If you have a Facebook account please vote for my entry in the competition link above!

Here’s the photo:

The aim was to send images from higher than my record of just under 40km, so the tracker was pretty much as light as I could make it. A launch was planned in April but but the wind predictions back then were poor for such a flight with a “landing” out to sea. That was the flight with Eben and Liz from the Raspberry Pi foundation, so rather than lose a precious slot in their over-full diaries I decided to launch a “floater” which was eventually lost over Switzerland and is probably still in hiding somewhere in France. For that I made a more suitable payload box, keeping the original for a later date.

That was a great flight with, as far as I know, the only live images sent from a flight of that type. It was also the first flight to use the new Raspberry Pi camera, and it got some good images though overall the success rate was much lower than I hoped for. Part of the reason was the solid white cloud cover which makes exposure very difficult, but also I didn’t realise that the camera was defaulting to spot metering which is the worst of all possible options for the challenging lighting conditions when the balloon is high. Spot metering basically makes the centre part of the image a certain luminance, which means that if it’s pointing at the blackness of space that turns grey whilst anything else in the image whites out. There are always going to be some useless images during a balloon flight, unless the camera is “aimed”, but on that flight almost all the downloaded images were poor (and that’s despite the on-board software selecting the better ones!).

So, I wanted to get that resolved plus do the flight that I’d intended to do before. Liz and Eben weren’t available but the predictions for wind and weather were both so good that I decided to launch anyway. I’d applied for permission from the CAA and that was granted a few days before launch, and the predictions continued to look good (though at one point I was going to need to use a different launch site to avoid a landing near Bristol). I ordered some more hydrogen as I wasn’t sure my cylinder had quite enough left, and with the weather forecast saying “sunny all day” everything looked good. Last but not least was to get the latest Pi camera software and configure it to use matrix metering mode. I also changed the code a little to take 3 types of image each about once per minute: small images for one radio channel, medium for the other radio channel, and large just for storage on the SD card. Oh, and also I replaced the main voltage regulator on the Pi for a better one so that the entire tracker could run off a lower voltage so it could run for longer on the batteries. I tested that down to a mere 3V from a power supply (that’s 3V going IN to the regulator) and the Pi booted and ran normally. Here’s the regulator soldered in place of the normal one:

I had some people coming along to see the launch, so they know a bit more about how things work before I help them on a commercial launch in a few weeks time. Originally I intended the tracker to run from 4 AAA lithium cells but the runtime on those was around 4.5 hours, which would mean connecting the batteries at the launch site shortly before launch. Instead I decided to use AAs so I could have the tracker running and glued up much earlier, keeping things simple for the launch itself and meaning that I wouldn’t have to worry about delays while I explained things. The extra weight does reduce the maximum altitude somewhat, but I thought on balance it was worth doing.

The day was sunny as expected, and the wind was mostly still but with some gusts. I showed my guests how the payload is attached to the parachute:

and then inflated the balloon, tied to the parachute, and re-checked that the tracker was still running. It was, and of course it had been transmitting images during our preparations, including this one overhead:

For launch the wind was very still so I just held the payload in front of the live streaming webcam and let go. Easiest launch ever!

The expected flight path showed a landing about 1 hour away by car:

And with an expected flight time of over 3 hours that gave us plenty of time to return to my house and watch the images come in. Thankfully the matrix metering on the Pi camera worked very well and the images were excellent thanks to the good weather and camera. Here are some of the low-resolution ones from the first radio channel:

I was pleasantly surprised by how well the other balloonists collected almost all of the image packets. Remember that my payload had 2 radio transmitters, meaning that people had to choose which one to listen to (unless they happened to have 2 receivers). Also I was using a fairly high baud rate which limits the range, but even so there were listeners in Northern Ireland, Holland and France contributing! The image quality was excellent and the matrix metering was certainly working well. Here are some of the larger images sent by the second channel, including this one of southern England and the Isle of Wight:

The flight path itself was interesting, with many changes of direction, including a few loops!

Whilst it’s great to sit and watch the images come in from your own flight, eventually you have to go and chase the thing down and recover it! Julie drove us in her car with me running the tracking:

Then I made a mistake. The original prediction (shown above) was for a flight earlier in the day, but the launch was delayed by more than 2 hours (partly to avoid a frequency clash with an earlier flight, and partly because we took our time). The later launch meant the balloon flew in more gentle winds so wouldn’t travel as far. Forgetting this, I asked Julie to drove close to the original landing spot. Also, it’s generally best in the first part of the chase to just get to a good position and wait for the burst, instead of assuming it will land where you thought. Once the burst has happened then you have a much better idea of the actual landing point. The end result of all this was that when the balloon did burst, and came down more quickly than expected (more on that later) we were in completely the wrong place.

