree/master/Pico92

The board will probably be redesigned as a single PCB.

Day 1

What a weekend! I’m happy to report an extremely successful pair of flights for both PIE & AVA. For a write up on PIE please visit Dave’s blog at http://www.daveakerman.com.

The morning started in a muddy field in Cambridgeshire, we’d agreed to meet nice and early about 9am, it was a pleasant if windy day :

Eben & Liz from www.raspberrypi.org had come along to assist with Dave launching the Pi Camera, in tow was a reporter doing an interview and Graham G3VZV had come along as well.

The plan for the day was to launch two 1600g Hwoyee balloons with very slow ascent rates (we were aiming for 1.5m/s). The slow ascent rate would increase the chances of the balloons entering a float (i.e not bursting) at altitude and extending the flights. The trouble with low ascent rates is measuring the neck lift of the balloon is very tricky, near impossible in any sort of wind whatsoever. Here Eben holds one balloon whilst Dave filled the second one :

The neck lift on the balloons was about 300g with my payload coming in at 180g and Daves a little more. The balloons with so little gas in also act like large sails pushing the balloon down into the ground. Here Eben and Liz hold the balloons :

At about 11:44 UK time with the balloons filled we decided to launch, the wind was dragging the balloons horizontal as you can see from this picture from my perspective :

With a break in the wind we took a run and released both balloons (Thanks to Alex from Raspi.tv for capturing the stream)

Daves rose slowly upwards and mine err didn’t. AVA got to about 160 meters and then stopped ascending. We watched as the payload just cleared the trees about 0.5km away, telemetry indicated the payload started to descend so we quickly threw everything in the car (sorry BATC.tv stream users that’s why the stream died).

Thankfully just as we were about to go chase it AVA decided to start ascending properly.This continued until just after 1pm at about 6200 meters in altitude where abruptly the 70cms radio packed up. We were unsure if it was just the cold affecting the transmitter, we wouldn’t know until it made landfall near the Netherlands where the second, APRS, transmitter would hopefully fire up.

At 15:30 as the payload entered the Netherlands airspace the APRS transmitter fired up and was igated up to http://aprs.fi. The HABhub team quickly transposed this on to spacenear.us. As the payload left the Netherlands and entered Germany the APRS continued to transmit.

Suddenly at 20:45 a Polish HAM SP3MCY started uploading telemetry from the 70cms transmitter, it had come back to life! At about the same time the APRS decided to stop working entirely (it wasn’t to come back for the rest of the flight). It was obvious something wasn’t happy as the 70cms kept stopping and then coming back. My code does regularly reboot the RFM22B so this is possibly what kept it working.

At about 22:00 stopped transmitting again. AVA had gone silent and dropped to about 22km in altitude as the gas had cooled down over night. With no APRS and no 70cms telemetry I decided to hit the sack happy the mission had been a success.

Day 2

I woke up at 7am and optimistically glanced at spacenear.us through blurry eyes. To my amazement, AVA was awake, over the Czech Republic and climbing slowly in the early morning sun. DL7AD had received the 70cms transmissions at 0615 UK time, for the rest of the flight the transmitter worked with no failures.

An hour later and AVA had receivers from Poland, Slovakia and Austria tracking it. Over the next few hours AVA climbed to a maximum altitude of 38984 meters, sadly it didn’t get back into a float and around 09:05 UK time AVA burst over Austria and started its decent.

With a rather leisurely decent the payload finally came to rest at the crest of an Austrian mountain 1600m above sea level. We really thought that would be the end of the story. Really I mean who would be able to recover that. We didn’t count on the Slovaks.

Enter Radim Mutina & friends from stsproject.net. Based out of Slovakia Radim (OM2AMR) and friends decided to go attempt the recovery of AVA from Austria. The terrain map showed the scale of the issue :

However undeterred the STS Project team headed out. Using APRS to report their position we watched as they took the slow climb up to AVA’s landing location :

In honour of this we changed their icon to a superman as I thought it was apt! As we subsequently saw from the pictures this was not an easy recovery :

Finally the location of Team STS and AVA were on top of each other, after an initial panic the payload may be in a tree OM1AMJ relayed a message back that AVA had been recovered!

