Windsond evaluated

A customer deeply involved in hot air ballooning and gas ballooning sent me the story of his first test of Windsond. Naturally the system has already undergone a lot of testing but it’s reassuring to read his account.

The launch

Today it looked like a nice evening for a test launch, so I readied my balloon chase car, got my old ACER Travelmate C312XMI charged – and brought my 220v inverter with me.

It is pretty dark out, but nonetheless, I would not need to rely on visual contact with the sond, since the system would tell me exactly where this device is floating.

Ok, system running in the car, antenna mounted, external power supply working, GPRS mobile data service, map loaded, all systems green, one more beep – and release (for flight)!

As I have hoped, very slow wind conditions.
As I have feared with saving precious helium, a very slow ascent (1 m/sec), but that doesn’t hurt.

Ok, we now see clearly in physical reality — and on the Google earth display — the typical fine weather easterly flow, my first Windsond drifting very gently to the west. Daddy was the driver, I was glued to the Laptop Display, my mother in the back row.

Now it was easy to predict where we should go (read the position off GEarth 😉
We went to a dark place, since then we would be able to observe (visually) better.
Then hopped to a next further stopping point, near the estimated landing positions, if we would release soon.

Bad thing was the local Fussball-game with its bright lights, and we were on the wrong side of the stadium, blinded by those beams. Anyway, I could not wait for the set altitude of 1500m AGL and sent the ‘cut’ command at about 1600 MSL or thereabouts. Too bad the wind shifted in those last 20 seconds of flight, although still slow, the track curved towards the stadium (we imagined trying to find the Windsond among 4.500 white foam cups of the visitors of the game …. 😉

So cutting was sensible, or wait for the Windsond to go over the city – I did not want to climb to any roofs in the night. We could even follow the fall quite nicely, the flashing makes sense (for night trainings, anyway)

The recovery

Then repositioned the car to a closer road, hopped out (with Laptop, GPRS Modem but no external power supply) and got our feet dirty… Estimated landing was about 90 meters short, but no problem, we could even hear the “find me” sound and see the flashing of the Windsond.

Retrieved in no time, it was like finding lost markers, better than Easter anyway.IT IS A GREAT SYSTEM, I TIP MY HAT FOR YOU, THE CREATOR, DESIGNER, THINKER.
It is well thought through, it is reliable, it is usable, it is a dream come true – for me.

The customer suggested some improvements for the software which I’ll incorporate. I’ll also look into improving the 90 m inaccuracy of the landing prediction.

Launching Windsond by a condom

The team and I launched a sond a few days ago by inflating a condom with helium. Condoms are much lighter than regular balloons but above all they improve the mood of the mission participants considerably. 😉 The helium volume was estimated to 17 liters. This gave an average ascent rate of 1.5 m/s. The ascent rate given the neck lift 15.1g should be 1.0 m/s so the shape of the condom probably helped to improve the speed.

The condom burst a bit prematurely at 2400m AGL, a few hundred meters before the sond was instructed to let go. The temperature had dropped from 0*C on the ground to -12.4*C at this altitude. Although it’s doubtful condoms are designed to “operate” at freezing temperatures, the gas expansion at the altitude is the probable cause of the burst. On the other hand, a ground test showed a condom of the same model could contain over 40 liters of air before popping but maybe there are variations between copies. I considered to write a complaint to the manufacturer over this blatant product failure. 😉

The launch was done in the evening, after dark. While the sond was still falling, I activated the beacon strobe by a button in the GUI. We were still 3-4 km away from the falling payload but the brief flashes of light were clearly visible, like a shooting star. The sond landed 65 m from the location originally predicted at 2400m altitude. In the dark, the blinking was very effective to locate the sond. The whole enclosure lit up and illuminated a meter of the ground around it. In the open landscape, the beeping was faintly audible at 200 m distance.

Another Windsond was also launched to compare the results of two data sources. They coincided very closely. The profiles of wind direction, wind speed, temperature and humidity were found to agree within tight margins. The second sond was launched while the first was still in the air which also demonstrated the ability of the system to handle multiple soundings at the same time. For the second sond I picked a landing location on a field close to a minor road. We picked up the sond 50 meters from the estimated position.

The double test was a big success and also very fun! A big thank you goes out to the team.

New Windsond screenshots

The graphics for Windsond is receiving a big overhaul. Here are two screenshots of the work in progress.

Windsond in Swedish winter

Gränna saw the gathering of hot air balloons during the weekend Andrée Memorial Meet in February 2013. The cloud base was on the low side but turned out quite ok for a nice competition.