By the time we got to the landing area, the payload had been on the ground for a few minutes. We had a rough idea of where it was and we went along both of the small roads nearby. Normally I’d get a radio signal within say 1 mile of the landing spot, but we got nothing at all. It seemed very unlikely to me that the radio transmitters had stopped working altogether – the batteries were soldered on so it would need a really hard impact to break the connection. The landing had been hard but not that hard. We tried a few highish points where I got the Yagi aerial out (much more likely to pick up a weak signal) but still got absolutely nothing. Very confusing.

I was about ready to declare the payload lost when my mobile phone rang. I thought it was going to be one of the other balloonists with advice, but instead it was someone who asked if I’d lost “a parachute with a raspberry and a big lump of latex”! My payload had been found! I asked for his address and said we’d be there right away. Typing the address into my phone it said we had a 14 minute drive – also strange! I reported to the other balloonists on IRC what had happened and they then saw my car travelling several miles into Swindon. They were probably as confused as I was as to how the balloon had managed to move that far from its last known position a few hundred metres up!

We got to the address and were happily reunited with the payload, parachute and almost all of the balloon! Then we found it what had happened – the lot had landed nearby (and right at the position given to me on IRC!) when he was out in the countryside at this place (exactly where I was told it would be on IRC – thanks Mike Willis!):

First he heard a loud thump as it landed, then he looked and saw a big lump of latex etc on the tarmac a few metres in front of him! He took it home and then, because of the smell of latex, put it in his garage. Of course I’d put my mobile phone number on the outside so once home he called me. We picked up the signal again when within a mile of his house, and were decoding when about half a mile away. The picture coming in was completely black, because the camera was face down! Now of course we understood why we couldn’t get a signal earlier – it had been driven several miles away from us! The reason for the rapid descent was also obvious – about 70% of the balloon (1.1kg) was still in one piece and had wrapped itself around the parachute. Here’s our happy rescuer with the remains of the flight:

The eagle-eyed may notice the small hole in the side of the raspberry, and that’s where the camera looked out, taking this photo of me as I photographed it:

Back home I copied all the images from the SD card, including of course all the full-resolution ones, including the ones below. The first and in my opinion the best is image file “snap314.jpg” – some coincidence! This is a view over Devon and Cornwall, taken at about 38.9km altitude and almost directly over the house of the guy who recovered it!

Finally, and thanks to Geoff Mather, here’s the flight path (click for an animated 3D model):

and the view from the payload, in Google Earth, with part of one of the actual images overlaid (again, click the image to play with the model):

Full set of Hi-Res images from the flight

Tom Gibson from Bloomberg came along to my last launch to video the proceedings and interview me for this nice little article that went live today. Click the above image to view on the Bloomberg site, or watch on YouTube below.

Probably my most unusual flight to date, this one began when I was contacted on Twitter by a media production company Sitcom Soldiers who asked for my help in launching a high altitude balloon for Wreck My Dress. They wanted to fly a wedding dress into the stratosphere, and I agreed to do it for them provided the dress was a small model, and not a full sized one (which would act as an unpredictable parachute – the payload could have landed anywhere!). After a few emails back and forth, they applied for launch permission from the Lake District and built a payload structure with 3 GoPro cameras on board. A date was set but the predictions didn’t allow for a safe flight, so we delayed till the following weekend. Even then, predictions weren’t ideal but an alternate launch site was found and the CAA kindly switched the permission to there for us. The end result is just great as you can see below:

Julie and I travelled up the day before, staying at a B&B in the village local to the launch site. After breakfast we set out to find … snow and sideways winds. Not ideal launch conditions! After meeting up with the film crew (a total of 18 people including cameramen, make-up etc.!), I was mostly heard to mutter “too cold”, “too windy”, and “can’t launch in this!”.

However, with all that lot having turned up, I didn’t want to disappoint, and a failed launch is only likely to waste a balloon and gas so why not just have a go? I then thought about how to give us the best chance of a successful launch, and decided to fly just the one payload (originally I was going to add a separate tracker and perhaps a separate camera/tracker payload). By putting 2 trackers into the camera rig I could have a backup to the main tracker without having a separate payload. Also, using shorter lines than normal would further reduce the chance of launching sideways into a tree. Finally, we found a partly sheltered launch spot where the wind would take the flight over a dip in the ground, giving it more clearance as the payload line swings down.

So, the aim of the day was to record enough footage for a promo video. The idea for the shoot was that a woman decides to “wreck her (wedding) dress” by sending it up into the stratosphere and then searching for it after it lands. My part of course was the launch and recovery, and as I said that was of a miniature dress specially made for the day. Here’s the make-up team working on the actress in a box van:

Here she’s being videoed preparing her dress (a real one) for flight:

While all this was going on I got the trackers and chase car set up and then payload line connected up, all in the relative comfort of the make-up van. It was then time to venture outside and fill the balloon. We’d positioned the filling area downwind of the van so it offered some protection, but even so I was nearly pulled over by the balloon a couple of times!