The payload was so high up even on the ground OM1ATS was still receiving it in Slovakia. The final landing position was 47.56405,15.91578.

From left to right  : Peter Vittek, AVA , Radim Mutina (OM2AMR), Juraj Marsalik (OM1AMJ) and behind the camera Brano Janicek.

For full images click here.

So all in an amazing flight covering a distance of 1256km in 22 hours and 29 minutes. My eternal thanks to everyone who turned out to track the payload, I won’t list everyone here as the post would be twice as long and I may miss someone as so many people have contributed to this project. I consider this launch more of a UKHAS group effort. Thanks to batc.tv for giving us the streaming.

Finally how can you say enough thanks to Peter Vittek, Brano Janicek, Juraj Marsalik, Radim Mutina and Dano Radovic liaising between the team on the ground and IRC. You went  way beyond the call of duty spending all day recovering AVA and documenting it so well for everyone to see. Amazing guys, amazing!

UPDATE : Radio failure appears to have been caused by the insulating foam wedged between the two boards, as the foam had expanded due to pressure it had popped the HX1 board off the controller. Also a nice video from Team STS :

Update 2 : People have been asking for technical details on the transmitter being used. It was a custom design described here : http://ava.upuaut.net/?p=551

I am planning on launching from Cambridge on Saturday the 13th of April 2013. With updated code and a map of where APRS in the air is permitted the payload being flow will include the APRS transmitter :

With a pack of 6 x AA batteries (3.6V out seen in pink gaffer in the background ) the payload should have about a weeks power possibly longer.

Assembled in a hollow polystyrene ball with antennas for 70cms (1/4 wave) and 2 meters ( Slim Jim suspended underneth the payload) comes in at fairly portly (for me) 180g :

This will be suspended under a 1600g Hwoyee balloon filled with H2 for an ascent rate of about 1.5m/s. Watch http://spacenear.us/tracker from about 10am on Saturday :

Dave Akerman kindly offered to launch a MAX7C based board for testing last weekend. I prepared the PCB and posted it out to Dave who sent it up under a 1600g Hwoyee with a slowish ascent rate. The plan was for a float but the balloon unfortunately burst rather than entered into a float.

It did however make it to France (just) :

it appeared the parachute tangled badly and it “landed” in a field at 15.9m/s (44mph) :

The impact although significant (it broke the battery off the board and bent the ICSP connector in the process) the transmitter continued operating. F5APQ kindly located the payload in a frosty field very early the next morning still transmitting :

Jacques kindly untangled the mess and returned the payload to me back in the UK ready to fly again :

So a massive thanks to F5APQ to going out of his way to do that.

Technically the uBLOX MAX7C performed as expected with no blips, the APRS Geofence correctly identified when it crossed into France, shame the balloon didn’t float but a good result in the end!

EDIT : The extremely long long times were very odd and I’m not sure something environmental wasn’t going on. I’ve subsequently tested both modules repeatedly and from cold inside (in a window) they are getting 3D locks from cold in sub 40 seconds consistently.

Finally the long awaited sample of the new offering from uBlox landed on my desk. The MAX-7C is the latest module from uBlox offering better power usage and performance. Additionally the MAX-7C can run from 1.65v to 3.6v meaning it can be used on both 1.8v and 3.3v boards with no problems. Additionally the UAV/RC crowd will love the 10Hz updates.

Downsides ? No more TCXO meaning lock times may be slower. I set up a testing environment mounting the MAX7 on a breakout with a chip antenna :

Along side was a MAX6Q for comparison. Firstly I ran the MAX6Q noting the power usage from 3.3v :

Aquiring 60mA (Average over 5 mins no lock)
Tracking 44mA (Average over 20 mins)
Cyclic      15mA (Average over 15 mins)

Its worth noting the GPS did seem to take an extremely long time to lock that evening, normally its nothing like 5 mins.