A Windsond sounding right before the morning briefing gave us the details of the wind conditions. As an anonymous pilot said afterwards; “I should have trusted the wind sounding more”.

Windsond ascent and fall in Gränna

Weather sounding before the flight

The system has matured a lot lately. The Windsond description is updated with the latest specifications.

Developing the next Windsond

Since the last update, I’ve developed most of the next version of Windsond. It will incorporate a great number of improvements:

  • Higher altitude
  • Longer radio range
  • Many concurrent units, both receivers and weather balloons
  • Lower weight
  • Longer battery life
  • Proper encasing
  • Optionally measure air pressure and humidity
  • Better adapted to automated production
  • … and more

Pictures and test results to follow.

Windsond graphics

So far I haven’t talked much about the computer software that controls Windsond and visualizes the data. The software is still not completely polished but I can give an idea of what it looks like.

For each launched sond, there are three views.

The first view shows individual data updates from the sond and gives control over sond settings such as detach altitude, radio output power, LED blinking and loudspeaker output. This part is receiving an overhaul right now so I’ll show a screenshot later.

The second view visualizes the status of the sond. This includes the present whereabouts such as battery voltage, altitude and distance travelled, but also graphs that show how radio reception, altitude and ascending speed has developed over time. There’s also a draft graph showing the shape of the flight path as seen from above.

View of sond status. In this example, the altitude curve shows how the sond ascended, then descended more quickly (after detaching from the balloon). When close to the ground, the radio reception was cut off until the chase car came within a few hundred meters of the landing site.

The third view shows weather data as measured by the sond by plotting wind direction, wind speed and temperature against altitude. The Y axis maps altitude and X axis the data at that altitude. The two right-most graphs have additionally a black curve-fitting function overlayed to disregard the uncertainty in individual measurements. The experimental left-most plot visualizes the wind direction at different altitudes using a polar representation, with radius being the altitude.

View of weather data

The sond can be tracked in real-time in Google Earth, something that proved very useful during the World Championship. Google Maps caches recently viewed areas and ready-made maps can also be loaded to enable viewing without mobile internet access.

Path of one launch, showing launch site at top and detach position, predicted landing and actual landing at bottom. This particular launch revealed vertical turbulence over the hill that a regular pibal couldn’t show.

The Windsond software also supports reviewing old sessions, either immediately or replaying the progress at different speeds.

After the Balloon World Championship

Early morning with Windsond and traditional theodolite

The Balloon World Championship is over and I’ve made my way back to Sweden. It was a great experience and I’d like to thank all the people that made it possible. The locals were also very hospitable and cheerful.

Windsond worked out well. It provided detailed wind readings to an altitude of our choice and immediately provided the pilot with a clear visualization. When other teams gazed in concentration on a pibal forming a dot in the sky, we could release a prepared balloon through the car side window and drive on.

Path of one launch, showing launch site at top and detach position, predicted landing and actual landing at bottom. This particular launch revealed vertical turbulence over the hill that a regular pibal couldn’t show.

Out of 13 launches, only 3 sonds were lost. The first landed deep in a thick corn field and we got tired of searching when we couldn’t even see a meter in front of us. After this experience, I improved the software to make such unfortunate landings improbable in the future. I’ll talk more about this in a future blog post.

The second sond had a faulty release mechanism and just sailed away. Having the first two sonds disappear was discouraging but after that we got the hang of it and could recover all of them except one which we lost radio contact with, probably from radio interference. Had I remembered to check the frequency before launch, I could have switched frequency and avoided the whole problem.

If you’re a balloonist reading this, I’d love to get in contact with you. The system works well as it is but there are many ideas for extensions and improved software. Knowing the level of interest and your needs helps me to motivate further development and drive down costs.

Reach me at or telephone +46 707 312608.

Balloon Worlds 2012

I’m now at the World Ballon Championships in Michigan, USA, helping the two Swedish balloons with weather observations using the Windsond system. It’s a day and a half left of the week-long competition. It’s a wonderous sight to see a hundred giant balloons suspended in the sky while the rising sun slowly dissolves the morning mist.

So far we released eleven Windsonds. There’s a lot of practical issues to consider and we learned some good lessons along the way. The first lesson was that corn fields are hard work to wade through… After that, I added a feature to predict the landing location if detaching the sond at different altitudes. It proved quite accurate and very useful to make the sond recovery easier.