And here’s the very nicely made miniature dress, hanging from the camera rig:

For the launch itself, we walked the balloon down to a partly sheltered area where the balloon had plenty of space to rise up over any trees. It was so windy I had a “minder” walking right behind me, making sure the wind didn’t pull me over as I held the balloon!

As you can see, it was pretty windy but eventually the wind calmed enough for a safe launch.

So with that done, we packed up the cylinders and launch kit and headed off cross-country to where the prediction said the flight would land, in the Yorkshire Dales. It was an “interesting” drive and I was glad of the fairly knobbly tyres on the 4×4!

I’d deliberately over-filled the balloon to get it to rise quickly and to burst early, in order to avoid it landing near some MOD establishments to the south-east. As we chased the balloon burst even earlier than planned, so we had to change our route. We then got to a closed road with no diversion signs, and found that the tiny roads we needed were unknown to the car sat nav! 3G and indeed any phone contact came and went all the time, so at one point we resorted to a proper paper map to plot our route! This then took us between some large water reservoirs, and past a small forest, at which point my heart sank as I realised that we were unlikely to recover the flight if it landed in either of those! In the event it flew over Hury Reservoir, avoided the trees and landed on a hillside within 150 metres of a road. Result! We were quite close at the time and it pretty much flew over our heads (but invisibly due to the low cloud cover). We parked up as my direction-finding app said the payload was directly to our right. Some of the locals seemed quite interested in what was going on:

Wellies donned, we set off up the steep, muddy hill, me carrying a yagi and receiver for direction-finding, plus a GPS app on my phone loaded with the landing position. It was a hard slog, with not much visibility, but we found the payload safe on the snow:

We brought everything back down to the cars so everyone (including the sheep) could see the rescued dress:

Now, on a regular flight that would be it – pack up and go home to view the photos! This though, as you’ve probably already gathered, was no ordinary flight, and the film crew had more to do to finish off the storyline. On route to the landing spot they’d spied a nice location with a stream and waterfall, and decided to video the recovery scenes there. So we drove a few miles back to there, parked up and warmed ourselves with tea/coffee.

Meanwhile, once make-up had done their stuff, it was down to the actress to “find” the dress in the snow …

Then another delay for make-up (if you think HAB time is slow, you should try waiting for make-up….) before a final video shoot …

Links

Behind The Scenes

Dress Stratos

As far as I’m aware, no previous floaters have carried cameras, and my plan was to do just that with a Raspberry Pi camera using the live image transmission system (SSDV) that I’ve used before. I decided to take fairly large images and to send those at 300 baud which is known to work over large distances, but to use 2 radio transmitters to effectively double the bandwidth. Essentially the hardware was the same as in my Raspberry Pi-shaped tracker (which will fly very soon) but in a better insulated container with more batteries for the longer flight. I was expecting a flight beyond Poland and for the tracker to run for about 24 hours by which time it would be out of range of our receiver network anyway.

With a workable flight I contacted Liz and Eben Upton of the Raspberry Pi foundation to see when they were free to to launch with me – something all of us have wanted to do for some time. Saturday was that day, and with predictions from my home looking a bit risky (a fly-past of some minor provincial airfields called Luton and Stansted) I asked Steve Randall if it would be OK to use his site in Cambridgeshire (convenient for the RPi guys of course!). He said yes, and Anthony Stirk was available to come and launch his AVA payload, so the plan had come together. I love it when that happens …

The weekend approached quickly and I didn’t get time to make the tracker till the Thursday, and then built the payload container on the Friday evening. No rush then! I even managed to find time to program in a late Easter Egg (more on that later). Liz publicised the launch on the RPi site and Lester Haines did the same on The Register. I didn’t get much sleep on Friday night, having to get up in time to finish packing the 4×4 and then leave for Cambs at 7am. We needed to get there in time to launch in the morning as the winds were going to get up in the afternoon.

We arrived first, a few minutes early to find the wind rather stronger than expected. Anthony was next and we waited for Steve to arrive as he’s launched there many many times before and knows the best spots to launch according to the wind direction. Here he is using his patent-pending wind-direction-finding technique:

Eben and Liz were next, then Tom from Bloomberg who was recording the event for a news item, Graham (another balloonist) and James (Pi Camera Coder).