Now the MAX7 :

Aquiring 22mA (Average over 20 mins no lock)
Tracking 17mA (Average over 10 mins)
Cyclic    6mA (Average over 10 mins)

Conclusion : In the very brief test I did the MAX7 did take longer to get a lock than the MAX6, however once it was locked the performance was very similar, I couldn’t tell any difference. However the big difference is the power usage. Reducing the usage in cyclic by over 50% is a massive improvement. I can’t wait to run test one of these on a the PAVA boards.

Additionally I tested 10Hz mode which seems to be pretty good, however further testing is needed. Once I’m confident with it I’ll have these for sale on http://ava.upuaut.net/store

When I sat down and designed the PAVA PCB I had in mind launching this under a small foil balloon. These are interesting for a number of reasons, firstly you don’t need a NOTAM, secondly under the right conditions they can float. Although alot of these foil pico launches have been done by members of the UKHAS community I decided it was time for me to do one.

Armed with some 36″ Qualatex balloons from Random Engineering I set about making a very small and light payload based on a single AA and the PAVA R7 PCB :

Coming in at 27g the antennas were made from the inside of CAT5 cables. The case was made from a hollowed out piece of insulation with additional thin insulation in close contact with the PCB. See here for further build images :

https://plus.google.com/photos/118244444241111963790/albums/5851582788244965537?banner=pwa

To fill the balloon I used a piece of thin tube. I’d been advised to aim for a neck lift of about 1g, to measure this I had a piece of blutac 1g heavier than the payload. I was suprised how quickly the balloon filled up and it may have been slightly over filled :

Launch was extremely windy :

I was happy to see it go as it was pulling towards the ground. Once in the air it floated up gently with an average ascent of about 1-1.5 m/s. Now in theory these things should float at about 4500m but PAVA didn’t it carried on upwards unfortunately too high for the balloon to remain intact. All of a sudden PAVA started to fall at an alarming rate.

Normally with Pico’s they are so light the balloon acts like a streamer slowing down the decent, however there was nothing slow with the 12m/s it hit the ground at. Fortunately it appeared to land in a field. About an hour later we started getting telemetry from the payload again, it seemed one of the people who had been tracking, 2E0KPI had decided to and have a look for it.

Despite the dramatic decent PAVA was still transmitting, however with fading light 2E0KPI was unable to locate the payload. He kindly agreed to go back the next day and have another look where he located PAVA, still transmitting, in a field. The reason for the descent rate was clear, the balloon had blown its valve entirely out :

A big thanks to 2E0KPI for locating the payload and to all the trackers. I’ll have another go at this soon. Congratulations to James Coxon M6JCX who launched an ATLAS/PAVA board around the same time. In a good demonstration on how to do it his balloon went on a 34 hour jaunt around the UK !

Dave Akerman kindly invited me along to launch on Wednesday the 27th of Feb. Dave was launching a Raspberry Pi powered Tardis and has a great write up >here<. Due to the lack of time to ascertain the legality of transmitting APRS in the air I took the decisions not to fly the ATLAS/PAVA board. Instead I ran the same code on my standard PAVA board. It wouldn’t transmit APRS (wasn’t a transmitter fitted) but it would transmit the current ITU country code via RTTY and should be a good test of the code if nothing else.

Dave kindly fabricated a polystyrene ball container and antenna whilst I was driving down and still managed to find time to paint it fluorescent pink (Thanks Dave! Here on the left shown next to PAVA the backup tracker for the Pi launch on the right) :

We decided to get the floater up first. To get the latex balloons to float you need to fill the balloon such that you get an extremely slow ascent rate, normally its suggested you aim for about 5.5m/s however to float we aim for less than 4.5m/s. In the end we decided to aim for 2m/s, which is a compromise between time taken to ascend and chance of floating.