The competition runs at about six o’clock in the mornings and about six o’clock in the evenings. That leaves a few hours for extra development in the day but in the end there’s not much spare time. Features such as multiple sonds at the same time and lower energy consumption have to wait to another occation.

Field testing Windsond

Last Friday evening we made the sixth field test of Windsond. We obviously already have some experience with it but this was the first test of the final circuit board and there’s always usability kinks to work out.

We connected the battery to the sond and closed the plastic bubbles that protect the electronics. The computer with connected radio dongle picked up the signal directly. The computer program made a sound every time a data packet was received to let us know the connection was good. (This feature is turned off by default.) We inflated the balloon with helium and attached with a thread to the detachment mechanism on the sond. After a minute or so, the sond made a sound to let us know the GPS fix was good enough.

Windsond and me

We let the balloon go after this brief photo shoot. It sailed away to slowly become a dot in the sky, and then invisible to the naked eye. The computer told us it was still ascending. 10 minutes and 40 seconds after start, the balloon reached 1000 meters altitude, the default top altitude. This was at a ground distance of 2651 meters, giving an average ground speed of 4.1 m/s and ascent speed 1.6 m/s. Previously we’ve filled the balloon more for a higher ascent speed. The radio connection was still good but since we’ve previously used the radio link at 5 km distance with no problem, we expected the connection to be strong.

A few seconds later, the sond released from the balloon as planned. It fell at a leisurely 4.7 m/s and continued to drift 647 meters with the wind during the free fall. During the fall, we lost radio contact with the sond. I had tested the sond with lower output power earlier and it might still be configured for lower power. I sent a command to the sond to increase output power again. This did the trick and we had good contact with the sond for the rest of the fall, down to 65 meters above ground. Due to lag in GPS updates, the real altitude was probably lower and the cut-off was caused by tree tops getting in the way. Still, this gave us an accurate GPS position of the landing site.

The pin shows the landing site. The difference in wind direction is discernible from the shape of the flight path.

A mapping application on a smartphone translated the coordinates to a patch of forest about 3 km away. We drove there and I took a hike into the forest to recover the electronics. A trip of 350 meters each way, according to GPS. I brought my netbook along and about 230 meters from the site, I started receiving radio updates from the sond again. This confirmed the coordinates. A smartphone GPS application provided a compass needle that pointed towards the position which took out the guesswork of the hike even though the vegetation was thick. Some 10 or 15 meters from the coordinates, I heard beeping from the sond and could walk right up to it. It was laying next to a spruce. The envelope cushioned the impact well and nothing was damaged.

All that remained was to disconnect the battery, walk back to the car and study the generated graph of the winds up to 1000 meter altitude. In a real situation, we would use the graph as soon as the balloon reached the peak altitude but for this test retrieving the sond was more important. The plan was to make a hot air balloon flight later that evening but the wind picked up so we chose to postpone the flight.The computer software could use some polishing but all in all, Windsond is ready for live action now. After this confirmation, we’re starting to make more of the sonds.

Announcing Windsond

During the last six months, I have cooperated with the hot air balloon club Ballongveteranerna to create an electronic solution to measuring winds at low altitudes, called Windsond. The development is sponsored by the Swedish Airsport Federation.

Wind conditions are of paramount importance for hot air ballooning, where the balloon drifts with the wind. The only means of flight path control lies in choosing the altitude. Interestingly enough, the wind direction and speed can differ a great deal at different altitudes so precise control is possible for a skilled pilot.

Windsond consists of a GPS, battery, microcontroller and radio transmitter. By attaching a Windsond to the small pilot balloon, a computer with a radio receiver can track the ascent without human involvement and present graphs of wind speed, direction and temperature. Once the sond reaches an altitude of 1000 m, it detaches from the balloon and slowly falls to the ground where it beeps and blinks to assist recovery. So far, we have released three balloons with Windsond. The software can be further polished but system basicly works well. Good radio reception was demonstrated from 2 km altitude and 5 km ground distance.

In mid-August, we will travel to the World Hot Air Balloon Championship  to use Windsond for the two Swedish teams. The championship will transpire in Battle Creek, Michigan, USA.

I will continue to improve Windsond to offer it as a commercial product for reusable, cheap and easy low-altitude weather measurement.

Also see the dedicated Windsond page.

Page 9 of 10« First...678910
(c) 2015 Sparv Embedded AB
Sparv Embedded AB
Ã…gatan 39
58222 Linköping, Sweden
Phone: +46 (0)707 312608