With the video streaming and tracking stations set up, we put our payloads together. Here Eben’s holding PIE5 for me whilst Anthony works on his AVA payload at the front:

With payloads working and tied to the parachutes, it was time to get inflating those balloons. You’d have thought that a hydrogen cylinder would be lightweight what with all that hydrogen in it, but apparently not …

Now, a normal flight (one intended to burst) has a “neck lift” in the range 1-5kg, typically about 2kg, and those are usually easy to measure by hanging a suitable weight from the balloon filler. here though we wanted a neck lift of around 300g, which becomes a problem when your filler weighs nearly twice that! So we had to use the scientific method known as “guessing”, followed up by removing the balloon and hanging a weight (2 reels of duct tape!) below. Here Anthony is helping me at a delicate part of the procedure:

We filled Anthony’s balloon first, and then asked Eben to hold it whilst we worked on my balloon:

Once my balloon was ready and both payloads tied to their respective balloons, I handed my balloon to Liz and we got the two of them to pose for us:

After checking that both payloads were still running fine, we were ready to launch. Steve called Air Traffic Control for clearance, and they said we had 5 minutes to launch in or wait a short while. We were ready, and after waiting a minute for the wind to drop we both launched:

(thanks to Alex Eames of raspit.tv for recording the video stream for us).

What you don’t see in the video is the resulting excitement and panic. Whilst my flight ascended at the expected rate, Anthony’s didn’t. See his exciting write-up for details, but suffice to say that it just cleared a tree 500 metres from the launch point!

With that over I did a piece to camera:

and Eben did a piece too. We all then packed up, and Anthony, Liz, Eben, Julie and myself made our way to Milton Keynes (why, you may ask?) to an excellent Dim Sum restaurant that Liz and Eben recommended (there’s your answer) and is probably the only reason to visit that place.

In between the many, many rounds of dishes we talked ballooning and Raspberrying, and kept tabs on progress of our flights. PIE4 was out over the North Sea transmitting some nice images as it went.

We then noticed that AVA had stopped transmitting and that uXABEN (Steve’s flight) had lost GPS. That was running a pre-preduction unit and a failure wasn’t a big surprise, but we were disappointed that AVA had hit trouble so early. After the meal we all set off to our respective homes, and during a short stop for fuel Julie and I noticed that AVA had updated on the map again and was near the Dutch coast. AVA carried 2 radio transmitters one of which is not allowed over the UK and was programmed to switch itself on only when legal to do so. We called Anthony with the good news, then went home.

At home I spent pretty much all the rest of the day watching the progress of both flights. PIE5 settled into a float at around 40km, gently going up and down with a peak altitude of 40,350 metres, comfortably above my previous “live images” record of just under 40km!

The images weren’t all as good as I’d hoped, principally because I’d used auto exposure and that faces an enormous challenge to find the correct settings with the vast difference in luminance between the clouds and the blackness of space. The flight was over a lot of cloud which makes that task almost almost impossible, and I’ve seen many poor images before with other cameras in this situation. It would have been better if I’d used manual settings, which I will do next time! Also, the flight seemed unusually stable (normally the payload swings around a lot) resulting in a lot of images with a lot of black and a little cloud. So in the even it would have been better to have the camera aimed downwards slightly. Nevertheless there were some interesting images including these apparently of a gigantic raspberry!

By now AVA had apparently died altogether, with neither RTTY or APRS transmitters running. After sunset my balloon descended as I’d hoped, achieving an excellent second float:

After that I waited eagerly for 9pm to arrive. Why? Because knowing that the sky would be completely black by then, I’d copied some specially chosen space-related images to the SD card the night before, and programmed the Pi to select and transmit images from the set. Needless to say this caused some confusion, consternation and amusement, as intended of course!

The missing segments are down to the fact that by now the flight was a long way from home and we only had a couple of listeners in range. I’d expected the flight to go towards Poland, but it had floated so high that the winds brought it south over Switzerland instead! When I awoke in the morning the last listener reported that PIE5 had finally stopped transmitting data, and that would have been because the batteries had given up. Here’s the last image it managed to transmit:

By now it was still over Switzerland and beginning to rise in the sun. Most likely it then burst, with a landing across the border in France. It does have contact details on it so it’s possible though unlikely it’ll find its way home!

Of course, if you do find the payload or know of its whereabouts, please contact me (details in the About page!

With PIE5 “lost in space”, attention turned to AVA which wasn’t as doomed as it had seemed. It suddenly appeared on the map over the Czech Republic, and was then tracked down into Austria where it burst and landed atop a 1500m high hill! Here’s the map showing the paths of both balloons – note how different they are despite both being launched from the same place at the same time!

We all assumed that was that, but we weren’t banking on International Rescue arriving in Austria in the form of a Slovak balloonist called Radim and his intrepid team. After driving from Bratislava they hiked up the snow-covered hill to rescue AVA! See Anthony’s account.

So, an awesome ballooning weekend, with records broken (highest live images, highest landing spot) and the most incredible recovery I’ve seen. Now we just need PIE5 to be found!

For more photos see my flickr set