With the Hwoyee 1600g balloon with H2 and the 65g payload this equated to about 320g of neck lift. This was a problem as the filler and hose weighed 550g. In the end we removed the filler and hung some gaffer tape rolls on the balloon, so the neck lift was about right. Attaching the pink ball the launch was quite calm and the balloon floated off upwards bang on target ascent rate :

At this point Dave and I concentrated on the Pi launch which I won’t cover too much here as Dave covers it in his blog. We occasionally glanced at Spacenear.us and the ascent rate was on track and AVA was heading out over the Bristol Channel as expected.

Our concentration taken by the Pi launch we didn’t noticed AVA creeping upwards slowly, it spend alot of time out over the Bristol Channel as it slowly turned direction. Sucessfully recovering the Pi from a field we directed our attention to AVA which was now tracking east towards London and the quite incredible speed of 110mph.

In the hour it took us to get back from the Pi landing site to Daves AVA had left mainland UK and was heading out over the channel at an altitude of 38.5km (126,000 feet) maintaining its extremely fast pace at times touching 120mph. A small (but critical) error in the code exhibited itself at this point, it seems the Geofence for the UK isn’t correct and had the payload had an APRS transmitter it would have started to transmit just east of London which would have been bad!

Hoping the rest of the Geofences would work I headed back to Yorkshire, Dave soon called me to advise as AVA had skimmed French airspace the ITU prefix switch to F, followed shortly by ON as it entered Belgium. Success! As it crossed the Netherlands this switched to PA and then to DL as it entered German airspace at 17:13.

As the sun set AVA was careering across Germany maintaining a corking speed, sadly at 17:51 the radio transmitted its last full sentence, shortly afterwards the radio stopped. This is a known issue with the RFM22B’s and is usually related to them getting cold. I have modified the code to power cycle the radio every 50 lines.

From the point where it turned over Weston Super Mare to the last received packet it managed to go 753km at an average speed of 167kmph, just over 100mph :

Many thanks to Dave Akerman for helping with the launch and providing the gas and to all the receivers who take the time to listen out for these payloads.

Images from the day here : https://plus.google.com/photos/118244444241111963790/albums/5851143941313777297?banner=pwa

Daves blog about the day : www.daveakerman.com/?p=873

Having completed the code for the APRS/RTTY Payload I got round to making a container and an antenna for it. The 70cms antenna was just going to be the normal 1/4 wave however the 2 meter one had to be slightly different. I settled on a Slim Jim made from twin core feeder wire. Using M0UKD’s Slim Jim Calculator I quickly made one up and made a choke out of some 40mm PVC tubing.

The estimates for battery life were a whopping 70 hours for each pair of AA’s in the payload (The operating battery voltage had to be > 3V or the efficiencies on the 5V stepup started to cause issues). In the end I decided on 3 pairs of AA energizers. Enough for continuous TX for about a week.

Here you can see the battery pack on the left and the transmitters on the right.
All fastened up note slim jim hanging below. Full build pics >here<

It was all ready to go, just plug in and launch then a spanner in the works. Apparently the operation of amateur radio equipment in the air isn’t permitted in France. Maybe. With the situation becoming unclear and the launch window rapidly approaching I made the decision not to launch the payload. Instead we’d launch a PAVA board with the same code on to test the Geofencing.

A floater is a balloon that reaches a certain height then doesn’t burst and continues to float at a given altitude. In practice the only balloon that can do this reliably is the Hwoyee 1600g with a slow ascent rate and the smaller foil balloons.

This was demonstrated to great effect (and accidentally!) by Apex Alpha‘s epic flight to Europe. Unfortunately Apex Alpha left the range of the tracking network and was never seen again. Our tracking network has expanded dramatically since then and thanks to SP9UOB’s efforts the number of stations in Poland (where Apex Alpha was lost) means we should have great coverage out there.

However the tracking network is an active thing, people need to be actively listening out for the payload. There is an alternative which is used alot in the US but not in the UK due to our laws. The APRS (Automated Packet Reporting System) is “an amateur radio-based system for real time tactical digital communications of information of immediate value in the local area“.

Its a passive network meaning if you transmit on this and there is a station listening in the area your packet will be digirepeated and you will get your location on the map. Unlike the RTTY based system this is all automatic and needs no one to retune or run custom software. Unfortunately although a network of APRS stations exists in the UK the use of Amateur Radio in any “Aircraft or other Airborne Vehicle” is forbidden. However there is away round this…

Inspired by Apex Alphas successful ”failure” Project Swift planned to do this with a dual transmitter board, the Swift board although functional was heavy and needs alot of batteries. In conjunction with James Coxon (M6JCX) we planned a version of the PAVA board that allowed the connection of a separate daughter-board with a HX1 transmitter on it.

The board would have the usual stepup regulator on it, the HX1 requiring 5V would have a separate stepup on it which to save power could be powered on and off as needed. So the ATLAS/PAVA board was born :

On the left is the microcontroller board running at 1.8V (GPS and RFM22B transmitter on the bottom) and on the right is the HX1 Daughterboard for APRS. The HX1 TXD was fed by a NPN transistor to take the 0-1.8v up to 0-3v. Additionally the GPS was fitted with a PFET to allow the GPS to be turned off and on if needed.

The microcontroller board could be run on its own or with the HX1 transmitter clipped to the top. To ensure compliance with the law code would be written to “Geofence” europe into countries, this was done in Google Earth. Code was then borrowed from Project Swift to ascertain if the payload was currently within a certain area :

From here we can take action such as turning the APRS off entirely to comply with local legislation or amending the ITU prefix to comply with the CEPT regulations. I.e when in France the call sign of the balloon should be F/M0UPU.

This map above is availble as a KML here : https://github.com/Upuaut/APRS_Projects/blob/master/Data/Europe.kml

Once I worked out a number of faults with the original board one was shipped off to James for testing, I hope to launch one soon as well.

With K3NG’s Rotator Control Software and my interface I am now able to remotely control my rotator so when there are launches in the day I can track them from work….if I remember to turn the rotator control unit on.

This seems to be the sticking point and I was constantly forgetting to power the control unit on before leaving for work. Therefore I decided to make a remote control power switch and link it into the rotator control unit. I ordered a 5V Relay Module from everyones favourite tat store dx.com.

When it arrived a few weeks later I wired this into a butchered 4 way extension :

After replacing the 13A fuses in the plug and 4 way with 3A ones (the relay is a chinese rated 10A), PAT tested it (passed!) I wired it up to a test Arduino to confirm it worked. The lead to the rotator controller was made from a head phone extension cut in two so it can be removed as needed. After soldering it to the Arduino 5V, GND and A4 pins I amended K3NG’s code as follows, changing all the “button” references to 0 (I don’t use the buttons).

#define IDLE_TIMEOUT 1 // Idle time out in mins
#define POWERSWITCH A4

unsigned long idle_timer;

void loop() {

check_serial();
 if( (millis()-idle_timer) > (60000*IDLE_TIMEOUT) ) {
 digitalWrite(POWERSWITCH, LOW);
 }
 else
 {
 digitalWrite(POWERSWITCH, HIGH);
 }


 void check_serial(){

if (Serial.available() > 0) {
 if (serial_led) {
 digitalWrite(serial_led, HIGH); // blink the LED just to say we got something
 }
 incoming_serial_byte = Serial.read();
 idle_timer=millis();

Now if there is no serial activity for 1 min the Arduino powers the controller off. Issue any serial command (i.e open PSTRotate) and it powers the controller on, and to boot turns the LCD back light on.

So now even if I do forget to turn the controller on I don’t have to revert to remote control V0.1 (“The Lovely Wife”)